Journal of Science Education and Technology, Vol. 9, No. 1, 2000

Virtual Solar System Project: Learning Through aTechnology-Rich, Inquiry-Based, ParticipatoryLearning Environment1

Sasha A. Barab,2,4 Kenneth E. Hay,3 Kurt Squire,2 Michael Barnett,2 Rae Schmidt,2

Kristen Karrigan,2 Lisa Yamagata-Lynch,2 and Christine Johnson2

In this manuscript we describe an introductory astronomy course for undergraduate studentsin which we moved from the large-lecture format to one in which students were immersedin a technologically-rich, inquiry-based, participatory learning environment. Specifically, un-dergraduate students used 3-D modeling tools to construct virtual reality models of the solarsystem, and in the process, build rich understandings of various astronomical phenomena.For this study, primarily naturalistic inquiry was used to gain a holistic view of this semester-long course. These data are presented as two case studies focusing on: (1) the role of theteacher in this participatory learning environment; (2) the particular dynamics that formedin each group; (3) the modeling process; (4) the resources used, specifically student-developedinscriptions; and (5) the role of technology and whether learning the technology interferedwith learning astronomy. Results indicated that VR can be used effectively in regular under-graduate university courses as a tool through which students can develop rich understandingsof various astronomical phenomena.

KEY WORDS: Virtual reality; modeling; constructionism; astronomy.

INTRODUCTION

Throughout the 20th century, undergraduate in-troductory astronomy courses have been taught inthe large lecture format. This format allows universi-ties to present astronomy material to numerous stu-dents and allows departments to generate large num-bers of credit hours. However, we are currentlywitnessing increased criticism regarding the effective-ness of the lecture format for university introductoryscience courses (Baxter, 1991; Carr, 1997; Gilbert,1982; Solomon, 1983; Tobin et al., 1988). The Division

1A version of this manuscript was presented at the 1999 AnnualMeeting of the American Educational Research Association.

2Instructional Systems Technology, Indiana University.3Learning and Performance Support Laboratory, University ofGeorgia.

4To whom correspondence should be addressed, School of Educa-tion, Room 2232, 201 N. Rose Ave, Bloomington, Indiana 47405.e-mail: SBarab@Indiana.Edu

7

1059-0145/00/0300-0007$18.00/0 2000 Plenum Publishing Corporation

of Undergraduate Education at the National ScienceFoundation (1998) recently produced a document,Shaping the Future: New Expectations for Undergrad-uate Education, that calls for university faculty tomake a transition from an emphasis on deliveringcontent through large-class lectures to getting stu-dents ‘‘involved in some way in scientific inquiry, notjust a hands-on experience.’’ In astronomy, inquiryhas always been difficult because the phenomena areso far out of reach—students obviously cannot visitthe Sun. However, the power of the modern daycomputer to do desktop virtual reality and computa-tional modeling has created a new opportunity forinquiry approaches to learning (McClellan, 1996;Sabelli, 1994; Stratford et al., 1998) and for teachingastronomy (Hay et al., in press).

It is our contention that astronomy educationshould make a profound transition from an emphasison delivering content through large-class lectures toa focus on supporting students as they engage in

8 Barab et al.

authentic inquiry that involves the construction ofscientific models. Lectures, even with slides, over-heads, and films help affirm to students that theknowledge is someone else’s and, potentially,contributes to the knowledge becoming inert(Whitehead, 1929). The challenge of our design workis how to develop an astronomy course that allowsstudents to directly and concretely engage in scientificinquiry of astronomy concepts that they see of value.Although students cannot visit the Sun, planets, andother objects ‘‘out there,’’ they can model these ob-jects and their dynamics on a computer using VRtechnology. These technologies allow students to en-act basic astronomy concepts (e.g., tilt of the earth,period of orbit, etc.) into dynamic, 3-D scale models.Student-created models can then serve as a vehiclefor posing inquiry questions (When will an eclipseoccur? What would happen if I changed the orbitalperiod?) as they come to understand the solar system.

In the past year, we have been exploring thepotential of using 3-D modeling to create a learningenvironment in which introductory astronomy stu-dents build sophisticated models of many astronomi-cal objects, and in so doing learn a great deal aboutastronomy in an exciting way (Barab et al., 1998; Hayet al., in press). We have developed our researchagenda as a series of ‘‘design experiments’’ (Brown,1992), in which we engineer various design modulesthat are introduced as curricular constraints and thatoffer new learning opportunities for our students.The interactions related to these modules are thencaptured using video cameras, interviews, and docu-ment analysis so that we can trace the impact of themodule and evolve the course curriculum accord-ingly. The purpose of this article is to present dataregarding learning in our hands-on, project-based,introductory astronomy course in which studentsused 3-D modeling technology to build virtual so-lar systems.

In this study, we use qualitative methods to un-derstand learning in this context. We begin with a de-scription of our pedagogical commitment and the po-tential of virtual reality to support students learningin a constructionist framework. A description of thecourse and research context then follows, with the lat-ter giving rise to a discussion of the five emergentthemes that were central to, and framed, this research.Focusing on the five themes, we first present a reflec-tion on the entire class and then two case studies arethen presented. Reflections on the overall class andthe presented case studies provide the backdrop for abroader discussion of the educational implications.

PEDAGOGICAL FRAMEWORK

Currently an increasing number of educators areabandoning predominantly didactic, lecture-basedmodes of instruction and moving towards morelearner-centered models in which students, fre-quently in collaboration with peers, are engaged inproblem-solving and inquiry (Land and Hannafin,1996; Roth, 1996). This movement is partly in re-sponse to the continued criticism regarding the he-gemony of the lecture format, especially with respectto university introductory science courses (Baxter,1991; Carr, 1997; Gilbert, 1991; Solomon, 1983; Tobinet al., 1988). Many educators have argued that thelecture format concentrates on memorization of fac-tual information and promotes the development ofsuperficial understandings of the concepts (Roth,1996; Ruopp et al., 1993). All too frequently, thesedidactic models promote the development of knowl-edge that is non-transferable and that will be forgot-ten soon after the tests (Barab and Duffy, in press;Cognition and Technology Group at Vanderbilt,1993; Whitehead, 1929). Still others have stated thatlarge lecture formats do little to correct the manyalternative conceptions that students have regardingthe foundational concepts of science (Pfundt andDuit, 1991; Wandersee et al., 1994). Further, it hasbeen argued that such approaches have the ancillaryeffect of stifling creativity and diminishing enthusi-asm (Cordova and Lepper, 1996).

