PA. 47p. - ERICsocial organizations situated within, and vi-tally affected by, other complex social...

ED 431 749

AUTHORTITLEINSTITUTION

SPONS AGENCY

REPORT NOPUB DATENOTECONTRACTAVAILABLE FROM

PUB TYPEEDRS PRICEDESCRIPTORS

IDENTIFIERS

ABSTRACT

DOCUMENT RESUME

SP 038 616

Cohen, David K.; Ball, Deborah LoewenbergInstruction, Capacity, and Improvement.Consortium for Policy Research in Education, Philadelphia,PA.

Office of Educational Research and Improvement (ED),Washington, DC.CPRE-RR-431999-06-0047p.R308A60003CPRE Publications, University of Pennsylvania, GraduateSchool of Education, 3440 Market Street, Suite 560,Philadelphia, PA 19104-3325; Tel: 215-573-0700; Web site:http://www.upenn.edu/gse/cpre/ ($10).Guides - Non-Classroom (055)MF01/PCO2 Plus Postage.Change Strategies; Educational Change; EducationalEnvironment; *Educational Improvement; *Educational Quality;Elementary Secondary Education; Faculty Development; HigherEducation; Preservice Teacher Education; TeacherEffectiveness; Teacher Improvement; Teacher Role; *TeacherStudent Relationship; *Teaching Skills*Capacity Building

This research report offers a theoretical frame forexamining recent instructional improvement efforts. First, it develops atheoretical view of instruction and an analysis of the environments ofinstruction. It analyzes the relations between instruction and itsenvironments and the circumstances under which intervenors must operate. Thereport discusses the problems and possibilities for intervention, offering aninstructional theoretical frame. It develops an interactive model ofinstruction to analyze both teaching and efforts to improve teaching. Thereport elaborates on this split-level frame as it goes, drawing outimplications for understanding instruction, instructional improvement, andresearch on the two. It concludes by summarizing the ideas and distinguishingthem from other approaches to understanding and studying school improvement.Section 1, "Instruction," focuses on the interactions among teachers andstudents around educational material, discussing issues related toinstructional capacity. Section 2, "Implications for Intervention andResearch," examines comprehensiveness and intervention, teachers' uniqueroles, and teachers' opportunities to learn. Section 3, "Instruction and itsEnvironments," focuses on schools, system organization, coordination ofinstruction, professional norms, resources for professional education,professional learning, the issue-attention cycle, and society and culture.Section 4, "Intervention," discusses design, specification, and development.Section 5, "Enactment," discusses relationships between intervenors andenactors. Implications for intervention and research are noted. (SM)

AcaLuch

Instruction, Capacity, and Improvement

David K. CohenDeborah Loewenberg Ball

GRADUA seHooL EDUCATION

BEST COPY AVAILABLE

GONSORTIU

alkeleogyUuormaiftaDittguaw

UoitmedVatiOnorift cf/

L2 I

U.S. DEPARTMENT OF EDUCATIONOffice of Educational Research and Improvement

EDUCATIONAL RESOURCES INFORMATIONCENTER (ERIC)

9/chis document has been reproduced asreceived from the person or organizationoriginating it.

0 Minor changes have been made toimprove reproduction quality.

° Points of view or opinions stated in thisdocument do not necessarily representofficial OERI position or policy.

RDM POLICY

PennsylvaniaHarvardStanford

Miehigan

RESEARCH

Wiseonsin-Madison

EDUCATION

Consortium for Policy Research in Education

The Consortium for Policy Research in Education (CPRE) unites five of the nation's top research institutions in an excitingventure to improve student learning through research on education reform, policy, and finance. The members of CPRE arethe University of Pennsylvania, Harvard University, Stanford University, the University of Michigan, and the University ofWisconsin-Madison.

CPRE conducts research on issues such as:

education reformstudent and teacher standardsstate and local policymakingeducation governanceschool financeteacher compensationstudent incentives

To learn more about CPRE, please call (215) 573-0700, and then press 0 for assistance.

In addition to conducting research as described above, CPRE publishes reports and briefs on a variety of education issues.The Consortium also sponsors regional policy workshops for state and local policymakers.

CPRE's website can be found at: http://www.upenn.edu/gse/cpre/

CPRE Research Report Series

Research Reports are issued by CPRE to facilitate the exchange of ideas among policymakers and researchers who shareaninterest in education policy. The views expressed in the reports are those of individual authors, and are not necessarily sharedby the U.S. Department of Education, CPRE, or its institutional partners. This publication was funded by the National Instituteon Educational Governance, Finance, Policymaking and Management, Office of Educational Research and Improvement,U.S. Department of Education (grant number OERI-R308A60003).

Copies of this report are available for $10.00 each, prepaid. Prices include book-rate postage and handling. Make checkspayable to Trustees of the University of Pennsylvania. Sorry, we cannot accept returns, credit card orders, or purchaseorders. Sales tax is not applicable. To obtain copies, write to:

CPRE PublicationsUniversity of PennsylvaniaGraduate School of Education3440 Market Street, Suite 560Philadelphia, PA 19104-3325

Quantity discounts are available. For more information, please call (215) 573-0700.

3

Instruction, Capacity, and Improvement

David K. Cohen and Deborah Loewenberg Ball

CPRE Research Report SeriesRR-43

June 1999

Consortium for Policy Research in EducationUniversity of Pennsylvania

Graduate School of Education

© Copyright 1999 by the Consortium for Policy Research in Education

-

Instruction, Capadty, and Improvement Cohen and Ball

Contents

Biographies

Acknowledgments

Introduction 1

Instruction 2

Implications for Intervention and Research 7

Instruction and Its Environments 9

Issues and Implications for Intervention and Research 14

Intervention 17

Implications for Intervention and Research 20

Enactment 22Implications for Intervention and Research 26

Conclusion 27

References 33

End Notes 35

CPRE Research Report Series, RR-43

5

Instruction, Capacity, and Improvement Cohen and Ball

Biographies

David K. Cohen is John Dewey Collegiate Professor of Education, and Professor of Public Pol-icy at the University of Michigan. In addition to his current work on educational policy and therelationships between policy and practice, his previous research includes studies on the effects ofschooling; efforts to reform teaching; evaluation of educational experiments and large-scale in-tervention programs; and relations between research and policy.

Deborah Loewenberg Ball is professor of education at the University of Michigan. With ele-mentary school mathematics as the main context, her research focuses on the challenges ofteaching for understanding and on efforts to support such teaching through policy, reform initia-tives, and teacher education. Her publications include articles on teacher learning and teachereducation; the role of subject matter knowledge in teaching and learning to teach; challengesembedded in trying to teach for understanding; and relations of policy and practice in instruc-tional reform.

Acknowledgments

The research reported in this paper was supported by a grant (No. OERI-R308A60003) to CPREfrom the National Institute on Educational Governance, Finance, Policymaking and Manage-ment, Office of Educational Research and Improvement, U.S. Department of Education. Theviews expressed in this report are those of its authors and are not necessarily shared by the U.S.Department of Education, CPRE, or its institutional members.

CPRE Research Report Series, RR-43 iii

6


Introduction

Since World War II, efforts to im-prove schools have numbered in thethousands. Most efforts have con-centrated on improving the curricu-

lum materials used in schools or on "train-ing" teachers in new instructional methods.Many of these efforts have gone under thebanner of "building instructional capacity,"a term that for decades has been featuredprominently in conversations about educa-tional reform. Unfortunately, three decadesof research has found that only a few inter-ventions have had detectable effects on in-struction and that, when such effects are de-tected, they rarely are sustained over time.

A review of research and professional expe-rience with school improvement suggestsseveral explanations for these dishearteningfindings. One is that schools are complexsocial organizations situated within, and vi-tally affected by, other complex social sys-tems including families, communities, andprofessional and regulatory agencies. Thelarger social environment of schools con-strains and shapes the actions of teachers,students, and administrators, often in waysthat greatly complicate the work of schoolimprovement. Challenges to school im-provement are particularly acute in high-poverty settings where recruiting well-qualified teachers is difficult and where theemotional and health problems of studentsoften deflects attention to educational issuesor impedes work on them. As a result, manyresearchers now believe that school im-provement involves much more than effortsto change interactions occurring withinschools. To succeed, school improvementinterventions also must attend to the com-plex relationships that exist among inter-vention agents, schools, and their social en-vironments. A new idea which has devel-oped in the last decade is that successful

school improvement in high-poverty schoolsrequires external interventions that are capa-ble both of making large and lasting changesin instructional capacity, and doing so underconditions that rarely support and often im-pede such work.

A second explanation for the typically smalleffects of school improvement interventionsis that most are not designed to provide theopportunities for teacher learning that wouldbe needed to change classroom instruction.Even when interventions explicitly introducenew curricular materials or provide teacher"training," they rarely create adequate con-ditions for teachers to learn about or developthe knowledge, skills, and beliefs needed toenact these interventions successfully inclassrooms. For example, new materials areoften brought into schools without sufficientguidance about how they are to be used withparticular students. Teachers are rarely of-fered opportunities to learn more about ei-ther the subject matter content they are be-ing asked to teach or about how studentsthink about that content. Neither is it com-mon to provide teachers with opportunitiesfor guided practice or reflection on how newteaching strategies are working in theirclassrooms. Instructional interventions arecommonly introduced into schools withouttaking adequate account of what it wouldtake to make them work in classrooms. In-terventions are often made as though theirmere introduction could change instruction.Despite the fact that this strategy has rarelyworked, adopting new curriculum materials,for example, is one of the most widely usedinterventions.

In the last several years, a number of newinterventions' have been invented and set inmotion that are different in focus, design,and approach than previous efforts to im-prove schools. All of them envision morecomprehensive strategies for school im-

CPRE Research Report Series, RR-43 1

7


provement. For example, many focus bothon preschool as well as elementary schools.Many try to involve parents in students'academic work. Many focus on improvingprofessionals' collegial interactions, or oncoordinated improvements in the teaching ofparticular academic subjects. The interve-nors vary, from state agencies to nationalreform networks, and the interventions alsovary, from new curricula to whole schoolreform. Despite their differences, they sharesome critical features. One is that they envi-sion much more comprehensive change ef-forts than those of the past. A second is that,in one way or another, they all seek to im-prove teaching and learning, and they allfocus on students who have been poorlyserved by schools. A third is that they workfrom a position external to schools to im-prove what happens inside.2

At the same time, differences among theseinterventions highlight alternative strategiesfor comprehensive instructional improve-ment. For example, the targets of their workdiffer. Some aim chiefly at reading instruc-tion; others target schools' decision-makingprocesses; still others focus on communitiesand children's welfare; and some target cur-riculum.

These interventions represent a new courseof action in school improvement, and thisreport offers a theoretical frame for exam-ining these efforts at instructional improve-ment. Since our analysis centers on instruc-tion, we begin there: we first develop a theo-retical view of instruction and then an analy-sis of the environments of instruction.These two elements offer a basis for ana-lyzing the relations between instruction andits environments, and for representing thecircumstances in which intervenors mustoperate. We turn then to a discussion of theproblems and possibilities for intervention.

The theoretical frame that we offer is in-structional. We develop an interactivemodel of instruction to analyze both teach-ing and efforts to improve teaching. Weelaborate this split-level frame as we go,drawing out implications for understandinginstruction, instructional improvement, andresearch on the two. We conclude by sum-marizing the ideas and distinguishing themfrom other approaches to understanding andstudying school improvement.3

Instruction

Instructional capacity is prominently fea-tured in the contemporary conversationabout educational reform. This capacity iswidely regarded as critical to good teachingand learning, and capacity-building is oftendepicted as the key to better education.Though reformers have frequently aimed toimprove what students learn, most efforts toincrease learningwhich number in thethousands since World War IIhave con-centrated on a single factor: improving cur-riculum materials, training teachers in newmethods, or adding new technology. Fol-lowing this logic, reformers seem to haveassumed that increasing the instructionalcapacity of schools depends on increasingthe capacity of either teachers or the materi-als they use. There is increasing evidencethat such efforts rest on very partial concep-tions of instructional capacity.

Capacity and Interaction. We focus on theinteractions among teachers and studentsaround educational material, rather thanseeing curriculum alone or teachers alone asthe main source of instruction. On thisview, each of the three elements is essential,but instruction requires all three. Instruc-tional capacitythe capacity to produceworthwhile and substantial learningis afunction of the interaction among these ele-ments, not the sole province of any single



Figure 1. The Internal Dynamics of Instructional Units

one, such as teachers' knowledge and skill,or curriculum. We briefly discuss each ele-ment and its relation to the others.

Teachers' intellectual and personal re-sources influence instructional interactionsby shaping how teachers apprehend, inter-pret, and respond to materials and students.There is considerable evidence that teachersvary in their ability to notice, interpret, andadapt to differences among students. Im-portant teacher resources in this connectioninclude their conceptions of knowledge, un-derstanding of content, and flexibility of un-derstanding; acquaintance with students'knowledge and ability to relate to, interactwith, and learn about students; and their rep-ertoire of means to represent and extendknowledge, and to establish classroom envi-ronments. All these resources mediate howteachers shape instruction. Consequently,teachers' opportunities to develop and ex-tend their knowledge and capabilities canconsiderably affect instruction by affecting

how well teachers make use of students andmaterials.

Most discussion of capacity has focused onteachers, but much research shows that stu-dents' experiences, understandings, inter-ests, commitments, and engagement are alsocrucial to instructional capacity. One way toconsider the matter is that the resources thatstudents bring influence what teachers canaccomplish. Students bring experience,prior knowledge, and habits of mind, andthese influence how they apprehend, inter-pret, and respond to materials and teachers.The same mathematics problem used by thesame teacher may produce a substantiallydifferent lesson with a group of students atone point than it might after students learnthings that affect their approach to the task.When teachers say, "My students couldnever do that," they do not recognize theways in which students could learn orchange. Studentsand interactions amongstudentsshape the resources for their ownlearning.


