Yesterday’s post on class size was a bit like a projective psychological test, in that readers saw what they wanted to see. Pro-class size reduction, anti-class size reduction, it’s all in there.Today I’m spotlighting the Tennessee STAR study, which, along with the High/Scope Perry Preschool Project, is one of the two most influential education research studies of the past quarter-century. All studies have their strengths and weaknesses, and no single study constitutes the last word on what is known on a particular topic, especially one as complex as class size. Still, the strengths of the design of the STAR study have made it a star. Funded by the Tennessee General Assembly, the Student/Teacher Achievement Ratio (STAR) study was a four-year experiment begun in 1985 in which kindergarten students were randomly assigned to one of three treatment conditions: a regular class of 22 to 25 students; a regular class of 22 to 25 students with a full-time teacher’s aide; or a smaller class of 13 to 17 students. Students stayed in the same condition through the third grade. Because both students and teachers were randomly assigned to the three conditions, we can be pretty confident that the three groups were equivalent at the start of the study. Therefore, any differences in student outcomes observed among the three groups can be said to be caused by differences in the conditions.
The results were unequivocal. Students in the small class condition scored significantly higher on reading and math tests in the primary grades than students in the other two groups, and the benefits were especially pronounced for African American students and students attending inner-city schools. Follow-up studies show that these effects persist on a range of behavioral and achievement outcomes as students move through secondary school, with students in the small class condition having a lower risk of dropping out of high school, and a higher likelihood of taking college entrance exams. STAR serves as a kind of “proof of concept”: the study demonstrates persuasively that placing students in small classes in the early elementary grades can have lasting beneficial effects on a range of student outcomes.
The challenge is in applying what we’ve learned from STAR to other settings, which is usually what we are interested in doing. STAR is not a license to say that any old class size reduction initiative will have the same effects. The more closely a particular policy initiative resembles STAR, the more likely we are to observe similar findings. If, for example, you believe that teaching and learning in high school are different than teaching and learning in the early elementary grades, you might be reluctant to make any inferences from STAR about what would happen if we reduced class size in high schools, since STAR didn’t do that. And, since STAR involved a contrast between a class of 22 to 25 and a class of 13 to 17, you might not want to speculate about the consequences of reducing class size from 28 to 21, since that wasn’t actually observed in the STAR study. (In fact, there’s evidence from other studies that it’s not so much that small classes are good as it is that large classes are bad.)
But perhaps even more important is how class size reduction is achieved—the mix of what policy wonks refer to as policy instruments that define how a policy is enacted. California tells a cautionary tale. On the heels of the STAR study, California embarked on an ambitious statewide class size reduction initiative. In 1996, school districts received an additional $650 for each student in a K-3 class of 20 or smaller; the figure rose to $800 per student the following year. This was a powerful incentive that few districts could resist. Studies of the initiative show that California districts scrambled to reduce class sizes by hiring teachers with intern or emergency credentials, and many of these teachers wound up teaching in large urban schools serving poor, minority students who were English language learners. (In contrast, all of the teachers in STAR were fully certified.) Moreover, districts had to cannibalize space for small classes that otherwise would have been used for other purposes, including special education, arts and music, and athletics. Perhaps as a consequence of how class size reduction was implemented in California, researchers were unable to conclude that it had positive effects on student achievement.
I’m persuaded by the STAR study that class size reduction can lead to better student outcomes. The key questions for me are: (a) under what conditions might class size reduction work? and (b) what mix of policy instruments can create those conditions? These questions focus attention on class size reduction in particular contexts, which, in the end, is what most of us care about.
