AAAS Conference on Improving Science Textbooks through Research and Development

Improving Science Teaching through Research and Development

Report of a conference sponsored by Project 2061 of the American Association for the Advancement of Science
1200 New York Avenue, NW
Washington, DC
October 17-19, 2001

Drafted by Charles W. Anderson

[Return to the conference documents]


On October 17-19, 2001, Project 2061 sponsored an invitational conference attended by researchers who have investigated and tried to improve science teaching and learning in K-12 classrooms. Staff members from Project 2061 and the National Science Foundation and developers of research-based curriculum materials also contributed to the conferences. These contributors shared their insights and perspectives and discussed the implications of research on science teaching and learning for the large-scale reform of science education in the United States.

The discussions were wide-ranging and reflected the variety of perspectives that participants brought to the conference. This report cannot represent those discussions in their entirety. There were, however, important areas in which the participants were able to achieve a substantial consensus about issues, problems, and significant research findings. The purpose of this report is to summarize some of those points of agreement among the participants in the conference. Additional information about the conference, including papers contributed by conference participants, is available at the conference website.

Like the conference itself, the report is organized into five sections. The first section summarizes the purposes of the conference. The second section discusses models of teaching and learning—the ideas shared by conference participants about what it means to learn science with understanding and the roles that schools play in helping students to learn. The third and fourth sections focus on the two primary mechanisms through which research could contribute to large-scale changes in science teaching—material support systems and professional development. The final section discusses issues of wide-scale implementation.


Purposes of the Conference

We are engaged in a reform movement that has “science understanding for all” as its goal.  We know on the basis of research done during the past 20 years how difficult this goal will be to achieve.  This research has produced a number of “existence proofs”—examples of successful teaching for understanding by individual teachers or by small numbers of teachers.  These existence proofs show that under the right conditions all kinds of students can learn science with levels of understanding that are currently achieved by only a small elite.

This research has produced new insights into the nature of scientific understanding and how it is achieved by learners.  The research shows that learning with understanding is a difficult and complex process of, in Niels Bohr’s words, “extending our experience and reducing it to order.”  To understand scientific ideas in ways that are coherent and useful, most learners need new experiences with the material world, social support from teachers and classmates, coaching on scientific ways of reasoning and communicating, and teaching that addresses topic-specific conceptual barriers.  It is difficult, but not impossible, for teachers to provide these experiences and support systems for diverse learners in American schools.

The existence proofs have not, however, led to successful scale-up efforts.  To date, the largest professional communities that support consistent teaching for scientific understanding have no more than a few dozen members.  While several factors have contributed to this lack of success, two essential factors are

  1. The inadequacy of material support systems for teaching and learning in American classrooms, including textbooks, technology, and materials for experiential learning.
  2. The absence of professional communities or networks that support teaching for understanding by their members.
In contrast, we can point to some examples in mathematics education where promising large-scale efforts are taking place—for instance, the NSF-funded programs, Investigations in Number, Data, and Space and Connected Mathematics, and accompanying professional development networks.  Some school districts have also shown promising gains in achievement through curricular improvements and professional development.  A third conference in the series, scheduled for May, 2002, will examine some of those efforts.

The purpose of this conference is to discuss where we should be placing our bets in science.  What kinds of systems give us the best chances of scale-up in large numbers of American classrooms, with large numbers of American teachers?  How do they balance investment in material support systems and the development of professional communities?


Models of Teaching and Learning

One theme for the discussions in the conference concerned the nature of our collective goals for reform in science education.  What would we take as evidence that students were learning with understanding or that a program was successful?  The discussion of models of teaching and learning was loosely organized around three general questions.

  • Goals for student learning: What do we mean by "understanding," and how do we express topic-specific goals for student learning?
  • Instructional models: What are the recurring practices of successful classroom learning communities?
  • Assessment: What do we take as acceptable evidence that students have learned with understanding?

These questions were discussed both in the conference and in the contributions of several of the participants to the course web site, particularly the papers by Berkheimer, Anderson, and Spees; Heller; Lehrer and Schauble; Carol Smith; Ed Smith; and Stewart, Cartier, and Passmore.