In response to the limitations of these teacher-centered or lecture-based learning environments,many educators are moving towards participatorylearning environments that support natural complex-ity of content, avoid over-simplification, engage stu-dents in the construction of products requiring prac-tices that embody complex concepts, encouragecollaboration, and present instruction within real-world contexts (Barab et al., 1998; Barab, 1999; Roth,1996, 1998). Predicated on a social constructivist phi-losophy, the role of ‘‘teacher’’ changes from one oftelling students correct answers to guiding and facili-tating learner activity (Bednar et al., 1992; Dewey,1963; Vygotsky, 1978).

In addition to requiring new roles of the facilita-tor, these environments are frequently collaborativein nature, requiring students to work with others asthey negotiate goals, tasks, practices, and meanings(Blumenfeld et al., 1996; Savery and Duffy, 1996).Working collaboratively involves the structuring ofthe learning environment so that students can worktogether towards a common goal (Nastasi and Clem-

VSS Project 9

ents, 1991; Sharan, 1994). As such, the role of groupdynamics and the need for supporting positive inter-actions in which all members have legitimate rolesto play become central (Johnson and Johnson, 1990,1994; Slavin, 1995).

Consistent with Papert’s (1991) constructionistpedagogical framework, our interest is in learningenvironments in which learners build understandingsthrough the collaborative construction of an artifactor shareable product. Constructionism builds on con-structivism in that it distinguishes itself from moretraditional instruction, in part, by the degree of activelearner engagement as well as the assumption thatlearners have the ability to create meaning, under-standing, and knowledge. Students are not passivereceptacles of the knowledge that teachers impart.Nor are they incapable of helping to develop learninggoals and discovering and developing meaning fromtheir authentic experiences. Papert (1991) arguedthat learners can and do construct knowledge andthat knowledge construction occurs most ‘‘felici-tously’’ when learners are engaged in the constructionof an artifact or shareable product. Thus, construc-tionism (for example, learning through the construc-tion of a virtual solar system), allows learners to de-velop their own reasoned interpretations of theirinteractions with the world. Perhaps more impor-tantly, constructionist learning environments allowlearners to share and collaboratively reflect upon theartifacts being built.

We refer to these environments as technology-rich, inquiry-based, participatory learning environ-ments for grounding understanding (TRIPLE-GU)(Barab et al., 1998). These environments take advan-tage of emerging technologies to establish participa-tory learning environments that immerse studentswithin contexts that challenge, ground, and, ulti-mately, extend their understandings (see Table I fora list of the central features). The emphasis of partici-patory learning environments is not the teacher’sfixed curricular objectives but rather the learners’emergent practices in relation to the need at hand.It is a move from a ‘‘teacher curriculum’’ to a ‘‘learnercurriculum’’ (Lave and Wenger, 1991), or from an‘‘acquisition’’ metaphor to a ‘‘participatory’’ meta-phor (Sfard, 1998).

Importance of Resources. Within participatorylearning environments, it is essential that teachersmake available, or support the development of, a richset of resources that students can use in their projects.These resources can be material, social, or concep-tual, and can include facts, instruments, phenomena,

Table I. Central Features of TRIPLE-GU

a. A central component of these environments is that they aretechnology-rich, integrating technology as a tool for facilitat-ing inquiry and/or other forms of authentic practice.

b. These environments must provide an opportunity for stu-dents to inquire into the phenomena they are learning, andnot simply receive information about the phenomena.

c. Rather than telling students about practices, our environ-ments are designed to support students in participating in do-main-related practices.

d. These environments are intentionally designed to supportthe process of learning.

e. It is out intention to establish rich environments (studios,workshops, construction spaces) where students work collabo-ratively, not isolated classes or places to listen to lectures.

f. These environments are intended to immerse students in a con-text that grounds their understandings to meaningful activity.

and theories that students use while doing their work(Roth and Bowen, 1995). By emphasizing the collab-orative nature of these learning environments, stu-dents are able to take advantage of the social re-sources (e.g., experiences of their peers, multipleperspectives on problems; division of labor) as theyinteract with peers. Newman, Griffin, and Cole (1989)have suggested that students working collaborativelyfrequently build understandings that go beyond theunderstandings that students generate in isolation.Further, vocalizing their current conceptions forcesstudents to organize their ideas, challenging the depthof their understandings.

Of particular interest to educators have beenresources developed by students themselves. Morespecifically, Roth and McGinn (1998a, 1998b) havefocused on students’ building of inscriptions, refer-ring to representations that ‘‘exist in material form(e.g., paper, computer screen) and can therefore beshared by several agents, [and can be distinguished]from mental representations, which are not publiclyaccessible’’ (Roth and McGinn, 1998a, p. 35). Inscrip-tions allow the organization, collaboration, and coor-dination of different group members’ contributions.The construction of inscriptions from available re-sources and group discussions increases students’competence in understanding scientific practices(Roth and McGinn, 1998b).

LEVERAGING ADVANCEDTECHNOLOGIES

Implementing 3-D Technologies in Learning.Current technological advancements make possible

10 Barab et al.

new types of learning experiences, moving fromtransmission models where technology functions likebooks, films, or broadcasts to environments in whichthe technology functions like studios and laboratoriesin which students immerse themselves within inter-active contexts that challenge and extend their under-standings (Allen and Otto, 1996). Many such technol-ogies have been discussed in the literature (CTGV,1993; Edwards, 1995; Jonassen, 1996; Koschmann,1996; Scardamalia and Bereiter, 1993; Winn, 1995).One exciting technology that has much potential inwhich to ground learning in rich environments is vir-tual reality (Dede et al., in press; Hay et al., in press;McClellan, 1996; Olson, 1998; Winn, 1995).

Virtual reality has the potential to immerse thelearner in various situations (the surface of the Moonor the delicate strand of the DNA molecule), visual-ize information (the temperatures of a frontal sys-tem), see hidden phenomena (forces directed on anobject or a tumor in a body), collaborate with peoplethousands of miles away (in adventure educationalgames or projects), and bring museum artifacts tothe hands of the learners. However, this technologyhas been, to date, mostly used by the military andthe aviation industry to help train soldiers and pilotsin complex simulators. In these contexts, soldiers andpilots are put into ‘‘real world’’ situations where theyhave an opportunity to experience and learn withoutlife-threatening risk. Only recently have educatorsworking with K-12 students begun to explore theeducational possibilities of VR learning environ-ments. Researchers examining VR in education havefound an encouraging array of positive learning out-comes in a range of projects and domains. A samplingof research findings includes better symbolic reten-tion of human cell organelle information (Gay, 1994),an understanding of atomic structure (Byrne, 1996),and an increase in low-achievers in drawing mentalmodels of ecology concepts (Osberg et al., 1997).

Using generic VR construction tools, we havebeen supporting students in building VR solar sys-tems and in the process challenging alternative con-ceptions (Hay et al., in press). In previous research, itwas shown that the process of building an interactive,dynamic virtual model of the solar system created apowerful and unique learning opportunity for stu-dents (Barab et al., 1998). In this pilot research, mid-dle-school students participated in a week-long campin which they built one of three projects: VirtualIndiana Statehouse, Virtual Theater, or the VirtualSolar System. Collapsing across groups, there weresignificant improvements in students’ knowledge

from the beginning to the end of the camp. It isimportant to note that these gains did not come fromdidactic lectures; rather, the increase in test scoresresulted from the completion of practices within thecontext of the larger project. With respect to theVirtual Solar System project, as students constructedtheir VR models, previous astronomical misconcep-tions were challenged, leading to the developmentof a more realistic sense of the relative interactionsof the Sun, the planets, and their moons. In additionto these astronomical understandings, students werealso engaged in and learning about the practice ofscientific modeling.

Science-Modeling as a Scientific Practice. Cur-rent computer advances are transforming science (Sa-belli, 1994) and the opportunities for learning science(Stratford, 1997; Stratford et al., 1998). Sabelli (1994,p. 197) called for ‘‘the science education communityto consider seriously the educational implications oftechnology-derived fundamental changes in scientificmethodologies.’’ In particular, the methods and pro-cesses of inquiry through computational-modelinghave dramatically changed science with the ever-in-creasing availability, power, and ‘‘affordability’’ ofgraphics computers. What started off in a few special-ized sub-fields has blossomed across the landscapeof scientific endeavors (Lehrer et al., 1994).

The advent of computational science brings withit new challenges for science educators (Jackson etal., 1994; Sabelli, 1994). We need models for engagingstudents in the process of computational sciences;more specifically, how do we construct learning envi-ronments where students build and visualize modelsto understand scientific phenomena? In a new erawhere scientific models and visualizations informpublic policy, illustrate points in the newspaper, andargue points of public interest, it is imperative thatstudents be familiar with the scientific process of com-putational modeling (Jackson et al., 1994; Rutherfordand Ahlgen, 1990).

THIS STUDY

For this study, primarily naturalistic inquiry wasused to gain a holistic vision of the semester longVSS course (Guba and Lincoln, 1983; Scriven, 1983;Stake, 1983). In addition to direct observation andfield notes, data were collected with four video cam-eras, one directed at each of the four groups. Eachresearcher attended each of the 25 one-and-a-halfhour classes, and was expected to continually main-

VSS Project 11

tain notes and, when appropriate, posed questions tovalidate interpretations. Also, in addition to two 20-minute interviews probing student understandings,students and teachers were questioned during thecourse to confirm and probe observations made inclass, and to gain better understandings about thevarious emergent issues.

Consistent with the work of Roth (1996), wecollected data that: (1) documented practices (e.g.,tool use, problem solving, student inquiry) and re-sources (e.g., concepts implemented, tools); (2) cap-tured the discussions among students and among stu-dents and teachers; (3) documented the progress ofstudent projects; (4) traced the same students, arti-facts, actions, and procedures over time; and (5) sup-ported and refuted emerging hypotheses about howpractices, resources, task constraints, task manifesta-tions, and student understandings evolved over time.The issues were continually refined during fieldwork,group meetings, and increasingly focused data collec-tion and analyses.

Lincoln and Guba (1986) recommended triangu-lation as one means of increasing the credibility ofinterpretations derived from naturalistic interpreta-tions. Data were triangulated using multiple datasources, including observations, interviews, docu-ment analysis, learner debriefing, and analyses of ref-erential materials. In meetings among the research-ers, field notes, learner interviews, and teacherobservations were discussed so as to generate asser-tions used to direct data collection efforts. In particu-lar, these meetings illuminated pertinent issues withrespect to the successes and challenges of the course.The issues deemed central to our pedagogical frame-work and which were most prominent at the end ofthe study were: (1) the role of the teacher; (2) theparticular dynamics that formed in each group; (3)the modeling process; (4) the resources used, specifi-cally student-developed inscriptions; and (5) the roleof technology and whether learning the technologyinterfered with learning astronomy (see Table II).

In addition to the researcher assigned to eachof the student groups, one researcher also carriedout two interviews and maintained field notes withrespect to the course instructor. In this presentationof the data, we begin with a general reflection on theclass as a whole, followed by two case studies that weview as representative of the four groups. Althoughthere were common emergent issues and a commonfocus, each researcher highlighted and presentedtheir case studies in a unique fashion. Following thesecase studies, we then reflect on the data to present

Table II. The Pedagogical Issues Investigated in this Study, Framedas Research Questions

Issue Research Question

Role of Teacher What was the role of the teacher in thisparticipatory learning environment?

Group Dynamics What were the dynamics that formed ineach group?

Modeling Process How did models facilitate students learn-ing astronomy?

Resources What were the resources, specifically stu-dent-developed inscriptions, used?

Technology What was the role of technology, and didlearning the technology interfere withlearning astronomy?

an overall summary of the course as well as the educa-tional implications.

THE COURSE CONTEXT

In this research, we have been exploringlearning/instruction within a collaborative, technol-ogy-based, student-centered learning environment—what we have labeled participatory learning environ-ments. The Virtual Solar System (VSS) project isan experimental undergraduate astronomy coursetaught at a large midwestern university. In the tradi-tional Introduction to Astronomy course, listening tolectures constituted the primary learning activity. Incontrast, in the VSS course, listening to lectures wasreplaced by students building 3-D models of differentaspects of the solar system using CosmoWorlds, aVRML editor, on average desktop personal comput-ers. Whereas immersive VR places students in thevirtual world, the software being used in this coursesimulates a 3-D environment on a 2-D screen, provid-ing the user with what McClellan (1996) referred toas a ‘‘window-on-the-world.’’

The curriculum was developed collaborativelyamong an astronomy professor, two educational psy-chologists, and a graduate student studying astro-physics and instructional systems technology. Twoprojects were engineered with the expectation thatstudents would model certain aspects on their com-puters during the semester. These were outlined inthe syllabus, which was passed out the first day. Theprojects were:

1) Project No. 1 was to model the Earth-Moon-Sun system. This included proper sizes, dis-tances between objects, surface features, cor-rect tilts of the bodies, and correct rotation

12 Barab et al.

and orbital periods. In addition, studentswere to provide a cut-away view or a trans-parent view that showed the interior structureof the Sun, Earth, and Moon.

2) Project No. 2 was to model the entire solarsystem, including both the terrestrial planetsand Jovian planets. Specifically, studentswere expected to make a model of the Sun,eight planets (Pluto and Ceres as options),six satellites (Moon, Galilean satellites of Ju-piter, Titan, and Triton), the Saturn ring sys-tem, and with the option of adding cometsand asteroids. Again, these bodies must havetheir proper orbits, sizes, colors, spin, dis-tances, and interior structures.

Student models were expected to address sylla-bus-delineated questions related to important astro-nomical phenomena. Each group negotiated plans toanswer the questions, identified resources (textbook,WWW, and scientists), designed and built their mod-els, evaluated them, used them to demonstrate an-swers to the initial questions, and to share them withother groups. Each project had four concluding activ-ities. First, teams created a joint paper describingthe features of their model. Second, each studentpresented and explained their model to students fromother groups in an automatic virtual environment(CAVE). The CAVE is a walk-in stereoscopic VRdisplay device that creates a total immersion experi-ence for the learner. Third, students engaged in grouppresentations in which they demonstrated the func-tionality of their model to the entire class, using anoverhead display in the regular classroom. Fourth,students wrote individual papers that compared andcontrasted their project with other projects in theclass and with the characteristics of the real solarsystem. This is a vital step in their learning about themodeling process. It is our position that if studentscan articulate the difference between their modelsand the real world they are demonstrating an under-standing of the astronomy they are describing at adeep level, as well as an understanding of modelingas a practice (Confrey and Doerr, 1994; Sabelli, 1994).

It was also necessary to ensure that students beprepared to pass a final examination on the same levelas students in other introductory classes. Towards thisend, students were given 60 questions from which thefinal examination would be constructed and whichwould serve as guidelines for their learning. Thesequestions, taken from review sheets used by the as-tronomy professor in his lecture-based classes, were

given to students on the first day of the class, just asin other classes, so that they could keep the courselearning goals in mind. Two of the authors wentthrough those questions and determined that stu-dents, potentially, would understand about 45 of the60 questions simply through building their models.

CosmoWorlds. Students used a virtual realitymodeling language (VRML) editor, CosmoWorlds,to build their 3-D models. CosmoWorlds is a multi-functional tool that allows students to create, manipu-late, texture, and animate shapes, group and ungroupobjects, create various view points from which to viewVR worlds, and add or modify light sources, amongother features. VRML was the WWW standard forVR and is a language similar to HTML in that itestablishes a common standard for making VR easilydistributed over the Internet. An example of aVRML 2.0 code is:

geometry Cylinder hheight 1 and radius 1j

This VRML line creates a cylinder 1 meter high and1 meter in radius that can be viewed on any computerplatform that has a VRML plug-in and a WWWbrowser. Instead of typing in the cylinder command,one simply drags a cylinder from the toolbox into theworkspace and sizes it with ‘‘handlebars.’’ Whereasadding a color or positioning the object anywhere inthe 3-D space would have taken four lines of codesimilar to the one above, this procedure takes theuser of CosmoWorlds only a few clicks and drags(see Fig. 1).

Fig. 1. Screenshot of the cylinder created in CosmoWorlds.

VSS Project 13

REFLECTION ON THE CLASS AS A WHOLE

During the first couple of weeks, just learningthe software occupied much of the students’ time.The decision was made by the instructors to allowthe students to explore the software and learn it asthey progressed through the projects rather than di-rectly teaching them the software. This decisionseemed to create anxiety among the students and,due to needing to learn the software, initially slowedtheir exploration of some of the richer astronomyconcepts. This was in part a product of our inexperi-ence with scaffolding appropriate astronomy goalstructures that would guide students in tackling andsystematically learning more difficult technologyskills (and astronomy concepts) within the affordan-ces of this new technology. However, it was also adesign decision based on the belief that this learner-centered approach would contribute to a creativemodeling building (versus replicating) ‘‘habit ofmind,’’ allowing students to develop skills that theywould be able to adapt more easily based on emerg-ing task requirements; thus, creating a classroom en-vironment that values student inquiry and creativityover teacher directed activity and student replicationof textbook knowledge.

The first project was intended to ground the stu-dents in both using the VR software and in funda-mental astronomical concepts. These goals were onlypartly reached, due to the instructor’s lack of under-standing of the software, lack of experience in teach-ing a constructionist-based class, and the conceptualand technical difficulties that the students were facedwith in the first project. Further, the use of the toolswas not ‘‘transparent’’ (Lave and Wenger, 1991), thuscreating a steep learning curve that potentially inter-fered and even competed with learning the astron-omy content. Yet, despite these problems, the stu-dents formulated insightful questions concerningboth astronomy and the effects of software limita-tions on their projects.

A particularly challenging problem for the stu-dents was the understanding of phases and eclipses.Several discussions were held with each group regard-ing what was needed for eclipses to occur. During oneexchange, there was much discussion on importanttheories, such as the Earth’s tilt, the Moon’s orbitalplane tilt, the Moon’s elliptical motion, and the rolethat the relative position of the Earth, Sun, andMoon plays in the occurrences of eclipses. Further,CosmoWorlds’ algorithm for handling the rotationof objects created a great deal of thought concerning

what needed to be done if their model was to accu-rately represent the rotation of the Moon. This dis-cussion was prompted by technical concerns, but re-quired students to grapple with astronomy concepts(see the design module discussion in the next sec-tion below).

At the end of the first project, the students weremore confident concerning their understanding ofCosmoWorlds and their astronomy knowledge asdemonstrated by their first papers and student inter-views. However, the increase in confidence was notuniversal among team members and, at least for thegroups of three, it was greater for the ‘‘team leaders.’’That is, in each team there was one individual whoemerged as leader, appearing to be central to themodeling process. These individuals submitted thebest papers and generally had more thought behindtheir arguments and comments than students whoplayed a secondary role in the development of theprojects. In addition to differences in overall confi-dence, there were also individual differences withrespect to their modeling responsibilities (e.g., mod-eling planetary orbits vs. modeling planet cross-sec-tions). Each student had the greatest degree of exper-tise with the content that they were responsible formodeling.

The second project began differently becausethe students learned from their mistakes and suc-cesses (as well as those of other teams) of the firstproject. Each team took some time before the secondproject to plan and develop strategies for buildingthe second project. The teams in general did morereading and researching before they began their mod-eling. From a science and learning perspective, thiswas refreshing because we were not only trying tofoster students’ understanding of the astronomy, butalso their problem-solving ability which starts withthe step of defining the problem at hand and devel-oping a strategy to solve the problem. In addition,each group developed inscriptions, a practice centralto doing science (Roth, 1998). The most commoninscriptions developed were tables in which they con-verted the distances and sizes of planets recorded inthe back of their textbooks to numbers that theywould use in CosmoWorlds.

As the second project progressed, the stu-dents become more and more disenfranchised withCosmoWorlds’ functionality and limitations. Themain problem was CosmoWorlds’ limitation of 3000frames (300 seconds) for any animation. Studentscontinually struggled with developing a scale that fitwithin the time limitation of the software and still

14 Barab et al.

demonstrated the important astronomical conceptsthat they wished to model. Further, the students hadto decide whether to create their models to scale,which was hard to visualize and work with inCosmoWorlds but better represented the real solarsystem than models not to scale. In determining thesize and distance of planets, students used mathemat-ical algorithms to instantiate numbers derived fromtheir textbook and the Internet into inscriptions offunctional value for building their models. We believethat working with issues of scale to build models hasmuch potential for understanding astronomy. Fur-ther, it allows students to see the value of mathemat-ics as a tool for addressing their needs, not thosedefined in a textbook. As one student said during thepilot work, ‘‘Wow, finally a use for math!’’

Another change in the second project was theamount of cross-project diffusion of knowledge/prac-tice. For example, students began to share their in-scriptions and to ask other groups how they accom-plished certain phenomena. On multiple occasions,all four teams would gather around one team’s com-puters so they could demonstrate how they tackleda particular modeling problem. One example of suchan exchange was with respect to modeling the propertilts of an eclipse. What was exciting in this case washow one group learned how to link objects and rotatethe linked Earth-Moon object, and another groupthen learned the subtleties of using local/relative cen-ter and view points from which to orbit and observethese rotations to experience an eclipse. They thenpooled their experience to produce their dynamicmodels. During this process, one of the students com-mented on the complexity of the interacting bodiesin our solar system: ‘‘It is amazing that they all don’tcollide.’’ This led to an in-depth discussion on gravi-tational pull.

On multiple occasions, the instructor was askedquestions he was incapable of answering to the stu-dents’ satisfaction (mostly concerning the software),and on several of these occasions, the students wouldinvestigate the problem further and develop theirown solutions. At different points during each proj-ect, the students would struggle with a particular as-tronomical concept or superficially discuss a topic.At these times the instructors would interject withshort, ‘‘just-in-time’’ lectures (under 10 minutes) torespond to student concerns. These lectures did notinvolve the instructor telling the students what to do,but rather helped frame the context and clarify theproblem. These lectures were usually proceeded by(and littered with) Socratic questions, facilitating stu-

dents’ formulating a better explanation or solutionto the problem at hand.

An exciting learning potential of models occurswhen students pose questions to their models (Con-frey and Doerr, 1994; White and Frederikson, 1998).In this manner, students can develop hypothesesabout the phenomena of interest and verify theseconjectures using their models. In our VSS coursewe saw this potential being actualized both throughstudent-posed questions and through Gedanken(thought) experiments posed by the course instruc-tors. An example of a Gedanken experiment oc-curred early in the semester. In this challenge, stu-dents had to use their understanding of the Moon’ssynchronous rotation with respect to the Earth (i.e.,that we always see the same side of the Moon) toanswer the following question: ‘‘If you walked out-side at night and could see a neon flag on the Moon,how many other clear evenings of the month wouldyou be able to see it?’’ In general, students wereunsure about the response, so the instructor devel-oped a model in which he placed a lit object on oneside of the Moon and then rotated the Moon aboutthe Earth. In this case, we used the CAVE so thatstudents could actually experience the Moon (andthe neon flag) orbit around them with the flag alwaysbeing viewable. However, one individual still was notconvinced that we would always see the same sideof the Moon, so he and his partner went back to theclassroom, put an object on their Moon, and put itinto orbit. It appeared that the student did not believethe instructor’s model and had to develop the repre-sentation for himself. At this point, the student statedwith confidence the correct response to the experi-ment: ‘‘every night.’’

Throughout the class, there were also stand-and-deliver sessions, intended to probe students’ under-standings of various astronomy concepts. Student pa-pers revealed deep understandings of astronomicalconcepts such as line of nodes, rotation of the Moon,and orbital motions. In addition, the papers revealeda considerable appreciation of the scientific process,in that students were developing, revising, and con-structing plans and hypotheses to improve their mod-els so as to better understand the concepts at hand.Surprisingly, students showed varying improvementin their geometric knowledge, and in understandingof scale. We have already begun the process of fine-tuning and adapting the class projects assigned andwe are confident that these changes will help thenext set of students in successfully mastering theseconcepts. One occasion in which students demon-

VSS Project 15

strated their understandings of scale occurred whenthey had to share their models with members of theother teams. During these occasions, students ques-tioned other groups, ‘‘How did you determine scale?’’or ‘‘Why did you choose those numbers?’’ At onepoint, one of the groups demonstrated their Earth-Moon-Sun system and a member from another teamimmediately saw what he believed to be a limitation,saying, ‘‘Your numbers can’t be good because theSun would be too far away . . . and you show it.’’The presenting group actually handed their table ofnumbers to the individual, and eventually she con-ceded, ‘‘the numbers look good.’’

The results of student performance on typicalmultiple-choice-question final exams and follow-upinterview questions were also examined (Keating etal., 1999). To this end, students were given the samefinal exam as a previous lecture-based class. The VSSstudents performed better on questions that requiredconceptual understanding rather than just memoriza-tion of facts in comparison to students from the tradi-tional class. For example, students developed richunderstandings of the relationships between colorsand temperature, of differences between sidereal andsynodic periods, and of the relationship between theorbital alignments of the Sun, Earth, and Moon, andwhether or not eclipses occur. The largest differenceoccurred on questions that challenged the studentsto change their frame of reference. For example, thestudents were able to predict what phases the Earthwould have when viewed from the Moon or evenanother planet. This ability cannot be understatedbecause it is fundamental to understanding much ofscience—particularly at a more advanced level.