9

3


By materials we mean what students areengaged in, as presented in texts and othermedia, as well as in problems, tasks, andquestions posed to students. Instructionalmaterials can mediate students' engagementwith the content to be learned, thoughsometimes the materials themselves arewhat is to be learned. They can be thoughtof as the material (as opposed to social)technologies of instruction, including print,video, and computer-based multimedia.

Curriculum is often developed in advance,but students' and teachers' interactions withthis material comprise the enactedwhich isto say, the actual or effectivecurriculum.4These material technologies influence in-structional capacity by constraining or ena-bling students' and teachers' opportunitiesto learn and teach. Features of these tech-nologies that seem likely to affect instruc-tional capacity are their complexity and thedesign of teachers' and students' intendedengagement. In the case of mathematicsmaterials, we would expect that the nature ofthe problems offered, the development ofthe ideas, the number and variety of repre-sentations, and the ways in which multiplerepresentations were coordinated wouldshape what teachers and students could doand learn.

It follows from this analysis that any givenelement of instruction shapes instructionalcapacity by the way it interacts with and in-fluences the other elements.

Capacity Not Fixed. If this last point isroughly right, then capacity is not a fixedattribute of interactions. One teacher's inter-actions with a class of fifth graders, for ex-ample, will yield greater instructional ca-pacity than those of a colleague who workswith the same class, because the first teacheris more adept at evoking and making use ofstudents' ideas. This means that speaking in

terms of what teachers or students "bring" tointeractions may be misleading, since whatstudents and teachers bring may be used tobetter or worse advantage by others. In dis-cussing what students bring to a task it isimportant to recognize that it depends in parton what teachers can see and use in students.One reason that different teachers elicit dif-ferent responses and work from the samestudents is that what teachers know, believe,and can do shapes their perceptions of whatstudents bring, the opportunities they subse-quently extend to students, and their inter-pretation of students' ensuing work.

Similarly, materials both depend on their useby students and teachers and affect such use.From one perspective, the use of readingmaterials would be shaped by the nature ofthe text they offer students, and the ap-proaches used to develop students' reading,comprehension, and interpretation. Fromanother perspective, materials are shaped bystudents' ideas and experiences. But herewe can see teachers' unique position in theconstruction of instructional capacity.Teachers' knowledge, experience, and skillsaffect the interactions of students and mate-rials in ways that neither students nor mate-rials can. That is because teachers mediateinstruction: their interpretation of educa-tional materials affects curriculum potentialand use, and their understanding of studentsaffects students' opportunities to learn.

As teachers learn new things about contentand students, they notice different thingsabout both, and are able to use them differ-ently. Change in students, teachers, or mate-rials has the potential to change the relationsof teachers, students, and materials, andhence affect instructional capacity. Butchange in teachers has unique potential, be-cause teachers mediate all relationshipswithin instruction.


1 0


Purposes, Professional Knowledge, andCapacity. Differences among interactionsare not the only source of variation in in-structional capacity. Capacity also seems todepend both on conceptions of professionalknowledge and on the aims, content, andmethods of instruction. Some uses of theterm "capacity" focus on space and storage,while others focus more on growth andchange. In the first case, capacity denotes afinite set of knowledge, skills, and commit-ments that are needed in order to producegood instruction, but in the second it denotesthe construction of new knowledge andskills in practice.

Though much instruction lies somewherebetween these two poles, they represent twoquite different conceptions of the relation-ship between knowledge and practice, andthus instructional capacity. Roughly speak-ing, the first view envisions capacity as astorehouse that contains fixed resourcesneeded for instruction. These would includeteachers' subject matter knowledge, skills,and commitments, their knowledge of stu-dents, and the content of instructional tech-nologies, among many other things.5 Fromthis vantage point, having capacity referschiefly to the extant body of teachers'knowledge and skill, the content of instruc-tional technologies, and the adaptation andapplication of that knowledge in particularsituations.6 On the second view, however,capacity is envisioned as a source and crea-tor of knowledge and skills needed for in-struction. Teachers would improve practiceby investigating teaching and learning, ei-ther in situ or in situations that derived frompractice. Rather than only drawing on ordelivering a fixed stock of knowledge,teachers would learn from practice in waysthat generated more resources for subse-quent teaching.' They would learn abouthow students think about particular ideas,how certain representations of content work,

what some common difficulties are that stu-dents encounter, and ways to mediate thosedifficulties.

The aims, content, and methods of instruc-tion also shape conceptions of capacity. Forinstance, philosophers of education and ob-servers of instruction regularly distinguishbetween teachers who cultivate students'reasoning or sense-making and those whoinculcate facts and skills. Such differencesapply to apparently different capacities.What instructional technologies would haveto contain, and what teachers would need toknow and be able to do, could vary enor-mously between these two aims. The differ-ences can be arrayed on several analyticallyand practically distinct domains. One ishow instructional technologies and teacherstreat knowledge: some take it as though itwas fixed, given, and settled, while otherstake it as open, constructed, and disputed. Asecond is how instructional technologies andteachers deal with students' thinking: someignore it, others focus on very limited chan-nels, and still others support active investi-gation of students' knowledge and ideas,using them as a central orientation for in-struction. A third is how instructional tech-nologies and teachers address and organizeclassroom discourse: some treat it as a mat-ter of one-way transmission from teachers tolearners, others as recitation, and still othersas creating conversations.8 In each of thesedimensions, instruction that was at one ex-treme would require different knowledgeand skills than instruction that was at theother; hence capacitywhat it takes to pro-duce instructionalso would vary.

Instructional aims, content, and methods in-teract with conceptions of professionalknowledge: different conceptions of profes-sional knowledge co-exist with differentmethods and purposes of instruction. Forexample, if capacity means learning in and


311


from practice, it can refer to instruction inwhich knowledge is seen either as facts andinformation or as intellectually ambitious,uncertain, and contested. In the first case,learning in and from practice would refer torelatively straightforward learning in a rela-tively bounded set of domains. Teacherswho employed Instructional Theory intoPractice would be an example of this alter-native: they would try to improve students'learning by learning in practiceadaptinginstruction to the students, monitoring stu-dents' work, and revising instruction in con-sequencebut they would do so within atype of direct instruction that focused onfacts, skills, and procedures. In the secondsort of instruction, however, learning inpractice would be much more complex; forif knowledge is interpretive, and thus oftenambiguous, it is more difficult for materialsand teachers to represent knowledge for stu-dents than if they define it as facts and skills.It would be even more difficult and complexfor teachers who held knowledge as inter-pretive and complex to try to apprehend andinterpret students' thinking.

This discussion suggests that capacity wouldin part be a variable function of the prevail-ing goals and methods of instruction and ofthe nature of professional knowledge. Thesethings would influence instructional capacitythrough the content of instructional tech-nologies and the ideas and beliefs of stu-dents and teachers.

Capacity, Time, and Mobility. Instructioncan only be examined in time. One reason isthat the interactions described above accu-mulate. Teachers and students interact overmaterials over the course of many days,hence instruction is a composite of relationsamong teachers, students, and materials overtime. Another reason is that learning occursover time. Learning to read text compe-tently, to calculate sums and differences, or

to interpret and solve problems are accom-plishments attained gradually. No singlelesson leads to achieving the main goal ofinstruction; learning occurs in fits and starts,sometimes clearly linked to instruction,sometimes less easily traced.

A third reason is that instruction and theelements that comprise it change over time.Some of this change is variability: a cross-section of instruction at one point in timecaptures only one slice of a highly variableset of slices. Materials may differ in subtleways from day to day, teacher and studentsmay interact differently, or what students orteachers bring to the topic at hand may vary.Given the unpredictability and variability ofthe interactions, it makes sense to think ofinstructional capacity in terms of centraltendencies or other attributes of streams ofinteraction, rather than a slice. However,some change in instruction is a product ofchanges in its elements or in their relationswith one another. Teachers and studentsmay learn, change their minds, or moveaway. Different materials may be used, ortheir role may shift. As they change, capac-ity also changes.

Precisely because teachers, students, andmaterials develop and change, instability iscritical. Two sources of instability may beparticularly salient in high-poverty schools:

Individual students often do not interacton a continuous basis with one teacherfor instruction in reading or mathemat-ics. Teachers may be absent, paraprofes-sionals or volunteers often are usedheavily, and low-achieving students areoften regrouped for instruction. Studentsshuttling among several instructionalsituations can threaten instructional con-tinuity and coherence.


1 26


Students often move in and out ofschools within a single year; studentmobility rates of some schools can behigher than 50 percent in one year. Suchmobility creates another version of thevariability in instructional situations dis-cussed above. It also affects instructionitself, for as students move in and out ofthe classroom, the social and intellectualresources of instruction shiftincludingsuch matters as students' relations withone another, norms of discourse, ideasand experiences available to the group,and patterns and substance of group in-teraction.

Implications for Intervention andResearch

We pause to note several implications of theanalysis sketched above.

Comprehensiveness and Intervention. Ifinstructional capacity is a property of inter-actions among teacher, students, and materi-als, then interventions are likely to be moreeffective if they target more interactionsamong more elements of instruction, ratherthan focusing on one element in isolationfrom others. Interventions that focus notonly on aspects of particular elements, butalso on their relations, are more likely toimprove capacity.

Interventions need not act directly on allelementson teachers, students, and materi-als. Those that work indirectly on all threeelements could be more effective than inter-ventions that work directly only on one ele-ment. For example, one intervention mightwork on teaching parents to improve stu-dents' motivation and to attend schoolevents. Another intervention might do thesethings and also help parents learn to workwith children on the curriculum that thechildren study. Still another intervention

might do all these things while also helpingteachers improve their knowledge of thestudents' curriculum and be more receptiveto working with parents. We expect thatchange agents of the second sort would bemore effective than those of the first sort,and that agents of the third sort would bemost effective of all, precisely because thesecond and third interventions have moreleverage on more elements of instruction.9

These points imply a new perspective on therole of curriculum in building capacity. Ef-forts to make change through materials havefrequently proved disappointing becausethey have failed to consider either teachers'or students' role in learning to use the mate-rials.1° But curriculum development couldtake account both of teachers' opportunitiesto learn and of students' likely approach tothe material. If intervenors designed materi-als to anticipate teachers' and students'needs, interpretations, and usethus attend-ing to students and teachers through the de-sign of new materialswe expect that teach-ers' and students' opportunities to learnwould increase, and that materials would beused to better effect."

Teachers' Unique Role. Interaction is cen-tral to our analysis; still teachers' knowl-edge, skills, and beliefs are distinctively im-portant for teachers interpret the materialswith which they work, and these interpreta-tions often change the aims, methods, andoutcomes of a curriculum.

One implication is that instructional capacityis partly a function of what teachers knowstudents are capable of doing and whatteachers know they are professionally capa-ble of doing with students. We do not meanthat any student can do anything a teacherbelieves he or she can or that all curriculacould be surpassingly good if only teachersused them well. Rather, we mean that every


13

7


student and curriculum is a bundle of possi-bilities, and teachers whose perceptions havebeen more finely honed to see those possi-bilities, and who know more about how totake advantage of them, will be more effec-tive. We use the term "know" what studentsare capable of doing rather than "believe,"because we suspect that this is more a matterof professional knowledge than general be-lief. Most teachers report that they believetheir students are capable of fine work, butwhat they think they know from daily expe-rience often hedges that belief with limitedexpectations.

These ideas have particular salience in high-poverty schools, where teachers tend to be-lieve that students are capable of only mod-est work, think that the students' families areto blame, and claim they can do no morethan they already are doing. These kinds ofideas about students and their abilities arelikely to mediate the implementation of anintervention that aims to improve studentperformance. But interventions that system-atically create opportunities for teachers tolearn different facts about their students maychallenge these beliefs. To do so, interve-nors would have to devise ways for teachersnot only to teach specific academic materialsdifferently, but also to see their students ascapable of performing well with them. Hardevidence that students were doing work thatteachers previously thought they could notdo might lead teachers to provide morechallenging opportunities for their students.

It follows that instructional capacity is cru-cially linked to interactions in which teach-ers and students gain knowledge about whatwork they can do. Being confronted withevidence of student learning provides teach-ers with evidence of student capability, andindirectly, of their own capability. Themore that evidence matches and exceedsconventional measures of student perform-

ance, the more convincing it will be. Forexample, mathematics assessments that in-clude measures of students' computationalfluency along with reasoning and problem-solving will be more persuasive than as-sessments that only measure elements ofmathematical knowledge and competenceuntapped by traditional assessments.

Teachers' ability to perceive and use stu-dents' capabilities is affected by their ac-quaintance with students' performance andby their understanding of the material. Be-ing able to hear what students are saying,and to see evidence of their thinking and un-derstanding, depends on teachers' knowl-edge of the terrain in which students areworking, and their knowledge of students'thinking. For example, knowing thatmathematics requires looking for patternsand making conjectures would affectwhether and how a teacher would hear astudent's comment that "the larger the num-ber on the bottom, the smaller the fraction."Knowledge of fractions would also shapethe teacher's response: that this conjecture istrue only when the numerator remains con-stant (1/5 is less than 1/3, but 3/5 is not lessthan 2/4).

A third implication of our analysis is thatsuitably designed assessment tools and cur-riculum materials can affect teachers' ideasabout what is important to teach, and how tounderstand students' ideas. A mathematicsassessment that requires students to justifytheir answers may not only offer evidenceabout students' reasoning, but also encour-age teachers to include mathematical expla-nation and reasoning as crucial instructionalgoals; similarly, developing rubrics for suchwork and scoring students' responses mayhelp teachers develop more refined ideasabout the elements of mathematical expla-nation and justification. Curriculum materi-als that unpack the subtleties of the ideas can


14

8


help teachers expand their knowledge of thesubjects they teach and thereby better equipthem to apprehend and interpret students'knowledge and performance.12

A fourth implication is that the teacher'srole in using students and materials to pro-duce instruction and learning is thereforedistinct, and distinctly prominent, in anyconception of instructional capacity. Formaterials do not use students or teachers,though they can be adapted for better orworse use by teachers and students. It istrue that students do use teachers and mate-rials, and that a part of instruction is teach-ing students to use them better. In fact,teachers who take the time self-consciouslyto help their students learn to use these ele-ments of instruction will more likely havestudents who can profit more from whatteachers say and offer. However, teachersare uniquely situated to mediate the interac-tions between and among themselves, stu-dents, and materials.

