Framework for articulating instructional practices and conceptions

We present and describe the development of a framework for assessing or categorizing instructional practices and related conceptions in the context of introductory physics instruction. The purpose of the framework is for articulating practices and conceptions of individual instructors or of curricula. It was developed based on the research literature and interviews with physics faculty and contains ideas with widespread acceptance as well as some outside mainstream rhetoric. We provide an example of how this framework can be applied by using it to categorize aspects of three well-known research-based curricula. The results of this categorization highlight areas that are common in research-based curricular designs and other areas that are less embraced but worthy of consideration.


I. INTRODUCTION
As McDermott and Redish allude in the introduction of their landmark resource letter, at the heart of all education research is the goal of "improvement." This resource letter is meant to contribute to the establishment of a research base that can serve as a resource for ongoing improvement and enrichment of student learning in physics ͑Ref. 1, p. 755͒.
After all, if one were of the opinion that the state of education was as it should be, then there would be no need to put effort into better understanding of how, when, or why students learn.The physics education research community has developed around this goal of improving physics education and believes that such improvement is possible.This goal is often more generally referred to as "reforming" physics instruction.
Although the idea of reforming physics instruction drives the field, rarely do we articulate exactly what is meant by reform.What does reformed instruction look like?What does it mean to improve?What changes are desired?Is it increased understanding of specific topics, 2 increased success at problem solving, 3 improved attitudes and beliefs about science, 4 more physics majors, 5 the building of community bridges, 6 or perhaps improved confidence in one's ability to critique and use science? 7hile words such as reform, improvement, and change are frequently used in the dialogue of the community, they are often not clearly defined, as if the physics education research ͑PER͒ community agreed on their meaning.The failure of the PER community to clearly articulate and discuss its beliefs, goals, recommended practices, and values can lead to misunderstandings and difficulties in the interactions among community members.
For example, a curriculum developer who views improvement as preparing more individuals to participate in the global economy would likely develop different curricula and ask different research questions than someone who views improvement as making physics more personally relevant to students.They may both embrace the use of cooperative learning, for example, but would likely implement it differently.In their conversations with each other, their unstated differences in values would likely lead to discontinuities in interactions.
In addition to talking with one another, curriculum developers must communicate effectively with non-PER faculty.If curriculum developers do not clearly articulate their goals, objectives, theoretical commitments, and resulting methods, this communication can fail.For example, faculty may not see the value of peer-peer interaction that is an important component of interactivity in some PER-based reforms, such as peer instruction, 8 and is based on the theoretical perspective of social-cognitive educational theories. 9Thus these instructors may attempt to implement reforms without this critical piece, 10 which would likely limit the effectiveness of the implementation.
There is nearly unanimous agreement that instruction can improve and many proposed ideas for how to accomplish this improvement.However, the specifics remain largely unarticulated and assumptions abound about what constitutes reformed instruction as well as the underlying goals and rationale for reform.In this paper we begin a discussion of these ill-articulated assumptions by presenting a theoretical framework for characterizing and discussing practices and beliefs associated with both traditional and research-based instruction.We briefly describe the framework and its development.We then provide an extended example of the use of the framework as an analysis tool for viewing three representative curricula developed within the field of physics education research.We do not intend for the framework presented here to be the last word on the categorization of the conceptions and practices of individuals or curricula.Rather we see this as an important first step in developing a tool to help in the articulation of these conceptions and practices.that there is no clear agreement about what constitutes reformed instruction and that articulating the extent of reform for a particular instructor or even a particular classroom was not an easy task.Much of the dialogue surrounding educational change is ill-defined.We therefore developed a framework of educational practices and conceptions in order to explicitly document the extent of alternative and traditional practices and conceptions of an instructor.While other frameworks, especially for instructional practices, are available in the literature ͑e.g., Refs.13-22͒, and we did incorporate ideas from these frameworks in our own work, we found that none of them were sufficiently detailed and inclusive.Such instruments are often focused on a specific vision of "good" instruction ͑e.g., "inquiry" 18 or "constructivist instruction" 13 ͒, rather than including a wide range of possible instructional practices.In addition, they tend to focus on actions that can be observed in a single class period, which can miss important aspects of instruction ͑e.g., grading practices, assessment items used, course design͒.