The participants in the conference were diverse with respect to their  backgrounds and scholarly vocabularies, leading to lively discussions as researchers worked to communicate across disciplinary boundaries.  It was clear from these discussions, though, that these researchers agreed on some important points.  First and foremost, the participants shared a deep dissatisfaction with what might be called the "culture of school science" in American classrooms .  The TIMSS video studies provide strong evidence that there are prevailing patterns of teaching practice in United States and in other countries.  These patterns encompass assumptions about the nature of scientific knowledge and practice, the roles that teachers and students play in classrooms, and beliefs about student learning and understanding.  It might be helpful to understand the shared models of the conference participants by contrasting them with currently prevailing practices in American science classrooms and with common interpretations of the goals of the reform movement.

The dominant tradition in American science classrooms (and in the textbooks and other materials used in those classrooms), packages and delivers science content to students as three kinds of knowledge:

  • Facts and definitions.  Students encounter many facts about the material world, such as the names of the planets in the solar system, the age of the earth, or the elements in the Periodic Table.  They also encounter many terms from the technical vocabulary of science and their definitions.
  • Sequences of events.  Students also encounter many stories or sequences of events, such as the stages of mitosis, the rock cycle, or the evolution of the peppered moth. 
  • Problem-solving procedures.  Students encounter procedures and formulas for solving particular kinds of problems, such as the gas laws, balancing chemical equations, or electrical circuit problems.

This system for packaging and delivering science content is convenient and efficient.  The facts and definitions can be memorized and stated in ways that are correct or incorrect.  The sequences of events have beginnings, clearly defined intermediate events, and ends (or returns to the beginnings in the case of cycles).  The problem-solving procedures have specified inputs, sequences of steps, and correct answers at the end.  Even scientific inquiry can be presented as a problem solving procedure:  The sequence of steps for defining problems, stating hypotheses, collecting and analyzing data, and stating conclusions in the "scientific method."  Because they are easily assessed, these kinds of knowledge are the basis for most science classes and for most assessments of student learning. 

However, the participants in the conference shared the conviction that these ways of delivering content and testing student achievement distort and misrepresent scientific knowledge and practice.  They render scientific knowledge sterile, virtually useless to most learners, and easily forgotten.  The conference participants were equally disillusioned with the most commonly proposed alternative-"inquiry" or "constructivist" approaches that in practice are virtually content-free, giving students new experiences with the material world but little access to the scientific concepts that would help them make sense of those experiences.

The participants in this conference have been active in developing alternatives to prevailing practice.  For example, the contributions to the website referred to above all describe design experiments in science teaching. These designs are based on a variety of scholarly literatures, including studies of history and philosophy of science, cognitive and developmental psychology, and research on science teaching and learning. Thus the participants brought with them a diversity of scholarly backgrounds and a rich array of analyzed experiences in science teaching and learning. What we find encouraging is that despite their differences in experience and scholarly vocabulary, the participants were able to see some important common themes in their work.

The most critical of these commonalities is a view of scientific understanding as something like reasoning about patterns in experience.  The participants in the conference did not discuss facts, definitions, or problem-solving procedures, either in their contributed papers or in the conference itself.  Their discussion of science learning and understanding tend to focus on terms such as the following:

  • Experiences, observations, or data.  Scientific understanding requires first-hand or vicarious experience with the systems and phenomena of the material world.  Not all experiences are equal.  Scientists create data by constructing representations of experiences that they can verify, reproduce, and describe or measure precisely; students can also create data from their experiences.
  • Patterns, representations, and visualizations.  Scientific laws and generalizations are statements about patterns that scientists see in their data.  Thus pattern finding is an essential scientific practice, a key step in reducing our experience to order.  Representations and visualizations such as diagrams, graphs, tables, and computer graphics help both scientists and science learners to link experiences to explanatory models and theories. 
  • Explanations, models, and theories.  Scientific models and theories explain patterns in experience.  Although they can be used to explain individual phenomena, their power lies in the fact that they provide parsimonious accounts of broad patterns that encompass many different systems or phenomena. The great scientific theories are beautiful in the elegant and parsimonious way that they unify and explain apparently diverse phenomena. 

