Commonly advocated by groups like the MEA, class-size reductions are aimed at improving teacher effectiveness by allowing teachers to focus on a smaller number of students. Teachers would have fewer papers to grade, behavior problems to manage, questions to answer and parents to consult; as a result, individual students would get more attention, learn more and improve their achievement.
In a September 2007 letter to The New York Times, the NEA’s Reg Weaver clarified the union’s position on student-teacher ratios when he stated that the federal government needs to “provide resources for programs that improve test scores, such as smaller class sizes. …” In support of a federal class-size reduction program, President Bill Clinton said: “Reducing class size is one of the most important investments we can make in our children's future. Recent research confirms what parents have always known — children learn better in small classes with good teachers, and kids who start out in smaller classes do better right through their high school graduation.”
Lowering class sizes does have intuitive appeal as a solution for low-performing schools. In a meticulous paper on class-size policies, Douglas Harris of the University of Wisconsin reported on survey results indicating that parents and the general public overwhelmingly support the idea of class-size reductions. Harris explained that extensive class-size reductions passed as a referendum in Florida even in the face of resistance from the state’s governor and of the general understanding that higher taxes would be needed to pay for the reform. He suggests that “one explanation for the popularity of small classes is that parents cannot easily observe many forms of educational quality.” Thus, he theorizes that parents may support class-size reduction policies because they are tangible reforms that can be enacted quickly.
Unfortunately, as Harris notes, class-size reduction may not be all that meets the eye. In a recent study of Texas student performance, Rivkin, Hanushek and Kain found that fourth- and fifth-grade students in smaller classes performed better in both math and reading. The effects of smaller classes got smaller each year, however, and were not apparent in grade seven. Nonetheless, these researchers do not advocate class-size reductions as a policy solution to low student performance. Instead, they focus on improving teacher quality. They explain that improving teacher quality by one standard deviation — i.e. getting a teacher who ranks in quality at the 85th percentile rather than at the 50th percentile — “is equivalent to a class size reduction of approximately ten students in 4th grade and thirteen or more students in 5th grade, and an implausibly large number in 6th grade.”
A randomized experiment of lowering class sizes during the 1980s in Tennessee — the famous Tennessee STAR project — also showed that lowering class sizes in early elementary grades raised student achievement. In this study, students were randomly assigned to three types of classrooms. The classroom ratios were 13-17 students to one teacher, 22-25 students to one teacher, and 22-25 students to one teacher and a teacher’s aide. Aside from potential shortcomings in the study itself,[*] several unintended consequences prevent this strategy from becoming a feasible solution in Michigan. As Rivkin, Hanushek and Kain note, the costs associated with class-size reductions do not simply result from the costs of hiring additional teachers. The need for extra classroom space and for more support staff cannot be ignored.
Perhaps the largest barrier to this reform is that the supply of high-quality teachers is limited, so the prospective gains from smaller student-to-teacher ratios would likely be undermined by staffing those classrooms with less effective teachers. As Jay Greene writes in his book “Education Myths,” “Even if class size reduction does produce improved performance under optimal conditions of a small, controlled experiment like the STAR project, labor pool problems may prevent this success from being reproduced on a large scale.” In other words, under class-size reduction policies, schools would be forced to hire more teachers, and those applicants may be the inferior teachers who were passed over in prior years.
In fact, this harmful substitution occurred during the late 1990s when California attempted widespread class-size reductions based partly on the perceived success of the Tennessee STAR experiment. California lowered the average number of students in a class from 28 to 20 in a program involving more than 1.8 million students, in contrast to the roughly 11,000 in Tennessee. The price tag was over $1.5 billion per year. Although some third-grade students showed slight achievement gains, other reforms undertaken simultaneously in California at that time make it difficult to attribute these minimal gains to the class-size reduction policy. Moreover, even if class-size were responsible, the performance gains were meager given the cost.
The California program’s evaluators were analysts from RAND Corp. and other leading research firms. Their report confirmed that principals had hired teachers of lower quality when the project was implemented. The evaluators found: “While [the project] was being implemented, the qualifications of California’s teacher work force declined. The proportion of teachers with full credentials decreased in all grades, … as did the proportion of teachers with the minimum level of college education (only a bachelor’s degree) and the proportion of experienced teachers (those with more than three years of experience).” Even though these metrics for judging teacher quality are questionable, it is still safe to say that the quality of California’s teaching work force declined.
The California program evaluators also reported that the class-size reduction project did not close the achievement gap between white and minority students and that schools serving disadvantaged students were the most likely to hire teachers with less desirable credentials. Thus, California’s class-size reduction policy was likely a failure, due at least in part to the low quality of additional teachers. Given that the project cost $1.5 billion dollars relative to a total state education budget of $34.9 billion, the modest and nonuniform gains simply did not justify the expense.
Although they were critical of California’s program, B.J. Biddle and David Berliner of the East Lansing-based Great Lakes Center for Education Research & Practice are supporters of class-size reduction. Regarding California, they suggest that lowering class sizes only to 20 students was not sufficient to realize significant gains. They also argue that there was not enough money to support the reform, and that, “[T]his inadequate funding imposed serious consequences on poorer school districts, which had to abolish other needed activities to afford hiring teachers for smaller classes.”