CASE STUDIES

The Three Ninja Turtles: LearningThrough Modeling

Group Description. This team consisted of threemale members: Donatello, Leonardo, and Rafael.They ranged in computer experience, astronomybackground, and educational level. Donatello was agraduate student with fair computer knowledge, butonly basic astronomy knowledge. Leonardo, an up-per classman, entered the course with good computerexperience and some astronomy knowledge, havingalready completed one college-level astronomycourse. Rafael, an underclassman, entered the course

as a relatively inexperienced computer user with nobackground in astronomy.

As the course progressed, each student began totake on specific roles within the group. After only afew class sessions, Donatello, the graduate student,emerged as the group leader. He initiated most of theplanning activities, and exerted a strong influence onthe direction of the project. A key point in Donatello’sdevelopment as a leader occurred early in the secondproject, when Donatello conceived, designed, and de-veloped a master planning document for the group.This document contained all of the data that the groupneeded to develop the second model, including thesizes, distances, orbital periods, and rotational periodsof the planets and major moons. Thus, Donatelloquickly became the gatekeeper of astronomy informa-tion for the group, although Leonardo also served asan astronomy resource throughout the semester. Do-natello also engaged in the most discussions with in-structors, both in and out of class, adding to both hisastronomy knowledge and knowledge of CosmoW-orlds. As a result, Donatello became a CosmoWorldsresource for the group, helping other group memberslearn to position celestial bodies, create animations,and create viewpoints.

In contrast to Donatello’s strong presence in thegroup, Rafael was much less involved in group pro-cesses. Rafael attended fewer classes, and during thesecond project he missed four of six classes. In bothprojects, Rafael focused on creating the cross-sec-tional views of the planets, which limited him toastronomy tasks that involved more direct input offactual data as opposed to the creation of orbits,rotations, and complex astronomical phenomena. Asa result, Rafael was called upon only as a resourcefor using the CosmoWorlds tool to create cross-sec-tions and very rarely for astronomy-related content.In the interviews, Rafael commented that he knewthe interiors of the planets much better than otherastronomy concepts, and would essentially have tolearn the other astronomy concepts on his own out-side of class time.

The final group member, Leonardo, was fairlyinvolved in team-building activities and the buildingof the virtual worlds. In Project One, Leonardoworked with Donatello on many of the model plan-ning and building tasks, although he tended to workmore independently in Project Two, serving as a lesscentral member of the group. Leonardo engaged inthe full range of tool-related practices and delvedinto a wide range of astronomy related concepts. Forexample, Leonardo created the complex Jupiter and

16 Barab et al.

Saturn moon systems, affording him opportunities towrestle with issues of relative scale, orbits, retrograderotation, and cross-sectional views. In building thesemodels, he engaged in all of the critical tool-re-lated practices.

Project One. On the first project, the team collab-orated on many of the model-building and planningtasks. Leonardo and Donatello were especially pro-ductive together, creating most of the group’s orbits,scales, and animations as a team, which affordedthem opportunities to discuss and reflect on conceptsas a dyad. For example, when attempting to modelthe Earth’s orbit around the Sun, Donatello andLeonardo experienced difficulty maintaining thecorrect tilt of the Earth. After much deliberation,they decided that this flaw in their model was impor-tant because it prevented their model from depictingthe seasons in an accurate manner. Before arrivingat a solution, they consulted texts, charts andtables, developed makeshift physical models withmousepads, and called upon the instructors to answerquestions. In doing so, Donatello and Leonardo be-came deeply engaged using astronomy resources tothink about astronomy-related content. Similar dis-cussions ensued as Donatello and Leonardo at-tempted to model the Moon phases and incorporatethe Moon’s 5-degree tilt from the plane of the eclipticinto their model. Rafael was very rarely a part ofthese discussions, as his work concentrated on creat-ing the interiors of the Earth, Sun, and Moon.

In interviews, the team agreed that much, per-haps even most of their energy in the first projectwas dedicated to learning CosmoWorlds, engagingin relatively little astronomy-related content. Learn-ing to create viewpoints, for example, was an espe-cially difficult skill for group members to learn, withseveral hours of class time devoted solely to learningthis process. On several occasions, Donatello andLeonardo had to redo entire animations because theydid steps out of order in setting viewpoints. Othertimes, Donatello and Leonardo both felt that theyunderstood the astronomy concept that they wereattempting to model, but CosmoWorlds limited theirability to efficiently build models. For example,Donatello and Leonardo spent considerable time at-tempting to model the Moon’s 5-degree tilt aboutthe plane of the ecliptic, redoing their entire modelseveral times in the process. After hours of experi-mentation, they developed a solution, giving theEarth an 18-degree tilt, then adding on the Moon’sorbit and finally tilting the whole system another 5-degrees to get the Earth’s proper 23-degree tilt as

well as the Moon’s 5-degree tilt around the Earth inthe same model (see Fig. 2 for a screen shot of thevirtual Earth with a 23-degree tilt).

While the limitations of CosmoWorlds impededtheir progress and caused some frustration, it alsoforced the team to grapple with astronomical rela-tionships in very meaningful ways. In another exam-ple, the group had difficulty with CosmoWorlds whenmodeling the orbits of the Earth, Sun, and Moon.Donatello and Leonardo spent most of a class periodattempting to track the Earth’s orbit around theMoon, but the relatively small size of the Earth andthe vast distance between it and the Sun made itdifficult to find. At one point, the group was unableto determine if they had modeled this phenomenoncorrectly, because they could not see the Earth intheir model. Finally, they reconstructed their modelwith the proper viewpoints in order to prevent losingcelestial bodies, and learned about key astronomicalconcepts, such as the ‘‘empty’’ nature of space, thedramatic size differences between the Sun, Earth,and Moon, and the vast distances between celestialbodies.

Project Two. Fresh from their experiences inProject One, the team took a very different approachto Project Two, spending multiple class periods de-fining the tasks and approach for completing thisproject. Drawing from their difficulties in incorporat-ing all of the relevant aspects of the Earth, Sun, andMoon system into one model, Donatello proposed afour-level approach to modeling the solar system.The first level attempted to model the relative size,scale, and distance of the planets without any anima-tions. Level two would also include the animationsof planets around the Sun, the accurate relative sizesof the planets and the accurate scaling of distances,but would dramatically shrink the distances betweenplanets so that multiple planets could be viewed on

Fig. 2. Screenshot of the Earth object with the 23-degree tiltbeing added.

VSS Project 17

the screen at once. Level three would contain themoon system of each planet, and level four wouldcontain the interiors of each planet. Each level wouldbe connected by links, making the model appear tothe end user as one seamless model. The team washopeful that this multi-leveled approach would allowthem to view important astronomical concepts andrelationships in their models without making fac-tual compromises.

The amount of energy this team invested in plan-ning Project Two also manifested itself in the team’splanning document of inscriptions. Immediately fol-lowing the completion of Project One, Donatello re-alized that it would be valuable to have all of the keyinformation for the project in one central locationthat the team could use to build their model. Hecreated a document containing information about theplanets, with all data converted into values that couldbe easily plugged into CosmoWorlds. This conversionprocess involved much planning and calculation. Forexample, in depicting the planets’ orbits, Donatelloneeded to develop the optimal mathematical factorby which he could collapse the distances of the plan-ets from the Sun so that the inner planets were view-able, while still retaining enough distance betweenthe planets so that Mercury was not swallowed bythe Sun. The understandings embodied within thesheet were developed by Donatello, with Rafael andLeonardo contributing little to this process. As a re-sult, Donatello soon became the gatekeeper of infor-mation and emerged once again as the leader ofthe project.

With the planning for Project Two completed(mostly by Donatello), the group functioned muchmore independently in the actual model building pro-cess than they had in Project One. Donatello handledthe most complex activities, creating the relative sizeand scales for the planets, the planets’ orbits, devel-oping interesting viewpoints that would depict as-tronomy relationships, and building complex extrafeatures into the model (levels one and two of themodel). Donatello spent several days thinking of andcreating viewpoints that would yield interesting as-tronomy insights, which led him to several engagingdiscussions with the instructor. Donatello furtherelaborated the model by creating the Neptune, Plutoand Charon system, which includes the elliptical orbitof Pluto and the twin system of Pluto/Charon. Theanimation attempted to model some of Kepler’s lawsby showing how Pluto’s velocity increases as it ap-proaches the Sun, and slows it as it moves towardthe furthest points of its orbit. Although Donatello

gained a deep understanding of those aspects he mod-eled, he later commented that he missed the opportu-nities to learn about planetary composition that camewith building the planets’ cross-sections. In inter-views, he lamented that in preparing for the finalexam he would have to learn those concepts and factson his own.

Leonardo also engaged in fairly complex model-ing activities, building the moon systems and cross-sections for the outer planets. He developed themoon systems for Jupiter and Saturn, as well as Ura-nus’ unique rotation. Leonardo elicited help fromDonatello with some of the more complex modelingprocedures, such as the Saturn moon system and Ura-nus’ rotation. In contrast, Leonardo only asked forhelp from Rafael in purely tool-related practices; heoften asked for help in operating the coloring andlabeling functions of CosmoWorlds. In the elab-oration phases of the project, Leonardo gravi-tated toward activities that demanded mastery ofCosmoWorlds, but not necessarily rich astronomyunderstandings. Leonardo added an asteroid belt forthe project, seeming to enjoy himself as he createdthousands of tiny objects and began animating themabout the solar system. Inspired by the movie ‘‘DeepImpact,’’ Leonardo even began an animation thatfeatured a rocket being blasted toward an asteroidthat was hurtling toward the Earth. This activity af-forded the instructor an excellent opportunity to pushLeonardo toward making a realistic model of thisevent. In doing so, Leonardo quickly understood whythe high velocity and relatively small size of asteroidsmake intercepting an asteroid that is headed towardthe Earth nearly impossible.

Rafael engaged in the least complex modeling ac-tivities of Project Two, specializing in creating thecross-sections of moons and the inner planets. Model-ing the inner planets allowed Rafael opportunities togain expertise in tool-related practices such as labelingand coloring, but very little experience in setting view-points, creating animations, and understanding orbits.

Conclusions. In closing interviews, the partici-pants were pleased with their performance as a group,and with what they learned in the course. All teammembers expressed satisfaction with the dynamics ofthe group, feeling that they got out of the group whatthey put into it. The team seemed to appreciate andvalue Donatello’s role as a leader, and no reserva-tions about team member roles were expressed. Allteam members expressed some concern that theymissed opportunities to engage in some areas of as-tronomy content by specializing in others. Likewise,

18 Barab et al.

each team member agreed that too much of theirfocus was placed on learning the tool during thiscourse. Optimally, the team would like to see moretutorials, job aids, and direct instruction on usingCosmoWorlds. As Donatello commented, ‘‘If thepoint of the course is to get us to learn to use comput-ers, then fine, but if it’s astronomy, just show me howto use it, and let me concentrate on the astronomy.’’

The group members also agreed that the directinstruction, such as lectures, on astronomy-relatedcontent, were quite valuable. They agreed unani-mously that the professor’s just-in-time lectures werean enjoyable aspect of the course, and they expresseda need for more lectures on other topics, especiallyover material that was not covered through buildingtheir models. In comparing their experiences in thiscourse to other more traditional courses, group mem-bers felt that they understood what they learnedmuch more in depth. For example, Rafael felt thathe never really understood tilts until he was forcedto model them. In summary, Leonardo commentedthat ‘‘Other students might learn more, but a yearlater, I will remember more of it. In other classes, Ididn’t remember much.’’

Butch Cassidy and the Sundance Kid:Technological Challenges

Introduction. Butch was a member of a fraternityand Sundance was a student from the local highschool. Butch was typically the leader of the groupand the one who was most excited about the innova-tion of the course. Butch often persevered throughthe project challenges that would flare up from timeto time. Sundance was eager to learn, but was muchquicker to get frustrated. They both attended classregularly and engaged the projects thoughtfully.Butch had some computer background through hisbusiness degree program. Sundance was comfortablewith using the technology, but did not have any spe-cial computer training.

