Does Smaller Class Size Improve Student Performance? A Brief Review of the Empirical Literature

Submitted to the Council of Economic Advisor Meeting
September 27, 2002
St. Petersburg, Florida

Russell P. Chuderewicz

Summary

Virtually everyone, students, teachers, and the concerned public would, without a doubt, prefer smaller class sizes relative to larger class sizes. Given these preferences and the scarcity of resources, the relevant question therefore is: Do the benefits of smaller classes exceed the costs associated with providing smaller classes? This brief focuses only on the benefits of smaller class sizes. Far from a specialist in this field, I proceeded to review the literature in this area and was surprised to see the lack of evidence connecting smaller class sizes to student performance. Therefore, the evidence suggests, given the substantial cost of mandating smaller classes in the state of Florida, that Class Size Amendment/Amendment 9 would not likely pass a cost - benefit test.

Contents:

A look a the empirical results – non-experimental
A look at empirical results – experimental
The Tennessee STAR Project
Class Size Reduction in California
Conclusion

Non- Experimental Empirical Results

Over the last half of 20th century, pupil-teacher ratios (in the US) have fallen from about 27:1 to 17:1, or 35% (US Department of Education and Hanushek (1999)). Note, importantly, that falling pupil-teacher ratios are not the same as class size reductions. Often, specialized teachers are hired or teachers take up administrative positions so that pupil – teacher ratios fall while class size remains unaffected. Lewit and Baker (1997), however, show that pupil-student ratios and class sizes tend to move together in the aggregate.

Non-experimental studies use actual data within the normal operations of schools as compared to experimental studies such as the Tennessee STAR project, where a controlled experiment is performed (more on the STAR project below).

Table 1 is reproduced from Hanushek (1999). Remarkably, of 277 estimates attempting to capture the effects of teacher- pupil ratios or class size reductions on student performance, only 15% of the estimates are significant and positive. Virtually the same percent, 13%, are negative and significant suggesting that raising the teacher-pupil ratio (or lowering class size) results in poorer student performance. As Hanushek points out, the results are even more striking if one considers the percent of studies that result in either negative and significant or statistically insignificant estimates (85%). In terms of the last two rows of Table 1, it appears that smaller class size (or larger teacher-pupil ratios) has stronger positive effects in secondary schools as compared to elementary schools (17% to 13% respectively). These results are evidence against the widely held belief that smaller classes are more effective during the early years of the education process.

Table 1

Percentage Distribution of Estimated Influence of Teacher-Pupil Ratio on Student Performance by Level of Schooling

Note. A positive sign implies that smaller classes enhance student performance.

Given the possible problems inherent in the distinction between class size and teacher-pupil ratios, Hanushek goes on to examine the estimates from studies that are limited to investigating the relationship between class size and individual students’ performance (coined value added models). Table 2, again reproduced from Hanushek (1999) provides the results.

Table 2

Percentage Distribution of Effect of Class Size on student performance, Based on Value-Added Models of Individual Student Performance

Note. A positive sign implies that smaller classes enhance student performance.

Again, the results are not favorable in terms of small class sizes increasing student performance. As Hanushek points out, of the best available studies, single-state, value-added studies of individual classroom achievement (bottom row), only 1 of 23 or 4% result in positive and significant small class effects. In fact, more studies indicate a negative and significant relation (13%), suggesting that small classes result in poorer student performance.

Hanushek Summarizes:

“The econometric evidence as a whole gives little support to the idea that smaller classes will lead to general improvements in performance.” (page 148)

And,

“The best studies that concentrate on differences in performance across individual classrooms with varying numbers of students and that separate out other possible influences on student performance offer no support whatsoever for general gains in achievement through class size policies.” (page 149)

Experimental Empirical Results

By far the most publicized Experiment in this area is the Tennessee STAR project (STAR = Student – Teacher Achievement Ratio).

Results – The following description of the results can be found on the world wide web http://www.wested.org/policy/pubs/full_text/pb_ft_csr23.htm

“Class Size Reduction: Lessons Learned from Experience” by Joan Mc Robbie, Jeremy D. Finn, and Patrick Harman

• positive results for small classes, year after year (K, 1, 2, and 3), in all subject areas and all school locales (inner city, urban, suburban, rural)
• similar results for boys and girls
• greater academic benefits – often about twice as great – for minority students or students attending inner-city schools
• lasting benefits through grade 7 or beyond, even though all students were returned to regular-size classes in grade 4.

Given these positive results after years of finding essentially no relationship between class size and student performance, much pressure was applied to policy makers to lower class size. Are these results believable?

From The Buckeye Institute: “Two new studies cast doubt on benefits of class-size reduction” on World Wide Web at http://www.buckeyeinstitute.org/policy/1999_5.htm

STAR’s exaggerations: STAR's poor methodology exaggerated benefits from class-size reduction. For more details on why STAR’s results are exaggerated, and other concerns about the methodology of the STAR project, see Hanushek (1999) and references therein.

Hanushek’s concerns about STAR’s results:

• between 20 and 30 percent of the students quit the project each year, with less than half the original number remaining at the end

• the students who quit tended to be below-average achievers, giving the smaller classes a perceived boost in achievement

• no pretests were conducted on any students upon enrollment, which provided no
benchmark to assess their level of achievement

• neither the teachers nor the schools chosen for the project were selected randomly

Hanushek also detailed that the data from the STAR project suggests that the gains in student achievement appeared to be virtually exclusive to kindergarten. For example, the difference in performance did not widen as students moved from first through third grade. One would hope the difference in performance would widen throughout the experiment since students continue to get more resources (smaller classes) that should keep adding an advantage (page 154).

