In lieu of an abstract, here is a brief excerpt of the content:

Brookings Papers on Education Policy (2006/2007) 15-47

Is Small Really Better?
Testing Some Assumptions about High School Size
Barbara Schneider
Adam E. Wyse
Venessa Keesler

Several years ago, I was in a meeting with a group of Chicago public school coaches and physical education teachers who were discussing the negative implications of one of Chicago's recent reform initiatives, the construction of smaller high schools. Much like other urban areas, Chicago had begun dismantling some of its large high schools to form smaller entities, with an "optimal" enrollment of 600 students. The coaches were deeply concerned that the small school movement was fostering the elimination of school-sponsored athletic teams, which sometimes acted as a magnet for marginal students, encouraging them to complete high school and in some instances enroll in college. From their perspective, intramural teams were unable to fill the void left by school-sponsored teams, which had helped some students obtain postsecondary scholarships and promoted a high school identity that instilled pride in the student body.

Reflecting on their comments, I was struck by how my work and that of others had championed small schools. Could we have been wrong? Small schools were generally viewed as places that fostered a strong sense of community and encouraged academic achievement and attainment. But many of us had not explored whether small schools were better for all types of students. More specifically, would the consequences of creating small-school environments prove to be detrimental, especially for low-income minority students enrolled in urban high schools? [End Page 15]

The case for small schools has been made in educational research since the 1960s, when scholars such as Barker and Gump argued that smaller schools provided students with greater opportunities for participation in various extracurricular activities.1 Within a smaller student body, the average adolescent would have a better chance of being on a team, taking a leadership position in the school, and developing stronger relationships with teachers and other adults in the school. The value of small schools was further supported in the 1980s by research on public and private schools that showed that smaller religious schools produced higher graduation rates and lower dropout rates than public schools.2 Analyses of the National Education Longitudinal Study of 1988 in the 1990s also showed that smaller public schools produced substantial gains in mathematics achievement for high school students.3 By 2000, the results of those studies were often used as evidence by policymakers and school administrators to support proposals to decrease school size as a strategy for increasing student achievement.

However, initial results from small-school reforms have been inconsistent.4 In light of those results and reviews of earlier work, serious questions are being raised regarding the methodological techniques used to study the effects of school size.5 Several concerns center on the use of inappropriate research designs for assessing causal effects, such as correlational analyses rather than random clinical trials. These concerns have led several educational researchers to revisit propensity score methods for using observational data to approximate experimental designs, methods formalized by Donald Rubin more than thirty years ago.6 As I reviewed my own work and that of my colleagues, it became increasingly clear to me that many of the reforms being advocated, particularly in today's high schools, had rarely been studied using Rubin's methods. Many of the cornerstones of high school reform, including better academic preparation programs, peer tutoring, and use of mentors, often lacked a rigorous evaluation component.

Uncertain of what we might find, my two colleagues, Adam Wyse and Venessa Keesler, and I decided to estimate the effects of high school size using Rubin's methods and conventional hierarchical models. This paper describes our efforts, using observational data from the National Education Longitudinal Study, to approximate an experiment on the effects of school size for several student outcomes: mathematics achievement, postsecondary expectations, college attendance plans after high school graduation, and number and type of...