A. Practices
The framework was developed through an iterative process.We began by cataloging and clearly articulating alternative ideas that have been proposed either within the PER community or within the larger educational research community.As stated above, some of these ideas came from existing frameworks, while others were based on our knowledge of the literature and discourses in the field.For example, education research communities generally promote interactive engagement. 23Hence the first general dimension in the practices section of the framework refers to the degree of interactivity.Traditional instructional practices are typically characterized by minimal interactivity while alternative instructional practices are typically characterized by significant interactivity. 23owever, we found it was not sufficient to place "significant interactivity" on the alternative practices side of the framework because this term is vague and used with varying meanings.For example, even within PER, interactivity is used to refer to teacher-student interactions, 24 student-student interactions, 25 student-artifact interactions, 26,27 or studenttechnology interactions. 28,29We needed to clearly articulate the various and specific ways in which the term is used.In addition we further clarified the meaning of each practice by placing it in contrast to a traditional practice.For example, the practice of "significant student-student discourse" is in contrast to the traditional practice of primarily teacherstudent discourse.
Table I illustrates the details of the interactivity category of the framework and Table II presents all the main categories of the practices dimensions.The full practices section of the framework is given in Appendix A. Note that references are provided for the alternative practices.These are not intended to provide complete documentation for each idea, but rather a suggested starting point for those interested in reading more about a particular practice.

