My years of attending different schools went like this: public, private, public, public, private. As the sequence suggests, I had opportunities to experience schools that were public and schools that were private. At least from my perspective as a student, a school’s classification as public or private did not make much difference. They were all schools to me.
I was reminded of the public versus private school debate, and my experience with the insignificance of this categorization, by a recent front-page article in Education Week with the headline, ‘Public Schools Outperform Private Schools, Book Says.’ The authors of the book (The Public School Advantage: Why Public Schools Outperform Private Schools), Christopher and Sarah Lubienski, argue that earlier research showing private schools—mostly Catholic schools—outperformed public schools was hampered by data limitations. When they analyze data from the National Assessment of Educational Progress (NAEP) and the Early Childhood Longitudinal Study’s (ECLS) kindergarten cohort, they find that after adjusting for student characteristics, the average student in an elementary public school had a higher math score than the average student in an elementary private school.
At least three issues come to mind about these findings. First is that costs are not mentioned, despite how much they matter. Second is how disconnected the findings are from the choices parents actually have to make. And, thirdly, the question of ‘public versus private’ is a sideshow diverting attention from a far more substantive inquiry: what can we do to make all schools become more productive?
First, costs. Recent data on Catholic schools reports per-student costs well below public schools. It’s about $6,000 for elementary schools and $12,000 for secondary schools. NCES reports average public school spending combined for both levels of $12,000. Over the K-12 span, then, and with most Catholic elementary schools spanning K-8, the total spent per student is about $102,000 in Catholic schools versus 50 percent more, $156,000, in public schools. By this metric, even if Catholic schools had the same test scores, their productivity (output per unit of input) is substantially higher than public schools.
Second, parents. Tables in the Lubienskis’ book show huge differences in public and private school test scores. Catholic and Lutheran schools have fourth-grade NAEP scores that are 10 points higher than public schools. Because of how the NAEP is scored, this difference is approximately an entire grade level. Fourth graders in public schools are scoring about what third graders in private schools score. Differences between public and private school students are even larger in eighth grade, ranging from 14 to 20 points.
These differences disappear when the authors do their analysis. But how? Scores are ‘adjusted for’ student differences using statistical models. I am not criticizing the use of statistical models, having spent decades using them. But there are limitations that need to be kept in mind. In the case of private schools, ‘adjusting’ for characteristics creates a hypothetical situation in which public and private schools are being analyzed ‘as if’ they had the same characteristics. For example, a local public school that has a large enrollment and many students on free lunch and a local private school with a small enrollment and few students on free lunch will be compared ‘as if’ they have the same enrollment and the same proportion of students on free lunch.
The fact that this situation doesn’t actually exist is the point. Parents see real schools, not hypothetical ones. Suppose a parent is considering whether to send their son or daughter to a private school or to a public school. For sake of argument, let’s assume transporting their child to either school takes the same time and energy. There is a huge difference in student test scores between the schools, which the parent recognizes might be partly because high-achieving students already attend the private school. The parent also learns that studies suggest students who attend private schools are more likely than similar students attending public schools to graduate from high school and to enroll in college. Suppose the private school is achieving these score differences and graduation outcomes while also spending less than public schools.
This scenario might lead many parents to choose private schools. Having high-achieving students in the private school as peers for one’s child is an attractive feature. Increasing the odds that one’s child will graduate from high school and attend college is appealing as well. The lower cost closes the deal.
Except in reality, parents don’t pay a lower cost for private schools. In fact, the cost of a private school is added onto the cost of a public school. Parents pay property and state income taxes that fund public schools, and then have to decide whether they can afford private school on top of that. So, private school becomes an expensive proposition. Some parents will nonetheless decide to pay for it, but it’s unsurprising that 90 percent of America’s K-12 students are in public schools.
The Lubienskis say their findings should undermine arguments in support of voucher programs and other market-based programs because these programs are based on the idea that students attending private schools will do better than if they attended public schools. Let’s expand ‘doing better’ beyond test scores and focus on the education attainment of older students.
The DC voucher study cited above found that using vouchers increased high school graduation by 21 percentage points. Applicants for those vouchers were low-income (families below 185 percent of the poverty level were eligible), and nearly all were African-American. That study did not follow students long enough to know whether students went on to college. However, the study of the New York voucher program cited above found that using vouchers increased college enrollment by 9 percentage points for African-Americans.
These seem like small numbers, but economists have estimated that compared to dropping out, lifetime earnings of high school graduates are $300,000 higher for African-Americans, and lifetime earnings of graduates who attend at least some college are $800,000 higher. Thus, the increased likelihood of graduating high school and attending college associated with the use of a voucher can add tens of thousands of dollars to lifetime earnings. This is likely an underestimate given that completing college is not accounted for (neither study explored college completion), but is associated with even greater earnings. And most voucher users do not use the voucher for long, usually only two to three years. The current DC voucher program provides $8,000 for elementary schools and $12,000 for high schools, which means public spending of somewhere around $20,000 to $30,000 could achieve an earnings effect three times larger or more.
Of course these are rough numbers that are influenced by data limitations: estimates of lifetime earnings necessarily involve many assumptions; the New York City study found beneficial effects of vouchers only for African-American students and not for Hispanic students; and only a couple studies have been done (though both studies referenced here use strong experimental designs). And if a much larger voucher program were created, it might serve other kinds of families and possibly have smaller effects.
The point is that comparing test scores of public and private schools can be a distraction from the bigger picture. In fact, the entire public-private debate is nothing more than a sideshow. How to improve schools generally is more fruitful, though less dramatic, than framing the debate as ‘public is bad and private is good,’ or the reverse. Certainly though, analyzing what private schools are doing to yield higher graduation and college-going rates while spending less is one way to begin answering that question. And if what’s discovered to be working in private schools can possibly be replicated by the public schools that nearly all students attend, there will be no need for future debate.