This column originally appeared in November 2006. Unfortunately, critics of parental choice and competition in education continue to cite the flawed study I discuss below.
RALEIGH – Parents in North Carolina and elsewhere who collectively spend many millions of dollars sending their children to private school are either dupes or bigots. This must be the assumption of those who eagerly embraced the findings of a recent federal study that purported to show public schools outperforming private ones. A little skepticism, informed by common sense about parental choices, would have come in handy here.
The report, issued by the National Center for Education Statistics (NCES), looked at 2003 performance on the rigorous National Assessment of Education Progress (NAEP) by several different groups of 4th and 8th graders: public schools, Catholic schools, Lutheran schools, and “Conservative Christian” schools. The latter three groups of private schools were also combined for analytical purposes.
No one doubts that in the aggregate, private schools have higher average test scores than public ones. The question is whether the private-school advantage is merely a reflection of selection bias – kids already destined to do well being disproportionately represented in private-school enrollment – or is instead an indication that there is something intrinsic to private settings that boosts student achievement. By the way, rejecting the selection-bias explanation doesn’t necessarily make you a private-school champion or voucher advocate. You could believe that any achievement-enhancing policies private schools use could and should be adopted in the public schools, or even that the private-school effect proves that more-generous funding allows schools to purchase better teachers and other inputs (although, as discussed below, that argument is based on a faulty premise).
The NCES study appeared to provide evidence for the selection-bias explanation. Researchers adjusted for student background by using sex, race, and school participation in federal programs such as Title I for disadvantaged students and free or reduced-priced lunch. Doing so essentially eliminated the private-school advantage for 4th-grade reading and 8th-grade math and revealed a public-school advantage of 4.5 points for 4th-grade math. Only in 8th-grade reading did a private-school edge remain, of 7.3 points.
The education establishment leapt at the findings. They should have looked first, at three important warning signs. First, the NCES researchers themselves cautioned readers not to jump to conclusions. They noted that snapshot studies do not prove causal relationships, that studies tracking the performance of students over time are necessary to see how the effects of particular schools or educational techniques affected their learning. Second, it should have been a red flag to those who follow education policy that the researchers used variables such as school participation in Title 1 and free/reduced lunch to adjust for student background. These are blunt instruments indeed, exaggerating the poverty and disability status of students in many public schools while failing to capture the poverty and disability status of students in many private schools.
Third, it should have occurred to commentators and analysts that a single study, operating at an analytical level of 50,000 feet, was not likely to prove what they wanted it to prove. Millions of parents spend millions of dollars to send their children to private schools. Unless you believe they are mindlessly throwing away thousands of dollars a year, or simply willing to spend that money to indulge cultural or religious prejudices, it is reasonable to assume that these parents, having direct experience with the institutions in question, know something that you don’t. This is not the kind of argument that leads to a logical or scientific proof, mind you. This is only about probabilities. If you amass sufficient evidence, you can certainly refute such arguments. But it is exceedingly dangerous to do so based on a single study. It is more likely that the flaw is in the study than in the judgment of millions of people, spending their own money repeatedly to improve the lives of their own children.
It turns out the flaws in the NCES study weren’t hard to spot. Harvard researchers Paul Peterson and Elena Llaudet conducted their own study of the same data, using better adjustments for student background that avoid the classification bias of the first study, and confirmed the existence of the private-school edge. Using three different models based on different adjustment criteria, they found higher student performance among the private schools in 11 of the 12 measurements (three models x four tests). The larger gaps were found for 8th grade, which makes sense, and for reading rather than math, which is interesting.
Even Peterson and Llaudet caution, properly, that these findings don’t prove that private schools are better educational environments than public ones, since they don’t randomly assign students from the same pool to either public or private schools, then track students over time. But studies that do meet these specifications – including one conducted in Charlotte a few years ago – provide compelling evidence for the proposition that private education is, indeed, more effective with a given pool of (in this case low-income) students.
Why? It’s not because private schools have more money. They don’t. Commentators often mistake prep schools, such as Raleigh’s Ravenscroft or Charlotte Country Day, as exemplifying private education when in fact preps are rare outliers – 80 percent of private schools, enrolling 75 percent of private-school students, charge less than $5,000 in tuition, many of them far less. Fundraising and church subsidies supply additional revenue, of course, but on the whole private schools cost significantly less per pupil than public ones. Their teachers usually make less money, but flock to private schools for reasons such as shared values, discipline, merit pay, and academic rigor.
There is a lot more to learn about the differences between public and private education. Champions of the latter get carried away, too, when they claim huge differences in outcomes even though the available evidence usually counts the private-school advantage in fewer than a dozen points. There are common problems inhibiting our children’s academic success: too many distractions, not enough reading in the home, socioeconomic factors, etc. But it helps students, not hinders them, when parents choose private schools. More ought to have that opportunity.
Hood is president of the John Locke Foundation.