In “Moving Beyond Spending Fetishes” (November 1995), University of Rochester Professor Eric A. Hanushek argued that increased spending for education does not equate to increased student achievement. To support his position, Hanushek cited Project STAR (Student Teacher Achievement Ratio), a statewide, longitudinal study of the effects of class size on student achievement, conducted in Tennessee between 1984 and 1990: Only a few of some 300 econometric investigations suggest that student achievement would improve with smaller classes, and an almost equal number confidently point in exactly the opposite direction.... Most studies give no reason to expect a change in student performance as a result of smaller class sizes. Moreover, direct experimental evidence from Tennessee reinforces these results by demonstrating that small classes at best make a discernible difference in kindergarten but not at later primary grades (p. 61).
Although Hanushek (1995) may have reviewed some secondary analyses and “econometric” investigations, his conclusions are incorrect. As a principal investigator of Project STAR, I welcome the recognition of our findings, but I am concerned that those findings be presented accurately.
Policy decisions must rely on substantial primary research, not on ideology, supposition, weak interpretations, or secondary reviews. In truth, our research shows the substantial positive benefits of early, small-class experiences for student achievement and development. Readers should note that Bracey (1995a, b) also has raised serious questions about Hanushek's logic and his conclusions.
Project STAR involved more than 7,000 students (K–3), who were randomly assigned to small classes (about 13–17 pupils per teacher), regular classes (22–26), and regular-with-aide classes. Teachers were assigned at random in approximately 330 classes each year, or about 110 of each type. Each school with a small class also had at least one class of each other type, so the in-school design controlled for building-level effects such as leadership, schedule, curriculum, and expenditures. Students were tested in controlled, monitored conditions. We employed a conservative approach, using the class as the unit of analysis.
Looking at Cause and Effect
Project STAR was an educational experiment designed to show cause and effect, not one of the less powerful correlational studies often cited in the literature. In contrast to Hanushek's assertion that small classes at best make a discernible difference in kindergarten, but not in later primary grades, Project STAR demonstrated substantial differences not only in kindergarten, but also in grades 1, 2, and 3.
The reading effect sizes by grade—.21, .34, .26, and .24 (Word et al. 1990)—were computed using the total data before correcting for out-of-range large classes that “shrank” or small classes that “grew” in grades 1, 2, and 3 due to pupil mobility. In this analysis, the greatest differences were seen in grade 1, not in kindergarten, as Hanushek erroneously stated.
When we removed the out-of-range classes, omitted the regular-with-aide classes, and analyzed distinctively small and regular classes, reading effect sizes became substantially larger than in the original analysis for kindergarten and grades 1, 2, and 3 (.52, .45, .63, and .63). The true effect of class size, then, probably can be found somewhere between the original effect sizes and those from the re-analysis.
Project STAR data provide a resoundingly affirmative response to the question, “Does class size make a difference in primary grades?” According to Finn and Achilles (1990), This research leaves no doubt that small classes have an advantage over larger classes in reading and mathematics in the early primary grades (p. 573).
Researchers in Tennessee have continued, with modest funding, to maintain the Project STAR database, to do exploratory analyses, to follow participating students in the Lasting Benefits Study, and to track the results of Project Challenge, a policy application of Project STAR findings in 16 of the state's poorer districts.
The Lasting Benefits Study results show that in 8th grade, students who had small classes in grades K–3 remain significantly ahead of those who were in regular classes. On average, poor districts participating in Project Challenge have moved from well below to somewhat above the state average performance in 3rd grade reading and math scores as they have reduced class sizes. Several other project-based studies are showing similar results. (The class size research is maintained at the Center of Excellence for Research and Policy on Basic Skills at Tennessee State University in Nashville.)
Using STAR Findings
It is time to stop misrepresenting Project STAR findings; such misrepresentation hinders the development of positive new policies. It is time to move on and to put the data to use. Project STAR researchers are concerned about better use of education funds, but our options will not be considered while ideology confounds the data.
- small classes ameliorate the effects of large schools;
- fewer students are held back a grade;
- while small classes benefit all students, minority students benefit the most;
- students receive more individual attention;
- smaller classes are friendlier and more intimate;
- there are fewer discipline problems in smaller classes;
- students are more likely to participate in activities.
In brief, STAR data show that small classes in early primary grades benefit students and provide a basis for substantial education reform without necessarily requiring massive infusions of funds. Consider some potential cost savings from using small classes in grades K–2 or K–3: fewer retentions, less need for remediation and/or special education, improved behavior, and increased achievement.
Project STAR and numerous subsequent studies conducted using its database provide the best research to date on class size effects. According to Frederick Mosteller (1995), Professor Emeritus, Harvard University: The Tennessee class size project [is] a controlled experiment which is one of the most important educational investigations ever carried out.... [It] illustrates the kind and magnitude of research needed in the field of education to strengthen schools (p. 113).
Leaders in 11 states have agreed with STAR findings and have enacted class-size initiatives. Space problems? Here's a challenge for the appropriate uses of technology. Personnel? Consider alternative certification. Parent involvement? Accept STAR's substantial class-size results, and a world of real restructuring options opens up. Let's start now.
