Professional networks and school improvement: the medical rounds model, applied to K-12 education, provides a community of practice among superintendents committed to better instruction.

You are a middle school principal. Today will be an unusual day for you and everyone else in your school. Your school will be visited by 12 superintendents who will spend the better part of the day visiting and observing classrooms, discussing issues of instructional practice in your school with you and your superintendent, and helping to frame a strategy to make progress on a specific issue of student performance in your school and in the district.

The particular problem the group is dealing with is lagging performance in reading and writing for students in the middle grades. The district has made a substantial investment in materials and professional development, but this investment is not evident in the performance of middle grades students on external tests.

The superintendents move in groups of three and four through classrooms, taking notes on what they see with an observational protocol. Today they will also observe some subject matter and grade-level teacher team meetings, where teachers are discussing the results of a recent schoolwide literacy assessment.

Prior to the visit, you and your superintendent prepared a detailed and specific problem statement, laying out the data that described the problem of lagging literacy performance and the specific conditions in your school. This problem statement will serve as the basis for a debriefing discussion that occurs in the school library, over lunch, after the classroom visits.

The debriefing follows a structured protocol. Groups of superintendents are asked to describe what they saw in the classrooms they visited. The focus is on description, not evaluation or solutions. After everyone has spoken and all the classrooms have been discussed, the discussion turns to the question, "What is the next level of work?" Superintendents suggest ideas for the improvement of organization and support for literacy practice, drawing on the evidence from classroom observations. At the end of the debriefing, you and your superintendent are asked to respond to what the superintendents have said.

At a later time, the superintendents meet for most of a day in a setting away from their offices and spend most of their time together exploring strategies for addressing the literacy problem. At the end of the discussion, your superintendent presents a tentative plan of action. Two of your superintendent's colleagues agree to a follow-up visit to your school in the future to gauge how the plan of action is working. Your superintendent returns and the two of you discuss the plan of action, both in your school and districtwide, and the next level of work on literacy begins.

A Common Body

This is how the Connecticut Superintendents Network operates. The network is designed to be a community of practice among superintendents who are committed to sustained instructional improvement in their districts. It is designed around a model of practice drawn from the medical rounds model, in which practitioners work in a disciplined way to develop a common body of evidence they use for diagnostic purposes and then work through a set of solutions.

The network also operates around a model of professional accountability. Superintendents are accountable to each other for their work--for the quality of their observations, analysis and advice. Superintendents are accountable to the discipline of the work and to improving their own practice around the work. And superintendents are accountable to each other for the consequences of the work.

The Connecticut Superintendents Network has been in existence since the fall of 2001. The original group of 12 superintendents was joined by a second group of 12 in the fall of 2005. The participants represent the full range of districts in Connecticut--urban, rural and suburban. All the participants share something important in common--they are committed to pursuing a strategy of large-scale improvement in instructional practice and student performance in their districts.

The network is supported by the Connecticut Center for School Change, which also has a broader portfolio of projects for school improvement in Connecticut.

In the fall of 2006, working with Superintendent Thomas Fowler-Finn, I also helped to form a school leadership network in Cambridge, Mass., following the basic design of the Connecticut Superintendents Network, except working within a single district.

The Cambridge network involves all the principals in the district, the high school unit administrators, the superintendent and deputy superintendent, curriculum and instruction support staff from the central office and the president of the local teachers' union. The Cambridge Network is a turn-key project in that it is designed to operate without external facilitation after two years.

Relevant Practices

The network model is designed to develop a model of practice for school administrators that is solidly anchored in instructional practice. The school leadership literature stresses the importance of instructional leadership, but academics and practitioners never have satisfactorily developed a model of practice that says explicitly what the leadership of instruction is.

Accountability pressure has drawn administrators more deeply into issues of instruction, and many practitioners say they are ill-equipped to organize and manage around the improvement of instruction. The network model is designed to provide a setting where school leaders can work together in a structured way on issues of instructional practice that are directly relevant to their work, developing their understanding and skill around practices of improvement.