B. Conceptions
Likewise, we developed a framework for categorizing conceptions consistent with traditional or alternative practices.Although conceptions are less tangible than practice, and therefore problematic to measure, it is worthwhile to understand and document not just what people do, but also why they do it.
0][41] We use the term conception.This term is used quite broadly to describe any mental construct of an individual teacher that potentially provides a rationale for a particular instructional practice.Thus consistent with other researchers 39 we do not attempt to distinguish between beliefs, attitudes, goals, and other similar mental constructs.These are all referred to as conceptions.
Although it may appear on the surface that a particular instructional conception would lead to a particular instructional action, most researchers have failed to document such a relationship. 39,40,42,439][50] To illustrate the potentially complicated relationship between conceptions and practices, we will consider a different context.Surveys show that almost all Americans hold the conception that smoking is detrimental to health. 51Yet, some people with this conception avoid smoking while others do not.A smoker might agree that smoking is unhealthy ͑negative outcome͒ but also believes smoking will attract friends ͑positive outcome͒ and so choose to smoke despite a conception that would predict another action. 52Likewise some people might smoke despite a conception that their health is at risk because it is encouraged by their surroundings ͑advertising, role models, physical addiction, etc.͒. 53aving stated that the relationship between conceptions and actions is unpredictable and that conceptions are often inconsistent, we assert that consideration of conceptions is still worthwhile.Conceptions have been shown to be instrumental in defining tasks and selecting cognitive tools with which to interpret, plan, and make decisions regarding such tasks; hence they play a critical role in defining behavior and organizing knowledge and information. 45,47,54Thus while conceptions may not consistently predict action, they are not unrelated to action.To continue with the smoker example, a smoker with the conception that smoking is unhealthy may one day be motivated to change this behavior when other circumstances also change.Thus conceptions can add insight to behavior and also suggest strategies that may be successful in changing behavior.
We include conceptions in the framework that are theoretically connected to practice.Often these conceptions can be found in the literature that promotes a particular alternative practice.However, it is important to note that we do not assume that these conceptions and practices are correlated with one another in real instructional situations.For example, there is a logical link between having a constructivist view of knowledge acquisition and supporting an interactive learning environment.Likewise, an instructor holding transmissionist conceptions about knowledge acquisition might be expected to utilize practices associated with low levels of interactivity as described above.However, we do not claim that just because there is a logical link between a particular practice and conception that any particular conception actually leads to a particular practice-as noted earlier, other research has shown that there can be a great deal of inconsistency between an instructor's stated conceptions and their actual practice.In fact, when using the framework to categorize the conceptions and practices of five physics faculty we found many such inconsistencies.These inconsistencies often appear to be related to situational constraints the faculty encountered ͑such as a requirement to cover content͒ and are discussed in another paper. 12imilar to the organization of practices, we organized the conceptions section of the framework under general categories, followed by specific supporting examples to clarify the description.Table III presents the main categories of conceptions.As with the practices, references are provided for those interested in reading further about a particular alternative view.The full conceptions section of the framework is provided in Appendix B.
Again, although we organize the conceptions around "traditional" and "alternative" we make no claim that the conceptions on the traditional side are conceptions held by traditional instructors.We only claim that conceptions on the traditional side are logically consistent with traditional practice.In other words, if you start with a conception on the traditional side and ask, "What kind of instruction would one expect from an instructor who strongly holds this conception and has few environmental constraints?" the answer would be mostly on the traditional side of practice, likewise for the conceptions on the alternative side.
For example, if an instructor held an extreme positivist view ͑C3͒ of knowledges he or she would view knowledge as objective, knowable, and separate from human thought.Instruction based on a positivist view of knowledge would ͑i͒ be content focused ͑in order to pass on the objective knowl-edge͒, ͑ii͒ be teacher centered ͑since the teacher is the expert about the knowledge͒, ͑iii͒ be focused on certification and credentialing ͑need to document that students have learned the knowledge͒, and ͑iv͒ place no particular emphasis on social interaction in the classroom ͑it is not particularly important since knowledge is separate from human thought͒.These actions are all consistent with a transmission approach to knowledge development and all fall on the traditional side of the practices framework.
Likewise if an instructor held an extreme post-positivist ͑C3͒ view of knowledge he or she would view knowledge as dynamic and socially constructed.Instruction based on a post-positivist view of knowledge would ͑i͒ be process focused ͑to learn knowledge development͒, ͑ii͒ include social interaction in class ͑since knowledge is developed through social interactions͒, ͑iii͒ involve the teacher acting as a facilitator ͑rather than a source of knowledge͒, and ͑iv͒ follow a constructivist approach to knowledge development.These actions are all on the alternative side of the practices framework.
We do not claim that any instructor actually holds these extreme views; in our experience most instructors hold views consistent with both positivism and post-positivism.We offer these designations not as positivistic representations of reality but rather as post-positivistic theoretical constructs to frame and entice discussion.
Once we had a draft framework, we used interview data to test, refine, and expand the framework.These interviews, with five tenured physics faculty from four different institutions ͑one small liberal arts college, two regional universities, and one major research university͒ who teach introductory level physics courses, were collected as part of a project to better understand instructors' experiences with educational change and are reported in more detail elsewhere. 11,12The interviews were semistructured, exploratory interviews containing questions about instructional goals, current and past instructional practices, attempts to change practices, and familiarity with educational research.
The interview transcripts were carefully read for all instances of identifiable practices or conceptions of the interviewee.When a practice or conception was identified it was coded according to the framework.If the practice or conception did not fit well within the current framework, then the framework was revised or expanded to accommodate the new practice or conception.In this way, new dimensions emerged and old ones were described in better detail.Overall, we found that the initial framework was able to account for most practices and conceptions of our interviewees.Throughout this process, one or both authors independently performed the analysis which was then verified by the other.
Therefore our final framework is the result of both theory and experiment.It is based primarily in the research literature and informed by both our own personal notions of important practices and conceptions as well as use in the analysis of instructor interviews.

C. Comments on the framework
The framework is not intended to be a continuum on which an individual can be located.Although it is set up with contrasting practices and conceptions, there is no evidence or reason to believe that an individual will fall on one side or the other.In fact, it is common for individuals to exhibit behaviors and conceptions on both sides.For example, an instructor may lecture while teaching some topics yet use highly interactive techniques for other topics.Likewise, the same instructor could treat learning as primarily an individual activity in one situation but in another situation treat learning as a social activity. 62e present the framework as a distinction between traditional ideas and alternative ideas.We do no intend to pass judgment on the various practices and conceptions, nor do we propose any "ideal" location on the framework.While all of the ideas presented on the alternative side of the framework are there because there is some reason to believe they are worthwhile ͑for example, most have been empirically shown to increase student learning, at least in some context͒ we do not mean to propose that all traditional ideas are bad all of the time, nor that the alternative ideas are always superior.For example, while it is well established that lecturing is an ineffective way to develop conceptual understanding, it may be effective in some circumstances, such as when the goal is to transmit factual information. 63Our goal is to present a range of ideas and a framework by which individuals or curriculum can be understood.
In addition, the framework includes many alternative ideas that are uncommon even in most "reformed" curricula.It was our intention to include a wide array of research-based alternatives, not just those commonly found in the reform discourse.This allows the framework to be useful not just for describing individual instructors but also for describing the extent to which whole curricula, approaches, and even research communities deviate from traditional practice.For example, using the framework, we can consider what kinds of reforms are strongly promoted by the PER community and what kinds of reforms are not strongly supported.This type of analysis is useful for framing the field in a broader context as well as identifying potentially productive new lines of inquiry.

III. USING THE FRAMEWORK TO VIEW ESTABLISHED CURRICULA IN PER
In a separate paper, we show that the framework can be a useful tool to examine the conceptions and practices of individual instructors. 12In the remainder of this paper, we demonstrate how the framework can be used to examine the conceptions and practices embodied in specific curricula.In viewing PER curricula through the framework, we find many alternative practices and conceptions embraced to a large extent while other alternative practices and conceptions are embraced only weakly within the PER community.Using the framework to articulate trends in PER curricula can help us to better understand the current PER movement and suggest possibly fruitful future directions.Since PER is a diverse and complex field we will attempt to make the discussion more concrete by focusing on three established curricula designed to be used in introductory level physics courses.These three approaches were chosen because they are widely known and respected, proven successful, well documented and disseminated, and represent a broad range of the products of mainstream PER.These approaches are described briefly below.
Washington Tutorials. 64,65In the tutorial approach, the recitation often associated with large lecture courses is replaced with tutorial sessions in which students work cooperatively on carefully structured worksheets designed to address common misconceptions and build conceptual understanding.Students typically participate in tutorial sessions after covering the given topic in a more traditional lecture setting.
Interactive Lecture Demonstrations (ILD's). 66In an ILD activity, some lecture time is set aside for microcomputerbased demonstrations of concepts.In these demonstrations students are shown the setup, asked to make predictions as individuals about the outcome of the demonstration to be performed, and then discuss their predictions in small groups before being shown the demonstration.Students are then asked to describe the result and discuss again in small groups.
Workshop Physics. 67,68In Workshop Physics, there are no formal lectures.All course activities take place in an integrated laboratory environment where students are free to explore concepts with real equipment and work at their own pace.An activity guide structures the students' hands-on explorations and concept development.Students are explicitly encouraged to talk to one another and to collaborate on course activities.The instructor and the written activity guide serve to direct and facilitate student work and discussions.
In the analysis that follows we take each of these approaches and locate it within the framework for two practice dimensions ͑P1: interactivity and P2: instructional decisions͒ and one conception dimension ͑C1: learning view͒.We chose to discuss the practice of interactivity and conception of learning views because these are often discussed in the PER community and it is often assumed that the PER community and curricula are in strong agreement on the alternative side of these issues.We chose to discuss the practice of student input into instructional decisions because there is very little discussion about this in the PER community.For each dimension analyzed we used published materials to document and justify our placement in the framework.
It is important to note that categorizations do not necessarily represent the people who developed the various approaches but rather the practices and implicit or explicit conceptions expressed in the curricular materials.This becomes particularly important to note when considering the conception's dimensions.Just as an individual instructor's actions are influenced by a complex and, perhaps conflicting, set of conceptions interacting with their perception of their instructional situation, 12,62 it is reasonable to assume that the practices embodied in a curriculum represent some mix of the developers' conceptions and the constraints from the institutions and traditions in which they work.For example, a developer may fully embrace constructivism yet design a curriculum to be used in conjunction with some component of transmission-based lecturing because of situational constraints such as large class sizes.In this case we would consider the curriculum to represent a mix of traditional and alternative conceptions even though the developers, if asked, might espouse only conceptions on the alternative side of the framework.Additionally, while it is difficult to ascertain the "conceptions" of an entity which cannot be interviewed ͑such as a curriculum͒ many guiding philosophies are explicitly described in the curricular documentation and it is possible to make reasonable inferences about others.We argue that these embedded conceptions are more important than the personal conceptions of the developers since they represent how the curriculum will be experienced by potential users.
Tables indicating the details of our placement of each curriculum within each category, including the subcategories, are included in Appendix C. Table IV provides a summary of these placements.Each curriculum is rated on each dimension according to the following scheme: Traditional-Practice or conception is overwhelmingly traditional.Semitraditional-Evidence of some significant alternative practices or conceptions along with predominantly traditional practices or conceptions.Mixed-Significant evidence TABLE IV.Summary of three PER-based instructional approaches along three categories of the framework.

Washington Tutorials
Interactive Lecture Demos Workshop Physics

Interactivity ͑P1͒
Although the Washington Tutorials promote a highly interactive environment during the actual tutorial sessions, these sessions only comprise one component of the overall course.Since the tutorials were designed, and assumed to be used as a supplement to a traditional lecture-based course, much of the course experience is still likely passive for the students.Therefore we rate the tutorials as being mixed within the interactivity category.
As with the tutorials, ILD's are designed to be used in conjunction with a traditional, passive lecture.Therefore it is assumed that most of the course is traditional and passive.Although the ILD's integrate some interactivity even during the ILD activity the instructor is still largely leading discussion and performing activities with only minimal discussion among students.Therefore we rate ILD's as being semitraditional within the interactivity category.
Workshop Physics was designed to be a fully interactive learning environment and incorporates all aspects of interactivity.Therefore we rate workshop physics as alternative within the interactivity category.

Learning view ͑C1͒
The developers of the Washington Tutorials express strong conceptions consistent with a constructivist view of knowledge development.However, the explicit structure of the tutorials indicates some transmissionist philosophy.Additionally, the tutorials are placed after traditional ͑presumably transmissionist͒ lectures on the same material.Therefore we rate the tutorials as being semialternative within the learning view category.
While the developers of the ILD's do clearly express constructivist conceptions, the ILD's are embedded in an otherwise traditional lecture which is assumed to be grounded in transmission conceptions.In addition, although some peer-peer interaction is integrated into the method, the description of the ILD's does not emphasize the importance of social engagement in the learning process.
Therefore we rate ILD's as being mixed within the learning view category.
The developers of workshop physics express clear constructivist conceptions and explicitly dismiss transmission-based conceptions.Therefore we rate workshop physics as alternative within the learning view category.

Instructional Decisions ͑P2͒
We find no practices consistent with student autonomy expressed in the tutorials.In addition, many of the practices described were consistent with teacher control.Therefore we rate the tutorials as traditional within the instructional decisions category.
We find no practices consistent with student autonomy expressed in the ILD's.In addition, many of the practices described were consistent with teacher control.Therefore we rate the ILD's as traditional within the instructional decisions category.
Although there is some opportunity for student autonomy in the workshop physics environment most of the course activities are determined by the teacher.It is the teacher who decides content, assessment, and the larger frame of pace.When students do have autonomy it is either optional ͑individual projects͒ or only occasional ͑student-designed experiments͒.
Therefore we rate Workshop Physics as semitraditional within the instructional decisions category.
of both traditional and alternative practice or conception.
Semialternative-Evidence of significant traditional practices or conceptions along with predominantly alternative practices or conceptions.Alternative-Practice or conception is overwhelmingly alternative.

A. Embraced alternative practice: Interactivity
An area strongly embraced by PER is the practice of interactivity discussed in the development of the framework section above.The discourse in PER strongly shuns the traditional lecture-based approach characterized by students sitting quietly taking notes in favor of a classroom where students talk to each other, share their ideas and even engage in hands-on activities during class time.Although the three PER-based curricula that we analyzed are all often used as examples of interactive curricula, 23,61 notice from Table IV and Appendix C that they actually vary quite substantially on the degree and type of interactivity.

B. Embraced alternative conception: Constructivism
A strongly embraced conception in PER is the alternative learning view of constructivism.Just as with interactivity, constructivist conceptions abound in both the discourse and practice of the PER community.Also, just as with interactivity, the three PER-based curricula vary significantly on the degree and type of constructivist learning views embraced.

