Legislative Analyst's Office
Analysis of the 2003-04 Budget Bill
In 1996, the state established the K-3 Class Size Reduction (CSR) program. In 1996-97, the state provided $971 million in incentive funding for school districts voluntarily to reduce K-3 classes to 20 students per class. By 2002-03, funding for this program had grown to $1.7 billion—making it one of the most costly categorical programs in the state and the largest CSR program in the nation.
Program Intended to Improve Student Achievement. When California designed its CSR program, considerable research already had been conducted demonstrating that CSR could improve student achievement. The now famous Tennessee Student-Teacher Achievement Ration (STAR) project, and subsequent research, already had arrived at three central conclusions: (1) students in smaller K-3 classes performed better than students in larger classes; (2) students in smaller K-3 classes retained their achievement gains in grades four, six, and eight; and (3) minority and inner-city children gained the most from smaller classes—experiencing effects that were two to three times greater than the effect on other students. These findings helped motivate California to establish its own CSR program.
Research Has Since Emphasized the Importance of Integrating CSR Efforts With Broader Reforms. Since the mid-1980s, 20 states have established CSR programs—providing additional opportunities for researchers to examine the effects of CSR on student achievement. Program evaluations in Utah, Texas, North Carolina, and Wisconsin have concluded that CSR is most successful when states gradually phase in the program and combine smaller classes with teacher development initiatives. Wisconsin has conducted the most extensive and methodologically sound CSR evaluation. In 1996-97, Wisconsin created a statewide pilot CSR program targeting schools with high poverty rates. The program was phased in over five years and required K-3 classes be reduced to 15 students per class. In addition to reducing class sizes, participating schools were re quired to implement a rigorous academic curriculum, provide before-and-after school activities, implement professional development, and adhere to accountability plans. As with Tennessee's program, the Wisconsin evaluation found significant achievement gains, particularly among minority students.
California Implemented Large-Scale CSR Program Almost Overnight. Unlike the Tennessee and Wisconsin programs, California rapidly implemented its CSR program in all schools at one time. Schools began reducing class sizes only a couple months after the legislation was passed and, by the next year, almost all first grade and second grade classes in the state had been reduced to 20 students. By 1999-00, almost all K-3 classes had been reduced.
Evaluation of CSR Program Conducted. Although California took a different approach than Tennessee and Wisconsin in implementing its CSR program, it did contract for a scientific evaluation of its program. Like other states' evaluations, its primary purpose was to determine whether California's CSR program was effective in improving student achievement and to identify possible program enhancements. The state contracted with a group of researchers, known as the CSR Research Consortium, to conduct the evaluation. The consortium submitted its final evaluation in September 2002. In addition to this evaluation, the Public Policy Institute of California (PPIC) released the results of an independently conducted study in June 2002.
The CSR Consortium evaluation and PPIC study uncovered three major findings.
Effect of CSR on Student Achievement Inconclusive. The CSR Consortium was unable to reach definitive conclusions about the effect of CSR on student achievement because of California's decision to implement the program across all schools at one time. Although the Consortium found that test scores did improve in California during the late 1990s, it could not attribute this improvement to CSR. This is because it did not have access to two systematically distinct groups of students—those exposed to CSR and those not exposed to CSR—and it could not control for other reform efforts that might have contributed to achievement gains. The Consortium only could conclude that CSR might have contributed to small achievement gains among third grade students. For example, it found the percentage of third grade students scoring above the national median on reading, spelling, language, and mathematics tests was 2 percent to 3 percent higher in the typical reduced-size class compared to the typical non-reduced-size class.
California's CSR Program Contributed to Decline in Teacher Quality. The implementation of CSR generated the need for many more K-3 teachers. Between 1995-96 (one year before CSR implementation) and 1998-99 (the third year of the CSR program), the total number of K-3 teachers increased by 28,886, or 46 percent. To meet the increased demand for K-3 teachers, many districts hired teachers who were not highly qualified (that is, they were not fully trained in subject matter and pedagogy or they lacked sufficient teaching experience). As a result of California's rapid and universal implementation of CSR, the percentage of K-3 teachers who were not fully credentialed increased from less than 2 percent before the program started to 14 percent in the third year of the program. The Consortium also found that the percentage of fourth and fifth grade teachers without full credentials increased significantly—from less than 2 percent in the year prior to the program to 15 percent in the third year of the program. Middle and high schools also experienced similar increases in the percentage of teachers not highly qualified.
California's CSR Program Generated Inequitable Distribution of Fully Qualified Teachers. California's implementation of CSR had a disproportionately adverse impact on disadvantaged schools because these schools had the most difficult time hiring fully credentialed teachers. Before CSR implementation, almost all California teachers held full credentials. Thus, little difference existed among schools in the percentage of K-3 teachers who were not fully credentialed. Even in the state's most disadvantaged schools, fewer than 4 percent of K-3 teachers were not fully credentialed. The state's CSR program changed this dramatically. By the third year of the CSR program, more than 21 percent of K-3 teachers were not fully credentialed in the lowest income schools whereas only 4.3 percent of K-3 teachers were not fully credentialed in the highest income schools.
