Is anyone recommending that General Motors repair itself by retraining its workers, or by adopting performance-based compensation for them? Are people advocating that we solve the nation’s health-care problems by altering the curriculum in schools of medicine?
The answer to both questions is no. Why, then, does it make sense for us to attack the problems of urban schools with equivalent measures: improving teacher training, adopting merit pay, or changing the curriculum and books?
General Motors got into trouble by being overcentralized in its management, and the debate over health-care reform is largely about whether centralization will make it better or worse. When big organizations get into real trouble, the issue is often about centralization.
Urban school districts are dysfunctionally centralized, and putting improved subsystems of selection, training, and compensation into them cannot overcome the larger problem. To see how inappropriate are the currently voguish palliatives, we need only to observe what happens when principals are given the freedom to spend their budgets as they see fit, staff their schools as they think best, and arrange curricula and schedules according to the desires of their teachers.
True decentralization is so difficult that there’s no reason to attempt it, except that it’s the only thing that works when a district is in a bad way.
My research team and I have now completed two studies of exactly that: what happens when principals are empowered to make the important instructional decisions on budget, staffing, curriculum, and schedule.
In 2003, my first book on this subject, Making Schools Work, reported on a study in which the team visited 223 schools in three traditional, centralized districts and compared them with three decentralized districts. We found that students in the decentralized districts outperformed their top-down counterparts.
In a new book, The Secret of TSL, and its companion statistical paper, we report on a study of 442 schools in eight U.S. districts that are now decentralizing in whole or in part, from New York City to St. Paul, Minn., to Oakland, Calif. The purpose of this new study was to find out what principals do when they are empowered, in comparison to what happens when all of the instructionally important decisions are mandated by the central office.
What we found is that empowered principals radically alter the staffing of their schools, greatly reducing the number of nonclassroom employees and hiring more classroom teachers instead. They also often use a combination of blocking and coring, combining English and history, for example, into a single two-period-long humanities course. In this arrangement, the same teacher might have 25 students for two periods of humanities (she teaches both English and history), followed by another humanities course for another 25 students, and then fill out her five-course requirement with an elective course for 25 more, thus having taught five periods but having a “total student load,” or TSL, of 75 students, rather than 125.
In our statistical analysis, we found that traditional forms of school improvement do not in fact have any effect on student performance, while reducing total student load is associated with a major improvement in performance. Thus, adopting new curricula, having more long-service teachers, reducing class size, and adding planning time or professional development have no effect, but giving principals full autonomy leads to lower TSL for teachers, and to significantly improved student performance. (“Management Guru Says ‘Student Load’ Key to Achievement,” Sept. 30, 2009.)
It should come as no surprise that a lower TSL means better student achievement. Theodore R. Sizer, in his Horace books of the 1980s and ’90s, argued that in order to personalize instruction, no teacher should have a total student load larger than 80. Arthur G. Powell, in his 1996 book on private schools, Lessons From Privilege, said that a lower TSL, unlike lower class size, creates the free time for students to seek out their teachers for the kind of one-on-one coaching that, at the right time, can make all the difference.
New York City, which fell into both of our studies, in 2001 and again in 2006, provides many examples of the effects of reducing TSL. In the first study, we interviewed 66 principals there and went over their budgets with them in detail. On average, we found that they controlled 6.1 percent of their budgets in 2001; most of the money was controlled by central-office dictates. In 2006, after Mayor Michael R. Bloomberg and Schools Chancellor Joel I. Klein had begun their decentralization, which has since spread to all 1,631 city schools, we found that the principals who had been empowered in 2004 and 2005 now controlled, on average, 85 percent of their budgets.
Some of these schools achieved administrative ratios (nonteaching employees to all employees) as low as 4 percent, compared with 44 percent in some traditional schools. As a result, average TSL in the empowered high schools was 87.7 students per teacher, and in middle schools 57.5—compared with the union-contract maximum of 170. After two years of principal empowerment, graduation rates in these schools had risen from 65.8 percent to 74.5 percent, and passing rates on the state Regents math exam had risen from 71 percent to 89 percent.
While we do not yet know whether the results for all 1,631 New York City schools will be of this same magnitude, other cities have shown similarly positive results. St. Paul, for example, achieved even greater decentralization than New York, with principals there controlling 87.2 percent of their budgets.
This is real decentralization. Most studies of “decentralization” have neglected to measure whether principals have decisionmaking authority or not. We have not, for example, found any study of “site-based management” that did so, nor any measure of whether the policy of decentralization adopted in Chicago in 1988 actually produced any decentralization of decisions to principals.
It would be premature, to say the least, to conclude that many districts have tried decentralization and found it does not work. It seems more likely that many districts have announced decentralization policies but failed to actually decentralize. The evidence in our research shows that true decentralization works well.
We also have found in our studies that implementing decentralization is difficult. It’s clear why so few of the nation’s 14,000 school districts have tried to do so. It requires the creation of an extensive accountability system over principals, and new methods for attracting, training, and supporting principals of empowered schools. Central-office staffs and information systems must be completely re-engineered as well, to support a decentralized district. In fact, true decentralization is so difficult that there’s no reason to attempt it, except that it’s the only thing that works when a district is in a bad way.
Training teachers, improving compensation systems, and developing curricula will continue to be important. But first things first. Even the most dedicated and best-trained teachers will fail if they are inundated with students and subjected to extensive central-office restrictions. There isn’t more money available, but there is the real possibility, as our research reveals, of reallocating budgets, hiring more classroom teachers, and personalizing instruction through true decentralization.