C. Nonembraced alternative practice: Student autonomy
Although evidence in the educational research literature points to student autonomy as an instructional practice associated with improved student learning, 69 this practice has not been widely embraced by PER.All of the curriculum we considered, including the otherwise radical Workshop Physics approach, were mostly or entirely traditional along the autonomy dimension.In fact, PER often removes autonomy over the learning process by strongly directing student behavior.

D. Discussion
While it might not be surprising to many readers that PER curricula often embrace interactivity and constructivist learning views, the differences between the three PER curricula in these areas suggest that it is important to clearly articulate these practices and conceptions.This facilitates discussion and comparison within the PER community as well as with those outside the PER community.The framework provides a tool that can be used for such articulation and also allows for distinctions to be made between different curricula.
One possible reason to articulate such distinctions is to compare results.For example, available evidence suggests that interactive practice results in higher learning gains 23 and that the more interactive the practice, the greater the gain 61,70 -see Fig. 1.While it may not always be possible to implement the most interactive curricula available, this evidence suggests that both instructors and curricula should attempt to be as interactive as possible given their other constraints.
Another possible use of the framework is to look for potentially fruitful future directions to explore.Since the development of interactivity has proven to be a very successful and important aspect of most PER curricula, it seems prudent to ask what alternative practices the PER community has not embraced.As described earlier, we find that student autonomy is not widely embraced in the PER discourse nor is it a practice evident in the PER curricula that we analyzed.
It is not our intention that this paper be focused on the issue of autonomy.However, having used our framework to illuminate a potential area of interest for the community, we believe it is illustrative to briefly discuss the implications of this finding.Therefore in the following section we will diverge for a moment from our discussion of the framework to argue that there is enough evidence in favor of the benefit of student autonomy to merit serious investigation by the PER community.

E. An argument for the consideration of student autonomy
According to our analysis of three PER curricula described above, all embody practices which are largely traditional with regard to who makes instructional decisions.There is, of course, nothing intrinsically wrong with being on the traditional side of practice.However, the extent to which practice is traditional or alternative should be an intentional choice.It is worth asking the question, are the goals of PER better supported by teacher control or student autonomy?As Weimer puts it, "Despite our extensive reliance on rules, requirements, and extrinsic motivators, we almost never ask whether these rules-oriented and require-that-you-do-it approaches are having an overall desired effect.We know they seem to work in the short term, but are they creating intellectually mature, responsible, motivated learners?"͑Ref.32, p. 92.͒Consider the goal of helping students become independent learners and thinkers who have the ability and confidence to critically examine ideas.For example, Redish states that "The task of the physics teacher today is to figure out how to help a much larger fraction of the population understand how the world works, how to think logically, and how to evaluate science.This is doubly important in democratic countries where a significant fraction of the adult population Student autonomy is more consistent with this goal because students need the opportunity to think independently if they are to become independent thinkers.In contrast, teacher control is more consistent with the goals of encouraging respect for authority and teaching discipline.As Karplus stated, "Students get a sense of control and responsibility ͓when given autonomy͔ that will encourage them to take more intellectual initiative in their studies and investigations in the future."͑Ref.71, p. 812.͒Although much of the rhetoric of PER is generally consistent with student autonomy, certain conceptions appear to pull most curricula toward teacher control.For example, common conceptions within the PER community are that students cannot effectively learn physics content without sufficient direction.Thus in practice, many PER curricula actually reduce the limited autonomy that students typically have under traditional instruction.Faculty that we interviewed voiced concerns about the higher degree of teacher control required by many PER curricula.They had reservations about forcing students to interact with each other, enforcing the use of a specific problem solving framework, or calling on specific students in class. 12For example, one instructor stated that he did not require peer to peer interaction as it limits student autonomy."That is my observation over many years of teaching, yes.That if I try to do it the ͓suggested way͔, I think they think they are being forced to do something they don't want to do…I think the different kinds of students want to do things in different ways" ͑Non-PER Faculty, Ref. 12͒.We note, however, that while traditional instruction may appear to foster student autonomy, students are typically given only superficial choices.While students may have the opportunity to opt out of participating in class, it is still the teacher who makes all of the important instructional decisions with little or no student input.
Thus it appears that student autonomy is commonly considered to be a practice that, while desirable in theory, is assumed to be not realistic to implement.However, there does not appear to be any research evidence to support this view.On the other hand, there is research-based evidence that not only can students effectively learn when given autonomy, but that their learning is improved.A review of the research on student autonomy indicates that when given choices students work faster, produce higher quality work, remember tasks better, are willing to work for longer periods, take greater pride in their work, are more creative, are more motivated, perform better on standardized tests, and improve their reasoning skills. 69In addition, there are a handful of schools that operate primarily around self-directed learning, such as Sudbury Valley, a school with no academic requirements.From a followup study of the graduates of Sudbury, Gray 73 reports that graduates "have had no apparent difficulty being admitted to or adjusting to the demands of traditional higher education and have been successful in a wide variety of careers.Graduates reported that…the school benefited them by allowing them to develop their own interests and by fostering such traits as personal responsibility, initiative, curiosity, ability to communicate well with people regardless of status, and continued appreciation and practice of democratic values."͑Ref.73, p. 182.͒Given the success PER has found by embracing other alternative practices, the research-based support for student autonomy in other areas, and the obvious contradictions between many goals embraced by PER researchers ͑e.g., critical, independent thinking͒ and the outcomes of teacher control, it seems reasonable to conduct research to determine what aspects, if any, of student autonomy might be profitably embraced by PER.
We do not believe this to be an easy goal to obtain.A student who has been told what and how to learn for 12 or more years will initially have great difficulty if placed in a strongly autonomous environment.Additionally, most instructors have institutional expectations ͑i.e., content coverage, inflexible structure of course meeting times͒ on them that make it difficult to give too much control to the students.However, moving towards instruction that supports student autonomy is possible.Small steps can be made, for example, by de-emphasizing grades as a motivation to learn 74 or by having students submit questions that they have about their assigned textbook reading rather than administer a reading quiz. 75While autonomy may be difficult to integrate, the potential benefits make the effort a worthwhile endeavor.

