Jeremy D. Finn, State University of New York/Buffalo, reviews the current status of class-size reduction programs in the United States. Finn summarizes the research base, calls attention to the misapplication of this research in some contexts and raises questions about small classes that need more research.
He also describes the long-term consequences of small classes and attempts to explain why they are effective. Studies carried out prior to the 1980s often suffered from methodological flaws and used small samples. However, these early studies indicated that
- Reduced class size (below 20 pupils) produces modest increases in academic achievement.
- Small classes are most beneficial in the early primary grades.
- Economically disadvantaged students are most likely to benefit from small classes.
Project STAR is one of the few large-scale, randomized within-school experiments in education. In 1985, Tennessee began to randomly assign kindergarten students to small classes (13-17 students), full-size classes (22-26 students), or full-size classes with a full-time teacher’s aide. Teachers also were assigned at random.
Pupils were kept in the same type of class for up to four years. In this very large scientifically controlled study (more than 6,000 students in 329 classrooms in 79 schools), students stayed in the same class all day, all year long with no other intervention. Teachers received no special training or curricula. Both norm-referenced and criterion- referenced achievement tests were administered in the spring of each year.
STAR pupils were followed after all of them returned to full-size classes in grade 4. Measures of academic performance were collected through high school, and information on post-secondary schooling and employment is still being gathered. STAR data has been reanalyzed by researchers not associated with the project, using a variety of statistical procedures. These analyses agree with the original findings with very few exceptions. (The most extensive summary of the short- and long-term academic outcomes of STAR appeared in Teachers College Record in 2001.)
Magnitude of effect in early grades
Results reveal that small classes have statistically significant academic benefits in every grade, in all academic subjects. The differences between small classes and full-size classes ranged from .2 to .3 standard deviations in each subject. By the spring of third grade, the advantage in reading was about half a school year.
The magnitude of the effects was greater for students who started early and spent more years in small classes. In every grade, the benefits were greater for minority students or students attending inner-city schools than for white students in non-urban schools. Statistically, the advantages were often as much as two or three times as great, thus reducing the white/minority achievement gap. Both minority and white students gained significantly by being in small classes, but minority students gained more.
After students returned to full-size classes in the fourth grade, the benefits of small classes continued to be statistically significant through all subsequent grades in all subject areas. Starting early and continuing in small classes for multiple years was related to the strength and duration of these carry- over effects.
Other class-size reduction programs
A large number of non-experimental class reduction initiatives have been undertaken since Project STAR. Those that have been evaluated include Wisconsin’s Project SAGE: the well-researched Burke County, North Carolina program; and California’s statewide initiative.
The Wisconsin and North Carolina projects targeted only low-income students. Findings from these programs are consistent with STAR. The North Carolina program differed in that it included intensive staff development. By reallocating existing staff members, the North Carolina schools reduced class sizes with no increase in per-pupil costs.
California’s statewide program reduced classes to 20 pupils. In two years almost all K-3 classrooms had been reduced. Evaluation of academic achievement was limited largely to third grade, and no comparison groups were studied. This program cost over one billion dollars per year. Achievement in third-grade classrooms showed a small but significant increase for students in reduced-size classrooms.
Teacher’s aides ineffective
In STAR and subsequent programs, there was no significant difference between full-size classes with teacher’s aides and those without aides. Yet more than 600,000 teacher’s aides were employed in American classrooms in 1998 at an annual cost of about 9 billion dollars.
All the research on this topic concludes that teacher’s aides benefit neither students nor teachers. When teachers were asked for their preference of teaching a regular-size class with a full-time aide or teaching a small class by themselves, 71 percent of teachers said they preferred teaching a small class.
No academic or behavioral advantages were reported for classrooms with teacher’s aides. Research on teacher’s aides points to deficient preparation, lack of clearly defined roles and the absence of training for teachers in utilizing their assistants.
Overall, this small-class research also reveals other benefits:
- Teacher morale is improved in small classes.
- There are fewer disruptions and fewer discipline problems in small classes and teachers spend more time on direct instruction.
- Students’ engagement in learning is increased.
- Retentions are reduced.
- More small-class students take SAT or ACT tests.
- Drop-out rates may be reduced.
Cost of class-size reductions
Policy makers want to know if these benefits of small classes are worth the costs. The most complete analysis of costs was performed by Dominic Brewer and colleagues, who estimated that reducing class size can be very expensive.
Finn reports, however, that none of the analyses looked at possible resource tradeoffs. Several school districts have used funds allocated for teacher’s aides to hire additional teachers. In financially poor Burke County, North Carolina, small classes were implemented in grades one through three by reassigning existing staff members with no increase in per-pupil cost.
The economist Alan Krueger analyzed other economic benefits. Data from STAR and other studies of school performance reveals that the cost of reducing class size is recovered in terms of students’ future earnings.
The wrong way to create small classes
Finn states that the California Initiative did things too quickly. It its haste to reduce class size, the state placed many individuals in classrooms without full teaching credentials or adequate experience managing students.
When inexperienced teachers are placed in classrooms without adequate preparation, the result is great disorganization that can take months to remedy. Such problems can offset the academic benefits of small classes.
Finn recommends that class reductions be undertaken slowly and with careful planning. Although Project STAR demonstrated the benefits of small classes without any special teacher training, the state had an abundant supply of credentialed teachers. Finn believes that professional development could help teachers in all class-size reduction initiatives take advantage of having fewer students by showing them how to cover content in greater depth and how best to take advantage of the increased sense of community that is typical of small classes.
Finn cautions against confusing pupil/teacher ratios with class size. Class size refers to the number of pupils regularly in a teacher’s classroom for whom that teacher is responsible. Larger classes that are team-taught by two teachers do not show the same academic advantages for students.
Despite this, the current U.S. Department of Education’s Class Size Reduction Initiative permits team teaching of large classes. The number of staff per school or district in relation to the number of students has only a weak relationship with academic achievement. It is the number of students per classroom that matters for academic achievement.
The best class size
Finn concludes that more research is needed in order to find out how small classes must be to produce positive benefits — is a class of 20 as effective as a class of 17? How effective are small classes in the middle grades? Can the effects of small classes be enhanced through particular instructional strategies?
Research is currently being conducted to determine the long-term impacts of small classes in the early grades. Information about college admissions, childbearing, delinquent or criminal behavior, and unemployment rates are being collected.
Some researchers speculate that small classes work well because teachers can change instructional styles, providing more one-on-one teaching and higher-quality instruction. Research to date does not support this hypothesis.
It is fairly clear that teachers of small classes spend more time on direct instruction and less on management or discipline. Not much else is known. Changes in instructional technique appear to be small and can not explain the consistent academic benefits of small classes.
A second hypothesis suggests that students become more engaged in small classes; they pay more attention to learning and display more prosocial behavior. In small classes, each student experiences continuing pressure to participate and can’t avoid the teacher’s attention as is possible in large classes.
Small classes also tend to encourage closer personal relationships between teachers and students and among students themselves. Teachers know each individual student better than they would in a large class and students tend to be more supportive of one other.
While this second hypothesis seems plausible, it has not been proved. Finn cautions schools and districts against undertaking class-size reductions without evaluation. He encourages teachers working in settings where class sizes are reduced to encourage the administrators to engage in evaluation and research for their own benefit as well as for the benefit of the broader education community.
“Small Classes in American Schools: Research, Practice, and Politics” Phi Delta Kappan Volume 83, Number 7, March 2002 Pp. 551-560.
Published in ERN April 2002 Volume 15 Number 4