Building New Knowledge

"The problem with most incentive structures is not getting people to do the right thing. It's getting people to figure out what the right thing is to do."

–Thomas Schelling
Distinguished Professor, University of Maryland School of Public Affairs

By Richard F. Elmore

In the next two or three years, a very large number of schools, most of them urban, with largely poor, minority student populations, will be classified as failing1 under the accountability provisions of No Child Left Behind (NCLB). This classification will trigger a series of increasingly harsh sanctions, ending with the shutdown of the school and/or its transformation into a charter, the handing off of its management to a private company, its takeover by the state, or another comparable, federally approved governance change. The law provides little assistance to these schools in their run-up to being shutdown. And, as more schools get identified (and they will), there will be even less assistance available to any given school. This policy appears to be driven primarily by an extraordinary belief in the power of incentives. The logic seems to be: If schools are threatened with closure and other sanctions, they will figure out how to improve themselves.

My theory is that these and similar policies toward failing schools are based on either faulty knowledge about school failure or no knowledge at all. I am a great believer in incentives; I've taught students about their power in numerous courses over the years. So, I'm not going to argue that they don't matter. They do. Nor would I argue that schools can't improve, to a degree, by just getting people more focused, encouraging them to make better use of their existing capacities, and making them work harder. But I also know that Thomas Schelling, one of the most astute economic theorists on the subject of incentives, was right. He told me, when I was a graduate student, "the problem with most incentive structures is not getting people to do the right thing. It's getting people to figure out what the right thing is to do." This is the problem I would like to focus on.

In this article, I want to argue that while incentives are important, they won't be nearly enough to bring about the dramatic improvements in student achievement that we all hope for. The performance targets set by current accountability policies are, for many failing schools, completely unattainable using their existing capacities. The knowledge simply doesn't exist in these schools to make the huge leaps in achievement that the law requires. Further, I want to argue that if we don't provide school staffs with what is necessary to make these leaps, that is, the knowledge and tools they need to raise student achievement, we will not only assure that schools don't improve substantially, we will increasingly sow cynicism and resistance toward the law. Asking people to do the impossible without helping them to master the skills necessary to do it is a formula for political resistance and ultimate failure. (Remember the old Russian saying, coined during the decline of the Soviet system, "We pretend to work. They pretend to pay us.")

I want to start with descriptions of two schools that are failing according to their state accountability systems and are very likely to end up on NCLB's "needs improvement" list. I think these portraits help illustrate the huge lift that is required to dramatically improve a school over a sustained period and the extent to which even motivated, intelligent staff lack the knowledge to bring this improvement about. From there, I will lay out a general framework for how I think schools do in fact build their capacity to improve. Lastly, I will suggest how policy could enable many more schools to build this capacity and, with it, substantially raise student achievement.

Portraits of Two "Failing" Schools

Thornton Elementary School

I am observing a second-grade classroom at Thornton Elementary.2 The teacher is working with a third of the students in one corner of the classroom doing guided reading, a form of literacy instruction in which teacher and students jointly read aloud and discuss a book with an explicit focus on the author's meaning, as well as the readers' responses to the text. Guided reading is new to the teacher, a veteran of 20 years. She is concentrating very hard. The students are also working very hard and seem to be successfully reading and responding to the book. Each student in the class will rotate through guided reading in the course of the literacy block—the 90-minute period every morning devoted to reading and writing at Thornton—in one of three groups. While the teacher is focusing on the eight students in the guided reading group, the remaining two-thirds of students in the classroom—about 16—are doing a variety of things. Two reading specialists are working individually with two students, obviously struggling readers, on specific problems of phonics and word identification. A classroom aide is supervising a group of students who seem to be filling out worksheets. Some students are reading on their own and writing in journals. There are books in considerable quantity available to students. Student writing is prominently displayed on the walls. In general, the classroom appears to be orderly, quiet, and efficiently run. Behavior problems are few. Students seem compliant and relatively happy. Above all, the adults seem to be very focused, working hard, and highly motivated.

The principal and superintendent have worked out a professional development strategy for the school that focuses time during the school day and during designated professional development days on priority instructional areas. The teachers uniformly say that this is the best teaching they have done. The teaching force at Thornton is a veteran group; the least experienced teacher has been there 12 years.

Over the course of the morning I visit several classrooms at Thornton. Each looks roughly the same in structure and texture. To the casual observer, it would be difficult to see why Thornton is a failing school. Teachers are working hard. Students are highly engaged. There are extra adults to work with failing students. The classrooms and hallways are orderly and clean. Thornton certainly looks nothing like the stereotype that laypeople might carry in their heads about failing schools—chaotic, disorderly classrooms, teachers obviously out of their depth with both content and student discipline, low-level student work, etc. In fact, most observers would probably say that, overall, Thornton represents a strong and positive environment for students.

I have been invited to the school by the superintendent and the principal because, after some initial modest success on the state reading and writing test, Thornton's test scores have gone flat. The student population at Thornton is more than 80 percent poor, with equal numbers of African-American and Latino students. The school is in an economically depressed city, and the patterns of student performance at Thornton reflect similar patterns in other elementary schools in the district. The state reading and writing test is a challenging test for even the highest-performing schools in the state. For Thornton, it is daunting. The superintendent and principal report that teachers in the school are heavily demoralized by their designation as a failing school. They feel that they have given the school's new literacy program their best shot. They feel they have dramatically changed their practice. The changes they have made are clearly visible in all the classrooms in the school, they feel, but they are still not making progress against the standards of performance they are expected to meet. After implementing the new literary program (which included in-class professional development for guided reading), they moved some students out of the lowest level of achievement, and they even increased the percentage of proficient students. But further improvement has eluded them and, perhaps tellingly, they never saw any increase in the percentage of advanced students.

Clemente Middle School

Shift now to Clemente Middle School, a school of about 1,000 students, grades six through nine, in a large northeastern city. Essentially all students at Clemente meet the income requirements for free or reduced-price lunch—the prevailing measure of poverty. They are predominantly Afro-Caribbean, Spanish-speaking immigrants, with a significant number of African-American students. A large proportion of the students are from families that might be classified as the working poor—they perform the basic services of the economy with very low pay. It is mid-morning, and I am observing a seventh-grade language arts class, taught by a novice teacher—a Teach for America corps member, one of several in this school. The teacher is a new graduate of a prestigious northeastern liberal arts college. She is young, energetic, highly engaged in her work. She is African American and obviously has a strong rapport with her students—about 15 of whom are in her class today. One student sits off by herself in a corner, focusing on something on her desk—probably some form of "time-out" discipline problem. The rest of the students sit comfortably at moveable desks, focused on the teacher. The lesson has to do with topic sentences and lead paragraphs, a key element of the state's middle-grades writing test, which these students will be taking next year. The teacher is carrying on a lively discussion of a topic that students are asked to use as the basis for their writing. As students volunteer ideas and write them down in their notebooks, the teacher actively engages them in a discussion of what they will write and how they will write.

This classroom is one of a number I have observed this morning and the patterns are similar: active teachers, highly engaged students, instruction targeted at skills that are, at the same time, useful on their face and included in the state reading and writing test. I do not see a discernible difference between the novice and experienced teachers on these dimensions. The principal takes me to visit a couple of classrooms where he knows he has problems with the teachers. These classrooms are noticeably less engaging places for students, the teachers are clearly struggling with the fundamentals of teaching; they also seem aware that they are not doing great work.

Clemente has four assistant principals, each of whom has instructional and professional development responsibility for a grade level in the school. The assistant principals are clearly present in classrooms. The principal and assistant principals have a strategy for professional development in key subjects with teachers. While time is limited, teachers participate and say the work is valuable to them in the classroom.

Clemente Middle School is a vibrant and exciting place visually. It is a relatively new building, with a large atrium as a central feature. It has a privately funded arts program in which students produce stunning examples of visual arts and writers from the neighboring city visit, while teachers and students conduct author studies of their work. The building exudes energy. Student work is visible everywhere, especially in the atrium.

Again, I am in the school because performance, after a brief gain, has gone flat—well below the target level required to keep the school from being classified as failing. The superintendent and the principal want me to see, hear, and feel what the school is like, not just examine the test scores. The superintendent thinks the principal of the school is one of the best in the district and is worried about losing him to a neighboring district with much higher-performing schools. The principal says, in passing, that he has had to learn to ignore much of the feedback he gets from the state in order to focus on the things that need to be done to improve the school. Teacher turnover in the school and district is about 15 percent per year. About 40 percent of the teachers in the district have four or fewer years experience. Virtually none of the Teach for America members stay after their two-year term is over. The district invests heavily in professional development in literacy and math, but the superintendent says that once the new teachers have received the basic staff development program, they are attractive recruits for neighboring suburban districts that offer them significantly higher salaries.

Again, to the lay observer, Clemente would not be thought of as a failing school. Although it has at least its share of marginal teachers, what you see as you walk the halls and visit classrooms are powerful examples of students doing interesting and creative work, teachers working hard to engage students in learning that is clearly connected to what the state tests measure, and students largely responding in the ways teachers want them to. Take away the discouraging test scores, and you have a school that most lay observers would say is a decent place for kids to learn.

*  *  *

I want to stress that based on my visits to failing schools in several localities, Thornton and Clemente are not atypical of many of the schools that are, and that very shortly will be, classified as failing under NCLB. These schools have been the object of intensive efforts to make them work better. People in these schools—teachers, administrators, students—are aware that they are in organizations labeled as failing, and, with certain exceptions, they are not happy or complacent about it. Liberal critiques to the contrary, failing schools are usually not resource-poor environments. They are heavily staffed, they have large numbers of specialists who work directly with students, and they have considerable access to outside guidance and expertise in most settings. They also frequently have access to community resources that bring considerable assets to the schools. Failing schools do not have uniformly weak leaders. Some do. Some don't. The point is that "strong" leaders—as in the case of Thornton and Clemente—are often just as baffled about what to do about their situations as "weak" leaders, though strong, competent leaders may have more motivation and ability to find out what to do.

To be sure, I have also been in failing schools over the past several years that more closely resemble the common stereotype of such schools: schools that show little or no evidence of consistent expectations around the quality of instruction or student performance; schools in which the adults assign responsibility for low student performance to families and communities rather than to themselves; schools in which the resources available to support student learning are managed in a chaotic and scattered way, if it all; schools in which teachers and students cannot answer the most basic questions about the purpose and direction of their work.

In general, what my colleagues and I have found in our research on accountability is that genuinely failing schools fundamentally lack what we have come to call "internal accountability."3 That is, they lack agreement and coherence around expectations for students in learning and they lack the means to influence instructional practice in classrooms in ways that result in student learning. In our research, high internal accountability leads directly to observable gains in student learning. Some failing schools lack internal accountability on anything but the most basic expectations—order in the hallways, for example. But, as the Thornton and Clemente examples illustrate, some failing schools are actually engaged in developing internal accountability and have had some success in generating increased student learning, but are still at risk of failure, and under the terms of No Child Left Behind, are likely to lose their franchise before they have an opportunity to meet the performance requirements of the law.

The conventional view that drives current policies regarding failing schools is that schools fail because they lack the proper incentives to succeed. These beliefs, I think, are embedded in accountability policies that focus on external rewards and sanctions as motivators for teachers, administrators, and students. These policies also focus on changes in governance and incentives (e.g. charters, school choice) on the theory that the "right" external incentive structure will "drive" schools and school systems to recruit and hire the "right" kind of people who will, in turn, lead schools toward the "right" kind of goals.

As I noted earlier, incentives matter. But one of the main insights I take away from my recent visits to failing schools is how clearly most of these schools have gotten the message that they are failing. The problem is that the message doesn't tell them what to do, other than to "get better." This is the problem that is exemplified by Thornton and Clemente.

Thornton and Clemente: What to Do?

As we debriefed our observations of classrooms at Thornton with the superintendent, the principal, and the lead teachers, a number of patterns became clear. First, while teachers were working hard to apply their new knowledge on literacy instruction, no one was paying attention to the overall instructional quality and intensity of what was happening to students while the teachers were doing guided reading. All one needed to do was to walk around the classroom and observe what students who weren't involved in guided reading were actually doing. Not much was going on with the other students when they were not in the group, with the exception of the two students who were working one-on-one with the reading specialists. But second, these two students presented another problem. The work they were doing with the reading specialists, while it was quite skillfully designed and done, was not explicitly connected to the work that these students were expected to do when they were not in remediation. That is, the intervention was successful, but it wasn't successful as a cumulative goal that would bring these students into the mainstream of the class. And third, when we asked students to describe to us, in real time, what they were doing when they weren't in a guided reading group, what its purpose was, and how they would know whether they had been successful at doing it, most of the students were unable to answer.

So it's not surprising that Thornton had some initial gains with its literacy program and then its performance went flat. What happened was that students were exposed to a potentially powerful reading intervention—the introduction of guided reading—that substantially increased the amount of time and the intensity of instruction for them relative to what they had been doing. This new activity got the teachers and students focused on coherent work around reading in a way that could be observed and improved. Clearly, the next increment in performance will come from increasing the level of intensity, cognitive demand, and coherence for all students, whether they're in guided reading or not. This will require the teacher to pay much more attention to the orchestration of activities in the classroom and to have much more clarity and agreement with students and support staff around the purpose of the work.

Several things are happening here. First, it often takes another set of eyes to see what principals, students, teachers, and support staff don't see, because they are working on solving the current problem, not on identifying the next one. At Thornton, the teachers were working so hard at mastering guided reading, they didn't have time to focus on what else was happening in the classroom. Second, teachers and students get more powerful in their practice, often against their own expectations, when they are brought to acknowledge a barrier and then put in the way of knowledge about how to get over it. Notice, it is important to understand that teachers and students don't get better by applying knowledge and skill they already have—they are stuck because their existing knowledge isn't enough. They get better by having access to new knowledge and discovering that they can use it in ways they did not fully appreciate before. Third, it is increasing the level of intensity, cognitive demand, and coherence around instructional practice that produces gains in student performance, and that process requires that everyone, including students, teachers, and support staff, develop increasing agreement about what the work is. This is what we have called internal accountability. If you walk into a classroom and sit down next to a student, ask him what he is doing and why, and you don't get a clear answer, it is highly unlikely that any powerful learning is taking place.

*  *  *

At Clemente, as we debriefed, a different set of problems emerged. This will sound odd, but bear with me. The teachers at Clemente were working too hard. Novice teachers and veteran teachers of students in the middle and upper grades often equate "good" teaching with teaching that keeps students amused, interested, and seemingly engaged—which usually means eyes forward, paying attention, not causing any discipline problems, and responding in a timely way to the teacher's questions.4 So what had happened at Clemente was that the "good" teachers in the building—including some novices and some experienced veterans—had adopted a style of practice in which the teachers were doing virtually all the work in classrooms and the students were doing very little. The teachers felt they were giving it their best shot, and to a layperson's eye they were doing even more than that. The students were engaged and amused, and they certainly weren't complaining. But when you looked at the classroom as a setting for student work, it was clear that not much was happening. A straight transcript of classroom discourse, for example, would reveal that, in order to keep students' attention focused on the front of the room, teachers were asking predominantly factual questions—questions that could be answered literally by the student pulling the information straight out of the text on the desk in front of them. When teachers did ask questions that required higher levels of cognitive demand—interpretation, argument, analysis—the overall pace of previous questions meant that waiting even a short period of time for a student response seemed like ages, so the teacher quickly moved on to the next question before the students could fully engage in the previous one. The actual written work that students were being asked to produce—remember, this is a class aimed at preparing students to pass the state writing exam, which includes open-ended writing prompts—was likewise short and truncated, apparently because the teachers had made the judgment that the students needed, again, a faster pace with more concrete tasks, in order to stay engaged.

What was happening at Clemente is what often happens in the early stages of instructional improvement—teachers are developing rudimentary norms of practice designed to signal their collective commitment to students' success. They are, in effect, developing internal accountability around student learning and performance, albeit at a very rudimentary level. In the absence of careful and thoughtful analysis of the kind of practice that would lead teachers and students to be successful on a demanding writing exam, teachers were doing what they thought they should do—working hard, being enthusiastic, demonstrating that they can hold the attention of the students—without much thought for the actual work that students were doing. An outside observer would see what most people would regard as "good teaching" going on in a significant number of classrooms and wonder why the results weren't more impressive.

More importantly, teachers were generally doing what they knew how to do, rather than doing what was necessary to produce the result they were trying to produce. In the absence of specific guidance that what they were doing wasn't going to get them where they wanted to be, they would, other things being equal, continue to do what they—and many others—regarded as "good teaching," without recognizing that it was precisely that kind of teaching that was producing the performance they were disappointed with. Doing what they regarded as the "right thing" was not enough. They would have to figure out what the right thing was to do and then figure out how to do it.

Building the Capacity for Improvement: What's Needed

Thornton and Clemente are both improving schools by any reasonable definition, but they are both failing schools under the terms of the current accountability systems in which they operate. They will both almost certainly be classified as failing schools under NCLB. They are not failing because the people in them don't, for the most part, recognize their limitations or fundamentally believe in the principles on which the accountability system is based. In fact, people who work in both schools accept that they are not doing as well as they should. This is why they have asked for help. Thornton and Clemente are not improving as much as they would like, I think, because of a fundamental design flaw in current accountability systems: the failure of policymakers to bring capacity-building measures into alignment with performance measures in the design of accountability systems.

As noted up front, the performance targets set by current accountability policies are, for many failing schools, completely unattainable using their existing capacities. Most schools, even nominally high-performing schools, couldn't do this work using their existing capacities. In order to meet these performance targets, schools have to develop successively higher capacities. Each new set of capacities speaks to the next level of problem. Each level of increased performance carries its own new set of problems. Each new level of capacity requires a period of consolidation. Acknowledging the gap between capacity and performance in accountability systems isn't, I repeat, isn't an argument for abandoning performance targets altogether. It is, however, an argument for a more knowledgeable approach to setting performance targets.

Improvements in school performance, as my colleagues and I currently understand them, probably take a form something like this:

• Schools recognize and internalize problems of performance by paying attention to evidence on student performance.

• They choose a proximate performance target—increasing reading performance, for example—and focus their work on improving their individual and organizational capacity to meet this target.

• If they succeed in choosing the right target and developing the initial knowledge and skill in teachers and students around that target, they typically see a modest bounce in student performance. Often, these initial moves, in very low-capacity schools, consist of very low-level changes: devoting a set number of minutes per day to teaching reading; realigning the curriculum so that the content that is tested is actually taught before the test is given; identifying students whose performance could easily be improved, thereby making the whole school look better, etc. I have come to call this the "low-hanging-fruit" stage.

• These improvements are real, but modest, and the capacity for further improvement is not there. But the critical moment here is that the school has decided to make some collective commitment to a goal that has to do with performance. This is the first stage of developing internal accountability.

• If the organization reads its performance well, it is at this stage that schools often try to tackle a more ambitious kind of instructional improvement. This improvement is often focused on the adoption of specific curricula and instructional practice. This stage almost always requires that the school receive some kind of external help, support, and professional development. Why? Because, by definition, people in the school don't know what to do, or they would have done it already. New practices take time to acquire and implement with any consistency. They also require people to organize and manage themselves around increasingly clear collective goals—another increment in internal accountability. But schools that go through this phase almost always see gains in student performance, in part because they are learning to work together more powerfully, and in part because they are actually teaching different content in different ways. Just as predictably, performance tends to go flat again almost immediately. This is where Thornton and Clemente were in the improvement cycle when I visited. As noted above, performance goes flat because the problems of improving student performance are more complex than the strategies adopted at this stage can cope with.

• If the organization diagnoses this problem—and schools usually require some kind of external help to do this—it then has an opportunity to examine the barriers to continued improvement. Typically, the kind of problems that schools work on at this stage are either problems of increasing the consistency and cognitive demand of instruction or figuring out why the instructional strategies they adopted earlier work for some students and not for others. Simply diagnosing these problems and working on their solutions creates new capacity for collective action, bringing the next stage of work and improvement.

• The problems of improvement become more complex and demanding as performance increases; the challenges to existing instructional practices and existing organizational norms become more direct and difficult. Often, schools go through some kind of crisis at around this time, where teachers and principals argue that the work has become impossible to do under existing resource constraints and that expectations set by external accountability systems are simply impossible to meet. This is a very tricky stage, because it is hard to argue that the demands of the external accountability system are reasonable—they often are not. But what teachers and administrators are also saying is that they simply don't have the capacity to make the next set of improvements, and they usually don't. I have come to call this the "impossible work" stage. The conditions for future improvement are present, but the capacities to make that improvement are not. It is critical for schools to receive high levels of support at this stage—to get help in diagnosing the next set of problems, to get help from people with expertise about problems of student learning and instructional practice, to broaden and deepen common expectations around high-quality instructional practice, and usually to see schools in similar circumstances that have managed to move through this stage. It is at this stage that the credibility of accountability systems, as systems of political authority, is tested. Here's why: My authority to command or induce you to do something you are not currently doing depends, in large part, on your capacity to actually do it. You may be motivated to do it. You may agree with me that it should be done. Or you may be willing to do it just because I have a legitimate grant of authority to require you to do it. But if you can't do it because you do not have the capacity to do it, then my authority is diminished because I have induced or required you to do something you cannot do. I can flog you harder, I can penalize you, I can threaten you, but I cannot make you do something you do not know how to do.

• Schools that make it through these crises typically emerge as much different organizations—stronger, more coherent, with responsibilities more widely distributed and with much higher morale around student learning and much higher cognitive demand in the classroom. But they often have difficulty demonstrating that these changes are consequential because going through a crisis saps the energy and commitment of people while it is going on, and often performance goes flat during these periods. So just as the school is feeling that it has a much better handle on student performance, its results often look less impressive than they should. This occurs because the school has built the capacity for higher-level instruction but hasn't yet seen its full effects. This is where more concentrated work on instructional practice—not less—is important, because it is important that teachers and administrators understand that not only have they changed the way instruction occurs in the schools, they have changed their own capacity to take responsibility for and manage their school's response to pressure for performance. Again, external support and assistance around targeted problems of student learning help to reinforce the idea that everyone has gotten here by developing new knowledge and skills.

• The next stage of improvement—one that very few schools achieve, even nominally "high-performing" schools—is where the school collectively takes over the management of its own improvement process, teachers and students internalize the values of managing and monitoring their own learning, administrators model their own learning for teachers and students, and individuals are empowered to ask for the help they need.

This is a highly stylized map of the improvement process, in part because we need much more research to get a deeper picture of what actually happens as failing schools improve, and in part because Leo Tolstoy was right—each unhappy family is unhappy in its own way. Each failing school faces, in addition to the general problems of improvement, a specific set of problems rooted in its own context.

*  *  *

Current accountability systems aren't built to do what they are supposed to do—to push and support schools in getting better. The systems exhort schools and localities to provide support and professional development for schools in need of help, but don't actually invest in the infrastructure required to make sure that that help gets to the right schools at the right time with the right technical expertise. They heavily underinvest in the development of the knowledge and skill required to rectify the problems that failing schools face. The systems are generally unresponsive to the systemic problems that prevent resources from getting to schools—resources necessary for high-quality work—and tend to view the problems of all low-performing schools as essentially the same. The most discouraging aspect of current systems is that they ignore and undervalue the struggles of people like those who work in schools such as Thornton and Clemente, creating the expectation that students would be better off in other settings, without understanding that moving students around is essentially moving the problems of capacity from one set of institutions to another, without remedying the underlying problems of how to raise the capacity of the institutions where the children are to begin with.

Schools don't suddenly "get better" and meet their performance targets. Improvement is a process, not an event. Schools build capacity by generating internal accountability—greater agreement and coherence on expectations for teachers and students—and then by working their way through problems of instructional practice at ever-increasing levels of complexity and demand. Right now, virtually no infrastructure exists to provide continuous support to failing schools.

Building capacity in failing schools is going to require a lot of feet on the ground—people who know something about school improvement and who know what they don't know. As my analysis suggests, I would look for these people in what I have called "improving" schools (not in nominally "successful" schools because a large number of "successful" schools are not improving schools5), where faculty and school leaders have worked through several stages of improvement, and in improving school districts, where district-level personnel have gained real knowledge about the kind of support and resources schools need to improve. Some of this knowledge can also be found in organizations with staff knowledgeable about school improvement, such as America's Choice, Core Knowledge, and Success For All.

But, finding, organizing, and deploying this expertise is going to require a kind of work that most educators aren't yet very good at and that most policymakers don't know anything about. It will require measures of instructional improvement and performance that are much closer to the ground than the state assessments that are the basis of accountability systems. It will require the creation of systems to find people with expertise in subject matter, instructional practice, and improvement, and getting them into the schools where they are needed. And it will require a new generation of people who are knowledgeable enough about instructional practice to be useful, but who are also interested in broader questions of designing and running support systems.

In short, it will require a lot of hard work in schools and a lot of investment by all of us in the infrastructure of capacity-building. But without this, improvements in school achievement will be small because we won't have provided what Thomas Schelling knew was necessary: We won't have provided people with the knowledge necessary to get the job done.

Richard F. Elmore is the Gregory Anrig Professor of Educational Leadership at the Harvard Graduate School of Education and a co-director of the Consortium for Policy Research in Education. He has written numerous articles, reports, and books, including materials for the Albert Shanker Institute. This article was adapted with permission from Richard F. Elmore, "Doing the Right Thing, Knowing the Right Thing to Do," School Reform from the Inside Out: Policy, Practice, and Performance, Cambridge: Harvard Education Press, 2004. Copyright © 2004 by the President and Fellows of Harvard College. All rights reserved. For more information, visit


1. NCLB does not use the term "failing"; it classifies schools as "in need of improvement." But the law treats the schools as though they are failing, and that is the term I'll use in this article.

2. All schools are identified with pseudonyms under assurances of confidentiality.

3. See: Elmore, R., Ablemann, C., Even, J., Kenyon, S., and Marshall, J. (2004). "When Accountability Knocks, Will Anyone Answer?" in School Reform from the Inside Out: Policy, Practice, and Performance. Cambridge: Harvard Education Press. Also, Carnoy, M., Elmore, R., and Sisken, L. Eds. (2003). The New Accountability: High Schools and High Stakes Testing. London: Falmer Press.

4. Elementary teachers, I have found, are usually much more knowledgeable and discriminating on this issue. I recently showed a videotape of a high school writing lesson to a mixed group of teachers and administrators from elementary, middle, and high schools. The tape showed a white teacher who had strong skills for engaging his largely minority students in playful and pleasant interactions in class. The work the students produced was, however, obviously very low level, and the teacher's expectations, revealed in his teaching and in a post-lesson interview, were very low. The middle- and high-school educators gave the lesson largely positive reviews. The elementary educators had strongly negative reactions to what they regarded as the insultingly low level of expectations for students and what they perceived as the teacher's condescending attitude toward his students.

5. Because No Child Left Behind requires schools to get all students to the same level of proficiency, schools with students who entered with high achievement levels have to make little or no improvement to meet NCLB targets. So staff in high-achievement schools may or may not know much about the process of improvement.

American Educator, Spring 2005