IV. CONCLUSION
We have introduced a comprehensive framework that can be used as a tool to articulate and compare the conceptions and practices of individual instructors as well as curricula.We have also shown an example of how this tool can be applied by categorizing PER curricula.We found that three PER curricula have generally embraced the alternative practice of interactivity over the traditional practice of passive students as well as alternative learning views of constructivism over traditional learning views of transmissionism.At the same time these curricula have not embraced the alternative practice of student autonomy over the traditional practice of teacher control although this practice appears to support goals ͑such as critical thinking͒ generally held by members of the PER community.
Much of the rhetoric in the physics education community centers on "improving" learning by "changing" instruction often by making the classroom more "interactive" or by focusing on "problem solving skills."But these terms are often used loosely and are ill defined.We offer our framework as a tool that can be used to begin to catalog and articulate many of these ideas.Specifically, the framework serves three important needs.

A. Document practices and conceptions of individuals and/or curricula
The framework can be used to identify practices and conceptions present in a particular setting.As shown in this paper, the framework can be used to document practices and conceptions of curricula.In a separate paper we show that the framework has applications for documenting the practices and conceptions of individual instructors. 12Additionally, it documents practices and conceptions in a way that both indicates the robustness of the practice or conception and identifies mixed modes.

B. Provide clear definitions
As mentioned previously, interactivity is one practice that has been strongly incorporated in mainstream PER research and development.Data collected by Hake indicates that "interactive engagement" is associated with higher learning gains. 23In his paper he defines interactive engagement as "designed at least in part to promote conceptual understanding through interactive engagement of students in heads-on ͑always͒ and hands-on ͑usually͒ activities which yield immediate feedback through discussion with peers and/or instructors."͑Ref.23, p. 65.͒This definition, however, is somewhat vague and does not address varying degrees of interactivity.As we have shown, our framework, with its detailed subcategories, offers specifics which can be used to more clearly define meanings of general terms such as "interactive." Further evidence ͑see Fig. 1͒ suggests that the "more interactive" the classroom, the greater the learning gain.Therefore it is useful to ask exactly what makes one curriculum more interactive than another.As we have shown, the concept of more interactive can be defined using the framework.For example, while a classroom that incorporates much student-student discourse is interactive, one that also utilizes student-artifact and student directed student-teacher interactions would be "more interactive."Additionally, the framework suggests that more interactive is also achieved by being "less traditional," i.e., less class time spent with only the teacher talking and the students taking notes.
These definitions help to identify specific aspects of a successful curriculum like Workshop Physics that lead to high learning gains.Clearly defining concepts also gives guidance to curriculum developers and instructors on specific aspects to incorporate in their curricular design.

