Does Class Size Matter

few years. The cry for smaller classes is DOES CLASS SIZE also resounding in Canada, Australia, the U.K. and even Japan, whose record of MATTER? secondary school performance is the envy of most other developed countries. Legislators are spending billions to reduce class sizes. Will the results Fewer students in a classroom be worth the expense? seem to translate into less noise and less disruptive behavior. By Ronald G. Ehrenberg and Dominic J. Brewer.

• In the U.S., Australia, Canada and other 4. Class-size reduction does have one countries, legislators have spent billions of obvious drawback: it costs plenty. It dollars to reduce class size in primary schools. requires more teachers and more In California alone, officials have already classrooms. These expenses can dwarf the invested about $1.5 billion. price of alternative schemes, such as testing • Hundreds of studies have examined whether teachers or increasing their pay as a means smaller classes really do improve academic of attracting better candidates. performance, but almost all of them have been 5. The surge of interest in smaller classes inconclusive. has spurred fresh analyses of the largest, • An exception, the Tennessee Project STAR most conclusive study to date, Project (Student-Teacher Achievement Ratio), found that smaller classes particularly benefit STAR, which took place in the late 1980s. minority students in early grades. At the same time, new data are flowing from various other initiatives, including the 1. EDUCATION IS THE safest route to a California program and a smaller one in life that is better in essentially all the ways Wisconsin. These results and analyses are that matter: intellectual, social, financial, finally offering some tentative responses to even physical. It is a pillar of modern the questions that researchers must answer society and the subject of endless, often before legislators can come up with policies passionate arguments about how it can best that make educational and economic sense: be improved. Do small classes in fact improve school 2. In the U.S., these debates have gained achievement? If they do, in what grades do intensity following revelations that the they accomplish the greatest good? What country's secondary school students kind of students gain the biggest benefit? perform poorly relative to many Asian and And most important of all: How great is the European students. The news coincided benefit? with increasing concern over the nation's urban and lower-income suburban schools, Why do Teachers Prefer Small too many of which have achievement levels Classes? far below those of middle-class and upper- 6. EDUCATORS HAVE a multitude of middle-class suburban schools. explanations for why smaller classes might 3. Of all the ideas for improving education, be expected to improve academic few are as simple or attractive as reducing performance, although frequently the ideas the number of pupils per teacher. With its are based on anecdotes. Fewer students in a uncomplicated appeal, class-size reduction classroom seems to translate into less noise has lately gone from being a subject of and disruptive behavior, which not only primarily academic interest to become a gives the teacher more time for class work public issue. In the U.S., more than 20 but also more freedom to engage students states have adopted policies aimed at creatively—by dividing them into groups decreasing class size, and billions of dollars for specific projects, say. And smaller have been spent or committed in the past classes make it more likely that the teacher 90 can bestow individual attention on bigger share of that time, but the increase is struggling students. Researchers suspect not nearly enough to account for any that the benefits of small classes derive significant differences in academic from this attention as well as from fewer performance. disruptions and more opportunities for 9. If teachers work in more or less the same teachers to use instructional methods, such way regardless of class size, what accounts as encouraging group discussions and for the benefits of smaller classes seen in assigning more written work, that usually experimental studies such as STAR? One do not succeed in large groups. likely explanation is that teachers who have 7. Smaller classes also allow teachers to naturally settled on methods well suited to encourage more discussion, assign more smaller classes— those who already like writing and closely examine their students' splitting the class up into small groups, who written work. This indicates that much of develop personal relationships with the benefit of reduced class size may students and who emphasize hands-on depend on the teacher’s method of projects—do very well when they are instruction. Finally, some analysts believe actually given small classes. Their that in the early grades, students in smaller improved performance pulls up the average, classes are more likely to develop good which probably also gets a lift from fewer study habits, higher self-esteem and disciplinary problems. possibly other beneficial cognitive traits— 10. This interpretation is consistent with which may very well persist for years, even findings that substantial performance gains after the students have gone back to from small classes occur in the early regular-size classes. All these ideas are elementary grades and do not accumulate largely speculative, however, because beyond first or second grade. Kindergarten hardly any research has attempted to nail and first-grade teachers in particular tend to down the ways that smaller class sizes may use small groups, hands-on projects and benefit children. personal relationships with students. To confirm this scenario, researchers would Ingredient X: A Teacher’s have to study the relationships among class Adaptability size, instructional activities and 8. THE CONVENTIONAL wisdom about achievement at various grade levels. That small classes is that they minimize no one has done this work is surprising, disruption and free teachers to bestow considering how useful it would be to individual attention and to use creative administrators in deciding where and how approaches, such as letting students work in to use small classes. small groups. Where discipline is not a significant problem, then, any achievement What the Record Shows gains resulting from reducing class size 11. ONE WAY