The networks make extraordinary demands on practitioners' time, and the work of the networks is not something to be undertaken casually. The Connecticut superintendents commit at least one full day per month to the network. Strong norms of attendance and participation require a high degree of preparation. Every superintendent commits to hosting at least one site visit over the course of a two-year cycle and to a consultation with a colleague in a follow-up visit.

The Cambridge network requires an even higher level of commitment. The cycle of school visits is more frequent, a segment of monthly principals' meetings is devoted to network work, and frequent sessions are scheduled around professional development work. The success of professional networks depends, to a large degree, on making a cultural transition from viewing instructional improvement as an added responsibility to viewing it as the work around which everything else revolves.

Power of Protocols

Over the course of our experience together, network members have learned some important things about the work of instructional improvement and about how to use networks productively to increase knowledge and skill in instructional leadership. Some but by no means all of this learning can be captured in a few central themes:

One of the biggest contrasts I have seen between medicine as a field of practice and education is that medical professionals use protocols routinely in their practice, and educators, for the most part, do not. We found in the networks that protocols are a central and powerful way to structure the work of instructional improvement. We use simple observational protocols to organize data collection in classrooms. We use protocols to present the results of observations and to structure discussions. And we use protocols to structure the problems we work on and to develop solutions to those problems.

Protocols provide a predictable structure to the work. They define roles and responsibilities in discussions, they provide group norms, and they keep the work focused in a productive way. But I have discovered, working with networks, the role of protocols is actually much more subtle and important than these structural dimensions.

Protocols separate the person from the practice. One of the things I have noticed about medical practitioners is they discuss problems of practice in a relatively impersonal way, as if their practice were an objective thing, apart from who they are.

Educators, on the other hand, tend to confound and confuse the practice with the person. Indeed, for most educators, their practice is who they are. It is hard to change your practice when your practice is central to your personal identity. Every change becomes a challenge to who you are, in some basic sense. In general, professionals can't afford this confusion of the practice with the person because they live in a world in which practices are changing rapidly in response to new knowledge and new problems.

So one important function of protocols is to depersonalize discussions of practice. For example, in the networks we spend considerable time and energy working with educators on developing a descriptive voice. We ask network participants to describe what they see in classrooms and to avoid evaluative or normative language. We do this to develop a body of evidence that can be used to diagnose the sources of the problem we are working on and to keep the network focused on solving problems at the school and system levels rather than evaluating individuals.

Educators find it extremely difficult to operate in the descriptive mode. It takes extended practice and repeated reinforcement to stay in the descriptive voice. Hence we use observational protocols that are deliberately descriptive and simple. And we conduct debriefings according to a protocol that requires participants to state the evidence before they draw evaluative or diagnostic conclusions. Participants get good at this practice over time, but it requires constant reinforcement because the culture of schools is relentless in its normative focus on practice.

Once we have developed a body of evidence, participants can move into the diagnostic and normative mode. But all diagnostic and normative statements have to be accompanied by reference to a body of evidence. So, for example, a participant might say, at the diagnostic stage, as is often the case, "the instruction was incredibly uneven across the classrooms we observed, which suggests to me that the professional development isn't having much influence on teaching practice." This statement has to be anchored in a body of evidence that makes reference to what work students were being asked to do, what teachers were actually doing, and what the major dimensions of variability were.

Another participant might say, "I didn't see the same level of variability that you saw, and I also saw quite a lot of evidence that teachers had internalized the main elements of professional development." The point of having a common body of evidence and a protocol for collecting and discussing it is that these disagreements can be argued out systematically with reference to actual observations. Notice also the discussion is not about the evaluation of teachers. It is about the conditions of instruction that produce the results that are problematical.