KEYS TO ACHIEVING A SMALL CLASS EFFECT (again from “Class Size Reduction: Lessons Learned from Experience” by Joan Mc Robbie, Jeremy D. Finn, and Patrick Harman)

• Adequate supply of good teachers. No organizational arrangement, including small class size, can compensate for poor teaching. In Tennessee, all STAR teachers were state certified and qualified to teach in their assigned grades. Even among the small classes, some teachers were more effective than others; researchers have yet to study what may have caused these differences.

• Sufficient classroom space. STAR’s participating schools had no problem finding appropriate space to create enough classrooms for the reduction in numbers of students per teacher.

• A representative student mix in each class. In STAR, the mixture of students in the class was determined at random and so mirrored the diversity in the school as a whole. Research has not revealed what would happen if, for example, 17 pupils with learning or behavior problems were assigned to a small class. In such a case, positive effects are less likely without the infusion of significant additional resources.

• Teacher access to adequate materials and services. STAR teachers had no change in the materials and services normally available to them. Small- and regular-class teachers had access to reading specialists, school psychologists, special
education programs (although there is evidence that the need for these services was reduced), and other school wide services.

• Small classes were not intended to serve as a substitute for other programs with demonstrated efficacy (including bilingual programs).

California’s CSR (Class Size Reduction) – The following was obtained from Class Size Reduction in California: The 1998-99 Evaluation Findings on the World Wide Web at http://www.classize.org/summary/98-99/

Results:

• CSR is Nearing Full Implementation
• Teacher Qualification Levels Declined in Grades K–12,
But the Decline Was Greatest in the Elementary Grades
• Decline in Teacher Qualifications Was Greatest in
Schools Serving Students Most in Need
• Reduced Size Classes Provided More Individual
Instruction But Not Different Curriculum
• Small Improvement in Third-Grade Student Achievement
Persisted into Fourth Grade
• Special Populations Fared Somewhat Worse Under CSR
The California CSR began in 1996 and was motivated largely by the STAR results outlined above. So far, the most positive result of the experiment is a “small improvement in third – grade student achievement.” There are plenty of negatives however. For one, teacher qualifications have declined, especially in the elementary grades. Furthermore, the decline in qualifications was most dramatic in the elementary schools serving low income, EL, and minority students. Another negative is that special education and EL students fared worse under CSR in California. Many teachers switched from teaching these two groups to teaching regular K-3 classes (about a 1000 teachers switched).

Important distinction between the California CSR and Tennessee STAR

From Class Size Reduction in California: The 1998-99 Evaluation Findings (page 3)

“However, there are substantial differences between Tennessee’s and California’s situations, and one cannot assume that the CSR program will automatically produce the same or better results for California’s students. The Tennessee program was a carefully controlled experiment involving about 10,000 students, whereas the California program is
implemented statewide and serves 1.8 million students. Another difference is that California’s program reduces its maximum class size of 33 students down to 20, whereas Tennessee took its class size of 22–26 students down to13–17. California also serves a student population that is decidedly more ethnically and linguistically diverse. Finally,
California schools lack two important ingredients that Tennessee schools did not—adequate space and enough qualified teachers for program implementation. These differences between the two programs are large. When considered together, they clearly indicate that California’s CSR program must be judged on its own terms rather than as a replication of the Tennessee experiment.

It is also important to understand that the CSR program is not being implemented within a static system. California’s schools are undergoing numerous other major educational reforms that involve changes in curriculum standards, state assessments, bilingual education guidelines, teacher certification procedures, and student promotion policies. These interventions and others interact in intricate and complex ways, making it difficult to attribute changes to any single effort, including the CSR program.”

And from Class Size Reduction: “Lessons Learned from Experience “ Policy Brief
No. Twenty-Three, August 1998 West Ed.

http://www.wested.org/policy/pubs/full_text/pb_ft_csr23.htm

California’s first year of implementation, for which staffs had almost no time to plan, required hiring 18,400 new teachers. Half were inexperienced; 30 percent were uncredentialed; and 21 percent were hired on emergency permits, meaning they had college degrees and had passed a competency test, but lacked any preparation for teaching. Finding 18,000 new classrooms meant turning libraries, music rooms, computer and science labs, childcare centers, faculty lounges, and even stages in auditoriums into primary classrooms, either temporarily or permanently.

Conclusion

The empirical results offered in this brief review suggest that the benefits of reducing class size are at best, hard to pin down and at worst, don’t exist. I strongly believe that we need to focus on improving the quality of K-12 education across the State of Florida. Passing Class Size Amendment/Amendment 9 is not the answer.

References / Citations

“Class Size Reduction: Lessons Learned from Experience” by Joan Mc Robbie, Jeremy D. Finn, and Patrick Harman. Policy Brief No. Twenty-Three, August 1998
http://www.wested.org/policy/pubs/full_text/pb_ft_csr23.htm

Class Size Reduction in California: The 1998-99 Evaluation Findings on the World Wide Web at http://www.classize.org/summary/98-99/

“ Class Size” The Future of Children,” 1997, 7, 112-121.

“Some Finding From an Independent Investigation of the Tennessee STAR Experiment and From Other Investigations of Class Size Effects.” Educational Evaluation and Policy Analysis, Summer 1999, Vol. 21, No. 2, pp. 143-163.

“Two new studies cast doubt on benefits of class-size reduction.” The Buckeye Institute: 1999: on World Wide Web at http://www.buckeyeinstitute.org/policy/1999_5.htm