The STAR project demonstrated that there is a significant causal relationship between reducing class size and improving student achievement. The effects were positive and durable, particularly for students who started in the smaller classes in kindergarten and stayed in them for three to four years. Such students continued to perform at a higher level on average than those students in the large class sizes. To articulate these differences, Dr. Finn used a measure called "months of schooling," which, as its name suggests, equates a child's achievement level to the number of months of schooling a child with that achievement level normally would have. In each analysis, the children in small classes outperformed the students in the large classes by this measure. Dr. Finn noted that the presence of teachers' aides was found not to affect student performance in the larger classes. However, he noted that the teachers' aides in the STAR study were persons—parents mostly—who signed on only for the duration of the study. He opined that teachers' aides with better training and a higher degree of professionalism could have a positive effect on student performance. [187 [187 Misc.2d Misc.2d 53] 53] The benefits enjoyed by the children in the smaller class sizes in K through 3 continued even after the STAR program ended in third grade. Of the 30 statistical analyses of STAR students taking tests in grades 4 through 8, 29 yielded statistically significant differences between the children who had been in small classes compared to the children in the large classes. Significantly, the advantages to minority students were greater than those exhibited by all students. The STAR project is considered especially probative by experts in the field because it was conceived as an experiment, rather than as a simple observation. Researchers are able to draw conclusions regarding the causal relationship between class size and student performance because of the STAR project's random assignment of children and teachers, and because of its isolation of two variables: class size—and, to a lesser extent, the presence of teachers' aides. The STAR project has been invoked by New York State and the Federal Government as justification for their efforts to reduce class size in the early grades. Defendants' expert, Dr. Hanushek, presented some evidence that the value of the STAR project was undercut by student transfers in and out of the schools participating in the project and by some schools leaving the project altogether. Dr. Hanushek conceded on cross-examination that these transfers did not demonstrate that the STAR results were unreliable. The court finds that Dr. Hanushek did not demonstrate that the transfers undermine Dr. Finn's conclusions regarding the STAR data. Dr. Hanushek did provide some probative evidence from California that that State's attempt to reduce class size in a very short time frame actually hurt student performance. According to Dr. Hanushek California may have hired too many unqualified teachers in order to speed the reduction of class sizes. This evidence does not undermine the STAR results. At most, it suggests that class size reduction must be carefully planned and coordinated with the hiring of qualified teachers. New York City's class sizes have been consistently higher than the State average and higher than the 18 to 20 student maximum that experts regard as the ceiling necessary to obtain the benefits seen in the STAR project. Dr. Finn presented the most comprehensive description of class sizes in New York City. As of October 31, 1998, average general education class sizes in New York City were: 23.8 for kindergarten, 25.12 for first grade, 24.97 for second grade, 25.46 for third grade, 27.34 for fourth grade, 28.05 for fifth grade, 27.62 for sixth grade, 28.13 for seventh grade, and 28.72 for eighth grade. [187 [187 Misc.2d Misc.2d 54] 54] These averages of course mean that a significant portion of New York City's school children are in much larger classes than the average. Thus, 27.1% of all students in classes K through 3 (or 89,139 children) were in classes of 28 or more; 66.6% of all students in fourth and fifth grades (or 102,347 children) were in classes of 28 or more; and 72.3% of the students in sixth through eighth grades (or 148,869 children) were in classes of 28 or more. New York City class sizes are consistently higher than the State averages at every level, including high school, for every year that data are available for the last 20 years. This data is contained in the April 1999 655 Report, the annual report submitted by the Regents to the Governor and the Legislature (see Education Law § 215-a). For example, from 1980-1981 to 1997-1998 the average kindergarten sizes in the City fluctuated between 24.2 and 25.4, while the State-wide average (including New York City) fluctuated between 21.3 and 22.4, an average difference of at least three students. In the same time span, class size for grades 1 through 6 ranged between 27 and 28.3; State-wide the range (including New York City) was between 23.5 and 24.2, a difference of nearly four students. Similar averages obtain for middle and high school classes. Obviously, the State-wide averages quoted in the 655 Report are inflated by the inclusion of New York City's students. The differences are more stark when the City is compared to the rest of the State. In sum, large class sizes in New York City's public schools have a negative effect on student performance. Conversely, smaller class size—particularly classes with a maximum of 18 to 20 students in grades K through 3—can have a positive effect on student performance, particularly that of at-risk students. Defendants argue that class size results from the way BOE deploys its teachers, and they argue that this deployment is beset by inefficiencies. Defendants point out that New York City's public school teacher student ratio in the 1999-2000 school year was one teacher for every 14.1 students. 28 This figure is substantially lower than the average for large school districts around the Nation. The City's teacher student ratio has declined in recent years, from 1 to 16.5 in 1997-1998, to 1 to 15.3 in 1998-1999, to 1 to 14.1 in 1999-2000. Defendants [187 [187 Misc.2d Misc.2d 55] 55] also argue that BOE employs thousands of school-based certified professionals, such as principals, assistant principals, attendance teachers, school psychologists, guidance counselors and social workers who may assist teachers in the instructional program. Defendants assert that these teachers and potential teachers could be put to more efficient use. Defendants claim that currently only approximately 3 hours, 45 minutes of the 6-hour, 20-minute teacher workday is spent on classroom instruction. It is true that the teachers' current contract provides that high school teachers must teach 25 of the 35 periods per week that they are expected to be present in school. For the remaining 10 periods teachers are assigned noninstructional activities. According to defendants, City teachers' workday is shorter than the average workday in a sample of 16 other school districts in New York State, and in several large cities scattered around the country. Finally, defendants argue that BOE has a large number of teachers' aides on payroll who add little or nothing to student productivity, and that money spent to pay these workers could be better used to pay for the salaries of more teachers. The court finds that teacher student ratios are not the relevant benchmark of adequacy. No matter how many teachers are on staff, class size cannot be