My beloved Freddie, who unfortunately unlisted me from his RSS reader the other day and will probably never see this, once argued that all of the benefit of "good" schools actually comes from the fact that the kids are high-IQ/SES, none of it from the school itself.  Somewhat less extremes of this argument frequently pop up in arguments against vouchers, arguing that private schools won't really do any better because private schools just skim the cream.

I think that this is rather extreme, but there's an element of truth in it.  Kids from high income homes are easier to teach.  They are less likely to come from broken or truly dysfunctional homes, given that at least one parent is together enough to keep a job.  Their parents are more likely to be highly verbal with them, and to read to them.  They aren't hungry or cold, haven't been up all night babysitting while Mom works, and a doctor regularly checks on their health.  They have higher IQs, some of which is probably genetic--as I recently heard one scientist say, it's hard to see how intelligence could have evolved if it weren't heritable--some of which is environment, and some of which is the excellent care their mother took of herself when she was pregnant.  Whatever you saw on that ABC After School Special, they are much less likely to be abused.  And their parents, who themselves are almost certainly college educated, place a very high value on things like homework and test scores.

But just as it's unlikely that IQ is either all heritibility or all environment, so with education.  Much of the educational benefit comes from the fact that the kids are easy to educate.  But to say that the school itself doesn't matter at all is to posit that one could get exactly the same results by parking the kids in a school library for twelve years.  There may be diminishing returns, but clearly there is some area within which improving educational techniques improves outcomes.  Are US schools really all above the red line?

Doubtful.  Anecdotally, I was better educated--at least in English literature--than college peers from very competitive suburban school districts, and not because I was noticeably better raw material or they were distracted by the gang wars in AP English. Nor were my classmates--lawyers kids are lawyers' kids whether they're in Bronxville High or Horace Mann.  My school had the resources to higher better teachers, have smaller class sizes, and expect more from its students.  Most of the people I know who went to private school say the same.

Now to the data.  What about those hard to educate kids?  We know that there are methods that work better than others, because in the 1970s the government commissioned a gigantic study of educational methods called Project Follow Through.   One method, Direct Instruction, consistently produced better methods, a result that has since been repeatedly replicated.  Educational approach does make a difference.  Unfortunately, this educational approach is hated by teachers, so even though we know it works better than almost anything else, adoption has been slow.

Of course, kids in the worst off schools often have a train wreck of problems--familial abuse, drug use, language barriers, poor diet--that schools also need to address.  Just jamming in direct instruction isn't going to solve those, and the social service infrastructure schools now provide won't and shouldn't go away, though perhaps it should be outsourced or transferred to family services.  Some of the problems will not be solved, at least not in the perfect way we fantasize about where no kid has to deal with problems that kids really shouldn't have to deal with.  I have no good heuristic, for example, that will ensure that no kid gets abused without removing some kids from their families unnecessarily.

More broadly, this kind of despair over composition effects makes me more likely to support vouchers, not less.  If the school really can't make any difference, then why not pay less and get higher levels of parental satisfaction, with at least no worse education for the kids?

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.