Teachers' Opportunities to Learn. Teach-ers' opportunities to learn are thereforelikely to be a crucial feature of improvinginstructional capacity. If interventions en-able teachers to change what they see in stu-dents' work, how they set tasks, and howthey interpret and deploy materials, they arelikely to have a distinctive impact on teach-ing practice and student learning.13 To doso, teachers would need opportunities thatwere rooted in specific academic content,that explored and tested out well-designedcurriculum materials for that content, andthat offered convincing information aboutstudents' thinking and performance. Suchopportunities would help teachers learnmore about their students and the materialsof instruction by grounding teachers' learn-ing in improved student performance of par-ticular content. Significant is that thesekinds of opportunities would coordinate ex-

periences with material and learners ratherthan omitting one or the other of those cru-cial elements, or leaving it to teachers toconnect separate knowledge of each."

Instruction and ItsEnvironments

Instructional interactions are situated inlarger environments, and intervenors whotry to influence instruction must work inthose environments. One cannot understandthe instructional relationships above, or in-tervenors' work in them, without under-standing the circumstances in which theyoccur and the opportunities and constraintsthey present. The environments vary withinand among systems and states in the U.S.,and they vary among national school sys-tems as well. These environments comprisepotential influences on and resources (bothpositive and negative) for instruction.Whether and how elements of these envi-ronments affect instruction, however, de-pends on how teachers, administrators, par-ents, and others interpret, respond to, anduse them.15

Schools. The most immediate environmentof instruction is the school, its departmentsand grade-level groupings. Most of the hu-man, financial, and material resources avail-able for instruction are made available byschools. The school also is the physical andsocial context within which teachers andstudents routinely interact. Schools also arethe key agent in mediating relationshipsacross classrooms and teachers and betweenthe school and other influences outside theschool.16

There is extensive evidence that U.S.schools with the very same formal structuredispose the central functions of instructionvery differently.17 A minority of schoolswork very hard to create unified instruc-


15


tional purposes and methods, while othersdo nothing of that sort, leaving it to individ-ual teachers to decide. A minority ofschools assiduously coordinate instructioninternally, but most others appear to do nosuch work, again leaving it to individualteachers to decide whether even to attend tothe matter. In a minority of schools theprincipal, who sits at the peak of the formalorganization, plays a prominent role in de-ciding instructional issues, but in many oth-ers principals leave such matters to individ-ual teachers' discretion. Some schools care-fully control the access that environmentalinfluences have within them, but most arevery porous. In such porous schools, in-struction is idiosyncratically open to influ-.ences beyond their schools' boundaries.18

This poses a great challenge for intervenorswho seek to improve instruction, in part, be-cause critical instructional resources are sooften lacking. For example, few schoolsshare common instructional purposes, andeven fewer employ instructional methodsconsistently. Standards for students' per-formance rarely are explicitly articulatedand concretely shared. Where some sucharticulation occurs, teachers still are usuallyleft to determine how best to meet commongoals and standards. Means are widely pro-claimed to be educationally distinct fromends. The idea that schools should coordi-nate instruction internallyto ensure thatstudents' opportunities to learn are coherentwithin and across grade levelsseems un-usual in schools' practice; few schools seemto have the means of establishing or sus-taining such coordination. In addition, fewprincipals enact their role as that of an in-structional leader; though many now claimsuch a role, in practice few know how to dosuch work and most leave such matters toindividual teachers' discretion. Intervenorsthat seek to improve instruction thus must

solve a host of problems, or find ways toenable schools to solve them.

System Organization. Larger environmentssurrounding schools do little either to fosterdemanding instruction or to encourage in-structional improvement. One reason is or-ganizational: the formal structure of theschool system is fragmented, which prolifer-ates interventions and messages about in-struction at many levels and from manysources. In fact, the school system is inmany respects a non-system, a host of morethan 100,000 schools situated in 15,000 in-dependent local governments, 50 state gov-ernments, hundreds of intermediate and spe-cial district governments, as well as multiplefederal agencies and countless private or-ganizations. Authority for schools is di-vided among federal, state, and local educa-tion agencies, and within those levels it isdivided among legislative, executive, andjudicial branches of government. And whileformal responsibility for schooling restswith the government, most agencies withexpertise in the core technologies of instruc-tion are in the private sector. States delegatemost responsibility for assessment to privatecorporations, and follow the same coursewith texts and other materials. Teacher edu-cation and continuing education are man-aged by a combination of colleges, universi-ties, private-sector professional entrepre-neurs, or professional standard-setting bod-ies.

Some states, school systems, and schools tryto exert a strong influence on instruction, butmost do not. A few states try to coordinatethe work of private sector agencies, but thisis unusual. The result is a general pattern ofpassivity and uncoordination, punctuated bysharply contrasting efforts at several levelsof government and in several private sectoragencies, to order instruction more ration-ally. The extraordinary fragmentation of


1610


organizations increases the probability thatmany different messages about instructionwill flow toward teachers and students fromliterally thousands of different agencies.That proliferation of different and oftencontradictory advice encourages diffusenessin the messages sent and received. From theintervenors' perspective, this means thatthey are likely to be seen as just one amongmany competing sources of advice and re-sources. If they are to help schools improveinstruction, they must find ways to maketheir agenda central to schools' work andscreen out many competing messages.

Coordination of Instruction. Despite thisfragmented organizational structure, if theagencies worked in close concert on thecentral functions of instructioncurriculum,assessment, and teacher educationguidance for instruction could be strong andconsistent. But there is little such coordina-tion, and that is another reason that the envi-ronments of instruction do little to fosterdemanding instruction or to encourage in-structional improvement. Some privatepublishers offer coordinated curricula andassessments, but many more assessments aredeveloped with little reference to curricu-lum. Professional development is usuallyunrelated to either curriculum or assessment,and requirements for teacher education andlicensure are similarly unrelated to standardsfor curriculum and assessment. Guidancefor instruction from these sources is profuse,often inconsistent, and only intermittentlysupports strong academic work. There is agreater volume of guidance in the UnitedStates than in other nations: teachers andstudents are deluged with assessments, pro-grams, policies, judicial decisions, instruc-tional materials, advice from pressuregroups, and much more. But the guidance isoften inconsistent and unclear, in part be-cause the volume of diverse advice over-loads cognitive capabilities and encourages

superficial acquaintance and misconcep-tions.19

Professional Norms. If professional norms,standards, and knowledge formation werestrong and consistent, the lack of coordi-nated guidance for instruction would be lesstroublesome for knowledgeable profession-als. Working in their particular contexts,teachers could nonetheless develop theirpractice within a frame of professionalnorms, standards, and knowledge. But pro-fessional norms are strong on individualismand weak on content, common expectations,and standards. So while most teachers' in-ventory of practical knowledge grows overtime, it typically does so idiosyncratically.The culture of teaching is individualistic,with each teacher developing his or her ownstyle, even within the same schools. Eventeachers who regularly talk with colleagueshave little concrete in common to discuss.Strong professional norms could help to so-cialize new teachers to high standards ofprofessional performance, but they rarely dofor the norms that have little to do with thecontent of teaching and learning, and insteadtend to support the notion that teaching re-quires basic technique complemented with alarge measure of personal style.

Lacking consistent social guidance or col-lective experience, knowledge for teachingin the United States arises mostly in individ-ual experience. Teachers' knowledge isnested in particulars, and they interpret andadapt in context, building ideas, habits, andpractices as they go. They primarily workalone, with their own students, and their in-terpretations and decisions are tailored to thespecifics of their situations. There is littlesense of an accumulation of practical profes-sional knowledge.

Intervenors that seek instructional improve-ment thus must not only find ways to buffer


.17

11


teachers from disparate guidance for in-struction, but also to work against the pre-vailing norms that maintain practice as anautonomous sphere of private discretion.

Resources for Professional Education.Preservice professional education couldbuild a strong foundation to help teacherscope with these problems, but it generallydoes not. Moreover, schools, districts, andstates offer teachers inadequate opportuni-ties to learn while at work. A good deal ofmoney is spent on professional education inthe United States, but most goes to sponsorinservice activities which are intellectuallysuperficial, disconnected from deep issues ofcurriculum and learning, fragmented, andnon-cumulative (Cohen and Hill, 1998; Lit-tle, 1993). Rarely do the inservice activitiesseem based on a curricular view of profes-sional development. Teachers are thought toneed "updating" rather than opportunities tolearn about curriculum, students, or teach-ing. Leadership for professional develop-ment is also scant.

One reason for this situation is that profes-sional education often promotes the messagethat each teacher has to figure out his or herown way, that teaching is as much a matterof personal expression as it is professionallystructured and normed (Huberman, 1980;Buchmann, 1993). Another reason is thatmany administrators and other specialistsare not able to offer help in instructional im-provement. Only a few of the many organi-zations involved with schooling have morethan a few specialized staff members in thecore areas of instruction: teaching, learning,students, curriculum, and assessment. Mostof these individuals are deeply committed toeducation and they work hard, but there islittle incentive for them to develop theirknowledge and skill and weak support fortheir work.

The absence of rich resources for profes-sional education, combined with the absenceof norms for common work on instruction,means that all sorts of professionals in-volved in U.S. schools lack languageanintellectual and practical vocabulary andsyntax, and knowledge underlying such lan-guagewith which to describe and analyzeteaching and learning. The absence of sucha language, and other resources for the studyand improvement of teaching, are anotherreason for instructional weakness in the en-vironments of schooling.20

As a result, when intervenors seek to im-prove instruction they must work withschools which have inadequate professionalresources. Efforts to enact schoolwidechange therefore require intervenors to de-sign ways for teachers and administrators toadapt and construct knowledge, and to de-velop language and norms of discourse thatenables concrete work on improvement.

Professional Learning. Instruction is notorganized to support learning and improve-ment. As a result there is often little com-munication among professionals aboutteaching or learning.21 To the extent thatteachers learn from their work, they learnalone, and in this sense the messages teach-ers receive in their professional preparationhave predictive validity. Rather than pre-paring teachers whose orientation is profes-sional and who expect to work with otherprofessionals, teachers are taught to see theirwork as personal and idiosyncratic. Teachereducation thus often reinforces the existingconditions of work. In addition, mostschools make little room for learning aboutteaching and learning, and other local andstate education agencies do even less.Teacher education rarely provides teacherswith common language, standards, andnorms that would join them with colleagues,linking their interpretations, judgments, and


1812


decisions. Instead, teachers are intermit-tently offered bits of theoretical and practi-cal knowledge, and left to learn mostly ontheir own. As a result, in most cases chancealone separates learning from merely "hav-ing experience."

Most comprehensive instructional interven-tions imply that teachers have a great deal tolearn, but neither instruction nor the appa-ratus of teacher education are organized tosupport such learning. The success of com-prehensive interventions thus depends onintervenors either making provision for suchlearning or finding ways for schools or otheragencies to do so.

Issue-Attention Cycle. The effects of weakand diffuse guidance for instruction arecompounded by rapid change in the instruc-tional environment. The attention spans andissue agendas of many agencies are short, sothe content of many activitiesincludingpolicymaking and school improvementtypically shifts frequently. Many electedofficials are more concerned with making amark by legislative or executive action thanbuilding programs or policies that mightbear fruit over a decade or more. Policyagendas thus shift rapidly, and policymakersoften flit from issue to issue in quick succes-sion. The culture of rapid innovation thatcharacterizes U.S. education complementsthe fickle nature of policymaking. One newfad rapidly succeeds another, curricula are"updated" on regular cycles, new vocabu-laries emerge and fade. The flush of noveltyanimates the practice landscape, but littlepermeates instruction (Cuban and Tyack,1995; Tyack and Tobin, 1994; Cohen, inpreparation). Practitioners have come toexpect that most policies and programs willfade after a few years, and have learned thatmechanical compliance or lip service is suf-ficient response. Rhetoric changes muchmore rapidly than practice.

Patterns of instructional innovation com-pound the problem. Schemes to improveteaching and learning have proliferated sincethe 1950s, with the result that, at nearly anypoint in time, American schools are awashin innovative curriculum materials, propos-als for teaching or learning, professional andstudent evaluation, ways to deal with trou-bled students, extracurricular activities, andmore. But most professionals have limitedresources with which to dig deeply into in-structional improvement, and most innova-tors adapt either by devising products thatrequire only quick and superficial attention,or by defining success in ways that arelinked to the production of products ratherthan their use. What results is a culture inwhich doing something new is valued, whiletaking instructional renovation seriouslyasa matter requiring sustained design, con-struction, and remodelingis rare.

One result of this situation has been a per-verse sort of social selection: the policiesand innovations that have the greatest appealare those least likely to produce any sub-stantial change in teaching and learning. Itis an environment in which something is al-ways new, and many things that were newlast year will soon be forgotten. Becauseagendas shift frequently, the opportunitiesfor substantial instructional improvement aremodest.

One consequence is that intervenors work inenvironments in which neither practitionersnor innovators have experience with deep orsustained work on the improvement of prac-tice. Instead there is a general expectationthat instructional improvement does not re-quire sustained effort, and school profes-sionals learn to marginalize interventions,treating them like peripheral ornamentsrather than opportunities for significantlearning and change. Intervenors who seekserious instructional improvement thus must


1913


find ways to change such expectations in theschools with which they work; failing that,they can only adapt to the schools' approachand hope that their efforts will meet a betterfate than most.

Society and Culture. A final reason that theenvironments of instruction do not fosterdemanding instruction or encourage instruc-tional improvement is social and cultural.Schooling is valued by many Americans, butthe social and economic supports for in-structional effortfrom parents' involve-ment with students' schooling to universi-ties' and business firms' attention to stu-dents' school recordsare relatively weak.22Only a minority of parents in the UnitedStates regularly spend much time with theirchildren on school assignments, compared toparents in other nations.23 In many nations,teachers are respected figures, teaching is arespectable career, learning is highly valued,and the popular culture supports hard workin school. In the United States, however,teachers are often figures of fun, teaching isnot a respected profession, and the popularculture features money, glamour, sex, andviolence more prominently than learning.

Business firms and institutions of highereducation send mixed signals about thevalue of hard work and high achievement inschool. Most colleges and some firms sendstrong messages about the value of hardwork and high achievement to a modest mi-nority of academically ambitious studentsand their teachers. But most students andteachers receive rather weak messages aboutthe value of academic achievement. Rela-tively weak social support for instructionaleffort reinforces rather than offsets theschool system's academic weaknesses.

Intervenors working in this environmentmust find ways to buffer the schools withwhich they work against messages from

popular culture and other agencies, to createa culture of hard work and serious academiceffort, and to attach or call attention to re-wards for school success.

The environments in which practice andpolicy intersect lack many of the resourcesthat would support fruitful interaction, in-cluding a shared language of professionaldiscourse, traditions of common work onteaching and learning, opportunities for pro-fessional learning, social and economic sup-port for demanding instruction, and profes-sional norms and incentives that support im-provement. Instead these environments aremarked by rapidly changing policy agendas,diffuse and often divergent guidance for in-struction, deep disagreement about the endsand means of schooling, and inconsistentsupport for instructional improvement.

Issues and Implications forIntervention and Research

If intervenors cannot rely on existing ar-rangements for consistent support in im-proving instruction, they must devise meansto solve such central instructional problemsas setting goals, devising means of coordi-nation among units of instruction, decidingupon and enacting methods of instruction.24

Intervenors that seek serious instructionalimprovement therefore face a dilemma. Ifthey can compensate for these environ-mental conditions they are likely to increasethe probabilities of success. But such workis time-consuming, expensive, and difficultto accomplish. Alternatively, intervenorscan try to use existing environmental condi-tions in ways that might bring some successin schools. This approach is likely to be lesstime-consuming and expensive, but its suc-cess in implementation is riskier.


0

14


Many interventions try to improve instruc-tion from the outside-in: they focus on for-mal organization, resource allocation, orregulation of instruction.25 Examples in-clude state standards-based reforms; state,local, or national accountability schemeskeyed to assessment results; state and localcurriculum regulation and resource alloca-tion schemes; and various organizationalreforms, including restructuring, decentrali-zation, site-based management, charterschools, and choice. But the intended in-structional effects are only likely to occur ifintervenors are able to carry the interventioninto the central instructional functions men-tioned above. For example, teachers'learning how to use improved curriculum isnot typically part of standards-based reform,school restructuring, and other interventions,but is needed to connect broad standards andassessments to the central functions of in-struction. Lacking those connections, en-actment is likely to repeat typical patterns ofhighly variable and frequently superficialeffects on practice and learning.

The recent Kentucky and Texas reforms of-fer useful cases in point. Both are versionsof standards-based reform, and both linkstatewide educational standards to annualtests, and schools' test performance to re-wards and punishments. Both are regulatoryreforms instituted at the state level, but thereare significant differences between them.The Texas reforms include assessments atevery grade, while Kentucky assesses stu-dents at only three grades. The Texas as-sessments appear to be a criterion-referenced test with each grade-level as-sessment embodying a set of desiredachievement outcomes for that grade; theKentucky assessments offer a more complexarray of outcomes and assessment formatsand are difficult to interpret.26 The Texasassessment instruments are released toteachers and the public on the assumption

that if the outcomes are desirable it makessense to teach to them while the Kentuckyassessments are secured.

Texas offers more opportunities to connectthe broad structural elements with the cen-tral instructional functions of schools.Texas teachers and families get evidence onstudent and school performance every gradeand every year, while Kentucky teachers andfamilies receive evidence on three grades.Publication of the Texas assessment instru-ments creates opportunities for Texas teach-ers and parents to teach to the test, which isto say, connect the assessment to curricu-lum. In Kentucky, test security and the lackof means to connect assessments and class-room work have made it difficult for teach-ers to make connections among assessments,standards, curriculum, and instruction. Weare not touting the Texas system, but notingthat even two such relatively crude regula-tory schemes can differ greatly in the extentto which a broad framework for regulatingeducation is carried into the central func-tions of instruction. The more complete thecarry-through, the greater the chance ofchange in teaching and learning.

Other interventions try to improve instruc-tion from the inside-out: they focus on ele-ments of instruction or closely related mat-ters, including professional development,new technologies, family involvement, newcurriculum, and improved pre-serviceteacher education. Rather than dealing withformal organization, regulation, and re-source allocation, these efforts occur in thevery domain that the outside-in approachestypically ignore. But these inside-out effortstypically focus on only one element in iso-lation from the others. If the central prob-lem of outside-in intervention is makingconnections to instruction, the central prob-lem of inside-out intervention is coordina-tion among the central instructional func-


2115


tions, and coordination across classrooms orschools. Examples of such coordination in-clude: setting common goals for instructionwithin or among classrooms, or amongschools; and coordination of instruction it-self, assuring that students and teachers dealwith the same or similar material withingrades, and that student work on successivegrades builds on rather than repeating or ig-noring what went before. Without the meansto set common instructional goals or to co-ordinate among such elements as instruc-tional methods, curriculum content, infor-mation about students' performance, andteachers' opportunities to learn, enactment islikely to be both variable and superficial.

The roles of professional development, cur-riculum, and assessment in the recent Cali-fornia mathematics reforms are illuminat-ing.27 Beginning in 1985, state officials andmath educators in California made efforts toalign math curriculum and teaching withdisciplinary knowledge and to pay more at-tention to students' thinking about mathe-matics. To support these ambitions, newcurriculum was written, professional devel-opment opportunities were created, and newassessments were devised. Research onthese efforts reveals that, like most other in-terventions in the core of instruction, theseinitiatives were typically carried out in anisolated and uncoordinated fashion. For ex-ample, many elementary school teachers hadsome professional development in connec-tion with the reforms, but most of it wasgeneral, not focused on curriculum for stu-dents, and not affording teachers conse-quential opportunities to learn about mathe-matics, math curriculum, math teaching, orstudents' mathematics learning.28 There wasno evidence that these professional devel-opment experiences had any consequentialeffect on teachers' practice. Similarly, mostteachers had few or no opportunities to learnabout new curricula that was consistent with

the reform effort. They would be poorlyinformed users if new materials showed upin their school or classroom. And, althoughthe state education department wrote newmathematics assessments that were consis-tent with the new frameworks, only a smallfraction of teachers had opportunities tolearn about the sort of work students did onthem, or to use that work to deepen theirknowledge of mathematics and how studentsthink about it.

Most California teachers lacked key ele-ments of instructional coordination: knowl-edge of what curriculum to teach; what de-sired student work would look like; opera-tional goals for instruction; and opportuni-ties to learn about such matters. Researchshowed that neither teachers' practice norstudents' achievement changed discerniblyfor such teachers, a result familiar to reportsof uncoordinated and superficial enactmentin many earlier studies of innovation. Butresearch in California also showed that aminority of teachers had opportunities tostudy the sort of curriculum advocated bythe state mathematics reforms. The cur-riculum of teachers' professional develop-ment was the curriculum that students wouldstudy: teachers had opportunities to learnabout the purposes and methods of the cur-riculum, and to consider how it might betaught and learned. Some teachers whoscored state assessments were also exposedto and had some opportunities to learn aboutthe mathematical work state officials wantedstudents to do. The curriculum for teachers'professional development in these scoringworkshops was the students' mathematicalwork; teachers had opportunities to learnabout the nature of the state assessment andthe curriculum implied in it, and about howstudents dealt with the tasks that were pre-sented in the state's new assessments.


ri216


In effect, a small minority of Californiateachers had occasion to coordinate cur-riculum with teaching, assessment withlearning and teaching, and professional de-velopment with both. Such opportunitiesare entirely atypical of instructional inter-ventions, but they were consequential inCalifornia: research showed that teacherswho had such learning opportunities appearto have revised their goals for math instruc-tion. That is not surprising, for there was aconsistent relationship between and among acurriculum for professionals' learning thatwas tied to the intervention, the purposes ofthe intervention, assessment and teachers'knowledge of assessment, and the studentcurriculum. Though only a small minorityof teachers had opportunities to create thisinternal coordination, it seems to havecounted for students: schools having moreteachers with such coordinated knowledgehad students with higher math scores on thestate assessments.

These points suggest that, given the uniquenature of instructional environments in theUnited States, effective intervention seemsto require more than managing the innova-tion. Intervenors may also have to findways to create alternative instructional envi-ronments that buffer many features of theexisting environments. Lacking that, inter-venors can try to encourage enactors tomanage the environment, or try to frame theintervention in ways that would enable it tosurvive. From an analytic perspective, then,the instructional environment is not simply a"context," a backdrop against which actionoccurs. It is also part of the content, becausethe content and process of instruction de-pends on its environments and because in-tervenors' and enactors' management of in-structional environments will be crucial fortheir success.

Intervention

From a perspective of practice, the previousdiscussion implies that intervention includesextensive work on two fronts: reconfiguringinstruction and its environments. Therewould be important costs to such work, for itexpands intervenors' management agenda,complicates problems of quality control, andincreases uncertainty. Every interventioncan be interpreted as an effort to find an ef-fective balance between comprehensivenessand manageability.

We expect that interventions will be morelikely to succeed the better they deploy theelements conventionally associated with in-struction. These elements include a teacheror teaching agent; actively engaged learners;a curriculum of intervention; framing theteaching and curriculum in light of an un-derstanding of the learner-enactors and whatthey bring (because that will affect interpre-tation and enactment); opportunities to learn,practice, revise, and reflect; examples ofsuccessful performance; support from otheragents in the immediate environment; andmore. Each of these elements can be seen asa set of influences on enactment, and inter-vention strategies can be interpreted as dif-ferent configurations of those sets.

Deploying these elements is more easily saidthan done. Some of the conditions refer tomatters that are or could be relatively wellunder the control of intervenors, including:a teacher or teaching agent; a curriculum ofintervention; framing the teaching and cur-riculum in light of understanding thelearner-enactors and what they bring; andexamples of successful performance. Wediscuss these below in terms of interventiondesign, specification, and development.


2317


But other conditions are relatively less wellunder intervenors' control, because they de-pend jointly on enactors, intervenors, andthe instructional environment in which theyboth operate. These include: engaged learn-ers; opportunities to learn, practice, revise,and reflect; and support from other agents inthe immediate environment. Despite theirrelative lack of control in these matters,managing them is a requirement of success-ful intervention. We discuss these later inconnection with enactors.

Design refers to both strategy (what interve-nors include as a target of their efforts), andtheir conception of improvement (the visionof the improvement process that intervenorsembrace and employ in work with enactors).Design bears on the "intended" intervention,that is, the intervention as it is conceived byintervenors. One element of design con-cerns the specific target of an intervention.Does it aim at all the elements of instructionor just one or two? Does it also take accountof the instructional environment? We ex-pect that intervention designs which aremore comprehensive and more focused oninstruction are likely to be more effective inchanging instruction.

The content of comprehensiveness dependspartly on assumptions about the purposesand means of school improvement. The keydistinction here concerns approaches to ca-pacity-building. Some intervenors empha-size the achievement of a given type or levelof instruction by providing teachers and oth-ers a defined body of knowledge and skill,while others emphasize helping teachers andothers learn to develop resources for im-provement by constructing knowledge inand from practice. Most interventions in-corporate some elements of both, of course,but the balance varies considerably amonginterventions, and those differences influ-ence the process of capacity-building. To-

ward one end of the continuum, instructionis improved by developing better ideas,practices, and material technologies andgiving these to teachers to implement. Tobuild capacity is to provide new technolo-gies, or to provide teachers with new knowl-edge and techniques which they can thenapply. On the other end of the same contin-uum, change agents help practitioners learnhow to improve instruction as part of theirpractice. Teachers then are able to developprofessional knowledge and skills, includingunderstanding of subject matter and knowl-edge of students, learning, and pedagogy, asthey learn in and from practice. Teacherswho work in this way are able to play a dif-ferent role in improving instructional capac-ity than those who receive and implementintervenors' or other experts' knowledge.Practice itself becomes a site in which togenerate knowledge for improving instruc-tion. Intervenors help teachers become boththe agents of instruction and the agents ofinstructional improvement.

The two approaches are likely to have dif-ferent benefits and costs. The first, althoughnot easy to accomplish, reduces the earlyand middle-run costs of specifying and de-veloping an intervention and is more likelyto produce interventions that many schoolscan adopt and enact, and that will producepositive change in many schools. But en-actors may have something like algorithmicknowledge and skills: they will be able to dowhat they have been taught, but may havedifficulty developing and improving in-struction beyond that. The second approachincreases the early and middle-run costs ofspecifying and developing an interventionand is less likely either to produce interven-tions that many schools could enact or toproduce positive change in many schools.But if schools did enact such things wellthey would be more likely to have deeperand more supple knowledge of instruction.


2 4

18


They would have a much more difficult timelearning to improve instruction, but thosethat get beyond that will be more likely todevelop further and improve instruction.

Instructional interventions are not mere de-signs. Each is accompanied by some effortto define, explain, interpret, and developtheir meaning and entailments for action.Some interventions have relatively littlesuch stuff; in our terms they are specifiedand developed in rather spare form. Othersinclude quite a bit of this, and in our termsthey are more fully specified and devel-oped.29 We expect that the sophisticationwith which interventions are specified anddeveloped will have a considerable effect ontheir implementation.3°

Specification refers to the explicitness withwhich an intervention is articulated andmappedto the plans for action, includingwhat the intervenor chooses to treat explic-itly and how. Hence we refer to the plans oreducational blueprints for intervention, in-cluding plans for a curriculum for enactors'learning, plans to collect and use evidenceon enactment, and much more. Historically,many interventions consisted chiefly ofstatements of principles and goals that sug-gested a general direction but little more;some deliberately restricted themselves tosuch on principle or for fear of inhibitingenactors. Specification in such cases wasspare.

Only a few recent intervenors offer much inthe way of specification. An example wouldbe setting out the terms of a curriculum forenactors' learning (what teachers and othersinvolved in carrying out the interventionwould need to learn and where or how suchlearning might be profitably arranged), andsetting out the content of that curriculumwould be further specification. Althoughspecification could mean detailed blueprints

for an intervention, it need not. For the na-ture of appropriate specification of an inter-vention would depend on its designthat is,its purpose and content. For instance, thecontent of an intellectually ambitious cur-riculum could be specified in terms of itsobjectives, its main themes, the types of ac-tivities entailed, and examples of the sort ofwork that should result along with examplesof inadequate work. Such specificationwould offer users a great deal of guidancewithout constraining their choice of materi-als, the order of operations, or other things.Specification need not impede practitioners'autonomy or inventiveness; like any materi-als for learning, specification of an inter-vention can either open up or constrain op-portunities to invent and exercise autonomy.

Analysis of specification would also includewhat is specifiedthat is, what do interve-nors spell out in detail and what do theyleave to local or individual wishes? For ex-ample, intervenors often provide explicitdetail about the curriculum to be used withstudents, but leave unspecified how teachersmight learn this new curriculum themselves.What interventions specify and how they doso will affect their role and influence in di-recting and encouraging learning, managingchange, and building instructional capacity.

Development refers to the action repertoireof interventions, including materials for en-actors; social processes such as professionaldevelopment; working models or examplesof adoption processes, or video materialsthat depict teachers' knowledge, norms, andskills in ways that would be educative forother teachers; and social processes for in-volving and educating parents.31 If thespecification of interventions is analogous toplanning for instruction, then development isanalogous to creating materials, occasionsfor instruction, and processes that wouldprovoke and support the learning thus


25APST rnpv AIM IC

19


planned.32 The effects of specification seemlikely to operate chiefly through processesof development, rather than independently.

Development refers to creating the organ-izational, social, and intellectual resourcesrequired to enact interventions. Buildingcurricula for enactors' learning is an exam-ple of relatively strong development; simplymentioning such a curriculum would be anexample of weak specification and no de-velopment; ignoring the matter entirelywould be weaker still.33 As with specifica-tion, development can be accomplished inways that offer extensive guidance withoutconstraining invention and autonomy, or inways that are quite constraining.

Complex social innovations offer manychoices for development, and which ele-ments of an intervention are developed is asimportant as the richness with which theyare developed. For example, one importantelement in development is whether it in-cludes a role for teaching in the intervention;another element is whether resources areidentified for those who would act as teach-ers for enactors. As in curriculum for chil-dren, the critical issue is the delineation ofthe teacher's role and the development ofassociated guidance for teachers, not justwhat opportunities are created for students.Another element is the extent to which cur-ricula for enactors' learning attends to thecommunity in which enactors would beworking and trying to learn. For example,interventions could be developed in waysthat provide resources for enactors to workcollectively, to share information, to encour-age and support greater interdependence ofeffort, or in ways that provide no such re-sources and assume that enactors wouldlearn and work alone.


We pause again to derive several pointsfrom the frame sketched above concerningboth the design of interventions and researchon them.

In order to have any hope of success, inter-venors must create designs for intervention,write some sort of specifications, and de-velop materials and processes that will helpenactors learn in light of those designs andspecifications. Such designs, specifications,materials, and processes for enactment arewhat educators will need as they try to im-prove teaching and learning. The more fullyintervenors design, specify, and develop in-terventions, the more guidance they offer toenactors, and thus the greater the probabilityof success for those enactors. Less is left forenactors to invent on their own. But suchinterventions may not always be the bestchoice, for the more fully intervenors de-sign, specify, and develop interventions, thegreater the intervenors' costs. Furthermore,many enactors may be averse to more fullydesigned, specified, and developed inter-ventions, either because the explicit designenables them to see that they do not like as-pects of the intervention or because they areaverse to such seemingly prescribed work.

There is a big trade-off between two ideal-typical approaches to intervention. Themore intensive types reduce uncertainty anddefine areas for rational action through care-ful design, specification, and development,which raise the costs of intervention. Themore conservative types moderate costs withmore loosely designed, specified, and devel-oped interventions; these allow enactorsmore autonomy to tailor interventions totheir own purposes and context, but alsothey increase uncertainty and reduce areas ofbounded rationality.


26

Instruction, Capacity, and Improvement Cohen and BaH

Intervenors who take the first approach usecareful design, specification, and develop-ment to detail enactment. They use knowl-edge of enactors to design the interventionso that it is as usable as possible. If interve-nors adopt this approach they assume exten-sive responsibility for teaching the interven-tion and seeing that it is learned. But howintervenors enact an intensive approach de-pends partly on the nature of the interven-tion. One version, for interventions that aresusceptible to direct instruction, is to detailenactors' learning and action. Intervenorswho take this approach do something analo-gous to curriculum developers who seek todevise curriculum for independent learn-ingthat is, they write programmed materi-als that closely specify what is to be learnedand the processes for learning it. The idea isthat learners will acquire the skills and un-derstanding that they need by following thecurriculum as written. At the extreme, suchprogrammed learning materials usuallyleave little or nothing to be developed by thelearner: the steps, the materials, the cogni-tive and other processes, the monitoring, areall incorporated.34 The aim is to reducevariability at every stage of design and en-actment.

Another version of this approach, for inter-ventions that are less susceptible to directinstruction, is to specify and develop materi-als and processes that enable intervenors tooffer rich experiences in which enactors canre-invent the intervention, but do so in waysthat satisfy openness to enactors' construc-tions and fidelity to the design. If the firstapproach is a version of programmed in-struction, this second is a version of guideddiscovery. There is elaborate design andrich specification and development in bothcases, and intervenors and enactors wouldhave close relationships in both. But thenature of these things would vary with thepurposes and content of the intervention.35

Enacting the conservative approach sketchedabove can vary appreciably. Intervenorscould design and specify the intervention byenunciating only principles and leave nearlyeverything else in the way of specificationand development to enactors. Some inter-venors take this path on the assumption thatthe nature of the intervention and the com-plexities of enactment would be violated bydetailed specification. A contrary approachis to design, specify, and perhaps even de-velop the intervention in detail, but to dele-gate enactment entirely to those who wouldadopt it. In either of these more conserva-tive approaches, intervenors limit their re-sponsibility to making some knowledge ofthe intervention available. Like manyteachers, they "put it out there" and let thosewho can "get it" do so, while those whocannot, do not. Intervenors do not acceptresponsibility either for enactors' learning orfor extensive curriculum and instructionresponsibilities which intervenors who takea more intensive approach actively embrace.The more conservative approaches allowintervenors and enactors to have a looserrelationship, and encourage local adaptationand invention.

Few intervenors design, specify, and de-velop interventions in one of the extremeways sketched above, but in one of the manydifferent.degrees that lie between the ex-tremes. One way is to shade design, specifi-cation, and development in a coordinatedway, in one of the many degrees that sepa-rate intensive from conservative approaches.Another way is to mix and match within anintervention, designing some elements in anintensive manner and others more conserva-tively. Purely intensive or conservative ap-proaches to intervention are unlikely. Thequestion for every intervenor is where theywill position the intervention: how muchintensive work can they afford to do, and


2 721


how much can they afford to leave to enac-tors?

Intervenors must manage the trade-offs be-tween reducing uncertainty and boundingareas for rational action through careful de-sign, specification and development on theone hand, and on the other hand moderatingthe costs and associated difficulties of inter-vention by requiring less design, specifica-tion and development, by allowing enactorsmore autonomy to adapt the intervention totheir purposes and situations, which also in-creases uncertainty and reduces areas ofbounded rationality.36 In the first approach,specification and development increase theopportunities for intervenors to teach andenactors to learn how to manage change andbuild instructional capacity, and they refer toextensive designs, materials, and processes.But such specification and development en-tail steep costs, hence some intervenors mayleave many particulars of the intervention toenactors to decide and invent locally.

This analysis implies that there exist differ-ent views of specification and development,some better suited to more conservative in-tervention and open adoption, and othersbetter suited to more intensive work. Thevery meaning of the integrity or fidelity ofenactment for those having a conservativeview could be very different from thosehaving a more intensive view of interven-tion. Specification and development couldrefer to loosely and weakly defined designs,materials, and processes. These differencesmay be related to the purposes and contentof intervention. An intervenor whose designfor change is premised on a behaviorist viewof learning is likely to have a very differentconception of specification and developmentthan an intervenor who adopts a radical con-structivist view of learning. Not only wouldspecification and development be more or

less salient, they might be salient in differentways, depending on intervenors' designs.

While we expect that the specification anddevelopment of an intervention will signifi-cantly affect its enactment, it will be impor-tant to relate the meaning of these terms tointervenors' strategies. For intervenorsseem to have different strategies. Some maysee their work as a matter of beginning tofamiliarize broad segments of Americaneducation with new ideas which, if success-ful, would take generations to develop.Others may strive to create models that canbe widely enacted within a decade or two.Such strategic differences imply varyingconceptions of what enactment means. Ifso, intervenors may hold different views onthe yield of their work. Some may see rela-tively weakly implemented versions of anintervention in many sites as preferable todeep implementation of a much smallernumber of versions in a few sites, becausethey envision a slow process of diffusion.While depth and fidelity would be the chiefcriteria of impact in some strategies theywould be less appropriate for others, at leastin the short run.

Enactment

Everything in the preceding section focusedon intervention, as if the only salient causalrelationships ran from intervenors to class-rooms, through instructional environments.But most intervenors deal with schools, andthey work in mostly voluntary relationships,so that all intervenors depend on the schoolsthat they wish to improve. In the case ofgovernment intervenors, the most importantsource of voluntarism is decentralizedschool governance, which leaves state andfederal agencies with less authority and in-fluence than localities in most educationaldomains, and thus politically dependent foracceptance of interventions by local educa-


2822


tional agencies. The ostensible objects ofpolicies and programs have enormous influ-ence over both higher-level executive andlegislative designs and enactment, whichcreates incentives for intervenors to respectenactors' wishes.

For non-government intervenors, the mostimportant source of voluntarism is that theycan only work with enlistees who choose todo so, and who therefore have a good deal ofindependence. Intervenors have no formalrelationship with schools or school systemsother than those that enactors choose. Thatcould create incentives for intervenors torespect enlistees' wishes, but intervenorsmight instead define the process of enlist-ment in ways that deliver only enlistees whowish to improve and seem eager to do it asproposed by the intervenor. So intervenorswould work with enactors who were deeplycommitted to the intervention and to high-fidelity enactment. From the perspective ofmost intervenors, those would be ideal en-actors. But whatever intervenors and enac-tors do, their mutual definition and man-agement of the process of enlistment islikely to be central, both to the formation ofintervenors' strategies and to enactors' re-sponse. 37

The forms of voluntarism and compulsionsomewhat belie the realities. Relationshipsthat appear to be compulsory in state andfederal policies have very strong elements oflocal autonomy and choice. And relation-ships that appear to be entirely voluntarymay have strong elements of obligation pre-cisely because they arise from mutualchoice.

A further and powerful source of volunta-rism for intervenors in and outside of gov-ernment arises from the nature of instructionitself. We pointed out earlier that interve-nors can offer resources of various sorts, in-

eluding inducements, money, education, andthe like, but such resources are used or not,and used well or poorly, by enactorsthat is,teachers, students, and others in their vicin-ity. Some researchers and reformers regardsuch resources and outside agencies ascauses of instruction, however, we treatthem as resources. They can becomecausesthat is, they can influence prac-ticeonly as teachers and students recog-nize, accept, and enact them. Such re-sources, including entire interventions, arepotential; only teachers and students can usethem, and they can use them only as they areable and inclined. Intervenors can onlydrive if enactors drive as well, and share thedriver's seat with them.

Against this background we see increasingnumbers of schools which have either strongincentives or explicit obligations to improve.In Kentucky, San Francisco, Chicago, NewYork, and elsewhere, the weakest perform-ing schools have been placed in forms ofreceivership, "reconstitution," or "crisis."There are strong elements of voluntarism inthe relations between intervenors and enac-tors, even in these seemingly extreme cases.One reason is that some compulsoryschemes have elements of local choice:many reconstituting schools are givenchoice among interventions, and someinKentucky, for exampleare encouraged tocompose their own remedies. Another rea-son is that state agencies have little capacityin a decentralized system to actually inter-vene in many schools; hence they tend towork with small numbers of the very worstschools, leaving most poorly performingschools to their own devices:58 Mutualchoice may be a stronger source of obliga-tion than compulsion in all but the most ex-treme cases.

In any case, our analysis implies that inter-venors must solve a central problem in


29


framing and managing their relations withenactors: will intervenors define improve-ment and the processes leading to it in waysthat center most responsibility in enactors orin themselves? Choosing the former coursewill create incentives to define tight rela-tions with enactors, to carefully select en-actors, and to define improvement and theprocesses leading to it quite explicitly.Choosing the latter course will create incen-tives to define loose relations with enactors,to admit enactors relatively freely, and todefine improvement and the processesleading to it in a more relaxed fashion.

Our analysis also implies that enactmententails a complex set of relationships be-tween intervenors and enactors, between en-actors and the environment, and betweenintervenors and the environment. Interve-nors and enactors need to manage all ele-ments of these relationships. Enactorsshould be understood as managing interven-tions and intervenors, just as intervenors areunderstood as managing interventions andenactors.

The ways in which intervenors and enactorsmanage each other will depend partly onhow intense or relaxed the interventionstrategy, but either approach implies con-tinuing communication and exchange of re-sources. And since their mutual manage-ment occurs in instructional environmentsthat offer few supports for extensive or en-riching instructional relationships, interve-nors and enactors do not begin by workingin anything like the stable relationships be-tween most teachers and students. Thecreation of stable relationshipsin whichcommitments and resources can be regularlyexchanged between intervenors and enac-tors, and in which they can conduct otherimportant transactionsturns out to be acentral task of intervention.39

If intervenors offer highly specified and de-veloped interventions, they will be morelikely to build extensive management sys-tems to support the intense relations withenactors that such interventions require.These management systems must be addedinto the costs and benefits of different inter-vention strategies; their existence will in-crease the probability of high-fidelity en-actment, but will add to the human and fis-cal costs of intervenors' work. Such man-agement systems also may raise the ante forenactors because they increase surveillanceand quality control.4°

Solving these problems of mutual choiceand continuing relations is further compli-cated by intervenors' and enactors' organ-izational settings. Most intervenors appearto be temporary: their funding is soft, oftenquite speculative, and subject to change infoundation and government agendas. Inter-venors' relations with recruits are voluntary,and they appear as an organization that mostschool professionals have seen come and gomany times. In contrast, enactors work inpublic schools that appear more permanent:schools are part of governments that typi-cally have existed for decades or more.Schools have relatively steady budgets thatare settled in apparently stable state and lo-cal budget lines. Despite intermittent fund-ing problems, schools' core budgets do notneed to be raised anew every few years.Although some schools close, some studentsdisappear, some parents are unhappy, andsome budgets shrink, these problems havenot been normative for the enterprise. Sev-eral decades of opinion research show thatthe overwhelming majority of parents arequite satisfied with their local school.

There are few strong incentives, as thissketch implies, for potential enactors to im-prove instruction. Therefore, potential en-actors have few strong incentives to enter


3024


into intense relationships with intervenors.Non-government intervenors operate inmarkets, and must find and satisfy clients,while public schools operate in the grantseconomy where few must search for clients.Government intervenors operate in a climateof political voluntarism, in which the objectsof policies and programs have great influ-ence, both over higher-level executive andlegislative designs and over enactment. Inboth cases there are strong incentives forintervenors to make connections with enac-tors that will not displease them and that willnot require either the termination of manyrelationships or repeated searches for newrecruits.

One implication of this analysis is that, fromthe enactors' perspective, the default optionhas been to marginalizethat is, to use theresources that interventions bring while re-taining the core of operations unchanged.41This strategy requires the least learning andchange and has worked very well in U.S.education.42 Marginalizing interventionskeeps the ratio between costs and benefits ofenactment in a manageable range and en-ables enactors to innovate and improvewithout basic change.43 In the typical U.S.instructional environment, it is entirely ra-tional. From the typical enactor's perspec-tive, the preferred relationship with interve-nors keeps things loose and permissive, withintervenors supplying resources and enactorsdeciding how to use them. Such a conser-vative approach could include some regularrelationsannual meetings, occasional vis-its, newsletters, and other informal commu-nicationsbut it would not include efforts tocheck on the fidelity of enactment. Interve-nors and enactors would be linked in a loosebut mutually beneficial network in whichknowledge, skills, and other resources couldbe exchanged without greatly taxing any-one's resources.

To have any hope of success, intervenorsmust devise intervention strategies to dealwith this situation. One strategy is to spec-ify interventions loosely and require onlyintermittent interaction with enactors; thatwould conform to extant preferences andpractice. Another strategy is to focus inter-vention on a select group of relatively high-achieving enactors for whom the problem ismore easily solved. Still another strategy,for those who contemplate intensive workaround carefully specified and developedinterventions, is to devise ways to attractenactors who would accept stable and tightrelations with the intervenor, fundamentalchange and extensive learning. Anotherstrategy still is to routinize large elements ofthe intervention so as to reduce costs to in-tervenors and uncertainty for enactors. Thereare other alternatives, including combina-tions of the strategies mentioned here.

An intensive intervenor-enactor relationshipis likely to require management of the in-structional environment. And the more ac-tive and complex that relationship, the closerenactors and intervenors would be to thecreation of new sub-systems of schooling.Though these sub-systems would be tempo-rary, unofficial, and nongovernmental, theycould serve several purposes: to screen outcompeting instructional guidance, to helpstabilize funding and political agendas, andto create some measure of professionalcommunity. But the prospect of such tightrelationships is likely to be unsettling tomany schools and the districts that sponsorthem. Intervenors that propose more con-servative and relaxed relations with enactorswould ease such worries by allowing enac-tors to cope with the instructional environ-ment themselves, and thus allowing inter-pretation of the intervention to vary.

Our discussion implies that incentives forschool improvement are a critical variable in


3


the relations between intervenors and enac-tors. The more that potential enactors areunder internal or external pressure to im-prove instruction, the greater the probabilitythat they will find it useful to accept interve-nors' specifications and close relationshipswith them. The more that potential enactorsare under pressure to demonstrate results,the more likely they are to accept highlyspecified and developed interventions whichpromise to produce such results. But to saythat incentives are important is not to saywhich incentives would be best. Manycommentators favor external incentives tiedto state or local accountability systems, butit is not yet clear whether these schemes ac-tually mobilize much improvement effort."Others favor incentives associated with localadministrative pressure and support for pro-fessional performance, but such endeavorsare few and far between, and there is littleevidence on their effects.


We pause once more to set out several im-plications for the analysis of both interven-tion and research on the relations betweenintervenors and enactors.

If we were to order the ideas about buildingcapacity discussed above in a formal causalmodel, we would represent intervenors' in-fluence on enactors as partly constructed bythose enactors. Causal arrows would run inboth directions between external agents andinfluences on one hand, and enactors on theother.45 This bi-directionality would applyto resources in the immediate instructionalenvironment as well as to more distant po-tential influences.

One critically important element of instruc-tional capacity is enactors' ability and dispo-sition to notice instructional problems, and

the ability and disposition to recognize anduse the resources that would be likely tohelp solve those problems. Yet when inter-venors work in poorly-performing, high-poverty schools, almost by definition theywork in situations in which those abilitiesand dispositions are weakly developed atbest. In this case, the people who have theproblem are the key agents for solving it, yetthe problem inhibits both their recognitionof it and their adoption of problem solvingstrategies. External intervenors cannot hopeto succeed unless enactors recognize prob-lems and adopt problem-solving strategies,but reaching that point would take greatchange for many enactors. Those in themost difficult circumstances are likely to befurthest from the required recognition and towork in the most troubled environments.46

Intervenors have several strategic alterna-tives in dealing with this situation:

Routinize interventions to ease the mag-nitude of the "recognizing problems andlearning to solve them" task for enactors,and the corresponding teaching task forintervenors.

Select only relatively apt enactors whoalready are well on the route to recog-nizing problems and learning to solvethem. This could be a matter of restrict-ing enlistment to entire schools based oncollective decision-making, or openingenlistment to individual teachers withinschools, or both restricting enlistment toentire schools and allowing dissentingteachers to opt out.47

Instruct enactors so that they learn andchangethat is, successively reconstructtheir initial dispositions and abilities insome instructional process.


32

26


Define intervention and enactment soloosely as to allow many flowers tobloom, thus allowing enactors to definetheir own levels of effort and imple-mentation, and the things they will learn.

Some combination of the first threestrategies noted here.

Building capacity can, like many otherschooling problems, be treated as a problemof selection alone, of learning alone, ofcombined selection and learning, or oftreatment definition. Recent interventionshandle this matter quite differently.

Conclusion

This report offers a significantly differentapproach to understanding instructional im-provement. We summarize the distinctiveelements and advantages of our approachand the implications for instructionalchange.

Capacity. First, we see capacity as specific.Most discussions of instructional capacityseem to assume that instructional capacity isgeneric, but our analysis suggests that ca-pacity always refers to what it takes to dosome specific thing. If so, the nature of ca-pacity varies with the direct object of thephrase ". . .capacity to. . ." In consideringinstructional capacity, then, one must recog-nize that instruction varies greatlyfor ex-ample, between teachers who cultivate stu-dents' ability to reason and those who incul-cate facts and skills. It seems likely thatsuch differences would entail substantiallydifferent capacities, for what teachers wouldneed to know and be able to do could varyenormously depending on whether instruc-tion is aimed at inculcating basic skills orcultivating rationality.

Since instruction occurs in interactions be-tween teachers and students around educa-tional materials, capacity resides in theseinteractions. Each of these elements is es-sential; instruction requires all three. Any ofthe three elements can influence capacity forinstruction, but they act only insofar as theyinfluence the other elements. In our view,then, isolated initiatives like curriculum re-form or restructuring are unlikely to improveinstruction. Improved capacity depends onaffecting the ways in which teachers, stu-dents, and materials understand, make useof, and influence one another.

Our perception implies a dynamic view ofthe endowments of instruction. Many com-mentators conventionally refer to whatteachers or students bring to instruction, ar-guing that capacity may be diminished whenthey bring less to the instructional table.While there is some truth to that, our analy-sis suggests that what teachers, students, ormaterials bring to instruction depends partlyon how well it can be discerned and used byteachers. In this view, endowments can beunderstood not as an absolute feature of ateacher, student, or bit of material, but as afunction of how well other interactors canmake use of the endowment. It follows thatto improve capacity would be to affect howteachers, students, and materials understand,make use of, and influence one another.Changing such understanding is likely to bea very potent intervention, apart from anyother change in instructional technology ororganization.

This perspective offers a more flexible con-ception of capacity, which seems bettersuited to the many different situations thatare presented in a single classroom in a sin-gle day. It also enables a more situationalunderstanding of capacity, which shouldhelp us make sense of the different interac-


3 3

27


tions among teachers, students, and materi-als.

Capacity and Teacher Learning. Manydiscussions of instructional capacity assumethat some individuals or institutions have itwhile others do notthat capacity can beunderstood as a store of knowledge and skillto be deployed at will. From this viewpoint,once capacity is built, it is there to be usedas needed. This helps explain why profes-sional development is often seen as a pri-mary strategy for capacity-building: ifteachers learn special methods and developimproved understanding of the curriculum,their capacity will increase. Teachers wholearn more are thought to have more capac-ity, hence the continuing appeals for moreand better professional development.

We have tried to show that instructional ca-pacity could not reside only in knowledge ofcontent or pedagogy, and that it could not bethe pure possession of teachers or instruc-tional technologies. If capacity arises in in-teractions among teachers, students, andmaterials, then any given element of in-struction shapes capacity, and thus perform-ance, by the way it interacts with and influ-ences the other elements. But teachers playa distinctive part in these interactions forthey guide and shape them. Their knowledgeof and skill at interpreting and making in-structional use of materials and student ideasis crucial.

It follows that improving knowledge andskill of instructional interaction is a particu-larly salient feature of instructional im-provement. This means that teachers' op-portunities to learn such knowledge andskills are likely to be more productive thanlearning content or methods alone. Helpingteachers hear and see more in student work,helping teachers learn how to intervene art-fully in student work and to motivate stu-

dents, all affect what students can learn todo. The most effective teacher learning islikely to focus on instruction-as-interaction,rather than on isolated elements of instruc-tion.

Capacity and Instructional Environments.Most capacity-building efforts appear to as-sume that interventions could be independ-ently effective; most efforts to improve in-struction in recent decades have focused oninterventions alone. Curriculum developersand teacher educators have acted as if theyassumed that teaching and learning could betreated as independent of their environ-ments. In contrast, our theoretical frameassumes that the environments of instructionare critical to intervention: to intervene ininstruction is to somehow manage those en-vironments.

The chief reason for this view is that theUnited States is inhospitable in many re-spects to serious instructional improvement.The environments in which capacity-building efforts operate lack many of theresources that support improvementashared language of professional discourse,professional socialization leading to sharednorms and standards of work, traditions ofcommon work on teaching and learning, op-portunities for professional learning, socialand economic support for demanding in-struction, and professional norms and incen-tives that support improvement.

Given the nature of these instructional envi-ronments, effective intervention must go be-yond managing the innovation. Intervenorsmust find ways to create alternative instruc-tional environments (and buffer out manyfeatures of the existing environments), theymust delegate such work to enactors, or theymust frame the intervention in ways that willendow it with survival value in the unusualenvironments of U.S. education. We there-


3 428


fore treat the instructional environment notas a backdrop against which action occurs,or as part of its context, but as part of thecontent of that action.

Managing the Environment. This analysisimplies that intervenors face imposing tasksif they seek to manage instructional envi-ronments. One high priority would be thecreation of new guidance for instructionwhile managing the extant busy and oftenchaotic guidance system. Another prioritywould be creation or adaptation of learningopportunities for enactors. Still another pri-ority would be building infrastructure forcommunicating about the intervention,which would require the establishment ofregular links between intervenors and prac-tice. And, intervenors and enactors wouldhave to find ways of managing persistentpolitical conflict and overcoming the lack ofstrong incentives for improvement.

Designing, specifying, and developing arethe three key elements of interventions. De-sign refers to overall goals and strategy.Specification refers to the plans or educa-tional blueprints of policies and interven-tions, including the curricula for enactors'learning, the nature of the discourse betweenintervenors and enactors, and more. Inter-ventions usually consist chiefly of state-ments of principle that amount to very weakdesigns; they suggest a general direction butlittle more. In such cases, specification isspare and thin. Only a few interventionsseem to offer more in the way of specifica-tion.

Development refers to the action repertoireof policies or interventions, including mate-rials, social processes (like teacher educa-tion), working models or examples of adop-tion processes, or video materials that depictteachers' knowledge, norms, and skills inways that would be educative for other

teachers, social processes for involving andeducating parents, etc. If the specification ofinterventions is analogous to classroomplans for instruction, then development isanalogous to creating materials, occasionsfor instruction, and processes that wouldprovoke and support the learning. Better-developed interventions are more likely toeffectively encourage learning, managechange, and build instructional capacity.

Careful design, specification, and develop-ment of an intervention can reduce uncer-tainty and define areas for rational action,and thereby increase the likelihood of suc-cessful instructional improvement. But thisgreatly raises the costs of intervention,which may reduce its attractiveness. Alter-natively, interventions can be designed withmodest specification and development, sothat enactors can respond with little or nochange in the instructional environment.That moderates the costs of intervention andis attractive to enactors because moreloosely designed, specified, and developedinterventions allow enactors more auton-omy. One cost of this approach is increaseduncertainty about effective intervention andenactment; a second cost is reduced areas ofbounded rationality. A third cost appears tobe variable and often superficial enactment,and thus a decreased likelihood of successfulinstructional improvement.48

Our theoretical frame also bears on theproblem of moving from small to large-scaleenactment. Most discussions suggest thatthis is a quantitative problem, a matter ofgetting broader adoption and implementa-tion. Our analysis suggests that these prob-lems are qualitative, not quantitative, butsolving them is a necessary condition forspreading innovations. One problem is de-signing, specifying, and developing inter-ventions in a manner commensurate with thetask at hand,49 and a second problem is


3 5,


building infrastructure for managing the en-vironments of intervention so that theywould support both intervention and enact-ment. Solving these problems would requireextensive qualitative change in interventionsand their environments. The move fromsmall- to large-scale enactment is as much aqualitative problem of designing moreelaborate interventions and building instruc-tional environments as it is a quantitativeproblem of more adoptions.

Enactment. Efforts at school improvementrest on profoundly different constructions ofthe situation. Most efforts to build instruc-tional capacity are made by agents, situatedoutside schools or classrooms, who try toimprove what happens inside. These agentshave a diagnosis of how and why schoolshave failed and methods for setting thingsright. While some enactors share the senseof failure, many locate the problem else-where and believe that no solutions exist.Schools cannot be improved unless thesedifferences can be somehow negotiated andresolved. What is more, intervenors haverelatively weak leverage in the situation, fortheir relationships with enactors are mostlyvoluntary in political, organizational, andtechnical terms. Intervenors can offer in-ducements and other resources, but these areonly potent as teachers and students recog-nize, accept, and enact them. Intervenors'influence thus is partly constructed by en-actors, and intervenors depend on schoolsbecause schools must choose to enlist andbecause schools must then make use of whatintervenors offer. Intervenors and enactorsoperate in something like a market in whichintervenors work only with enlistees whochose to work with them, and who have agood deal of independence.

These relations are further shaped by timehorizons and resource flows. Most interve-nors have only temporary funding and few

or no political sponsors, while schools havemore permanent funds, are sponsored by thestate, and have clients who are mostly satis-fied with the school their children attend. Insuch a situation intervenors and enactors arelikely to define their relations quite differ-ently. Intervenors would prefer only enlist-ees who wish to improve and seem eager towork as the intervenor proposes, becausethat greatly simplifies the tasks of interven-tion, checking compliance, and the like. Incontrast, enactors are likely to prefer accessto the resources that interventions bringwhile retaining the core of operations un-changed, because that reduces the work tobe done and enables educators to adoptmany interventions while keeping all ofthem on the margin. Any school improve-ment effort must deal with these very differ-ent constructions of the situation and coursesof action.

Intervenors that seek high-fidelity enactmentand thus cultivate intensive relationshipswith enactors will be drawn to more activemanagement of those relations and the in-structional environment. The effect wouldbe to create something like new sub-systemsof schooling that could screen out competinginstructional guidance, stabilize funding andpolitical agendas, and create some measureof professional community. Such workwould further raise the costs of interventionwhile possibly reducing some risks of vari-able and low-fidelity enactment. The lessintervenors cultivate intensive relationshipswith enactors and seek high-fidelity enact-ment, the less need there would be for elabo-rate sub-systems. This approach would re-duce the costs of intervention while in-creasing the risks of variable and low-fidelity enactment.

This analysis implies that incentives for im-provement will be a critical variable in rela-tions between intervenors and enactors. The


3 6


more that potential enactors are under inter-nal or external pressure to improve instruc-tion, the greater the probability that they willaccept intervenors' specifications and makeclose relationships. The more that potentialenactors are under pressure to demonstrateresults, the more likely they are to accepthighly specified and developed interven-tions, which are more likely to produce suchresults. But it remains unclear which incen-tives will be most effective. For instance,there are schemes to mobilize external in-centives by holding teachers accountable forstudent performance on assessments, butthere is little evidence yet that such schemesalone mobilize much effort at improvement.Other sources of incentivesthose associ-ated with local administrative pressure andsupport for professional performance andthose associated with enlistment in inter-ventionshave been investigated less, butmay have elements that support improve-ment.

Intervention and Instruction. We haveframed intervention as a form of instruction.If enactors are to make use of the resourcesthat intervenors offer, intervenors must helpenactors learn how to recognize and diag-nose problems, identify and use resources,solve the problems, or find enactors whoalready are able to do those things, or both.Intervenors' approaches vary, but all actfrom an external position in schools orclassrooms in order to encourage the profes-sional actions that will improve instruc-tion." Such action by external agents canbe seen as a sort of teaching.

Intervenors and enactors work together onan agenda for implementation, which com-prises a species of curriculum. Most inter-ventions require considerable learning forenactors, and no intervenor can rely purelyon improvised tutorials. Whether we con-sider a huge state reform like the California

math frameworks, QUASAR'S work with ahandful of schools, or a new curriculumadoption in a district, there is some explicitor implied curriculum for enactors. But thenature of that curriculum varies, from rudi-mentary to elaborately developed and self-conscious designs for learning an interven-tion.

But curriculum is not enough. Just as teach-ers cannot cause students to learn, interve-nors cannot cause enactors to do the interve-nors' will. Only enactors can learn an inter-vention, and their efforts to do so will beinfluenced by their knowledge, skill, andwill to learn, and by the instruction that in-tervenors offer. Intervenors working withhigh-poverty schools face a distinctiveproblem: the school professionals, parents,and others whose schools perform poorly arealso the key agents for improving thoseschools. Yet the problems of such institu-tions and of poverty inhibit the recognitionof problems and the adoption of problem-solving strategies. External intervenors can-not succeed unless enactors recognize prob-lems and try to solve them, but reaching thatpoint implies enormous change for manyenactors. We identified several strategicalternatives that intervenors have in ad-dressing this situation:

Routinize interventions to ease the mag-nitude of the enactor's task of recogniz-ing problems and learning to solve them,and the corresponding teaching task forintervenors;

Select only apt enactors who already arewell on the route to recognizing prob-lems and learning to solve them;

Instruct enactors so that they learn andchangethat is, successively reconstructtheir initial constructions in some in-structional process;


3 7

31


Define intervention and enactment soloosely as to allow many flowers tobloom, thus allowing enactors to definetheir own levels of effort and imple-mentation and the things they will learn;

Some combination of the first threestrategies noted here.

Building capacity, like many educationalproblems, can be treated as a problem ofselection alone, of learning alone, of com-bined selection and learning, or of treatmentdefinition.5i

This frame focuses attention on both in-structional relationships and mutual selec-tion in intervention. It implies that inter-ventions will be more likely to succeed ifthey effectively deploy the elements con-ventionally associated with instruction.These elements include a teacher or teachingagent; actively engaged learners; a curricu-lum of intervention; framing teaching andcurriculum in light of an understanding ofthe learner-enactors; learning opportunitiesfor professionals; and more. The more ex-tensive intervenors' instructional designs,the more opportunities that enactors have tolearn and the more likely that enactors willlearn. Some interventions deploy these ele-ments in relatively spare ways, while othersdeploy much richer versions; in our termsthe latter are more specified and developed.We expect that the richness with which in-terventions are specified and developed willaffect their implementation.

Another implication of this analytical frameis that ability and disposition to notice andact on instructional problems, and abilityand disposition to recognize and use re-sources that could help solve those prob-lems, are critically important elements ofinstructional capacity. Cultivation of theseabilities and dispositions in enactors is one

important element in any effort to improveinstructional capacity. The effectiveness ofthe instruction that intervenors offer is likelyto depend heavily on their ability to helpeducators learn very different things aboutwhat they and their students can do.

This analysis implies a new view of the coretechnology of interventions. Interventionspropose to change what teachers and stu-dents know, believe, and can do, hence theyoperate by means of ideas, beliefs, profes-sional norms, and intellectual practices.These are central to intervention becausethey are the stuff of instruction. If knowl-edge, ideas, beliefs, and intellectual prac-tices are the key agents for enacting inter-vention, that implies a distinctive view ofhow politics, organization, and other suchfactors influence instructional improvement.Instead of considering politics and organi-zation as influences only in their own right,we expect their influence to be mediated bythe instructional relationships summarizedabove. That influence can occur in at leasttwo ways. One is by impeding or enablingideas and practices, for example, the im-pediments that the fragmented organizationof U.S. schools offers to the formation ofprofessional communities and thus to delib-erate learning about professional practice.In this case, the macro structure of schoolpolitics constrains opportunities for profes-sionals to learn in and from practice. Poli-tics and organization also influence howproblems are framed; for example, the bat-tles over basic skills and critical thinkinginfluence decisions about the content, adop-tion, and evaluation of interventions. Theseculture wars place constraints on the aimsand content of tests, curriculum, and teachereducation. In both examples, the importanceof politics and organization is expressed inthe context of instructional relationships.


38


References(Incomplete; many are listed in endnotes only)

Buchmann, M. (1993). "Role over person: Morality and authenticity in teaching." In M.Buchmann & R. E. Floden (Eds.), Detachment and concern: Conversations in the phi-losophy of teaching and teacher education, (pp. 145 157). New York, NY: TeachersCollege Press.

Cohen, D. K. (in preparation). Teaching and its predicaments. Manuscript in preparation:University of Michigan, Ann Arbor, MI.

Cohen, D. K., and Hill, H. C. (1998). Instructional policy and classroom petformance: Themathematics reform in California. Philadelphia, PA: Consortium for Policy Research inEducation, University of Pennsylvania.

Cuban, L., and Tyack, D. (1995). Tinkering toward utopia: A century of public school reform.Cambridge: Harvard University Press.

Huberman, M. (1980). Finding and using recipes for busy kitchen messages: A situationalanalysis of routine knowledge use in schools. Prepared for the Program on Research andEducational Practice: National Institute for Education, Washington D.C.

Little, J. W. (1993). "Teachers' professional development in a climate of educational reform."Educational Evaluation and Policy Analysis, 15 (2), 129-151.

Tyack, D., and Tobin, W. (1994). "The 'grammar' of schooling: Why has it been so hard tochange?" American Educational Research Journal, 31, 453-479.


39


End Notes

I We include as "intervention" state policies, private initiatives (like new text series or the Na-tional Board of Professional Teaching Standards), school reform networks, and other efforts toimprove instruction.

2 External" here refers to being outside the direct interplay of teachers, students, and curricu-lum; hence it may refer to staff members in the local central office, state or federal legislators, orprivate sector agents.

3 Readers who want a quick overview of the ideas should read the conclusion first. We are de-signing a study on instructional improvement that is informed by the ideas in this report. A de-tailed research design and proposed sampling frame are available under separate cover.

4 Doyle (1984) highlights the importance of distinguishing the "enacted curriculum" as it is con-structed by teachers and students.

5 Recent scholarship on teacher knowledge covers more than these elements and domains. Seefor example: Ball, D. L. and Wilson, S. W. (1996). "Integrity in teaching: Recognizing the fu-sion of the moral and the intellectual." American Educational Research Journal, 33, 155-192;Schon, D. (1987). Educating the reflective practitioner: Toward a new design for teaching andlearning in the professions. San Francisco: Jossey-Bass; Sockett, H. (1987). "Has Shulman gotthe strategy right?" Harvard Educational Review, 57, 208-219; Tom, A. (1984). Teaching as amoral craft. New York: Longman. For purposes of the distinction we are making here betweenresources for teaching and resources for the construction of knowledge in teaching, we deliber-ately compress and streamline our map of teacher knowledge here. In developing instrumentsfor our research, we will want to both elaborate and broaden the lenses with which we examineinstructional capacity.

6 We use the word "knowledge" here to refer to understandings, conjectures, theories, and ideas.For example, when we say "knowledge about the discourse of classrooms and how that discourseaffects what students learn," we mean theories about such, and acknowledge that ideas about thisimportant domain differ among researchers and practitioners, as well as among those who aim toreform schools.

7 In the moment, of course, one often cannot know that something being learned is general ratherthan specific to the moment and the situation.

8 Each of these alternatives implies certain attributes of learning, and, taken together, they tell usa great deal about how teachers try to construct students' learning in a classroom: that is, howstudents will treat knowledge, how and how much they will open their thinking to teachers, andwhat sort of discourse they will engage in. Several of these categories are taken from Cohen,D. K. (in preparation). Teaching and its predicaments. University of Michigan, Ann Arbor, MI.

9 The sheer number of leverage points may not be crucial; more is not necessarily better. Whatmatters is how and on what intervenors work, with what emphasis and intent, and how those de-


4 0


cisions impact instructional capacity. For example, new mathematics materials may influencecapacity more than similar materials might in reading, because math is less dependent on the so-cial environment. We want to learn what sorts of interventions on which elements precipitatewhat sorts of changes in instruction. It may be that certain strategic points of intervention pro-duce more powerful effects on instruction than other, more voluminous and varied interventionstrategies.

I° See for example Peter Dow's (1993) account of the development and implementation of Man:A Course of Study. Dow, P. (1993) Schoolhouse politics. Cambridge, MA: Harvard UniversityPress.

I I For a longer discussion, see Ball, D.L. (1997a). "What do students know? Facing challengesof distance, context, and desire in trying to hear children." In B. Biddle, T. Good, and I. Good-son (Eds.), International handbook on teachers and teaching (Vol. II), (pp. 679-718). Dordrecht,Netherlands: Kluwer Press.