Thus the participants in the conference shared a view of scientific understanding in which learners (a) expand their stock of experiences with the material world, (b) find patterns in those experiences, and (c) use theories and models to explain those patterns.  This kind of reasoning is both a social and an individual process; it engages learners in first hand experiences in creating and representing data, in arguing about the meaning and significance of data, and in using scientific models to make sense of their experiences.

This view of understanding has important implications for the interpretation of reform documents such as Benchmarks for Science Literacy or the National Science Education Standards and for assessment  of student learning.  It is possible, for example, to treat the benchmarks as statements of facts to be learned and to assess students on their ability to recite those facts on demand.  Many participants in the conference knew educators who interpreted the benchmarks in these ways.  They were vocal in their concern that such interpretations could lead to "reforms" that do more harm than good, locking the nation into curricula that incorporated some of the worst aspects of current practice.

The idea of scientific understanding as model-based reasoning about patterns in experience suggests a very different interpretation of the Benchmarks and other reform documents.  In this view, each benchmark is a theoretical statement which students understand only when they can connect it with a substantial set of experiences and patterns in experience.  For students of any age, understanding is grounded in data created from their experiences with the material world; students understand when they can represent these data in ways that show patterns and can make sense of those patterns using developmentally appropriate theories or models.

These ideas about the nature of scientific knowledge and practice led to extensive discussions of the "developmental trajectories" through which students of any age could develop scientific understanding.  There was general agreement that these developmental trajectories were more complex than the metaphor of "replacing misconceptions with scientific conceptions" would imply.  Learning with understanding involves

  • Extending the range of students' personal and vicarious experiences with the material world and their abilities to create data out of those experiences.
  • Developing increasingly sophisticated ways of finding patterns in data and of representing those patterns in ways that enable others to see them.
  • Developing theories or models that account for patterns in data and using those models to explain and predict new experiences.
  • Developing personal epistemologies for making sense of the world and of a meta-level awareness that there are different ways of making sense of the world.

Participants in the conference also discussed the nature of classrooms that support the development of scientific understanding in students.  These discussions were largely based on the experiences of the participants themselves, most of whom had taught in or worked with other teachers to create successful experimental classroom communities that supported learning with understanding by their students.  These classrooms differed in many respects: They included students of different ages, races, and cultures; they used textbooks, hands-on investigations, and instructional technologies in different ways; they differed in the degree to which teachers controlled the content taught and the activities of students.  In spite of these surface differences, however, there were deeper characteristics that the experimental classrooms all shared.

Some of these shared characteristics of successful experimental classrooms involved the practices of teachers and students.  Students participated in communities of inquiry in which they extended the range of their experiences with the material world, developed new ways to represent and communicate about their experiences, and developed models to make sense of patterns in experience.  The students were actively involved in reasoning around their experiences with the material world.

Other shared characteristics involved the resources that supported classroom practices.  All of the successful experimental classrooms had teachers who understood science content in terms of model-based reasoning about experience and who were insightful about their students' scientific thinking.  The experimental classrooms also had material resources that gave students access to real or vicarious experiences with the material world and that supported their efforts at data representation and theory -building.  (For a deeper and more thorough discussion of these practices and resources, see the papers on the website referred to above.)

Although the orientation of these experimental classrooms is "constructivist" in the broad sense that students are actively involved in constructing knowledge, these classrooms differ from many models of "constructivist" teaching in their emphasis on topic-specific strategies and resources that support the development of canonical scientific knowledge.  The participants in the conference were generally convinced that there is no substitute for topic-specific development.  General models like those discussed in this section can suggest important questions about the nature of important data, patterns, and models and about students' developmental trajectories for that topic.  However, the answers to these questions and the resources to support classroom communities must be painstakingly developed separately for each topic in the curriculum. 

Thus "scaling up" from the design experiments conducted by the participants-generally a few weeks or months in duration-to curricula covering years of schooling will necessarily be a long, labor-intensive process.  The participants were also generally convinced that scaling up to cover all of the Benchmarks would be impossible.  All American students cannot achieve understanding of all the benchmarks in the time currently allocated to science in American schools.  There was also general consensus among the participants about how this dilemma should be resolved.  In the words of one participant, "there are too many benchmarks."  It would be better for students to achieve understanding of the experiences, patterns, and models associated with a subset of the benchmarks than to "cover" all of the benchmarks by presenting them to students as facts, definitions, and procedures.  The consensus on this point was, unfortunately, not accompanied by a consensus about which benchmarks should be included and which left out.

The participants in the conference were also aware of the difficulties inherent in scaling up these instructional models from a few experimental classrooms to real schools on a large scale.  The alternative models proposed by these researchers require fundamental changes in the social organization of classrooms: Teachers and students must take on new roles as participants in new kinds of classroom communities where teachers function as coaches for students who are actively engaged in scientific inquiry and application.  The question of how this might happen on a large scale is discussed further in the sections on material support systems and professional development.


Material Support Systems

Classrooms like those envisioned in the discussion of models above require substantial material support systems, including textbooks, materials for hands-on inquiries, and information technology.  The conference included many participants with extensive experience in the development and evaluation of these material support systems.  The discussion of these systems was loosely organized around the following questions:

  • What material or technological supports are needed for productive classroom learning communities?
  • What are the qualities that distinguish good material support systems from poor ones?
  • How do we develop good material support systems?

These questions were addressed both during the discussions in the conference and by many of the papers on the conference web site, including the contributions by Berkheimer, Anderson, and Spees; Heller; Lehrer and Schauble; Carol Smith; Ed Smith; and Stewart, Cartier, and Passmore.

Qualities of good material support systems

The first two questions concern the qualities of support systems for productive classroom learning communities.  Four general themes emerged from the wide-ranging discussion of these questions.  The first of these concerned the role of material support systems in providing a central classroom role for students' scientific reasoning.  Material support systems can and should help teachers to build class activities around students' thinking.  This means going beyond "misconception alerts" or invitations for students to express their ideas.  Material support systems can (and should) support serious engagement with students' ideas as ways of making sense of their experience in the world.  This means promoting dialogues in which students' ideas are compared with scientific ideas and their ability to account for a range of experiences is explored.

A second theme concerned the ways in which materials support productive patterns of practice in classroom communities.  Good materials encourage students to learn from one another and provide models for how scientific ideas can be constructed and used.  They encourage students to be reflective about their learning.  They include embedded assessment activities that inform both teachers and students about the nature of the students' understanding and the limitations in their experiences and models.  They encourage classroom activities that take advantage of local resources or apply scientific ideas to local problems.

A third theme concerned the multiple types or levels of student learning that are supported by good material support systems.  In addition to conceptual learning (i.e., model-based reasoning about data), students can become aware of, and learn to talk about, the nature of science and scientific reasoning.  They can learn how to solve problems cooperatively with their peers.  They can learn how to learn-how to act productively when they encounter new topics or new problems that are unfamiliar to them.  They can develop self-knowledge and see connections between school learning and their lives outside of school. 

A final theme concerned the roles of classroom materials in supporting teachers' professional development.  Good materials help teachers set reasonable goals for their students' learning.  They help teachers learn more deeply about the science content that they are teaching.  They alert teachers to likely conceptual barriers to student understanding and include assessment activities that provide evidence about the existence of those barriers in their own students.

A corollary to this final theme is that "teacher-proof" materials do not exist.  Most teachers take several years before they can take full advantage of the affordances of good material support systems.  These are years of intense learning and of change in their classroom cultures along the lines discussed in the sections on models above-the development of classroom learning communities in which students are creating data from experience, finding and representing patterns in those data, and using scientific models to account for those patterns.  None of the participants in this conference believed that teachers could work their way through these difficult changes without social support.  Material support systems are useless without adequate professional development and supportive professional cultures.

Developing good material support systems

The conference also included extensive discussions of how good material support systems can be developed, with particular emphasis on the problems of scaling up from the experimental systems that many of the conference participants had developed.  This scaling up process involved two dimensions.  The discussions focused on the development of material support systems that (a) could be used by large numbers of teachers, including teachers who had no direct contact with the developers of the systems, and (b) covered substantial periods of time-a year or more of the curriculum.  Themes that emerged from this discussion involved the importance and the complexity of setting goals for student learning, pilot testing, and building professional communities and organizations.