Participants often have to subject their own practice to scrutiny and analysis, and protocols help to de-personalize those discussions. We often find ourselves in situations where it is clear the reason why we are not seeing instructional improvement in classrooms is that the organizational supports and expectations are not clear and focused. In these situations we find ourselves talking about the practice of participants, not the practice of the teachers we observe. Separating the person from the practice is essential to being able to say, in a collegial way, "I don't think the evidence supports your approach to running teacher conferences and leadership staff meetings; people don't seem to be getting the message."

Norms of Confidentiality

Needless to say, if people are going to feel secure in an environment where their colleagues are speaking to them candidly about their practice, there have to be ground rules about what stays in the network and what can be discussed outside.

In general, the networks have agreement on a norm that nothing that can be associated with a particular person or school can be discussed outside the network. General issues in the district, examples drawn from school visits that are not identifiable with a particular school, and patterns of practice that have been observed across settings are all fair game for discussion.

The ideal environment is one in which colleagues can speak candidly with each other in a way that focuses attention on the improvement of practice. In the culture of American schools, there is a strong current of "niceness" that makes productive discussions of practice nearly impossible. People talk about their own practice and the practice of others in convoluted and obtuse ways, always accompanied by something superficially positive.

Candor is essential to the effective operation of professional networks and norms of candor have to be taught and practiced in network settings, so that people feel comfortable with the practice. Confidentiality provides the support necessary for this work.

Theories of Action

Focusing school leaders' work on instructional practice can create a serious disorientation. Superintendents move from a world in which they are managing political conflicts among members of their boards, dealing with unhappy parents and trying to balance the budget into a world in which they are talking with their colleagues about specific classroom practices and how these practices relate to problems of student performance in their districts.

Likewise, principals have a similar disorientation. To connect the specific work of instructional improvement to the broader picture of organizing and managing complex school systems, we ask school leaders to develop a personal theory of action.

A theory of action is a brief statement of how the practice of the individual--at the system or school level--leads to increases in learning and performance for students. Part of the discipline of this work is to try to put the theory into the form of a series of "if, then" statements. For example: "If I meet regularly with principals and focus on common instructional issues, then we will develop a common understanding of what we are trying to achieve." "If I model good pedagogy in the way I organize principals' meetings, then I will communicate the core values of the organization and its focus on good instructional practice."

The idea behind theories of action is to provide leaders with the opportunity to rise above the specific instructional issues we focus on in the network and try to state, in a useable form, what they think they have learned about their work from being in a collegial network. We make time for participants to talk to each other about their theories of action and to advise each other about particularly important dimensions of their practice that may or may not be present in their colleagues' work.

Moving to the Field

It is one thing to create a healthy and productive professional community among school leaders. It is quite another to have the work of the community move out into the systems and schools that the participants manage.

We have learned this stage of moving practice into the field takes time and patience. We more or less thought, in an inchoate way, the transfer would be automatic. If you have a powerful learning experience, you will want to carry it back to colleagues in your district or school. What we have found, not surprisingly given the research on innovation, is that this movement into the field is initially spotty, uneven and slower than we wanted.

Messages we thought were clear in the network meetings about the usefulness of network practices for the participants back in their districts and schools, it turns out, were not clear at all to participants. We discovered it is important to have strong group norms about what we expect to transfer and time to practice in network meetings some of the knowledge and skills we expect to transfer.

The slogans of instructional leadership have had little effect of the core practices of educational leaders. What we have tried to do with professional networks is to focus on the development of a body of professional practice--tangible behaviors, protocols, structures and processes--that model what instructional leadership might look like.

We also have tried to create a setting in which people can learn a new kind of professional practice, be connected to peers who are in the same learning process, and, over time, take control of their learning and make it relevant to the specific problems they face. The work speaks for itself.

Richard Elmore is Gregory Anrig professor of educational leadership at Harvard Graduate School of Education, Gutman Library 448, Cambridge, MA 02138. E-mail: [e-mail]