12 These ideas are drawn from Ball, D.L. (1997b). "Developing mathematics reform: What don'twe know about teacher learningbut would make good working hypotheses?" In S. Friel and G.Bright (Eds.), Reflecting on our work: NSF Teacher Enhancement in K-6 Mathematics, pp. 77111. Lanham, MD: University Press.

13 As above, we are not arguing that all learning opportunities need to be direct or explicit.However, interventions that leave teacher learning to chance are unlikely to work as well asthose that attend in some way to teachers' learning opportunities.

14 Some of our other research shows that when learning opportunities for teachers are groundedin the students' curriculum, the learning affects both teachers' practice and their students' learn-ing more significantly than other sorts of professional learning. Cohen, D. K., and Hill, H. C.(1998). Instructional policy and classroom petformance: The mathematics reform in California.Philadelphia, PA: Consortium for Policy Research in Education, University of Pennsylvania.

15 We develop these ideas about resources in Cohen, D. K., Ball, D. L., and S. Raudenbush(1999, April). "Educational resources and instruction.". Paper presented at the annual meeting ofthe American Educational Research Association, Montreal, Canada.

16 Moreover, this structure is remarkably homogenous: there is little variability within the U.S. inthis structural dependence. All schools constitute classrooms and other internal units (such asdepartments and multi-grade groupings). The chief area of variability is grade structure: all highschools are internally departmentalized by school subject or academic discipline, while all ele-mentary schools are not.

17 Some researchers and intervenors appear to assume that because classrooms are structurallydependent on schools they are also functionally dependent. They argue that the school's formalpre-eminence means that it is or should be the central unit for investigating instructional proc-esses and effects. Some intervenors argue, in roughly parallel fashion, that schools are the pri-


41


mary unit for improving instruction. But there is no necessary relationship between structuraland functional dependence.

18 School systems differ in this respect. In some national systems, schools' formal organizationis closely linked to the arrangements for instruction. State schools in France, for example, havedealt with the central functions of instruction quite consistently, with relatively little variationwithin or among schools in instructional goals, methods of coordination, and the like. See forexample, Broadfoot, P., Osborn, M., Gilly, M., and Bucher, A. (1993). Perceptions of teaching:Primary school teachers in England and France. London: Cassell. Because the system is cen-tralized nationally, and because professional norms seem quite powerful and homogenous,teachers and administrators appear to agree on these matters, and to agree as well on the relationsbetween formal organization and instruction. The system appears relatively homogenous bothwithin and among schools.

19 There is some coordination between some governments and some agencies that produce andmanage these elements of the core technology: state agencies set standards for teacher licensesand certification, and some states regulate text purchase, but many do so only weakly, and othersdo nothing.

20 One cannot imagine physicians and other health care professionals improving either patients'health or medical practice if they lacked the descriptive and analytic languages of physiology andanatomy, of disease processes, and treatment. Such professional language and the knowledgeunderlying them are at the heart of medical practice, but they are very weakly developed in edu-cation.

21 This communication could occur through sharing student work, presenting lessons to otherpractitioners, discussing videotapes of teaching, collective examination of curriculum, collectivecurriculum construction, or joint analysis of student performance.

22 On families and schools, see Stevenson, H., and Stigler, J. (1992). The learning gap: Why ourschools are failing and what we can learn from Japanese and Chinese education. New York:Touchstone; and Cohen, D.K. and Spillane, J.P. (1992). "Policy and practice: The relations be-tween governance and instruction." In G. Grant (Ed.), Review of Research in Education, 18, 4-49. Washington, DC: American Educational Research Association.

23 Stevenson and Stigler, op. cit.; Cohen and Spillane, op. cit.

24 A special case of this situation is that a central task of many interventions is to constitute orreconstitute the instructional environment that we call "school." Schools' passivity with respectto instruction means that schools' formal or structural resources cannot be translated into re-sources for instructional improvement. For this reason many interventions devote significantresources to leadership development, improved decision-making at the school level, and relatedmatters.


4 2


25 Instructional regulation includes graduation requirements, text adoptions, standards-based re-form, and other efforts to use regulatory mechanisms, usually at the state or federal level, to in-fluence classroom work.

26 See Elmore, R.F., Abelman, C., and Fuhrman, S. (1996). "The new accountability in state edu-cation policy." In H. Ladd (1996). Holding schools accountable: Petformance-based reform ineducation, pp. 65-98. Washington DC: The Brookings Institution.

27 This account is based on Cohen and Hill (1998).

28 This seems typical of most professional development in the United States today: teachers en-gage in a variety of short-term activities that fulfill state or local requirements for professionallearning but are rarely deeply rooted in the school curriculum or in thoughtful plans to improveteaching and learning.

29 Since policies and interventions often are aimed at students but depend on professional edu-cators, what follows applies both to the elements of policy that centers on student learning and tothe elements of policy that centers on educators' learning.

30 Both terms thus refer to and modify intervention design.

31 Development is not the mere.enactment of specification. It is not difficult to imagine a directinstruction scheme that is very highly specified but for which the materials and teach& educationare detailed, thin, and mechanistic. The results of development may not express the promises ofinnovative specification, or they may exceed those promises.

32 We do not assume that specification and development exist in linear form; often specificationis refined as a consequence of work on development.

33 Instructional interventions of very different sorts can be strongly and weakly specified and de-velopedspecification is not just a matter for more didactic interventions. But specification anddevelopment would look rather different for interventions that had modest instructional goals anddidactic methods than for interventions that had ambitious instructional goals and flexible meth-ods. When the instructional purposes and methods of policies and interventions are quite simple,then more complete specification and development are possible. Specification and developmentare more complex and difficult when interventions are more intellectually ambitious and childthinking-centered, and must be less complete because more complex social interactions will en-sue. But the greater difficulty and incompleteness of specification and development is not reasonto conclude that only broad principles can be enunciated. A great deal can be specified and de-veloped for more complex and ambiguous policies and interventions, but there will be morecontingency. Weak specification of such interventions has been more common, both because thework is difficult and because of the mistaken idea that specification and development inhibitcreativity, autonomy, or both.

34 Developers who create such materials sometimes design them in such a way that differences inwhat learners bring (and the effect of these differences on what they would learn) are muted by


4 3


so closely specifying the desired outcomes and so fully developing the processes and materials toattain them.

35 Each approach entails efforts to manage the instructional environment, the first by creating anew environment that would both guide instruction and buffer other competing signals, and thesecond by accepting the existing environment and trying to compete with it.

36 The less well-specified and developed instructional policies and interventions are, the morethey leave to be invented, improvised, and figured outor unwittingly ignored. Such interven-tions delegate much difficult work to enactors, but they do so tacitly, without exposing muchabout the nature, extent, or difficulty of the delegated work. What is left underspecified or un-derdeveloped matters. Recent research suggests that there is a bi-modal response to weaklyspecified and developed interventions: a small population of enactors deeply engages the ideasand struggles with very complex and difficult learning in the absence of much guidance, while amuch larger populationbelieving that it has engaged the ideas deeplyresponds either withmodest, superficial, and often distorted practices, or responds not at all. In most cases, then, littleguidance from intervenors does not promote enactors' autonomy or creativity, but weak learningand enactment.

37 In the case of schools in "reconstitution," "crisis," and "receivership," intervenors work withthose who did not choose to improve, and who, almost by definition, did not recognize theirproblems. Even if they chose to enlist with intervenors, the choice resulted from external com-pulsion. Work with such schools may create different dynamics between intervenors and enac-tors.

38 In the decentralized U.S. political system, even when state or local governments require orstrongly urge enlistment on a school, schools will have considerable political autonomy. Theforms of voluntarism and compulsion may belie the realities: what appears to be compulsorymay have strong elements of choice, and what appears to be choice may have strong elements ofobligation.

39 These points are implied or explicit in most analyses of implementation, although few analystsgive any consideration to teaching and learning as a frame for analysis, and most give differentweights to the three considerations above. Some treat all of the considerations as less importantthan the intervention or policy itself.

40 The management systems that enactors and intervenors create could include the followingmeans of managing relations between enactors and intervenors:

Professional and technical knowledge aloneintervenors offer professional education,technical assistance, the collection and feedback of data on enactment or some combina-tion of these (which can be very loosely or tightly specified);Compliance reviews, in which standardized data collection and/or site visits are used todetermine the nature and extent of symmetry with the intervenor's design;Extensive mutual exploration and contracting in advance of any intervention, to set ex-pectations, define roles, and mobilize commitment;


4 4


Placement of an agent (site facilitators, coaches) of the intervenor in enactor's organiza-tions, to both assist with enactment and check progress (site facilitators, coaches);Make side bargains with local school agencies to manage the environment (stabilizefunding, reduce conflicting messages, etc.); andGive enactors strong and detailed guidance for instruction.

41 See Consortium on Chicago School Research (1996). Charting reform: Teachers take stock.Chicago: Author.

42 It is worth recalling that such rationality depends in part on an instructional environment inwhich the incentives for improving instruction are modest at best.

41iIt s possible that some enactors would use interventions to fend off the invasion of other, less

desirable, influences or forces. There are some reports of such a response in many professionaldevelopment schools. Deborah Schifter has reported a similar response to her work, and there issome sign of this with several of the California math professional development projects.

" See, for example, Elmore, Abelman, and Fuhrman (1996).

45 That is not to embrace a subjective view of external resources; we do not assert that resourcesare mere figments of teachers', students', or observers' imaginations. The presence of such po-tential resources can enable practitioners to reduce the difficulty of good work, just as their ab-sence makes it more difficult to do good work. If two equally able and energetic teachers work intwo very different schoolsone that has an equally energetic and able faculty and another thatdoes notthe teacher in the better situation will have many more resources to use in doing goodwork that her colleague in the poorly endowed school. The lack of such colleagues does not pre-vent good work, but it does make it more difficult. Similarly, the presence of good colleagues orother resources cannot cause a teacher to do good work, though it may create some conditionsthat would enable such work if the teacher is motivated and able. While institutions can offerteachers rich or poor social resources of practice, only teachers and students can see and usethose resources, or fail to see and use them.

46iThis s a persistent dilemma of teaching: many teachers and leaders who instruct (whether they

work in classrooms, larger organizations, or some body politic) are confronted by many potentiallearners who recognize no need to learn. It is a condition of their success that these teachersmust somehow mobilize the requisite will and skill to learn among these potential learners. Atleast several of the interventions that we propose to study are such cases, and their success de-pends heavily on finding ways to solve problems of enactors' recognition of their own problemsand adoption of problem-solving approaches.

47 The proposed research will require that we further develop hypotheses about how the selectionmight work. This would prominently include the attributes that intervenors would be expected tolook for and select ongeneral aptitude? Affinity for the intervention? Quickness of learning inearly acquaintance? Other things?

CPRE Research Report Series, RR-43 4 5 40


48 These two tasks also have been nearly entirely ignored. For example, in their book, TinkeringToward Utopia, David Tyack and Larry Cuban appear to take innovations as given, as thoughproblems of design and specification did not exist. Yet most of the reforms that they discusswere extremely weakly specified and quite underdeveloped. A similar point holds for managingthe environments of interventionit simply does not enter most analyses.

49 In the case of simple organizational changesextending the school day by an hourthis is notvery difficult, but efforts to promote intellectually ambitious instruction could require extensivespecification and development.

50 External here refers to the classroom; hence it may refer to staff members in the local centraloffice, state or federal legislators, or private sector agents.

51 Other educational problems include admission to private schools or to programs within publicschools.


4 6

Nondiscrimination Statement

The University of Pennsylvania values diversity and seeks talented students, faculty and staff from diverse backgrounds. TheUniversity of Pennsylvania does not discriminate on the basis of race, sex, sexual orientation, religion, color, national or ethnicorigin, age, disability, or status as a Vietnam Era Veteran or disabled veteran in the administration of educational policies,programs or activities; admissions policies, scholarships or loan awards; athletic, or University administered programs; oremployment. Questions or complaints regarding this policy should be directed to Executive Director, Office of AffirmativeAction, 1133 Blockley Hall, Philadelphia, PA 19104-6021 or (215) 898-6993 (Voice) or (215) 898-7803 (TDD).

Suggested Citation

Cohen, D. K., and Ball, D. L. (1999). Instruction, capacity, and improvement (CPRE Research Report No. RR-043).Philadelphia: University of Pennsylvania, Consortium for Policy Research in Education.

4 7

C:.

U.S. Department of EducationOffice of Educational Research and Improvement (OERI)

National Library of Education (NLE)Educational Resources Information Center (ERIC)

NOTICE

REPRODUCTION BASIS

RIC

This document is covered by a signed "Reproduction Release(Blanket) form (on file within the ERIC system), encompassing allor classes of documents from its source organization and, therefore,does not require a "Specific Document" Release form.

This document is Federally-funded, or carries its own permission toreproduce, or is otherwise in the public domain and, therefore, maybe reproduced by ERIC without a signed Reproduction Release form(either "Specific Document" or "Blanket").

EFF-089 (9/97)

C RE,

June 23, 1999

CONSORTIUM FOR POLICY RESEARCH IN EDUCATIONUniversity of Pennsylvania Harvard University Stanford UniversityUniversity of Michigan University of Wisconsin-Madison

Dear Colleague,

The Consortium for Policy Research in Education (CPRE) is pleased to send you a copy ofInstruction, Capacity, and Improvement, a new research report by David Cohen and DeborahLoewenberg Ball of the University of Michigan. I hope you will find this report useful. If you have anyquestions-or need additional information, please contact me via phone at 215-573-0700, x225 or viaemail at [email protected].

Best wishes,

(96

Robb a SewellDissemination Coordinator

Enclosure

PENN Graduate School of Education, University of Pennsylvania, 3440 Market Street, Suite 560, Philadelphia, PA 19104-3325Phone 215.573.0700 0 Fax 215.573.7914 0 [email protected] 0 http://www.upenn.edu/gse/cpre/

PA. 47p. - ERICsocial organizations situated within, and vi-tally affected by, other complex social...

Documents

Transcript of PA. 47p. - ERICsocial organizations situated within, and vi-tally affected by, other complex social...