Conference participants viewed the processes of deciding on appropriate goals for student learning and using them as a basis for program development as critically important, but very difficult and complex.  Although documents such as Benchmarks, the Atlas of Science Literacy, and the National Science Education Standards provide useful starting points, they do not provide the research-based developmental trajectories that are needed to design material support systems.  Available research literature is also useful, but incomplete.  The goals for a program must also do more than make developmental sense.  They must be expressed in forms that make sense to teachers, students, parents, and the general public.  They must justify the investment of time and money that are required by a program that involves substantial changes in the culture of school science.  So deciding on appropriate goals involves extensive planning, consultation with stakeholders, and modification during pilot testing.

Materials that will be used on a large scale require pilot testing on both small and large scales, with teachers and students who represent the range of eventual users of the materials.  The pilot testing requires multiple iterations and multiple forms of feedback from teachers and students.  It is particularly important for pilot testing to attend to:

  • How students are making sense of scientific ideas and reconciling them with their own experience.
  • How teachers are making sense of the content they are teaching and the implicit and explicit expectations about their roles in the classroom.

Finally, materials development requires organization building.  At the core of the development process is sustained cooperation between writers and the teachers who pilot test the materials.  Beyond that, most successful materials development efforts require other people with other kinds of expertise-scientists and programmers and artists and advisors and evaluators.  All of these people must learn from one another and work cooperatively in order to develop successful programs. 

Although these points may seem like common sense, they are extremely difficult to achieve in practice, requiring large budgets, extended time spans, and teams of developers who can work together while blending a diverse array of knowledge and abilities.  The participants in the conference were impressed with Glenda Lappan's and Betty Phillips' videotaped accounts of how they had managed to accomplish these feats of scholarship and organization for the Connected Mathematics Program, and they were in agreement that no one in science education had yet managed such a sustained and successful project. 


Professional Development and Communities of Practice

Classroom learning communities that produce student understanding must be led by knowledgeable and accomplished teachers. The conference included many participants with extensive experience in working with individual teachers and developing professional communities.  The discussions in the conference about professional communities and professional development were loosely organized around the following questions:

  • What kinds of personal qualities and professional knowledge do teachers need to teach for understanding?
  • How can those qualities be nurtured in communities of practice?
  • What professional development practices can help to create those communities?

The participants agreed that, even under ideal conditions (including both excellent supporting materials and professional development programs), teachers develop the knowledge and skills they need to transform their classroom cultures only through a difficult and complex learning process.  In general, participants in the conference were not optimistic that this process could be substantially simplified or speeded up.  Teaching for understanding is a complex enterprise that requires a lot of learning  (Cohen & Hill, 2001; Garet, et al., 2001; Kennedy, 1998; Loucks-Horsley, et al., 1997). 

Many participants felt that they had achieved important, if incomplete, insights into the developmental trajectories through which teachers became more successful at teaching for understanding.  Teachers need to learn on a number of levels, including the following:

  • Understanding of the science content that they teach.
  • Understanding what it means to understand.
  • Understanding student thinking and students' developmental trajectories, including student thinking about the specific topics that the teachers are teaching.
  • Mastering strategies for creating and sustaining classroom communities with transformed roles and practices.

The participants in the conference agreed that teachers could change their classroom cultures only if they learned about these ideas in ways that explicitly connected general principles to their personal experiences in their own classrooms.  Opportunities for teachers to share and analyze stories and artifacts (such as examples of student work or videotapes) from their own classrooms are essential to any successful professional development program.  Many conference participants also reported successes in developing cases and material support systems for professional development, including the following:

  • Role of cases (video, personal experience) to help people develop and use theories and a precise technical language for talking about practice (Matt Koehler DVD, presentation at
  • Materials to use in classrooms that support good practice, such as those described in the papers on the conference web site
  • Information about student thinking, developmental trajectories
  • Rich problems of teaching and learning/student thinking around which professional discourse can take place

The discussions also addressed questions of transfer from one topic to another.  Clearly, material support systems must be developed separately for each topic that they address.  Is that necessary for teachers, though?  If teachers have successfully taught some topics for understanding, and if they are supported by good materials,, then could they move on to teaching other topics relatively quickly and easily?  The participants agreed that there is partial transfer from one unit to another for teachers.  Teachers can develop general belief systems  and classroom cultures that support learning and problem solving for new topics.  When teachers have advance knowledge of the problems that have to be solved to teach effectively, they solve those problems more quickly and efficiently.  There remains, however, a substantial amount of topic-specific learning about content, students, and pedagogy that teachers must go through anew for every topic that they teach  (Cohen & Hill, 2001; Kennedy, 1999).