Yet simply increasing the budget for education is no trivial undertaking, and working within the reality of budgetary constraints, policymakers should consider the trade-offs involved with class-size reduction policies. As Jay Greene notes: “Any serious reduction in class sizes would require us to invest a very large amount of money, so we could only produce small classes by taking resources away from other educational priorities. … Smaller classes would almost certainly leave insufficient funds left over for other, much more promising reform strategies. Success in reducing class sizes would be a Pyrrhic victory — more would ultimately be lost than gained.”
The cost of this trade-off is real, no matter what policies one prefers. The University of Wisconsin’s Douglas Harris notes, “Resources that go to small classes and small schools cannot be used to buy laptops for teachers, raise teacher salaries, increase professional development, add pre-kindergarten programs, or purchase new textbooks.”[†]
Harris also provides a helpful guide for evaluating the costs and benefits of an education policy. He describes three ways to consider trade-offs: cost-benefit analysis, cost-effectiveness analysis and an “optimization” approach. Cost-benefit analysis is the most straightforward approach; it monetizes total costs and benefits and subtracts the former from the latter to determine the viability of a policy proposal. Cost-effectiveness analysis involves dividing incremental benefits by incremental costs, where incremental costs and benefits are the costs and benefits that accrue when looking at the next unit in a series. (The term “incremental” is equivalent to the economics term “marginal.”) The higher the ratio of incremental benefits to incremental costs, the better the solution. The reason that cost-effectiveness analysis is helpful is because it allows us to compare the efficiency of multiple policy proposals even when the total costs and benefits of those proposals are of very different sizes.
The optimization approach improves upon cost-effectiveness by considering the concept of diminishing marginal returns. In the optimization approach, the incremental costs and benefits are not assumed to be linear — or constant for each new unit — as they are in the cost-effectiveness analysis. In other words, as Harris explains, the incremental benefit of reducing a class size from 23 students to 22 students is not assumed to be the same as reducing the class size from three students to two students, for example. Under the optimization approach, the ratios that are calculated will point to the most cost-effective solution for class-size reduction by signaling the point at which reducing the class size by one more student is not as cost effective as the prior one-student reduction.
Without any budgetary constraints, Harris explains, the optimization approach would be the most helpful. However, since budget constraints do exist, and since these can easily preclude achieving the optimal solution, the cost-effectiveness approach, even though it assumes linear costs and benefits, is preferable.
Harris reports on his earlier analysis, which “suggests that increasing test scores by 0.05 of a standard deviation by reducing class size would require $1,287 in additional expenditures per pupil, much more than the apparent $163 cost per pupil of achieving the same test score increase through an increase in teacher salaries.” According to Harris, his own findings are consistent with earlier research that he claims “suggest[s] that the broad-based trend toward smaller classes in recent decades has probably resulted in lower student achievement than would have been possible if other uses had been made of the resources available.” Harris is careful, however, to qualify his claims. He asserts that since incremental costs are in fact not linear, there may be situations where class-size reductions would be warranted. For example, Harris states that moving from an exceptionally large class may be a good idea. Still, Michigan’s student-teacher ratios do not suggest extremely large class sizes. According to NEA estimates, the average student-teacher ratio during the 2004-2005 school year in Michigan was 17.8 students per teacher, compared to the national average of 15.8. In 2007, the average student-teacher ratio in Michigan was down to 17.4 students per class.[**] The incremental gain to be captured by reducing student-teacher ratios by one or two students to get to the national average — i.e., moving from approximately 17.8 to 15.8 — would probably not be cost-effective.
Harris has made a comparison of the resources involved in increasing teacher salaries and decreasing class sizes. As we have shown above, across-the-board salary increases are not a particularly compelling solution. Harris’s comparison is most helpful in demonstrating that class-size reductions, though popular, contain hidden costs. His calculations give some indication of the magnitude of costs associated with class-size reductions and lend support to the arguments of those who advocate looking at policy proposals from all angles.
One parting thought on class-size reductions: Policymakers should also recall that self-interest may be involved when teacher unions advocate class-size reduction policies. Terry Moe suggests that teachers unions support class-size reductions because they want more teachers, who in turn will become fee-payers or union members. Douglas Harris disagrees with this notion, but argues that since class-size reduction policies are extremely popular among teachers, unions are simply “representing the wishes of their members.” Perhaps Harris is right, but satisfying these wishes may not improve student achievement, even if they make teachers happier.
The point to take from this extended discussion of class-size reductions is that once again, the research suggests that policymakers should focus on ways to increase the number of highly effective teachers in the schools. As Jay Greene notes, students will do better in a larger class with a great teacher than they will do in a smaller class with an average or below-average teacher.
[*] For a discussion of the potential shortcomings in the Tennessee STAR study, see Greene, Education Myths: What Special-Interest Groups Want You to Believe About Our Schools — and Why It Isn't So.
[†] Research does not generally suggest that the alternative programs Harris mentions have a significant impact on student achievement. Nevertheless, Harris’ point remains: Costly class-size reduction initiatives inevitably drain resources from other possible reforms.
[**] The NEA correctly notes that student-teacher ratio is not the same as average class size, but they do concede that “no state-by-state ‘actual’ class-size information exists.” See “Class Size - NEA's Efforts to Gather Accurate Class Size Data,” National Education Association, www.nea.org/classsize/datacollection.html (accessed May 17, 2008).