Brookings Papers on Education Policy

In recent years, the large comprehensive high school has been a subject of growing critique by researchers and reformers. "Factory model" schools have been criticized for their impersonal structures, fragmented curricula, segregated and unequal program options, and inability to respond effectively to student needs.¹ Some studies have found that, other things equal, smaller schools appear to produce higher achievement, lower dropout rates, lower rates of violence and vandalism, more positive feelings about self and school, and more participation in school activities. These outcomes appear more pronounced for students who are traditionally lower achieving.² In addition, the belief that large schools are necessarily more cost-effective has been challenged by studies finding equivalent operating costs and lower costs per graduate in smaller schools.³

However, there are competing findings about the effects of smaller schools for different groups of students and about the effects of school size and organizational features in diverse contexts. This review examines these findings across a wide range of studies over the last thirty years. We conclude that the influences of size appear to be mediated by other features of school organizations that are sometimes, but not always, associated with size, making the relationship between school size and many desired outcomes an indirect one. These other features are associated with aspects of school design, including how adults and students are organized to work together, the nature of the curriculum, [End Page 163] and how access to knowledge is organized. Smaller schools may provide the opportunity for important educational conditions such as stronger relationships, greater student involvement, and greater academic press, but they do not, by themselves, guarantee that those conditions will exist.

Furthermore, the processes used to create smaller schools or units within large schools have been highly varied, including many different models (for example, schools within a school with varying degrees of autonomy, house plans, mini-schools, learning communities, clusters, magnets, and charters within larger schools or as stand-alone schools) that have been launched under widely differing conditions. We would expect—and research has confirmed—variable levels of success among school restructuring efforts launched with different designs, varying amounts of planning time and resources, and varying levels of staff experience and quality, as well as under different political conditions, collective bargaining agreements, and other factors.

Finally, there are questions about what "smaller" means and about what school size may be optimum, since comparative studies have examined different size ranges. For example, research looking at data from different systems in different eras has resulted in recommended maximum high school enrollments ranging from 500 to 1,000.⁴ Some analysts suggest that optimum size varies with the socioeconomic status of the community, with 1,000 probably the upper limit for schools serving affluent students and substantially smaller sizes for schools serving low-income students.⁵ Our review of the available evidence suggests that, in addition to student backgrounds, important variables influencing high school outcomes for different groups of students include organizational structures that create more coherence and "communal" orientation, reduce curriculum differentiation, increase instructional authenticity and rigor, and enhance personalization (that is, the extent to which students are well known by adults).

While we have found no research support for the large urban comprehensive schools serving several thousand students that still exist in many cities, many questions remain about what kinds of organizations should replace these gargantuan institutions. We believe that interpreting competing findings in ways that can guide productive decisionmaking requires a much more nuanced analysis of the interacting elements of relative school size, student body composition, organizational design, curriculum content, and instructional features than is typically undertaken. As Bickel and Howley note, "Seldom have policymakers or researchers asked 'Better for whom?' or 'Better for what?' or 'Better under what conditions?'"⁶ In what follows, we seek to address these questions. [End Page 164]