School Class Size and Pupil Achievement

1990

School Class Size and Pupil Achievement

Gene V Glass

The relationship between school class size and pupil achievement has been one of the most researched topics in education. Studies appeared in the U.S. in the first decade of the 20th century; and new articles still appear. Class size is an issue of great fiscal importance. It is a key teacher workload concern.

School class sizes differ substantially around the world and within nations. The average elementary school student in Norway may attend a class with 10 or 12 age-mates. In Bangladesh, a class may have 40 students. Classes in North Dakota are much smaller than classes in Los Angeles. Class sizes reflect both wealth and population density. They have declined steadily around the world. The pupil-to-teacher ratio in U.S. public schools dropped from about 40 in 1900 to about 23 in 2000.

The research results on whether children learn more in smaller classes are quite clear. Student achievement increases as class size drops. Dozens of studies support this conclusion. My colleagues and I performed a meta-analysis of some 75 studies of class size reduction and published the results in 1982 in School Class-size: Research and Policy. This synthesizing of data on thousands of students of all ages in many different subjects confirmed what educators long suspected: children learn more in smaller classes, and decidedly more in very small classes. One of the most ambitious and best experiments in the history of educational research—the Tennessee Student Teacher Achievement Ratio (STAR) study reported by Frederick Mosteller in 1995 in The Future of Children: Critical Issues for Children and Youths—has confirmed it.

Even though the research on class and achievement is clear, the leap to policy recommendations to reduce class sizes is not easily made because achievement gains from class reductions may not be as cost effective as other methods. The relationship between class size and achievement is what mathematicians call “geometric.” Each unit decrease in class size predicts an increase in achievement, and each subsequent unit decrease predicts a larger increase until size “1” is reached. Decreasing class size from 30 to 29 may produce a very small, indiscernible increase in achievement, whereas decreasing from 10 to 9 is likely to to increase achievement more markedly.. The question is often asked, “What is the optimal class size?” The person asking the question wants an answer like “19” or “14.” In fact, the only answer that research gives is “1.” Most elementary school classes in the U.S. have about 25 students. The gains in achievement from even very expensive size reductions, say from 25 to 20 students, would not be impressive. Compared to other approaches, such as cross-age tutoring, class size reduction is not particularly cost-effective.

It is often argued that even though achievement gains from class size reduction are not large in general, impressive gains could be obtained by reducing classes at certain grades or subjects. By and large, such assertions are mere conjecture with little empirical research to back them up. The Tennessee STAR experiment seemed to suggest that smaller classes were more beneficial at the early elementary grades, e.g., Kindergarten and first grade. The researchers also found that splitting a class in half was more beneficial than merely introducing a teacher’s aide to reach the same pupil-to-adult ratio. However, these qualifications were not as well established as was the overall finding that smaller classes learn more. Similarly, it has never been convincingly shown that gains in achievement from class size reduction occur for one subject (e.g., math, reading, science, social studies) and not another, or for one type of student and not others. The STAR study suggests that class size reduction benefits the achievement of very young children (K-3) more than older students (4-6).

Teachers want smaller classes because they are less work. Students want smaller classes because they feel better served. School boards and taxpayers often want larger classes because they reduce the cost of schooling. The two sides frequently argue about class size when they meet at the bargaining table.

Further Reading

Henry M. Levin, Gene V Glass, and Gail R. Meister, “Different Approaches to Improving Performance at School,” Zeitschrift für Internationale Erziehungs und Sozial Wissenschaftliche Forschung, 3, No. 1 (January 1987), pp. 156-176.