investigators have would be expected to derive mainly from attempted to analyze the effects of class the teacher's use of methods that take size is by reviewing existing data, such as advantage of smaller classes. But study records kept by the U.S. Department of after study has found that educators rarely Education. These show that between 1969 change their instructional styles to match and 1997, the average number of pupils per the size of their class. In fact, data from teacher in American public and private Project STAR, the best study of small elementary schools fell from 25 to 18, a classes to date, show how hard it is to decline of greater than 27 percent. In change the way teachers practice their craft. secondary schools, the number also fell, Moreover, educators seem to devote the from 19 to14. same overall amount of time to individual 12. How much did academic performance instruction in small and large classes. With change while these steep drops in pupil- fewer kids in a class, each child gets a teacher ratios were occurring? Not a lot. 91

Data from the National Assessment of whose average amount of experience also Educational Progress—a series of tests that went up. is the only U.S.-wide indicator of student knowledge in reading, mathematics, science Demographic shifts make it very and other subjects—show no significant or difficult to determine the effect of consistent gains. In some specific age and subject categories, such as 17-year-olds and reductions in pupil-teacher ratios. science, performance actually decreased slightly. 16. The bottom line is that demographic shifts make it very difficult to determine the How Reliable are the Findings? effect of reductions in pupil-teacher ratios. 13. DO THESE findings mean that class Well-designed experiments strive to rule size makes no difference? Not necessarily. out the influence of those other factors by For a variety of reasons, most researchers, randomly assigning students and teachers to including us, pay little attention to those different class sizes and by including a figures. For instance, schools strive for large sample. Over the past 35 years, more than just high test scores; they also hundreds of studies and analyses of existing usually try to keep their dropout rates low. data (such as the Department of Education And indeed, the dropout rate for students records) have focused on class size. Most aged 16 to 24 fell from15 to 11 percent over found some evidence that smaller classes that period. Because dropouts generally benefit students, particularly in the early come from the low end of the achievement grades, and especially kids at risk of being distribution, a reduction in the dropout rate underachievers. Unfortunately, most of could be expected to pull down average test these studies were poorly designed. Teacher scores in the upper grades. and student assignments were rarely 14. Another reason for discounting those sufficiently random; a number of studies data goes right to the heart of the were simply too brief or too small, and too difficulties in this field of study: it is hard to few had independent evaluation. isolate the effects of class size from the myriad factors that influence student Project STAR performance. Ideally, U.S. students would 17. The notable exception was the STAR all come from families that are financially project. This Tennessee study was a state- well off, with two highly educated, English- sponsored, $12-million demonstration speaking parents who are involved in their program. Students entering kindergarten children's schooling. Teachers would all be were randomly assigned to one of three creative and have complete mastery of their kinds of classes: a small class of 13 to 17 subject matter. Schools would be nicely students, a regular-size class of 22 to 26, or outfitted with libraries, computers and other a regular-size class with both a teacher and resources. a full-time teacher's aide. The students 15. The reality is that in 1995 only 68 remained in whatever category they had percent of American students came from been assigned to through the third grade, families with two parents in the home— after which they joined a regular classroom down from 85 percent in 1970. The number in the fourth. To ensure that teaching of children who had difficulty speaking quality did not differ, teachers were English rose from 2.8 percent in 1979 to 5.1 randomly assigned to small and regular-size percent in 1995. And the percentage of classrooms. Few teachers received any children living in poverty increased from special training for working with small 14.9 in 1970 to 20.2 in 1995. The only good classes, and there were no new curricular news was that the level of education among materials. parents increased a bit during that time 18. After the study ended in 1989, period, as did the level among teachers, researchers conducted dozens of analyses of 92 the data. One of the few points analysts 21. He and others have also shown that agree on is that the teacher's aides did not during the study, too many children make any difference. Researchers disagree migrated from the regular to the small about how long students have to be in classes, probably because school personnel smaller classes to set a benefit, how big that caved in to parent demands. Furthermore, benefit is, and when it becomes noticeable Hanushek asserts that STAR had —in other words, the collected findings insufficient checks to ensure equality in have yielded no consensus on the issues of assessment of teacher and student real interest to policymakers. assignments. These are good points, but 19. Jeremy Finn of the State University of they do not really undermine STAR'S New York and Charles M. Achilles of finding of a statistically significant benefit Eastern Michigan University found "an of being in a class having 13 to 17, rather array of benefits of small classes" in their than 23, students. review. Finn calculated that students in the small classes outperformed their counter- Criticism does not undermine the parts in regular-size classes and that this findings of a statistically significant jump in achievement generally appeared by the first grade. Best of all, this advantage benefit of being in a small class. seemed to persist into upper elementary grades even after students returned to larger Two Views: California classes. Finn and Achilles also found that and Wisconsin the effect