Could we improve student achievement by lowering class size? Parents and teachers have always thought this idea made a lot of sense, but most researchers have found little evidence to support it. And because reducing class size is expensive, policymakers have cheerfully embraced what researchers said. Now, it looks as though the parents and teachers were right all along.
In an upcoming article, Frederick Mosteller, a professor of statistics at Harvard and one of the country's most distinguished statisticians, reports on a little-known study that proves the value of smaller classes. Mosteller's article, "The Tennessee Study of Class Size," which will appear in the journal The Future of Children, says the study leaves no doubt that children in early elementary school achieve at a significantly higher level in classes of 15 than they do in classes of 25. For poor, minority children, the gains are even greater.
Student Teacher Achievement Ratio (STAR), as the experiment was called, was commissioned by the state of Tennessee in 1985. It included some 7,000 children from 79 elementary schools, representing a geographical cross-section of the state. Children were randomly selected from those entering kindergarten in September 1985. Then they were randomly placed in one of three types of class: a small class (13 to 17 students with an average of 15 students); a standard-size class (22 to 25 students with an average of 25), in which there was an aide to assist the teacher; and a standard-size class with a teacher only. The third type served as a control group. The experiment continued for four years, as the children moved from kindergarten through grade three.
Did smaller class size make a real difference in student achievement? To find out, STAR researchers compared scores of children in the two experimental classrooms and the control group on both standardized exams and curriculum-based tests. The results were striking. Students in classes of 25 with aides did slightly better than students in the classes with teachers only. But the real differences showed up in the smaller classes. On standardized tests of reading and math, these children gained .25 of a standard deviation over children in classes of 25 with a teacher only. This means, as Mosteller explains, that a pupil achieving at the 50th percentile--better than half the kids at his grade level--would now be achieving at the 60th percentile--better than 60 percent at his grade level. During the first two years of the experiment, the results were even more striking for poor, minority students: Their gains in the smaller classes were twice that of white children. In grades two and three, they were about the same.
STAR researchers continue to monitor the achievement of the students who participated in the experiment. So far, they have found that the gap narrows, but children who spent the K-3 years in the smaller classes enjoy a significant advantage over children from the larger classes. For example, in grade five, children from the smaller classes scored at the 51st percentile in reading compared with the 41st percentile for children who had been in the regular-size classes. In grade six, children from the smaller classes achieved in the 54th percentile in reading, and children from regular-size classes at the 45th percentile.
The state of Tennessee took these findings seriously. In 1989, it directed that class size be reduced to 15 for K-3 students in 17 mostly minority districts, where poverty was high and student achievement low. The results look very promising. As Mosteller says, "Before the small classes were introduced, these districts had been performing well below the average in the state for mathematics, and after the intervention, they moved above the average."
This is an extraordinary story in a number of respects. STAR is a rarity in education research--a real experiment, well-designed and carefully enough carried out that we can have confidence in its results: Mosteller calls it "one of the great experiments in education in United States history." And it offers solid findings about our biggest problem in education--how to raise student achievement and particularly that of poor, minority students.
It is also extraordinary that the media, which have given currency to studies that discount the importance of class size, have consistently ignored the Tennessee experiment. Now that Mosteller has flagged its importance, will we see it widely cited as a breakthrough in our knowledge? And will we see policymakers and educators acting on its findings?
School boards and cities and states say they are desperate to find ways of improving student achievement. The ideas they grab for--charter schools or for-profit management schemes or voucher proposals--are all untried and unproven. But they don't have to guess any longer. The Tennessee experiment offers proof that smaller class size works.