Class Size: Does it Matter?

  • Research into class size has been mixed partly because few studies have directly manipulated class size and successfully removed any other factors that could influence learning, and partly because of there has been no consistency in what constitutes a "small" or "reduced" class size.
  • Evidence points to a class size of 15 students or less being necessary to show clear benefits.
  • Small class size is more important in the early years.
  • Small class size may have greater benefit for disadvantaged students.

While parents and teachers have always strongly supported small class sizes, their belief has not always been supported by evidence. Part of the problem lies in that word “small” — what constitutes a small class? Different interventions have looked at reducing class sizes from 40 to 30, or 30 to 25. It may well be that such reductions are not sufficient to show clear benefits.

The STAR Project

The project everyone talks about, the STAR project (Student Teacher Achievement Ratio), looked at class sizes well below these. The longitudinal study was undertaken in the American state of Tennessee and involved over 7000 students from 79 schools. For three years, from kindergarten through grade 3, students were placed either in small classes of 13-17 students; regular classes of 22-25 students; or regular classes with a teacher aide. Those in smaller classes performed significantly better on tests than those placed in regular classes. The largest gains occurred in inner-city schools.

Excitingly, the advantage was not only maintained in subsequent years but actually increased: in grade 4, students who had been in smaller classes were 6-9 months ahead of regular class students in reading, math, and science; by grade 8, they were a year ahead. Later, almost 44% of small class size students took college entrance exams, compared to 40% of regular class size students — the difference was greatest for African-Americans; 40.2% compared to 31.7%. 72% of small class students graduated from high school on schedule, compared to 65-6% of regular class students. They were also more likely to complete high school, to graduate with honors, to complete advanced math and English classes.

The SAGE Project

More recently, in Wisconsin, the Student Achievement Guarantee in Education (SAGE) program has reduced the student-teacher ratioto 15:1 in K-3 classrooms in 30 schools, comparing their performance to 14-17 matched schools. The benefits seen were again particularly great for African-American students, who reduced the achievement gap with white students by 19% — in comparison schools, the achievement gap widened by 58%. Interestingly, the results of having 2 teachers in a class of 30 were the same as having 2 classes of 15.

Why have different studies found different results?

Let’s look a little further at why there has been confusion about what educational research has told us about class size, given that this is one of the most studied issues in education. Howard Blake in 1954 reviewed pre-1950 studies. He found 85 that were based on original research, and of these 35 found benefits of small classes, 18 found benefits to large classes, and 32 found no difference. But Blake analyzed the studies further, looking for scientific acceptability. He found only 22 studies that reached this standard (a not surprising result for educational research in this time period). Of the 22, 16 favored small classes, 3 favored large classes, and 3 were inconclusive.

A meta-analysis of 77 studies in 1978 (Glass, Cohen & Smith) concluded that the greatest benefits occurred when class sizes were reduced to 15 students or less. A follow-up study suggested that the benefits were greatest for those below the age of 12 (Smith et al, 1979).

Unfortunately, educational experiments such as STAR — where students are randomly assigned to different treatments — are rare. More usual are attempts at indirectly investigating class size by comparing different situations. This, obviously, has many problems. You can get a feeling for these by reading a British analysis at: Apart from anything else, it shows you how one type of statistical analysis in studies of this nature can come up with no clear benefits of class size, while another type shows a very clear benefit.

It also seems that the principal benefit of reduced class size lies in its effect on the teacher; clearly some teachers will be more affected by this than others.

It is also worth noting the considerable international variation in class size -- a variation showing no correlation with performance -- indicating that class size cannot be considered out of the context of teaching method. The TIMSS international study, for example, found that although the average eighth-grade mathematics class was 31 students, there was considerable variation even among the higher-performing countries –- from 42 students in Korea to 19 in Belgium.


Many policy-makers argue that, while class size may be of value, the benefit doesn’t warrant the huge amount it would cost, given that there are other ways to spend the money — more and better trained teachers, for example. And, certainly, there would seem little benefit to reducing class size if you can’t put qualified teachers in the classes. Class size isn’t a factor that can be considered in a vacuum. But it does seem clear that:

  • class size can be an important factor in learning outcomes,
  • it is more important in the early years,
  • it is more important for disadvantaged students,
  • benefits may not be seen unless the class size is reduced to around 15.


For more about the results from Project STAR, see

The National Education Association (U.S.) has information about class size at:

TIMSS has an interesting international comparison of class size and math achievement at: