Home Articles Reader Opinion Editorial Book Reviews Discussion Writers Guide About TCRecord
transparent 13

Rethinking the Quest for School Improvement: Some Findings from the DESSI Study

by A. Michael Huberman & Matthew B. Miles - 1984

A review of the Study of Dissemination Efforts Supporting School Improvement (DESSI) field study indicated a need for reorganization of the conceptual paradigms used to account for school improvement. Current paradigms do not account for the rational and conflict theories of social change. (Source: ERIC)

Definitive formulas—seen in craft and science terms-for achieving school improvement have often proved elusive. Elusive because the normative nature of the term makes it slippery; one persons’s version of improvement is another’s version of wastefulness, or even of worsening. Elusive also because researchers have ridden off in all directions, armed with different conceptual and methodological trappings, and have returned with formulas that often appeared incompatible or incommensurate. Elusive, finally, because of implementation problems: even though we may know what successful school-based innovation looks like, delivering it is another question. Too many of the key variables seem unmanipulable, and too many success stories look more providential than intentional.

It is dangerously easy for school people and researchers to conclude that the quest is hopeless, and turn to other matters. The risk, of course, is that of abandoning “school improvement” to the opportunists and zealots, making it into little more than an ideological banner, thus paving the way for rhetoric-laden “programs” with little real benefit for schools.

Yet we believe the quest is not at all hopeless, on two counts. First, it is increasingly clear that we know a good deal more about implementing changes that improve school-based practice than we let on. If one compares recent studies, whether they define improvement as increased achievement scores, as classroom mastery, as use of new instructional modules, or as improved academic self-concept, the findings across studies1 are far from inconsistent. The same macro-variables crop up, from “adaptation” to “assistance” to “involvement,” “ problem solving” and “institutionalization.” Often, the configurations of these variables are constant across studies and settings, and one begins to get a glimpse of which families of predictors will influence which families of mediators and of outcomes, in which families of settings. In short, we are getting a progressively better fix on school-improvement processes and outcomes, though it is a far more differentiated and multilayered fix than we expected and might have preferred. We may not have a technology, but we do have a series of very promising cottage industries.

Our second reason for optimism about the quest is grounded in the belief that we can reorganize the way we have been going about it. Many researchers in the “dissemination and utilization of knowledge” tradition, of which the study of school improvement is a subset, have been guilty of a certain hubris. We have been imbued, often unconsciously, with the rationalist paradigm of change, and this has oriented us toward the quest for an uncontingent technology of “innovations” that make for “improvement” that may not exist empirically. We have tried to derive context-stripped “factors” or “variables” that somehow lay behind the welter of phenomena in idiosyncratic school settings. And we have inferred as well that the process within individual settings was orderly and intentional—or at least could be made so if educational affairs were managed properly. In short, we leapt too quickly and perhaps illegitimately from the descriptive to the prescriptive, and nurtured the Promethean illusion that by following our conceptual tracks, policy planners and administrators could get there operationally as well.

Correcting these biases requires us to reorganize the quest, but not to abandon it. The first, most obvious need is to abandon the master plan and accept a more modest scenario of generating smaller parts, multiple formulas, and clearer specifications of the contextual conditions under which the formulas may be expected to produce desired results. The second need is for improvement researchers to move deliberately into another paradigm, the “conflict” paradigm, in which the explanation of social change is rife with power, uncertainty, continuous negotiation, loose-endedness, and local history. We need only to survey the drift in the literature on organizations to get a sense of the exodus from the rational paradigm to the conflict paradigm in the past decade.2 Clearly, there is a fund of explanatory power here, and, obviously, a good deal of school improvement has to do with careerism, interpersonal conflict, politics, posturing, bungling, arbitrary power plays, temporary and opportunistic coalitions, and the like. But how much? And does this necessarily rule out a more planful or consensual approach? For example, conspiratorial behavior can be both technocratic and consensual in a sense very close to that of the rational model of change; the assumptions are different, but it is not obvious that they are incompatible. Adherents of both paradigms like the term “strategy,” for instance, and although one paradigm may be naive and the other overly conspiratorial, it is not clear that the meanings they impute to strategy are incompatible.

The point is that the rational, meliorist approach to educational innovation can and should be enlarged—made more conceptually streetwise—by the more conflictual, open-ended perspective in order to yield more powerful and useful studies. The search for “usefulness” may be harder than that for conceptual powerfulness—for example, we may find out that a lot of the critical variables can only be “managed” under exceptional conditions, and that those conditions fluctuate over time—but we need to know these things, too. Knowing what not to try is as serviceable as knowing what to try. And enlarging the meliorist approach does not oblige us to fully integrate the two perspectives. Rather, it saves us from having to choose between competing conceptual paradigms that are incompatible only at the extremes, and to concentrate on the middle ground—where the bulk of the data and the meaningful concepts are.


For the better part of the past five years we and our colleagues have been trying to take a serious crack at that conceptual and methodological enlargement, with some promising results. The chief product of that work has been the ten-volume Study of Dissemination Efforts Supporting School Improvement,3 whose findings have been distilled at successive American Educational Research Association (AERA) sessions and in practitioner journals.4 The DESSI study itself was of gargantuan proportions: 45 innovative programs, analysis at the federal and state levels, studies of 146 schools and school districts in 10 states with 5,000 interviews and questionnaires and a parallel, multiple-site field study keyed to the sampling frame and research questions being addressed in the survey component. While the size of the study created many logistical complications and problems in data interpretation, it did allow for a comprehensive analysis of federal and state dissemination activities throughout the country, and it brought together the contributions of five research centers active in the field of school improvement research.5

We were involved in most facets of the DESSI project, but most intensively in the field study, and the “learnings” we shall be reporting on come primarily from there. We did, however, work closely enough with the survey data base to see the substantial degree of overlap in findings, so the generalizations we offer can go beyond the twelve-site sample comprising the field-study component.

In many ways, the survey component of the DESSI study embodied the “rational” paradigm of school improvement, and the field study the more “conflict-theoretic” paradigm. To some extent, this was a methodological artifact. Surveys are inappropriate vehicles for picking up on subterranean career agendas or internecine rivalries or people’s incoherent behaviors, and when they do get such data, the statistical analyses often yield interpretations that border on the surreal. Field studies, on the other hand, can handle only a few settings and can get so mired in local-setting variables that they lose the programmatic thrust of the study initially undertaken. Surveys and field studies combined not only extend and deepen the data set; they also keep one another analytically honest and on target.

The field study mobilized four people for eighteen months, with about twelve months of spaced visits to schools. One of its purposes was to compensate for a survey’s typical weaknesses (predesigned instrumentation, one “snapshot” pass at a site, difficulties in unraveling over-time processes, clumsiness in the face of unanticipated or unequivocal findings). Another objective was historical and descriptive: that of “telling the story,” and of identifying and documenting typical patterns and local determinants. There was the additional hope of validating, or at least of lending more plausibility to, survey-analytical findings. More fundamentally, the field study was designed to capture the everyday properties of local sites and to follow how they change and are changed by the process of implementing new practices—what school improvement is an improvement of or from, what the “baseline” or “receiving context” is. Capturing these properties invariably entails getting underneath the official, noncontroversial, and socially desirable renditions often given by survey respondents, and dredging up more latent issues, agendas, conflicts, and contextual features.

The field study sample itself comprised twelve sites drawn from the survey sample and varying along the dimensions shown in Chart 1. For program sponsorship, only two programs were studied (National Diffusion Network, or NDN, and Title IV-C), so as to compare locally generated (IV-C) and externally introduced (NDN) innovations.

The conceptual model underlying the field research roughly paralleled that of the survey, although it took more deliberate aim at motives, conflicts, and organizational issues characterizing the everyday life of the school. It hypothesized a series of reciprocal “transformations” among users, the properties of the innovation, and the organizational unit affected—much along the lines of the mutual-adaptation phenomenon described by Berman and McLaughlin.6 As a set, however, the thirty-four research questions forming the basis for field-study data collection overlapped and mapped well onto the main foci of the survey.

We visited sites between 3 and 7 times each for a total of 5 to 11 days, depending on the site. Researchers worked with a general interview guide, keyed to the research questions, and supplemented this with observations and document collection. In all, there were some 440 interviews, 75 observations, and 259 document analyses. The interview and observation data, along with researchers’ comments, were dictated and transcribed (field notes for one site ran about 250 pages), then coded. Raw survey data were also fed into later fieldwork. The coded field notes went through a series of interim analyses, and were written up in identically formatted case reports and fed into the cross-site analysis. We used a variety of conventional and experimental methods for formatting, reducing, displaying, and analyzing qualitative data, including qualitative analogues of cluster and causal analyses, and have since extended them into a methodological sourcebook.7The field study itself was published both as a technical volume (vol. IV) in the DESSI report series and in an abridged form for a wider public.8


Finally, we should note that though we used many familiar ethnographic methods, the nature of this field study—short, interrupted time on site, presence of multiple sites; integration with a survey data base—moved it toward another species, one closer to the investigative reporting and investigative social research tradition.9 A distinctive quality of that tradition is its “recycling” or “forensic focusing” property. Field researchers use successive waves of interviewing and observing to establish and flesh out what appears to be the most plausible version of local events. Through a series of progressive revisions and cross-checks, that version undergoes a fairly merciless verification before field researchers and their critical readers settle on it. These procedures are also described in detail in Huberman and Miles.10


There is no fruitful way of summarizing the corpus of findings briefly; it would read like an incoherent shopping list. Let us try to be discriminatingly selective. In particular, we want to follow through on the arguments and the agenda laid out in the initial section by presenting carefully three main findings that are consonant with either a conflict paradigm or a rational paradigm, then try to show how they can be consonant with both paradigms at a conceptually superordinate level. These are also findings that may not have been brought home sharply enough in prior studies; in addition, they indulge our—probably decadent—bent for highlighting controversiality. Let us begin at the “front end,” with the question of why innovations are initially implemented. Here is the finding:


This is a finding that a more conflict-theoretic analyst might take in one direction and a rationalist-meliorist in another. For instance, Yin et al. found that the main motives for adopting public service innovations were not primarily service- or client-oriented but involved facilitating working conditions on the part of the adopters. 11But there was a mix of motives, and the configuration was different for different groups of adopters and users. There was, in many cases, a problem to be solved and the innovations tended to constitute solutions. So motives are multiple (for a given adopter), different (for different users), and even contradictory (wanting better working conditions and higher service payoff when both may not in fact be possible).

We found a similar pattern, with a few surprises. For example (see Chart 2) nearly two-thirds of the “users” in the sample (usually teachers) adopted primarily because of administrative pressure.

There were degrees of pressure. In some projects just under way, teachers said that they had been “strongly encouraged” to volunteer. One respondent put it concisely: “The principal wanted it in; we got the message.“ There was sometimes an element of seduction at this phase, but underneath it was the clear understanding that demands for compliance could not be countermanded.


Still, being pressured is not inconsistent with wanting to adopt. If the innovation looked good, and the supporting conditions (preparation for implementation) were satisfactory, there was no strong reason not to adopt. There were relatively few users who felt “violated” by the pressure, and some who saw such an exertion of pressure as a legitimate leadership role for the administrator.

The findings beg immediately for some contextualizing. What kind of innovations imply that teachers are essentially targets? Obviously, these are not classroom-generated changes, but rather changes initiated from without and affecting several classrooms at once. As such, these are exceptional efforts. They are also, as we have noted, efforts involving outside assistance, for example, federal or state seed money, provision of outside consultants or materials. So this is only one species of school-improvement project, albeit one whose potential impact is likely to be greater (more funding, more political clout, more local units affected) than that of isolated efforts.

Less than a third of the users said that their reasons for adoption had to do with instructional improvement. Of these, very few mentioned any chronic or severe instructional problem to be solved. Rather, they liked the added resources, the enrichment value for the curriculum they taught, the overall appeal of the unit or technique or project.

Another fascinating finding: Users’ motives often centered around professional growth. Many respondents thought these innovations could counter professional stagnation through their novelty value, could stretch existing instructional repertoires, could add to professional qualifications. When these individual items are grouped, they exceed the frequency of “practice improvement” motives. This underscores the need to broaden perspectives when we look at incentives and rewards. Although innovations are supposedly introduced for instrumental reasons—for example, to increase achievement scores—they are not necessarily construed this way at the classroom level, at least not in this sample. For example, no teacher mentioned “achievement gains” as a motive for adoption, although one administrator did. On the other hand, professional growth was a strong motive in many cases, and acted as a catalyst for the introduction of new practices, many of which did, in fact, improve academic performance. This pattern intersects nicely with the results of previous studies.12

Administrators gave more diverse reasons for adoption and configured them differently from users’. Pressure was far less frequently reported, but was still there, notably among principals, and sometimes had a sharp bite:

Just another dumb program being shoved down our throats.

For a principal to cross the district openly [and refuse to adopt] is almost like suicide.

Instructional and schoolwide improvement was a salient motive for administrators: The new practice “had merits”: it was “a good alternative”; it “could be used elsewhere” in the system and would improve working conditions in the classroom. Here again, however, local problem solving was not a dominant theme—nor was it more prevalent in the locally initiated change (IV-C) sites than in the externally introduced (NDN) change sites. Nor was there—with one exception in forty-one cases—an explicit expectation that achievement scores (or any specific student outcome) would improve as a result of program adoption. The emphasis was on school capacity-enhancing characteristics of the innovation (for example, enriching curricula, providing remedial materials) that administrators believed could mean achievement gains further down the line.

As with the teachers, administrators had multiple motives. Obviously, the greater the number of incentives and their relative importance to a key administrative actor, the more assertive and consequential the decision to adopt. And the closer we got to the full agendas of principals and central office personnel, the more numerous and various the incentives. Here is an illustration from the field notes:

Summary of inferred and explicit motives for her: [the innovation] resolved the local problem of low achievement, [helps her] compete with elementary-level rival in central office; increases relative clout of elementary education sector; meets [her] emotional needs [attention to disadvantaged]; enhances professional image, responds to pressures external to the district. All of roughly equal importance to her. (Supervisor, Banestown site)

Career Pursuit as Motive

The “conflict” paradigm makes us attentive to the fact that, in making available new resources, an innovation often alters career opportunities.13 First, it often creates new roles facilitating upward mobility. Teachers can become local trainers. Administrators are appointed program directors. Second, innovations also attract attention: Teachers involved with their execution frequently get higher visibility than in their conventional work, and higher visibility can enhance professional advancement. Finally, central office administrators, often active advocates of new practices, come into more direct contact with principals and teachers, thereby multiplying the opportunities for school-level personnel to forge useful interpersonal links. In short, innovations can accelerate promotions or other career shifts for both teachers and administrators.

We found many career-related incentives for adopting innovations at the twelve field sites. But we should be careful. At only one site was career advancement the paramount motive for adoption; it was usually one of several reasons. Also, career incentives were not always obvious at the time of adoption, but materialized later, when local actors saw more clearly what the institutional implications of the new practice were. So local personnel were neither overly cynical nor Machiavellian; they simply looked out for themselves and capitalized on opportunities crossing their paths. The important thing to remember is that paths exist.

What kinds of career shifts did site informants mention? We chose to sort out these motives as derived from what we called “trajectories.” School people may have been going nowhere or going someplace in their career progression, so the innovation was an asylum, a way station, a vehicle, or a final destination. Logically enough, there were four gross trajectories: The innovation provided as opportunity to moue in (for example, from a prolonged leave or from a distasteful job elsewhere), to stay in (securing, solidifying a new role; avoiding demotion), to move up (e.g., from aide to teacher, teacher to principal, principal to central office administrator), or to moue away (to a lateral job, to another line of work).

Roughly half of the administrators and teachers for whom there were data had no discernible career motive. Either they were not interested in changing roles, did not see the innovation as a good vehicle for career shifts, or simply focused fully on the contribution of the new practice to their immediate settings without regard to their own career progression.

The remaining half had plans. For both teachers and administrators, these plans frequently involved promotions, but teachers’ career motives were quite varied, whereas administrators were more often setting up a promotion or solidifying one. Informants usually spoke to us openly about these plans, but were often discreet in their formulations, speaking of the innovation as “an intermediate step” or “a way to get more responsibility.“14 Sometimes, there were two-step trajectories;

I figured it was a good way to get back in after my leave. The program had a good reputation and I liked the one-on-one work in it. . . . It was an interesting job while I was waiting to get my own class, (Teacher at Banestown)

“Moving someone else in” was a dominant motive in only one site, but a sub rosa one at two others. In two cases (Proville, Banestown) this led to intense jockeying that took its toll on program execution. The innovations did not create these perturbations, but amplified interpersonal tensions already present, adding noise and friction to the system.

In contrast, at three of the five sites where administrative careerism was low (Carson, Masepa, Tindale) implementation was smoother: more task-focused, more collaborative between central office administrators and building-level personnel, and less perturbed when staff turned over. (In passing, the two other sites—Dun Hollow and Lido—seemed to suffer from a lack of central office investment, which might have been stronger if a career motive had been present.)

In the five remaining sites, administrative careerism was strong, without being manic or exploitative; it appeared to inject energy into the adoption process by accelerating the time line, mobilizing resources, and, in some cases, generating enthusiasm at the school-building level.

So we can see that the configuration of motives is both less disinterested and less problem-focused than the rationalist paradigm would have it—but it also has meliorative, task-centered components. Moreover, even some of the more apparently opportunistic phenomena (e.g., career pursuit among administrators) have positive results and can probably even be incorporated into an innovation “strategy” from a rationalist’s perspective. A larger, more ecumenical conceptual framework is helpful.

Let’s move on to the next proposition:


This may be an objectionable proposition, but it is also a logical one. It nests well in the rationalist paradigm in assuming that school improvement can be managed and achieved as the product of a social technology. But the rationalist paradigm also assumes, with disarming naivete, that people will achieve and be managed because they are convinced of the merits of the project, along with its problem-solving potential. One needs the conflict-theoretic paradigm here to get much closer to the ways in which organizational changes are actually played out. Put crudely, organizational change occurs in a complex, continuously negotiated power field, in which some parties wield more influence than others, but the others are never powerless.15 Senior administrators in a school system generally have the power to adopt innovations, to mandate them, to finance and backstop them, even to institutionalize them. This goes a long way toward actually delivering an innovation. But managers don’t execute innovations, teachers do; and if teachers decide not to execute the innovation, managers will find themselves institutionalizing placebos, perhaps unwittingly. So administrators have to leaven an enforcement strategy with something that teachers are interested in and that, at the same time, will facilitate execution. This turns out, logically enough, to be sustained assistance. Administrative bullying—the superintendent or principal as thug—is an ineffective strategy, but assistance-rich enforcement—tough-minded help—appears to pay off.

All of this needs empirical substantiation. First, however, let us spell out the first part of the proposition. We are saying that a more power-coercive strategy, when blended with a normative-reeducative strategy in the classic sense of Chin and Benne,16 is potentially a surer path to school improvement because powerful people tend to be able to exert directional control over the environment—to shape the surround, to reduce the uncertainties, to reduce the degree of freedom of actors having countervailing plans—and to offer assistance resources.

In assessing the nature and the degree of school improvement, we came to focus on five normative outcomes—normative in the sense that having achieved more of them at higher levels was perceived as more desirable than having achieved fewer at lower levels. They are:

Stabilization of use of the innovation at high levels of technical mastery

Percentage of use: proportion of actual users relative to the total pool of eligible users of the innovation

User capacity change: increments in teachers’ professional competence as a by-product of the innovation

Student impact: the effects of the innovation on pupils

Institutionalization: the “built-in-ness” of the innovation in local regulations, procedures, budget and procurement, cycles, training programs, and the like

The final outcome variable is an especially important one. New educational practices that do not hang on beyond one or two years are clearly not worth the investment, and the gist of relevant research is that hangers-on are fewer in number than one had thought. 17The notion that a technically mastered, well-administered, impactful innovation will lock automatically into the local setting is, once again, a Panglossian and potentially self-delusive view of the world. We found that technically mastering a new program, even one that demonstrated better pupil performance than had its predecessor, did not ensure and sometimes had no bearing on the likelihood of its continuation. But we also found that continuation can be ensured when administrators attend to the requisite parameters.

The Supported-Enforcement Scenario

Let us focus on the “institutionalization” outcome, with attention to some of the others as well. In the DESSI field study, we drew a “causal network” for each of the twelve sites describing the course of local events in terms of a core list of empirically derived variables.18 We were analytically able to sort the twelve sites into four different “families” according to the configural clustering of the key variables. The first family performed well on all the main outcome variables. We came to call it the “supported-enforcement” scenario, in that it achieved stable, institutionalized use through administrative mandating and sustained assistance that led through teachers’ energies to stable use. Figure 1 shows one of the cases in that family:

Here, a powerful central office administrator (2), the director of curriculum and special projects, working from a centralized power base (3), put considerable pressure on users (5) to implement the new, locally developed reading program. Initially, this lowered users’ commitment (8); they resented and feared the pressure. But substantial assistance (1) was supplied, which increased users’ practice mastery (4) a good deal and subsequently their commitment (8). In addition, organizational rearrangements (9), including scheduling, pupil rotation, and teacher teaming, were made, increasing student impact (11). User mastery and commitment, along with stability of program staff (6), led to stabilized use (IO), which both increased percentage of use (12) and led to institutionalization (13). Stability of program leadership (7) also aided institutionalization. The general picture is one of administrative decisiveness, accompanied by enough assistance to increase user skill, ownership, and stable use in the context of a stable system. The cases in this supported-enforcement family share some characteristics that are worth underlining:

1. Sustained administrative commitment. Senior administrators, often central office staff in a coordinator or curriculum slot, wanted to see the project implemented. They therefore exerted pressure, made the necessary procedural and structural changes, and provided sustained backstopping: materials, training, authorizations, consultancies, planning sessions on school time. They did not get the project going and then turn their attention elsewhere, which is the modal pattern in this and in most empirical studies. Nor did they assume that assistance at the front end (prior to program execution) would suffice, but rather redistributed these resources to make proportionately more of them available later on.


2. Technical mastery as a stimulant. Some teachers initially did not like the innovations in this family and did not appreciate the pressure. They did, however, make a serious attempt to get on top of the new practice instructionally—in part owing to the pressure, in part to the assistance, in part to the challenge of mastering a demanding technical task. As that level of mastery rose, two things happened. First, as the figure shows, user commitment (8) increased. When one follows this process in the field, it is more the mastery that drives the commitment than the reverse; commitment is not a precondition, but rather an achievement, the result of effortful and technical application. Teachers talked of “being on top of it,” of “getting it to go my way.” Second, heightened practice mastery (4) led to stronger student impact (1 I), in terms of achievement or attitude increments. The innovation had started to deliver, and the users, increasingly effective in manipulating it at the classroom or school level, could increase, differentiate, and more consistently produce those impacts.

3. Deliberate focus on institutionalization. Innovations are highly perishable goods. New practices that get built into the training, regulatory, staffing, and budgetary cycle survive; others do not, notably when their original champions move on. The administrators in this family attended specifically to these issues and did not move on until the innovation was satisfactorily locked in.19

The Overreaching Scenario

We said that administrative enforcement was a family in our data; there were others, at least one of which could also lead to positive outcomes. It was, however, a less certain path, and it entailed slightly lower levels of outcome. We called these projects “overreachers.” They also entered the district through central office advocacy. Administrators provided initial and some follow-up support, and authorized users to make execution more manageable by changing the innovation—something the administrators in the first family of projects did not allow in the early stages. Most of the support, however, was peer support rather than administrative support, and it brought users through to fairly strong levels of commitment and mastery. Project outcomes were fairly high, but users tended toward exhaustion, and responded by reducing the gradient of practice change (which lowered the outcome levels) or by reviving career pursuits that led out of the district. Personnel turnover, both among staff and project leadership, reduced commitment to the project and thereby made for an uncertain longevity of use at the site.

This family of projects illustrates the fact that new practices can be introduced and consolidated without administrative strong-arming. The problem, however, is that as users reach masterful ownership of the practice, some burn out, others revive career ambitions, and most are working with a somewhat watered-down project relative to its initial scale and impact. Nor do administrators appear to build lasting commitments to the projects they sponsor or lead, so that success or environmental turbulence are prompts to move on. We shall return to this point in a moment.

So this family, while more democratic and lateral than the first one, is also less “managed.” It achieves somewhat weaker results than the first set-but in a more adventurous way. These projects are not “implemented” as much as they are collectively “crusaded.” They have a lot more uncertainty, in part because the senior administrators run them with a lighter hand. They run the innovations even less, however, in the later stages, when institutionalization issues are paramount. The scenario here is more appealing to those of us nurtured on nondirective leadership, but it is also less reliable than the first one, which tends more to deliver the improvements it pursues. In other words, implementation can be a social technology—school improvement can be enacted—but one pays a price for it in the degrees of freedom allotted most of the main actors.

The Dark Side of Adaptation: Blunting and Downsizing

We are now ready to spell out the second part of proposition two, in which we endorsed intelligent, supportive coercion as a surer path to significant school improvement. The claim we are making here is that innovations that are managerially handed over to teachers often become ”blunted” or “downsized” in such a way as to trivialize their outcomes.

We found a third family of projects with a basic theme: Innovations that might have been major ventures, either at the district or classroom level, had been progressively reduced in scale and scope to the point where outcomes could be, at best, only modest. Typically, the reductions took place at the instructional level. Given wide latitude by administrators to make changes, teachers redid the innovations, stabilizing them as minor add-ons or drop-ins. To redo them, teachers unbundled external (NDN) projects, using only those components that were congenial to personal teaching styles and that called for few changes in ongoing instructional routines.

An additional thread tied this group together and connected it thematically to the second family of cases for three of the four sites: Continuing administrative leadership was low. Key advocates at the building and district level backed off, left, or were reassigned.

As an example of the scenarios, in the Astoria site the districtwide mandate for using an early childhood screening and skill-development program created new and stiff demands on the kindergartens. This made for a poor fit between program and building characteristics, already potentially poor because of the heavy requirements of the program design. The central office administrator then gave wide latitude to building administrators to “adapt” the project, thereby restoring the teacher-administrator harmony that had been compromised by the innovation and its requirements. As a result, there was substantial program change, resulting in less impact on pupils than its developers expected.

The scenario at other sites was similar. At the Burton site, the central office advocate considered the program to be in a “pilot” year, when teachers could “pick and choose” promising aspects, then, supposedly, build them into the curriculum. But teachers quite naturally shied away from the more demanding or adventurous components (for example, emphasis on activist community experience and more student-directed learning activities). Because these aspects of the program were central to its success, the impact on pupils was minimal.

The point here is not that local adaptation is a poor policy, but that fiddling with a project to improve the fit between the school and the innovation can often trivialize results. The whole point about innovating is that it creates a discrepancy between local practices and the demands of the new practice—and in that discrepancy lie the changes that can produce significant results. Reduce that discrepancy and you throw away important outcomes. The logical—and understandable—response of teachers is to improve the fit, to begin gently, and to avoid situations of high uncertainty, all of which create pressures on administrators to give teachers their heads. Such latitude usually leads to down-sizing—which is rarely followed by up-sizing later on. In particular, administrators who give a lot of leeway (often in order to get the innovation accepted) and then turn their attention elsewhere are asking for placebos. More often than not, that is precisely what they get.

There are, of course, many caveats here. Supported enforcement is, as we have said, not the only way to get significant practice change, although it might be the most literally manageable way. Also, we are talking about innovations intended for several schools and therefore engaging the central office directly. There is also the assumption that the innovation’s quality is high—or at least that it can be made so. Enforcing lemons, with or without assistance, is a flawed enterprise, and the projects comprising our fourth family (i.e., cases of failure) are of that kind, with a strong dash of administrative indifference thrown in.

A final caveat: Our discussion of the second proposition has drifted into the realm of the Promethean: we have depicted muscular administrators sagely manipulating resources to enact impactful, well-rooted changes. The assumptions are that muscular administrators will not abuse their muscles and that the resources are there; both assumptions are questionable in many instances. There are also two other problematic assumptions: (1) that the achievement of some outcomes does not undo the achievement of others, and (2) that the external environment is benign. In other words, human affairs may be manageable—but only up to a point. Uncertainty and perversity have a way of confounding our best-orchestrated efforts. That brings us to our third proposition.


Let us address this proposition, which might be captioned “the failure of success,” with the two remaining outcome variables we have not yet illustrated: the normative outcome, “user capacity change,” and a more descriptive outcome that we called “job mobility.” The former has already been defined; the other warrants some added discussion.

As we saw in the analysis of informants’ motives, about 40 percent of users and 50 percent of administrators had career plans tied in with the adoption of the innovation. One is naturally curious as to whether and when people got where they wanted to go. How much job mobility occurred? And how much of it was related to the innovation, or was adventitious? Finally, what were its local effects?

This is the kind of variable that surveys either ignore or are ill-designed to capture, but that field studies can identify and document when it looms large empirically, as was the case in our study. When we arrayed the projects by program type, numbers of key actors involved, job shifts at different moments in the life of the project and the degree of program-relatedness of those shifts, there was a good deal of moving around related to the innovation (63 moves among 123 key actors—people closely involved with the innovation—for our 12 sites, of which 83 percent were clearly innovation-related). Activity was high at the outset, when the project created new, often upwardly mobile roles. But activity was also high much later on, when administrators moved out and up or simply within the district, and teachers associated with the project took over their places. About 35 percent of the innovation-related job mobility was upward.

Some job mobility (14 moves, or 17 percent of the total) was not related to the innovation. It was of two types: (1) project staff had planned to leave or retire, or were moved elsewhere independently of the project; (2) environmental turbulence, usually in the form of sudden budget cuts, ejected people from the project, usually into unwanted jobs. More on this latter type in a moment.

At four of the twelve sites, there was little evidence of innovation-related job mobility. At the eight others, career moves were clearly tangled up with the fate of the innovation. Within these eight, we identified four somewhat distinct families. Let us illustrate one case that straddles two families and, thereby, accounts for five of the eight high-mobility cases. Figure 2 presents the causal network:

The first stream leading to job mobility is an ominous one; it goes from low internal funds (2) to environmental turbulence (3) in the form of unexpected budget cuts to job insecurity (5) and to job mobility (7) for two of three core staff. Job mobility was mostly unwanted; people were essentially casualties of external events over which virtually no one had much control. Three of the eight cases have similar casualty scenarios.


However, there is another, more upbeat stream to job mobility: the one leading from user practice mastery (6) through user capacity change (8) to job mobility. Through progressive mastery of the project, users and administrators developed new skills they were interested in using elsewhere—preferably but not necessarily at a higher level. Paradoxically, it was the success of the project at the individual level that triggered mobility and, in so doing, lowered its institutionalization (10) at the local level. A similar phenomenon is afoot in the stream leading from career advancement motivation (1) through dissemination effort (4) and external network development (9), then up to user capacity change and job mobility. In brief, disseminating a largely successful program brought on opportunities for self-advertisement and new jobs that, when taken, reduced the local stability of the innovation. Some actors won, some lost (the “casualty” cases), but both kinds moved and the project’s continuation was made uncertain by their departure.

Lest we be accused, as some qualitative researchers have been, of overusing irony as an analytical device, we should note that nondisruptive cases of job mobility did exist at two of our eight high-mobility sites. Both, interestingly enough, were highly effective supported-enforcement sites. At Tindale, users moved into building-level administrative jobs, a department head went to the central office, and a central office administrator shifted to another district. But the movement was seen as a legitimate payoff for successful program execution, and the key staff remained, continuing to provide leadership for the innovation. At the Masepa site, two users moved into full-time teaching, a principal’s “visibility” was aided, and another user became first a turnkey trainer and later a Title I coordinator. No one left the district because of involvement with the innovation. So career advancement needs and innovative success are not incompatible. The question is whether this balance can be planfully managed. In most of our high-mobility sites it was not.

It could be argued that mobility that disrupts local programs contributes to another educational enterprise; those who move will use their new capacities, or as one of the principals we studied did, replicate the innovation in a new position. Nevertheless, it is important to maintain social capital. On balance, we might wish for school-improvement programs that could accommodate individuals’ needs for capacity development and career advancement without destabilizing the local gains achieved.

But educational change is manipulable only up to a point. Project success at the individual or even the institutional level can create the kernel of longer-term failure by unsettling the project, and there may not be much that can be done about it. Environmental turbulence can do in a well-executed, impact bearing innovation in a matter of months; innovations are typically the first victims when budget crises strike. We would do well to be modest, not only in the claims we can make within a rational, social-engineering paradigm, but also concerning the control over events that is possible within a conflict paradigm. Within the bounds of that modesty, as we have seen, there are identifiable and probably replicable ways of bringing about school-based improvement.


In conducting this selective tour of the findings of the DESSI field study, we have tried to show that the conceptual paradigms used to account for the school-improvement process need enlargement and articulation if they are to mirror the real world and provide useful explanations of it. The advocates of highly managerial, meliorist perspectives need to wallow around more in the untidy, often disruptive empirical realities we find in the field. Those pushing conflict-theoretic perspectives may be equally simplistic in their endemic emphasis on conspiratorial strategizing, an emphasis that obscures much of the goal-centered activity to which different actors actually commit themselves within schools. Being more ecumenical within and between these paradigms strikes us as both conceptually plausible and empirically robust.

Finally, we have suggested that school improvement is a reachable objective, and that we are beginning to understand the conditions for reaching it. Most of our field-study (and survey) cases achieved moderate to strong positive outcomes. Though administrators and teachers live in different worlds and have different agendas, they can, under the right circumstances, complement each others’ efforts productively. Committed administrators can push, protect the innovation from casual adaptation, and supply supportive assistance. Users can follow their own professional growth motivation to master the innovation, help each other, extend their own classroom practice, and deepen their commitment.

Getting all the pieces of that formula in place, however, is difficult and only partly manageable. The conditions for attaining school improvement are stringent—and sometimes undesirable for one or more of the parties involved. Perhaps that is why there is less school improvement going on than we would like. If it were easy, we could simply follow the craft and mold our concepts around it. As it is, the craft of school improvement is unevenly successful, and we need newer, tougher concepts to reorient the quest.

Cite This Article as: Teachers College Record Volume 86 Number 1, 1984, p. 34-54
https://www.tcrecord.org ID Number: 928, Date Accessed: 1/16/2022 5:59:19 PM

Purchase Reprint Rights for this article or review
Article Tools
Related Articles

Related Discussion
Post a Comment | Read All

About the Author
  • A. Huberman
    University of Geneva

  • Matthew Miles
    Center for Policy Research, New York City

Member Center
In Print
This Month's Issue