was stronger for minority 22. THE CHALLENGE for legislators now students. Black and Hispanic children is to come up with sensible policies based improved their scores slightly more than on sound interpretations of STAR and other did other students—a significant finding studies. Unfortunately, the largest public from a policy standpoint, because program so far, California's multi-billion- minorities typically score below non- dollar effort, begun in 1996, stands more as minorities on standard tests. a model of what not to do than as an 20. A few analysts, among them Eric initiative worthy of emulation. That state is Hanushek of Stanford University, criticize trying to reduce classes in kindergarten STAR and some of the key conclusions through grade three from a maximum of 33 reached by its proponents, Hanushek agrees to a maximum of 20 in rich and poor that students can gain an initial benefit from districts alike—despite a shortage of small classes. But, he argues, the STAR qualified teachers, especially in low-income data cannot be used to prove that the gains areas. This across-the-board approach may persist for years after a student has returned be politically expedient, but it seems to to regular-size classes. If a child is still have actually exacerbated the disparity in doing well years later, it is hard to know resources available to rich and poor schools how much of that performance stems from in California, other factors, such as a supportive home. 23. Not surprisingly, the program triggered Hanushek also disagrees with an analysis an increased demand for teachers in almost indicating that the benefits of small classes all California districts. The better-paying, accumulate—that students who stay in such more affluent districts got the best teachers classes for several grades widen the —including a fair number that came from performance gap with their peers in large the poorer districts, which were already classes year by year. When he studied the having trouble recruiting and retaining good four-year gains of STAR students who were teachers. These mostly urban districts ended in the smaller classes from kindergarten up with inexperienced teachers who had no through grade three, he did not find the credentials. The rapidly growing urban gains to be larger than those in regular-size districts also had little space to build new classes. classrooms, and some of them could not 93 reduce class size to 20, which they had to little forethought and insight, can consume do before they could qualify for the state billions of dollars and, at least in the short funds. run, produce only minuscule gains and even 24. The California experience has not led some losses. to any firm conclusions about whether class 27. Alternatives need to be considered. size affects performance. There was no What little work has been done on teacher state testing system in place initially and no competence suggests that students perform across-the-board evaluation procedures. better with more able teachers and, at the Nevertheless, several researchers who secondary school level, better knowledge of looked at the first years have managed to their subject matter. Astoundingly, make a few points concerning third-grade however, when choosing among applicants students, the only ones for whom statewide for teaching positions, school districts often test-score data are available. The evaluators do not select the candidates with the found a small but statistically significant strongest academic backgrounds and the achievement advantage in reading, writing highest scores on aptitude tests. Rather, and mathematics for students in classes that school officials tend to favor teachers who had been reduced to 20 or fewer pupils, as live nearby, graduated from local colleges compared with the classes of more than 20. and possess proved classroom management In contrast to the STAR findings, though, skills. Emphasizing aptitude and subject- the tiny effect did not appear to vary for matter competence in hiring decisions students of different races or ethnic or wouldn't cost anything, although getting socioeconomic backgrounds. more high-aptitude candidates to go into 25. The second program, Wisconsin's teaching would probably require higher Student Achievement Guarantee in salaries. So far no one has studied the Education (SAGE), also begun in 1996, relative costs of attracting better teachers. was a five-year study. It was small—class 28. Legislators and administrators need size was reduced in just 14 schools—but more solid information on the relative costs noteworthy because it targeted schools in of the other options before they can make which at least 30 percent of the students sensible policy decisions. Let's hope they were below the poverty level, compared get it before they commit billions more to with California's one-size-fits-all approach. reducing classes across the board—and It brought down the average number of before millions of kids get covered by pupils per teacher in kindergarten through blanket policies that may be less effective third grade to 13. Analysts have so far than they could be. compared first-grade students in SAGE schools with first-grade students in a group RONALD G. EHRENBERG is professor of economics of schools serving populations that have at Cornell University. DOMINIC J. BREWER, who specializes in the economics of education and similar family income, enrollment and analyzes of programs and policies on education racial compositions. The results from the issues, is a visiting professor at the University of first two years are in line with those from California. STAR: first-grade SAGE students made gains that are statistically significant—and MORE TO EXPLORE that are considerably larger than those Standards for Our Schools. Mark S. Tucker and calculated for the California initiative. Judy Codding. Jossey-Bass, 1998. Research: Sizing Up Small Classes. Linda Jacobson What are the Implications? in “Education Week on the Web;” February 28, 26. STUDIES such as STAR and SAGE 2001, www/edweek.org/ew/ewstory have made it hard to argue that reducing Class Size and Student Achievement. Ronald G. class sizes makes no difference. On the Ehrenberg, Dominic J. Brewer, Adam Gamoran and other hand, the California initiative has J. Douglas Willms in Psychological Science in the shown that the strategy, applied with too Public Interest, Vol. 2, No. 2, pages 1-30;