Sunday, June 12, 2005

Public and Private Schools’ Performance: Does Governance Matter?

June 7, 2005 | Volume 3 | Number 12
Public and Private Schools’ Performance: Does Governance Matter?

The Question
Does a difference in school governance (public versus private) result in a difference in students’ achievement?

The Context
Researchers and activists have long debated the different effects of school governance on student achievement. Some studies have purportedly found that students in private schools significantly outperform their public school counterparts. In fact, much of the argument for market reforms in education—including vouchers and charter schools—revolves around the assumption that private governance results in higher student achievement at similar (or lower) cost when compared to public governance. As highlighted by the Heritage Foundation, a conservative think tank that advocates for market-based education reform, as far back as 1981 research by James Coleman suggested that private school students performed much better than their public school counterparts.

Research on voucher programs—which are reliant on such governance assumptions—remains mixed, however (see ResearchBrief, Volume 1, Number 6), and some researchers have argued that once student background traits are taken into account, the difference in achievement between public and private school students virtually vanishes. In effect, these researchers say, the difference in achievement may have little to do with how a school is funded and governed; rather, it may be an artifact of other variables, including student socioeconomic status (SES), parental education, community support, or peer group characteristics.1  The study highlighted in this ResearchBrief adds to the research base focusing on public versus private school governance, using a particularly powerful measure for determining the effects of SES variables on student achievement.

The Details
Sarah and Christopher Lubienski conducted the study highlighted in this issue of ResearchBrief (see below for full citation). The study was actually the by-product of a larger research project the authors were conducting using National Assessment of Educational Progress (NAEP) data from the year 2000 to study issues related to mathematics instruction and equity. As they were analyzing the data for the larger study, they unexpectedly found that public school students seemed to outperform their private school counterparts in mathematics, once SES variables were held constant. As a result, the authors embarked on a more detailed analysis of the data to explore the strength of their initial findings.

The National Assessment of Educational Progress is a nationally representative assessment of student academic performance. In this case, the authors used data from the NAEP assessment for mathematics in grades 4 and 8. The 4th grade sample comprised more than 13,000 students in 385 public schools and 222 private schools; more than 15,000 students in 383 public and 357 private schools were included in the 8th grade sample.

The authors created a measure of SES by combining student responses to a number of questions on a survey administered as part of NAEP. Previous research on SES has tended to focus on the number of students in free and reduced-price lunch programs, but the authors felt there was an opportunity to create a richer measure of SES by combining multiple measures that prior research has associated with poverty. At the 8th grade level, the authors used student responses to eight items to create their measure:

* Types of reading material in students’ homes (newspapers, magazines, books, and encyclopedias)

* Computer access at home
* Internet access at home
* Extent to which studies are discussed at home
* Free and reduced-price lunch eligibility
* Title I eligibility
* Mother’s education level
* Father’s education level

The composite measure for 4th grade students did not include the parent education item, because that information was not collected from 4th grade students in 2000. Most of the measures were based on student self-reports, but the authors obtained school lunch and Title I eligibility information from school records. After creating this measure, the authors split the population into four quartiles representing low to high SES. As has been the case with previous SES measures, there was a higher proportion of white students in the high SES groupings and a higher proportion of black students in the low SES groupings. Finally, to create a measure that emphasized the schoolwide and individual effects of SES, the authors combined the student-level SES data with two school-level items: the percentage of students eligible for school lunch programs and the percentage of students qualifying for Title I services (for this school-level variable, the authors emphasized school lunch participation over Title I participation because of variation in defining eligibility for Title I services).
Once the SES variable was created, the authors compared the performance of students in public and private schools on the mathematics assessment in NAEP. As a whole, students in private schools outperformed students in public schools; however, private schools also tended to have many more high SES students than did the public schools. To find out whether the high SES population of the private schools was skewing the average, the authors analyzed performance by quartile (low SES, low–mid SES, mid–high SES, and high SES). When broken into SES subgroups, the researchers found Simpson’s Paradox at work. At each level—low SES through high SES—the researchers found that students in public schools slightly outscored their private school peers.

The Bottom Line
When the mathematics performance of students in public and private schools was compared while controlling for individual and school-level socioeconomic status, students in public schools were found to outperform their private school peers at each SES level: low SES, low–mid SES, mid–high SES, and high SES.

Who’s Affected?
Students in public and private schools at the 4th and 8th grades were the focus of this study.

The researchers noted that because the SES data are based largely on self-reports and represent only the 2000 administration of the NAEP, the correlation between SES and achievement may not be causal. In addition, because the researchers focused only on mathematics achievement, this study indicates little about achievement in other subjects. Although the SES measure the authors created is based on a set of characteristics associated with a low SES, each component is not necessarily associated with only SES (for example, although research suggests that low SES children are less likely to have computer access at home, not all children without a computer at home are low SES).

The Study
Lubienski, S., & Lubienski, C. (2005). A new look at public and private schools: Student background and mathematics achievement. Phi Delta Kappan, 86(9), 696. Retrieved May 15, 2005, from (Note: Some of the details in this ResearchBrief were found in the larger study of NAEP conducted by the authors, Reform-oriented mathematics instruction, achievement, and equity: Examinations of race and SES in 2000 main NAEP data.)

Other Resources

Voucher Effects Revisited. ResearchBrief 1(6)

Simpson’s Paradox and Other Statistical Mysteries. American School Board Journal


1  This is not to say that schools have no impact on student achievement. If school-level practices were similar across all schools (e.g., teachers use similar pedagogy, teach similar content, and work through similar structures—such as individual grade levels, subject-specific classes, and similar contact time), then differences in achievement based on academic variables would be hard to track. Instead, measures where there is variation (such as SES) would become more apparent.

No comments:

Post a Comment