For the last 15 years, I have been studying the geological accumulation of education reforms in U.S. schools—the sedimentation of the last two or three geological eras. In a book I wrote with Penelope Peterson and Sarah McCarthey on the structure and restructuring of schools, the main finding we report is that changing structure does not change practice. In fact, the schools that seem to do the best are those that have a clear idea of what kind of instructional practice they want to produce, and then design a structure to go with it.

My favorite story, which is now increasingly confirmed by the aggregate analysis of block scheduling—the current structural reform du jour of secondary education—involves a high school social studies teacher I interviewed recently. I asked him, “So what do you think of block scheduling?” He said, “It’s the best thing that’s ever happened in my teaching career.” I asked, “Why?” And he said, “Now we can show the whole movie.”

That captures my take on structural reform. We put an enormous amount of energy into changing structures and usually leave instructional practice untouched. Certainly that message has been confirmed by Fred Newmann’s work at the Center on Organization and Restructuring of Schools, and other research. We’re just now getting the first generation of aggregate studies on block scheduling, which, shockingly, show no relationship between its adoption and any outcome that you can measure on student performance. Of course, this is exactly what one could have predicted, given the previous research on structural reforms.

The reasons for this are pretty straightforward. Notice that I didn’t say structural changes don’t matter. They often matter a lot, especially when you’re talking about U.S. high schools, which are probably either a close third or tied for second as the most pathological social institutions in our society after public health hospitals and prisons. There are problems in high schools that cannot be solved without making dramatic changes in structure, but in the vast number of cases there is no instrumental relationship between any change in structure, any change in practice, and any change in student performance. That is the big problem with the usual approaches to school improvement. We are viscerally and instinctively inclined to move the boxes around on the organizational chart, to fiddle with the schedule. We are attracted and drawn to these things largely because they’re visible and, believe it or not, easier to do than to make the hard changes, which are in instructional practice.

The pathology of American schools is that they know how to change. They know how to change promiscuously and at the drop of a hat. What schools do not know how to do is to improve, to engage in sustained and continuous progress toward a performance goal over time. So the task is to develop practice around the notion of improvement.

Weak Theories

We can talk about what’s wrong with the state accountability systems that are springing up everywhere. But the fact is that school improvement strategies are being driven by performance-based accountability systems. These systems involve setting standards about what constitutes good practice, a solid curriculum, and acceptable student performance. They entail various kinds of stakes for students and for schools—and virtually none for teachers and administrators. (Interestingly, the stakes tend to fall most heavily on the kids, who have the least representation in state legislatures.)

The problem, however, is that the organizations we work in aren’t built to respond to this kind of performance pressure. We may know what to do theoretically, but I have serious doubts that we know what to do at the level of practice. For example, I’ve been in enough high school math classes over the last five years to know that there is no developmental theory of how students learn algebra. The kids who don’t make it and don’t respond to the kind of instruction they’re receiving are simply not included in the instructional model. And teachers in the classrooms I’ve observed take no responsibility for the lowest-performing students. That’s because the prevailing a theory of learning suggests that teaching mathematics is not a developmental problem but a problem of aptitude. Some people get it, some don’t. (In this regard, literacy is perhaps an exception.)

People do not believe that these problems can be solved by inquiry, by evidence, and by science. They do not believe that it is necessary to have a developmental theory of how students learn the content and how the pedagogy relates to the development of knowledge and content. Nor are most teachers interested in addressing the intellectual challenge that some students learn the content and some don’t. As a result, we are asking schools to make improvements in the presence of an extremely weak technical core.

Also, schools are not organized to support problem-solving based on cooperation or collaboration. The ethic of atomized teaching—teachers practicing as individuals with individual styles—is very strong in schools. We subscribe to an extremely peculiar view of professionalism: that professionalism equals autonomy in practice. So when I come to your classroom and say, “Why are you teaching in this way?” it is viewed as a violation of your autonomy and professionalism.

Consider what would happen if you were on an airplane and the pilot came on the intercom as you were starting your descent and said, “I’ve always wanted to try this without the flaps.” Or if your surgeon said to you in your pre-surgical conference, “You know, I’d really like to do this the way I originally learned how to do it in 1978.” Would you be a willing participant in this?

People get sued for doing that in the “real” professions, where the absence of a strong technical core of knowledge and discourse about what effective practice is carries a very high price. Instructionally, we know what works in many content areas. But the distribution of knowledge is uneven, and we resist the idea of calibrating our practice to external benchmarks.

School systems are also characterized by weak internal accountability. When I use that term, I mean the intersection between the individual’s sense of responsibility, the organization’s expectations about what constitutes quality instruction and good student performance, and the systemic means or processes by which we actually account for what we do. How frequently do we observe teachers? How do we analyze performance data? How do we think about teachers’ performance? The schools in which these things are aligned have very powerful approaches to the improvement of instruction. When they are not aligned—and in most cases they are not—schools have extreme difficulty responding to external pressure for improved performance.

Meanwhile, the usual remediation strategies we employ when kids fail to meet the statewide testing requirements are to give them the same unbelievably bad instruction they got in the first place, only in much larger quantities with much greater intensity. This is what we call the louder and slower approach.

Better Benchmarks

This brings me back to the notion of improvement versus the notion of change. Improvement is a discipline. It requires picking a target that has something to do with demonstrated student learning, one that’s ambitious enough to put schools in “improvement mode.” If you’re a school leader whose students are scoring consistently in the 95th percentile, you need another performance measure because that one is doing you no good— except to help your marketing. For improvement purposes, you need a new ceiling, a goal to push for that’s quite a distance from where you are. You also need some kind of external benchmarks.

If the only benchmarks you have come from your own connoisseurship—your particular opinions and ideas about what good practice is—then you’re in trouble. Real improvement comes when you visit a classroom where somebody is doing the same thing you are—only much better. That’s when the real conversation, the tough conversation about improvement takes place. Whether you’re a novice or an expert, the important thing is to focus on the next stage of improvement and to determine where that increment of knowledge and skill is going to come from.

The norms and values that go with ambitious conceptions of learning and improvement grow out of practice, not vice versa. School improvement doesn’t happen by getting everyone to come to the auditorium and testify to their belief that all children can learn—not if it means sending everyone back to the classroom to do what they’ve always done. Only a change in practice produces a genuine change in norms and values. Or, to put it more crudely, grab people by their practice and their hearts and minds will follow.

Finally, instructional leaders need to know and model the knowledge and skills needed to do this work. This includes knowledge about performance, knowledge about development in content areas, knowledge about the improvement of instruction. Leaders need to create structures for how they learn in schools. If you can’t model the norms and values you expect others to adopt, it’s unlikely that any real improvement will take place.

Richard F. Elmore is Anrig Professor of Educational Leadership at the Harvard Graduate School of Education and the Faculty Editor of the Harvard Education Letter. This essay was drawn from an address given by Professor Elmore at a recent institute on leadership and policy hosted by The Principals’ Center at Harvard University. It has been edited for this issue.