As Christine Massey argues in her contribution to the conference web site, professional development in this sense involves the development of new professional cultures and professional communities in addition to the development of the skills and knowledge of individual teachers. These professional communities enable teachers to learn from one another and provide teachers with social support for difficult and sometimes frustrating learning.  Thus successful professional development necessarily involves community building as well as the development of individual teachers.  Within those professional communities, teachers also play a variety of different roles.  They are learners; they are collaborators with one another and with professional developers; they are change agents in their schools.  Large-scale reform will require teachers to play all of these roles.


Scale-up Issues

A final topic of discussion in the conference involved issues of scale-up from successful experimental programs to large-scale implementation.  This discussion was necessarily more hypothetical than the discussions of material support systems and professional development.  Although many conference participants had participated in large-scale programs, they did not generally feel that those programs had been successful in changing large numbers of participating classrooms.  The discussions of scale up, therefore, focused more on problems and issues and less on accomplishments than other discussions.

The participants in the conference recognized that issues of scale-up involve, in part, questions about how to make intelligent use of limited resources.  Large-scale programs will necessarily have fewer human and material resources per teacher.  How can those programs be designed so that they produce, in the words of one conference participant "half-baked apples rather than half-baked pork chops?"  How can we distinguish, in other words, between useful if incomplete halfway steps toward systemic  change (half-baked apples) and programs that provided teachers with resources so inadequate for the task they faced that they were likely to do more harm than good (half-baked pork chops)?  In general, participants were discouraged about the current prevalence of inadequate and oversold programs that will inevitably produce disappointing results.

Participants generally agreed that any successful large-scale program would have to include both material support systems and professional development, but there was no consensus about the appropriate balance between those forms of support, or about how they could most effectively be combined.  A closely related issue concerned adaptation to local conditions.  The successful experimental programs worked in part because they made use of local resources and were adapted to local conditions.  How can those qualities be preserved during large-scale implementation?  Which aspects of a program can successfully be published or franchised, and which aspects will have to be created locally for each professional community?

Adaptation to local conditions involves political as well as educational issues.  Local district leaders will have to be convinced to allocate their budgets for teaching materials and professional development for new purposes.  The reform ideas about science teaching and learning will have to compete successfully with other visions both in school professional communities and among students and parents.  Longstanding relationships between schools and communities will have to be renegotiated.

As Ruiz-Primo argues in her contribution to the conference web site, program evaluation will also play an essential role on large-scale implementation.  Formative evaluation will be essential for adapting programs to local conditions, and convincing evidence of student learning will be needed for programs to compete successfully for local resources.  This evidence of student learning will need to take a variety of forms, ranging from close assessments demonstrating that programs are achieving their stated goals to performance on large-scale assessment programs that are accepted by political leaders and the general public as evidence of science learning.

Finally, participants recognized that fundamental changes in the culture of school science such as those discussed in this conference would have to be in part generational changes.  Thus programs of preservice education and credentialing would need to play an important role in the development of large-scale changes in science education.



The participants in this conference came from different personal backgrounds and scholarly traditions.  The debates in the conference were lively.  It appears, though, that in those debates the participants found substantial common ground in their views about the models, materials, and conditions needed to support teaching science for understanding.  For example, the conference participants reached substantial consensus about the following points:

  • Students learn science with understanding when they engage, individually and collectively, in model-based reasoning about patterns in experience.
  • Students of all ages and social classes are capable of learning science with understanding in classrooms that help them to extend their experience and reduce it to order.
  • Good teaching materials can support learning with understanding in many important ways, but the development of those materials is necessarily long, complicated, and topic-specific.
  • Developers will have to choose between "covering" the entire contents of Benchmarks for Science Literacy (or other curriculum frameworks) and supporting learning with understanding by most students.
  • Teaching for understanding requires both substantial topic-specific learning by individual teachers and changes in professional cultures that would make science teaching a more communal and reflective enterprise than it is now.

