The book's title is bold and its cover even bolder—all caps, black and red, occupying the entire cover. A month before its recent publication, The Atlantic interviewed the authors, Christopher and Sarah Lubienski, for a story headlined to shock: "Are Private Schools Worth It?" Other major media, including The Washington Post and Washington Monthly, provided prominent coverage.
The book takes aim at what the authors see as the dominant education reform strategy of the last 20 years, reliance on the market. Voucher programs provide opportunities for low income students in a number of cities to enroll in private schools. Charter schools, now numbering 6,000 nationwide and serving over 2 million students, offer choice to parents while increasing competitive pressures on all schools to improve. Both reforms enable more students to attend more schools that operate with autonomy from the politics and bureaucracy that allegedly cripple traditional public schools.
Why do policymakers believe autonomy is a good thing? Because that is one of the distinguishing attributes of private schools. And it has become conventional wisdom among researchers and policymakers that private schools outperform public schools. But what if the conventional wisdom is wrong? Then all of the market-oriented reforms of the last two decades can be called into question. The Lubienskis' book title all but screams: the conventional wisdom is wrong. Public schools are the superior achievers and it is their practices that should be followed.
Does the Lubienskis' research have any merit? The core of the book is an analysis of 2003 NAEP data first reported in 2006. Private schools have higher test scores than public schools on average, but they also serve students from more affluent and better educated families. Because family attributes affect student achievement, those effects should not be credited to schools. Once family attributes are taken into account, public schools, the book concludes, have higher test scores than private schools for students from similar backgrounds.
How family background should be taken into account is the big question, and the Lubienski book just gets it wrong. Researchers have worked for decades on methods for teasing out the effects of schools from other effects on student achievement, such as families, peers, and student ability. The preferred methodology by far is randomized student assignment, much like randomized medical trials. This technique has been used in charter and voucher research, and has identified pretty consistent advantages for schools of choice. Where random assignment is not possible—and it most often is not—researchers have used "propensity scores," "interrupted time series," and other methods more sophisticated than the linear regressions used in this book to obtain estimates considered state-of-the-art among researchers.
The issues here, though, are even more basic. The authors control for family and student characteristics in multiple ways that over-represent student disadvantage in public schools and underrepresent it in private schools. The authors include student participation in the federal free-and reduced-price lunch program, students on federally mandated IEPs, and students classified as Limited English Proficient by federal standards as controls for student background. Yet private schools are not required to participate in these programs and do so at rates about one-fourth of what public schools do. The rates differ, however, not because the needs are not substantially present in private schools; they are just often met outside of federal programs. Other measures of student need in the two sectors show student differences that are much smaller. The book also controls for peer effects, as measured by the composition of the student body. Leaving aside the question of whether the school should or should not be credited with creating positive peer effects, the book exaggerates negative peer effects in public schools by controlling for IEP and LEP percentages once again—meaning once at the student level and once at the school level.
If this seems all too technical, two Harvard researchers simplified matters by replicating the book's 2006 analysis, but replacing its federal program participation measures with alternate measures comparable across sectors. They found that once controls equally appropriate to both sectors were applied, the public school advantage disappeared and a private school advantage returned, for both grade 4 and grade 8 mathematics. The Harvard team also point out an embarrassing lacuna in the Lubienskis' NAEP analysis. The book and the 2006 study look only at mathematics, because that is the subject that schools influence most. The Harvard team used the book’s methodology to estimate sector-effects on reading scores. There was a clear private school advantage in reading, even with erroneous controls. The Lubienskis owe it to readers to acknowledge this, not hide it.
The "dog bites man" story here is not, as the authors would have it, from looking at more recent data than extant research. The first public-private research (of which I, in full disclosure, was part) did begin with data from the 1980s. But steadily more sophisticated research has examined data from private schools and charter schools right up to the current time. The Lubienski book offers outlying findings not because it is more current, but because it makes assumptions and uses data that most researchers reject.
The book concludes with a look at the classroom practices and cultural influences that are associated with higher math gains in elementary schools. This is territory already well explored. The Gates Foundation’s MET Project has validated highly effective instructional models in reading and math and science. The importance of culture and social capital has been underscored in work on schools of all types. The book corroborates much of what is already accepted. The Public School Advantage suggests that public school teachers benefit by comparison to Catholic school teachers (*note about the study sample below) from certification and modern professional development. These topics have been studied far more rigorously elsewhere, and to far different conclusions.
The book is needlessly polarizing. The nation's education future does not lie in a choice between government- or market-dominated schools. Parents should and do have choices among public and private schools. Local control of public schools has always provided choice based on residence, and charter schools have added to the options, especially for families with limited economic means and residential options. Private schools have long served about a 10th of the population, but not without some measure of licensing or government oversight. Innovative and successful practices have been incubated in schools of all kinds. Smart reform asks how the country can take best advantage of a pluralistic system.
This book ultimately fails because it wants to pick a fight, not solve a problem. Our nation needs to understand better what works for students, and especially what works for students from diverse backgrounds in an age where the knowledge and skills that schools should teach is not clear. Schools of every type have things to learn, and things to teach. After years of high quality research—and much higher quality than found in this book—we know that the important questions are not "which is better, public or charter, or public or private?" The important questions are when we find substantial variations in school quality, "why?" Why do some private schools outperform public schools? Why do public schools sometimes come out on top? If educators and policymakers better understand what great schools do, they can more successfully drive improvement in schools of all kinds. That may not grab headlines, but it is the kind of research that deserves them.
*Note: because participation in NAEP and other federal studies is voluntary -- for schools of all types -- it is not known how many NAIS schools may have participated in the studies used in the Lubiesnki book or other analyses using these federal data. The Lubienski NAEP analysis examines four types of private schools—Catholic, Lutheran, Christian, and Other Private. NAIS schools are most likely to be represented in the Other Private category if represented at all. The participation rates for schools in the Other Private category of NAEP do not satisfy the federal government’s requirement for representativeness and should be viewed with great caution, though the book reports them anyway. The Early Childhood study groups private schools as Catholic and Other Private. NAIS schools may be among the latter—a group that public schools do not outperform.