Research on grouping and class size: The need for sustained study

Schoolchildren and teacher in science classQuestions about the impact of grouping and class size on student achievement have been raised for decades. Yet a study of the educational literature points up striking contrasts in the research on these two important topics. A well-controlled, large-scale, long-term study concerning class size was carried out in Tennessee — Project STAR. Continuing results of this study are definitive and provide educators with valuable information. However, the quality, size and longevity of this study are exceeding rare in the field of education.

The available studies on grouping, in fact, could serve as examples of the deficiencies in educational research efforts. Frederick Mosteller, Harvard University and the American Academy of Arts and Sciences, and Richard J. Light and Jason A. Sachs, Harvard University, studied all available research on grouping and class size and they concluded that educators have to make decisions every day for which they do not have the information they need.

Skill Grouping

According to a 1988 study by the National Educational Longitudinal Survey, 86 percent of public school students in middle schools and high schools in the United States are placed in skill-grouped classes for mathematics. Yet these researchers found only a handful of well-designed studies exploring the academic benefits of this practice, and the results of the available studies were equivocal.

The practice of separating students into low, medium and high classes is referred to as XYZ grouping. Mosteller et al. found 10 reasonably well-designed studies of XYZ grouping carried out between 1960 and 1975. Each took place in a single school and only one lasted more than one year. Results indicated a modest benefit for the highest-skilled students, while low- and middle-skilled students appear to do a little better in mixed-ability classes. However, the findings were variable and all the studies were small and short-term. Overall, remarkably little is known about the impact of XYZ grouping.

Cross-grade grouping

Another form of grouping, cross-grade grouping, is often referred to as the Joplin Plan. This is a way of grouping student across grades so that they work in small groups with other students who have similar skill levels. The goal is to enable teachers to advance students quickly as their skills improve. Although only two high-quality experiments were found, these suggest that the Joplin Plan gives students significant learning benefits. A three-year study of reading development also favored cross-grade grouping. These findings are very encouraging, but the evidence is limited and larger-scale studies are needed.

Only three well-controlled experiments of within-class grouping were found. The high-, medium- and low-skill groups each gained about half a grade more than students taught in whole-class instruction. However, in one study in which the same teachers each taught one semester of whole-class instruction and one semester of skill groups, the results showed almost no difference. More extensive studies of within-class grouping are needed.

Overall, the 15 studies of various forms of grouping offer little evidence that it has a major impact, either positive or negative, on students’ learning. The positive effects of cross-grade grouping need further study.

These researchers could not find a single large-scale, well-designed experiment following students over several years to evaluate the effect of any form of grouping on their learning.

In these short-term studies the initial effects tend to be larger in the first period of treatment than later. This may indicate that once the novelty of the experiment wears off, grouping has little effect on learning.

Class Size

In contrast, there is definitive data about the effects of class size on student achievement.

Project STAR (Student/Teacher Achievement Ratio) is an example of the kind of experiment needed in appraising school programs, and it is proof that large, carefully designed studies in education can be carried out in the United States. The Tennessee state legislature authorized and funded a four-year study of the effects of class size and teacher aides on children’s learning in the early grades. In the experimental classes, Tennessee reduced class size to about 15 students in kindergarten, first, second and third grades.

In fourth grade, all students moved into regular-size classes. The learning of students in small classes was compared with students in regular-size classes with a teacher’s aide and students in regular-size classes without a teacher’s aide. The experiment was carried out in 79 schools. A second phase of the study followed participating students into later grades to measure their academic progress.

Results showed substantial improvement in early learning in such subjects as reading and math when students were placed in small classes. And the positive effects persisted into grades four, five, six and seven, after pupils moved to regular-size classes.

Gains by poor and minority students

In particular, poor and minority students benefited from small classes. In the first two years, they gained twice as much as other students and then matched others’ gains in subsequent years. As a consequence of this study, the Tennessee legislature decided to implement the small-class program in seventeen school districts with the lowest per-capita incomes.

The results of this third phase of the program are encouraging. In smaller classes, children in these poor districts are performing better on both standardized and curriculum-based tests than students in the same districts in earlier years.

End-of-the-year performance has raised their district ranking in arithmetic and reading from far below the average for all districts to above average. The presence of teacher aides, though beneficial, did not produce improvements comparable to small class size and did not have any lasting effects once students moved into regular classes without aides.


Comparing the research on these two educational practices reveals the need for more large, well-designed educational studies. Currently, the United States has no systematic national program for evaluating the impact of different educational policies or teaching practices.

Research and development in the areas of health, energy and defense are funded at much higher percentages of their total budget. Twelve percent or more of their total budgets is spent on research. Less than one percent of money spent on education is for research.

Mosteller et al. make a convincing case for the need for large-scale research in education. An industry that spends over $300 billion each year, employs 2.5 million teachers and serves 44 million students needs programs for research on the effectiveness of its methods.

Successful innovations are important, but they often rely on charismatic leaders and exceptionally skilled and motivated teachers. These innovations must prove effective with different populations of students and teachers, and only large-scale studies can determine if such innovations will be of general benefit to our students.

“Sustained Inquiry in Education: Lessons from Skill Grouping and Class Size”, Harvard Educational Review, Volume 66, Number 4, Winter 1996, pp. 797-849.

Published in ERN January/February 1997 Volume 10 Number 1

Leave a Reply

  • (will not be published)