Smaller Is Better
Ask a parent or teacher to name the one thing that could improve student learning, and you're likely to hear about small classes. Ask a principal or superintendent the same question, and you'll probably get a different response. That's because many administrators play down the benefits of reducing the number of students in classrooms. Research, they say, has not conclusively shown that small classes are linked to improved student learning. Others point out that learning may improve but only when classes get down to 15 kids.
Well, not quite, say researchers from the Student/Teacher Achievement Ratio Project--Project STAR for short. As part of a two-phase study, they have been tracking classes of 15 to 17 students in 79 schools scattered across Tennessee since 1985. Their efforts make up the largest and longest-lasting experiment ever conducted on the effects of small classes on student achievement.
Not only do students in the early grades learn more in smaller classes, the project's investigators say, but they also continue to have an edge over the rest of their peers years after they return to normal-sized classrooms. What's more, the data they have collected are beginning to show that every time a student is added to a classroom, learning is diminished for the rest of the class.
"This is one of the great experiments in education in United States history,'' says Frederick Mosteller, a Harvard University statistics professor. "It definitively answers the question of whether reduction from this size to that size does make a difference, and it clearly does.''
But some education researchers are skeptical that Project STAR's findings make the definitive case for smaller classes. They point out that more than 1,100 studies on class size have been conducted over the years and that the results have been mixed. "I don't think a single study proves all that much,'' says Herbert Walberg, a professor at the University of Illinois at Chicago. "You can find some studies that indicate that bigger classes have better effects on learning.'' Walberg suggests that other kinds of interventions, such as cooperative learning, may be more effective and less expensive.
Before Project STAR, the landmark study used to argue for small classes was an analysis by Gene Glass and Mary Lee Smith in the 1970s. They reviewed dozens of studies on the subject and concluded that reducing the number of students in a class does have a modest effect on learning. But there was a catch: Their findings indicated that the benefits would not show up until classes had been reduced to 15 or fewer students. The study was controversial, partly because it mixed in many different grade levels and kinds of classes, including graduate seminars and one-on-one tutorials.
When the class-size debate erupted several years later in the Tennessee legislature, lawmakers decided to test out the hypothesis themselves. Spurred by then Gov. Lamar Alexander's pro-education agenda, the legislature allocated $3-million to four of the state's top universities for Project STAR.
"All educators and parents know that, with fewer children, you can do a better job,'' says Helen Pate-Bain, who lobbied for the project when she was an associate professor at Tennessee State University and later became one of the study's four principal investigators. "But when you talk to the people who hold the purse strings, they all said to us that research says small classes don't make a difference. The reason for doing the study was to once and for all show that class size does make a difference.''
Unlike the studies that came before it, Project STAR was to be a true experiment in the most scientific sense of the word. Pate-Bain and her partners decided to focus their efforts on students in kindergarten through 3rd grade, reasoning that small classrooms could have the biggest effect on young learners. "If you can give a child a good beginning, if they learn to read, nobody can take that away from them,'' Pate-Bain says.
Of the 79 elementary schools that took part in the study, 25 were located in urban areas, 16 in suburbs, and 39 in rural areas. To participate, schools had to have at least 57 students in a grade--enough for one small experimental class of 13 to 17 students and two normal-sized classes of 22 to 25 students, one with an aide and one without. "That way, whatever else might be happening at the school would happen under all three conditions,'' says Barbara Nye, director of a follow-up study that became the successor to Project STAR.
Through all four years of the study, students in the smaller classes at each grade level outscored their peers in both types of larger-class settings on a battery of standardized tests, including the Stanford Achievement Test, Tennessee's Basic Skills Criterion Tests, and another test developed especially for the project. Students in the inner-city schools made the greatest leaps. But their counterparts in suburban and rural schools made gains, as well.