Bottom Line—Any Positive Impact of Smaller Classes Likely to Be Offset by Deterioration in Teacher Qualifications. The results of the PPIC study confirmed and enhanced the Consortium's findings. Similar to the Consortium, PPIC also found that reduced-size classes were associated with small achievement gains. The PPIC, however, was able to measure the offsetting effects of having a new, less experienced teacher in a reduced-size class. It found that having a first-year or second-year teacher in the classroom essentially eviscerates any achievement gain likely to result from a smaller class size. Moreover, it found that certain groups of students were much more likely to be taught by inexperienced teachers, which offset the achievement gains possible from smaller classes. In short, many students, particularly those in the most disadvantaged schools, were likely to experience no net benefit from California's costly CSR program.
The Consortium and PPIC's research findings offer three policy lessons. The Legislature might consider these lessons when devising improvements for the CSR program and developing future education reforms.
We recommend the Legislature include the K-3 Class Size Reduction (CSR) program in a new Academic and Instructional Improvement Block Grant. Placing the K-3 CSR program in a block grant would allow districts greater flexibility to select the academic and instructional improvement strategies that best address local needs. It also would enhance accountability by: (1) requiring districts to report relevant fiscal data to the State Department of Education, (2) creating a measure of districts' instructional improvement, and (3) tracking both academic and instructional performance.
The Governor's budget proposes to consolidate 58 existing education programs into a single categorical block grant, but it retains the K-3 CSR program as a separately funded categorical program. Earlier in this chapter, we recommend the Legislature create five block grants. One of these block grants would consolidate 22 existing academic and instructional programs, including the K-3 CSR program.
Academic and Instructional Improvement Block Grant Would Enhance Flexibility. In our categorical piece, we detail the potential benefits of allowing districts to use improvement strategies that best address local needs rather than having to use state-dictated strategies that might not be well suited for particular local contexts. We argue that this flexibility is particularly important given the lack of data on the relative effectiveness of particular improvement strategies. Although research suggests that K-3 CSR, under certain conditions, can improve student achievement, research has not yet been able to compare the relative cost-effectiveness of K-3 CSR compared to other reform efforts, such as high quality subject-based professional development programs, comprehensive support programs for inexperienced teachers, or specialized one-on-one reading programs for struggling K-3 students.
Block Grant Would Not Sacrifice Accountability. In our categorical piece, we argue that block grants can be structured so that local discretion is increased without sacrificing accountability. We recommend that school districts provide expenditure data to the State Department of Education (SDE) and identify the types of strategies they are using to improve performance. For the Academic and Instructional Improvement block grant, we suggest several outcome measures the Legislature might use to assess teacher performance. These performance indicators might include measures of school districts' ability to hire qualified teachers, retain competent teachers, and involve veteran teachers in high quality professional development programs.
If the Legislature retains a separate K-3 Class Size Reduction (CSR) program, we recommend it: (1) establish a pilot program to compare the cost-effectiveness of various CSR and teacher initiatives; (2) allow school districts voluntarily to participate in the program, in which they could use existing K-3 CSR funds to implement one of several allowable educational reforms; and (3) designate $500,000 in federal Title VI funds for the State Department of Education to conduct the evaluation of the various reform options.
If the Legislature chooses to retain K-3 CSR as a separately funded categorical program, we suggest it create a pilot program that would allow districts to use existing K-3 CSR monies to implement a different reform strategy if they agreed to be evaluated. Specifically, we recommend the Legislature create a voluntary program in which school districts would be able to select one of several allowable education reforms. As a condition of receiving this programmatic flexibility, districts would agree to provide relevant outcome data to SDE.
School Districts Voluntarily Would Select One of Several Allowable Improvement Strategies. School districts choosing to participate in this pilot program would be required to select one of several allowable cost-neutral CSR and teacher improvement strategies. In addition, SDE would establish a control group consisting of districts that continued to use their CSR monies to reduce every K-3 class to 20 students. The allowable reform strategies offered to participating districts should include those that have significant potential for improving academic achievement. We suggest that participating schools be allowed to select from among the following five reform strategies. Three of the strategies would involve some variant of class size reduction, in which participating school districts could reduce certain classes even further (for example, to 15 students per class), while ensuring remaining classes were enlarged by no greater magnitude (for example, no more than 25 students per class). The other two strategies would involve teacher training and staff development.
Federal Funds Available for These Types of Evaluations. Because school districts would be redirecting existing K-3 CSR monies, this program would require only a modest amount to pay for data collection and research. If established, we recommend the Legislature include a budget bill provision designating $500,000 in federal Title VI funds to pay for the evaluation activities. Although the administration designates the bulk of Title VI funding for specific data collection and assessment projects, it does not have a well-developed plan for using $1.5 million of it. We recommend setting aside $500,000 of Title VI funding annually for several years to allow SDE to develop a meaningful longitudinal database and arrive at conclusions about the relative effectiveness of CSR and teacher initiatives.
In conclusion, if the Legislature chooses to retain K-3 CSR as a separate categorical program, we recommend it establish a program to compare the cost-effectiveness of various CSR and staff development initiatives. We further recommend the Legislature designate $500,000 in federal Title VI funds for SDE to manage the evaluation.