C. Provide direction for future research and development
Not all alternative ideas proposed in the education research literature are common in the rhetoric or products of PER.As we showed in the discussion on student autonomy, something which may at first seem unworkable ͑giving students more control over their learning͒ may, in fact, be beneficial if properly implemented.Additionally, as we discussed, the abandonment or embracement of a particular idea can be based on conflicting conceptions ͑for example, that it is important that students become independent thinkers but that the teacher needs to tell students what they need to know͒.Use of the framework can make these conflicts ex-plicit as a first step towards their possible resolution.
As another example of the use of the productive identification of conflicting ideas, consider the subdimensions under the conceptions category C3: knowledge view: Knowledge is developed through the scientific method.
Knowledge is developed through social interaction.
While these two ideas are in opposition to each other, it is a rare instructor who will fall completely on one side or the other.Most people will exhibit a mix of these ideas, perhaps displaying one more prominently than the other depending on the context in which they are embedded at the moment.For example, in his classic book, Aarons discusses the role of logic, reason, hypothesis testing, and observation in the development of scientific knowledge. 76But he also repeatedly emphasizes that scientific knowledge is based in social constructions."͓Galileo's development of the concept of velocity dis-plays͔ several significant facets of the scientific enterprise: the roles of inductive and deductive reasoning; the fact that scientific concepts are created by deliberate acts of human intelligence and imagination and are not 'objects' discovered accidentally; that choice may be necessary and that there might be room for aesthetic criteria such as elegance and simplicity."͑Ref.76, p. 295.͒Aarons' statements indicated that he would probably agree with both of the above views, with one or the other gaining importance depending on the context to which applied.These views of knowledge, of course, have important implications for instruction.Through a framework offering oppositely articulated constructs we are able to identify such conflicts, which are likely to be productive areas for future work.
The framework presented in this paper can both identify areas of potential development of practice while also identifying conceptions that may hinder or support such development.Discussions of desired outcomes, new practices towards the achievement of these outcomes, and the identification and articulation of conflicting conceptions regarding these outcomes are healthy and beneficial.

FIG. 1 .
FIG. 1. Force concept inventory ͑FCI͒ results from nine schools with varying degrees of interactivity: traditional lecture classes ͑TRD͒, lecture classes supplemented with research-based activities ͑RBC͒, and Workshop Physics classes ͑WP͒.͑From Ref. 70, used with permission.͒

TABLE I .
Interactivity category of the practices framework ͑P1͒.

TABLE II .
Main categories of practices.
Formulaic problem solving: Problems assigned to students are well-defined and similar to problems students have previously seen Creative problem solving: Problems assigned to students are novel to solver and may have unknown or open-ended solutions ͑Refs.37,38͒

TABLE III .
Main categories of conceptions.
"Well they actually can't do it, the openended things completely overwhelm these students."͑PER Researcher No. 1, Ref. 72.͒There is also the conception that students must be forced to learn, "If you want students to learn something, you have to test them on it."͑Ref.61, p. 75.͒In addition, many argue that students are not capable of deciding what they need to learn-they need an expert to do that for them."But the students I don't think ͓should͔ decide what they need to learn and at what pace, because one, they don't know what they need to know." ͑PER researcher no. 2 Ref. 72͒.

TABLE V .
Practice categories of the framework.

TABLE VI .
Conception categories of the framework.
Students are best motivated intrinsically ͑make learning meaningful, connected to student's lives͒E Students cannot make good decisions about their learning needsStudents can make decisions about their learning needs