In reading and mathematics, the size of the gain was about a quarter of a standard deviation. To understand what that means, Mosteller says, think of a child who, without any special treatment, might score at about the 50th percentile on a test. A gain in score of a quarter of a standard deviation would raise that child from the 50th percentile to the 60th percentile. In other words, now 60 percent of the testing population scored lower than that child.
"And we didn't do anything to schools other than reduce class sizes,'' adds Jayne Boyd-Zaharias, who took part in Project STAR and is also part of the follow-up study. Having a teacher's aide in the classroom produced only slight improvement in student achievement over the large classes that had no aide.
By the end of Project STAR, researchers had collected data on 7,000 students and spent $12 million. But questions arose over whether the documented gains would last. So, with continuing support from the state, the researchers launched the follow-up project, now called the Lasting Benefits Study. Through it, researchers were able to track student progress after they returned to normal-sized classes in the 4th grade and for years afterward.
In grades 4, 5, 6, and 7, students who had been in smaller classes in grades K-3 continued to outscore their peers who had been in larger classes. But the advantage, the researchers discovered, diminished as the years went on.
The researchers found that the benefits were not limited to reading and math. Students from smaller classes outscored their peers in science, social studies, and other subjects, as well. Other studies suggest that those students also participated more in class and took part in more extracurricular activities than their peers from larger classes.
Researchers are still tracking more than 4,000 of the students and hope to continue doing so after they leave high school and go to college or start a career.
Charles Achilles, professor of educational administration at Eastern Michigan University and a principal investigator on both the initial and follow-up projects, is analyzing the data to see whether 15 or 17 students is the "magic number''--the threshold at which real learning gains start to occur--or if any reduction in class size helps. Would adding or taking away a child in a class of 24 or 25 make a difference? "It looks as though the addition of a child to a class decreases the class's average scores,'' he says. But he warns, "We're still tinkering with this.''
The results from Project STAR and the initial findings of the Lasting Benefits Study have already translated into policy changes in Tennessee. In 1989, state lawmakers set aside funds to reduce the size of K-3 classes in 17 of the state's poorest school districts. More recently, the American Academy of Arts and Sciences, spurred by a favorable review of the project by Harvard's Mosteller, has begun to tout the study as proof that small classes matter.
"It's time that we quit asking the question, 'Does it make a difference?' and begin to ask why it makes a difference and how we can begin to use this information,'' Achilles asserts.
But the question keeps being raised, most recently by Eric Hanushek, a University of Rochester economist and
public-policy professor. Hanushek, who analyzed 300 studies, argues that across-the-board reductions in class size are not worth the expense. Although Project STAR wasn't included in his analysis, Hanushek does have an opinion about the study. "Those people are zealots,'' Hanushek says of the STAR researchers.
The problem with their findings, Hanushek explains, is that there is not much "value added'' beyond the achievement gains that come about in kindergarten. The size of the effect in grades 1 through 3 is about the same. The upshot, he points out, is that the gain comes at a considerable expense. "Dropping a class from 25 to 22 students increases classroom expenditures by more than 10 percent,'' he writes in his 1994 book Making Schools Work: Improving Performance and Controlling Costs.
Project STAR's investigators concede that the greatest gains come in the first year or two that students attend smaller classes. But the point, they say, is that those gains remain just as strong as long as classes continue to be small. The researchers also point out that by 1st grade the sample of students in small classes included students who were retained, as well as 250 students who had not gone to kindergarten. (Kindergarten attendance was not mandatory in Tennessee at the time.) If anything, the STAR researchers say, their numbers are probably conservative.
They suggest that, contrary to what Hanushek says, the benefits are cumulative. In kindergarten, 55 percent of the top-scoring classes in the project's sample population were small classes. By 3rd grade, small classes accounted for 78 percent of the top 10 percent.
As for the cost, the STAR researchers point out that reducing the number of students that are held back each year or that require remedial service through the federal Title I program for disadvantaged students translates to cost savings in the long run. "Is it worth it to spend $1,500 extra for a child not to fail,'' Nye asks, "rather than spend $10,000 for a child to repeat a grade?''