In another respect, though, the conference was not as successful.  The participants in this conference did not find many "half-baked apples"-courses of large-scale action that would be seen as reasonable and practical by curriculum developers and policymakers while being acceptable to the researchers at this conference as reasonable halfway steps toward learning with understanding for all students.  The participants were aware that such a position leaves them vulnerable to the accusation that they are ivory-tower idealists, willing to participate only in projects that are impractical and unattainable in real schools on a large scale.  Although they share this concern, the conference participants were also concerned that the proposed "practical" alternatives simply won't work-they will not achieve scientific literacy for most students. 

To illustrate the dilemma faced by both researchers and policymakers, consider the fourth point above-the idea that teaching for understanding is incompatible with covering the entire contents of reform documents such as Benchmarks for Science Literacy.  Policymakers are already sensitive to the accusation that, since it includes less content than most textbooks, Benchmarks advocates a "dumbing down" of the science curriculum.  Suppose a group of researchers devoted years out of their professional careers to developing a program that supports true learning with understanding for a subset of the benchmarks while ignoring others.  What kind of market penetration could such a program achieve when schools must be responsive to high-stakes assessments based on all the benchmarks?  Would a funding agency be wise to support development of such a program? 

On the other hand, would a funding agency be wise to support development of a program that covers all the benchmarks by packaging them as facts and procedures?  The researchers' case that such programs are doomed to failure is well supported by both theoretical argument and empirical data.  We clearly need a policy-making mechanism for setting realistic priorities and support for research and development to act on those priorities.  It did not seem to the researchers at this conference that either the mechanism or the resources are currently in place. 

Thus this conference revealed differences in priorities and world view between researchers and policymakers that make coordinated action difficult.  Some aspects of current policy (including current high-stakes assessment programs) demand that our science education system achieve goals that are, in the view of the researchers at this conference, either impossible or undesirable.  The question of how researchers can best participate in a system where such policies play a dominant role was not resolved in this conference.   



References from the conference web site

Building a Sound Macroscopic Theory of Matter and Deeper Epistemological Understandings of Science Among Elementary & Middle School Students
Carol L. Smith, University of Massachusetts at Boston

Designing Systems to Support Learning Science with Understanding for All: Developing Dialogues among Researchers, Reformers, and Developers
Andy Anderson, Professor of Teacher Education, Michigan State University

Developing Understanding Through Model-based Inquiry
James Stewart, Jennifer L. Cartier, Cynthia M. Passmore, Department of Curriculum & Instruction, University of Wisconsin-Madison National Center for Improving Student Learning & Achievement in Mathematics and Science

Lessons Learned in the CIPS Curriculum Project
Pat Heller, Curriculum and Instruction, University of Minnesota - Twin Cities

Modeling Nature
Rich Lehrer and Leona Schauble, University of Wisconsin-Madison

Science Learning in the 21st Century: A Perspective from Cognitive Science
Christine Massey, Institute for Research in Cognitive Science, University of Pennsylvania

Some Ideas for the Conference Paper
Maria Araceli Ruiz-Primo, National Science Foundation

Strategic Approaches to Achieving Science Learning Goals
Edward L. Smith, Michigan State University

Using Conceptual Change Research To Reason About Curriculum
Glenn D. Berkheimer, Charles W. Anderson, and Steven T. Spees, Institute for Research on Teaching, Michigan State University

Other references

Cohen, D. K., and Hill, H. C. (2001) Learning and policy: When state education reform works. New Haven: Yale University Press.

Garet, M. S., Porter, A. C., Desimone, L., Birman, B., F., & Yoon, K. S.  (2001, Winter).  What makes professional development effective?  Results from a national sample of teachers.  American Educational Research Journal 38 (4), 915-945.

Kennedy, M. M. (1998).  Form and substance in inservice teacher education (Research Monograph No. 13).  Madison:  University of Wisconsin-Madison, National Institute for Science Education.  (available at

Loucks-Horsley, S., Hewson, P., Love, N., & Stiles, K. (1997).  Designing professional development for teachers of science and mathematics.  Thousand Oaks, CA:  Corwin Press.