Opening Challenges. One of Butch and Sun-dance’s first challenge was to define the center of thesolar system. After developing skills at creating andsizing spheres in CosmoWorlds, they began movingthe planets with the direct manipulation interface.Sundance created the Sun and put it into the centerof the screen, which he interpreted as the center ofthe solar system. However, when he used theCosmoWorlds ‘‘thumb wheel’’ tool, the Sun wobbledlike a poorly-centered piece of clay on a potter’s

wheel. He reasoned that it was not in the center, sohe moved the thumb wheel to the point where it waswobbling the least and dragged the Sun toward thecenter. Sundance spent a lot of time on this andcame close to centering it. He asked Butch periodicquestions and at one point Butch used a ruler to helpfind the center. It was revealed that this practiceemerged as a result of a particular artifact of thesoftware. CosmoWorlds allows the user to ‘‘grab’’the entire virtual space and spin it like a globe. Thisis one of the first features Butch and Sundance cameto understand and they used it as their first ‘‘model’’of how to create planetary orbits—it allowed themto create a sphere that was not at the center of the‘‘virtual space’’ and to make it spin around its center.However, this model had a number of operationalflaws. The first flaw was that the orbit speed wasdetermined by the speed which Sundance or Butchwould click, drag, and release the mouse button. Thatis to say, it was not based on data about a particularplanet’s orbit. Another flaw in their first model wasthat unlike a globe, the virtual space could be spunaround a single point instead of an axis. Thus, thedirection of the click, drag, and release of the mousewould determine whether the virtual space wouldspin about the X-, Y-, or Z-axis or somewhere inbetween. Finally, the model was flawed because whenmore planets were added, the period of their orbitswould be exactly the same, which is in fact not thecase.

Although these flaws are profound, at this earlypoint they did not concern Sundance because oncehe accepted this model, his primary concern was toput the Sun in the center of the world. Sundancedefined the center, in his mind, as the point whereyou could place an object and it would not movewhen he used the thumb wheel to rotate the virtualspace about the X-, Y-, or Z-axis (the center of the‘‘potter’s wheel’’ in 3-D). This proved to be difficultbecause the center of the ‘‘potter’s wheel,’’ and of 3-D space is hard to find. Also, the Sun was the centerof the current viewpoint of the world, not the absolutecenter of the virtual space, adding a second elementof complexity. This second issue created much frus-tration because when Sundance moved to a new loca-tion that changed the scene, he basically had to startover because the viewpoint and its center changed.Conceptually, Butch broke through the shortcomingsof this approach when he challenged Sundance tocreate the Moon’s orbit. Adding a second object be-came problematic because this process of creatingorbits relied on the creation of a new center in the

VSS Project 19

virtual space. In other words, every time he addedan orbit he had to change the center of the universe,which affected the accuracy of previous orbits. TheMoon orbits the Earth, not the Sun. The challengeforced Sundance to abandon his first model of a plan-et’s orbit.

Project Two. Project Two was marked by severalevents. First, there was an intense planning sessionwhere Butch and Sundance sat down and created anelaborate plan where the planets, moons, and, mostimportantly, viewpoints were going to be placed. Theexperience of Project One gave them a clear under-standing of what they wanted to create and the neces-sary approach to take. The second event was thatButch and Sundance’s skill level was at a point thatthe technology became ‘‘ready at hand’’ and even‘‘transparent’’ (Lave and Wenger, 1991). That is tosay, it no longer was the primary focus, but ratherwas a tool to see and model the rest of the solarsystem. They successfully moved from frustrated nov-ices to rather accomplished modelers.

Technical Struggles. One of the more difficulttasks Butch and Sundance engaged in had very littleto do with astronomy. This activity was the labelingof cross-section layers and arrows connecting labelsto the layers. This seemingly easy task took consider-able effort because they were working in three dimen-sions. When they put a label and a connecting arrowon an object, it would frequently not be on the sameplane as the planet. So Butch and Sundance wouldwork and create all the labels, but when they movedto another perspective, the labels and arrows wouldbe floating off into space, not pointing at anything.This could have been avoided through the software,if it were to default to putting labels on one of theorigin planes (X, Y, or Z planes) or the plane ofthe object. However, Butch and Sundance eventuallybegan to use all three dimensions, but time was takenup in trying to position these labels.

Tool-Related Practice. The practice of settingviewpoints created an interesting dilemma for Butchand Sundance. Viewpoints are the points that directthe view of the person looking at their model.CosmoWorlds uses the metaphor of a camera to de-fine these viewpoints. You place a camera at a specificspot and name it, then when people go into yourworld, they select the viewpoint by name and arequickly transported to that position looking in thedirection that was pre-specified. Conceptually, view-points are very important for students’ understandingof astronomy. Like most scientific observation, find-ing the correct location is the key to understanding,

and in the huge vastness of space, you need to look atsomething in a particular way to understand specificrelationships. For example, a quarter-moon does notmake sense from just any position. The observer hasto be in the right place, and looking in the rightdirection at the right time to see a quarter-moon.Viewpoints create the opportunity for students toconstruct observations of the Earth-Moon-Sun sys-tem in the quarter-moon position as we see it fromEarth, from the Sun, from the Moon, and from outerspace. In constructing these viewpoints and viewingthe Earth, Moon, and Sun from these positions, stu-dents can build understandings. In other words, theyused viewpoints to ask questions of the model.

Butch and Sundance found the use of viewpointsto be difficult. They quickly mastered the settingof global viewpoints and moving between them inCosmoWorlds. Global viewpoints are viewpoints thatare fixed at an absolute position relative to the world.The challenge came when they wanted the camerasto move with the objects. These types of viewpointsare called local viewpoints. Local viewpoints are criti-cal to see most of the interesting relationships of theEarth-Moon-Sun system because these interesting re-lationships involve the movement of the Earth andMoon around the Sun. If you put a viewpoint lookingat the Earth and Moon at the starting position, assoon as you enter the world, the Earth and Moonwill immediately disappear and will only return toyour view once a ‘‘virtual year’’ has passed.

Butch and Sundance worked collaboratively,further refining their abilities by learning how to‘‘make visible’’ the viewpoint icons (cameras) in theirworld. While it is not important to see the ‘‘camera’’in setting global viewpoints, you can just move your-self into position and set the global viewpoint to yourcurrent position, when you create and animate localviewpoints seeing if and where the camera is movingis critical. Another snag is that local viewpoints aregrouped with an object, but, if the object is animated,it will not move in the same way the group moves.This caused Butch and Sundance much frustration,to the point where they almost gave up. At one point,Butch said ‘‘Let’s just set the dynamics and screwthe viewpoints. I hate viewpoints, I hate this, I HATEYOU!’’ to the computer. There was some significantintervention by the instructor to help them workthrough the issues, to attempt several different fixes,and finally to come to a resolution. The solution wasto simply treat the camera as any other object androtate it at the same rate as the objects they wantedthe viewpoint to illuminate.

20 Barab et al.

Conclusion. Sundance and Butch had a conver-sation somewhere in the middle of the course thatwas recorded on videotape. During a moment of frus-tration, Sundance asked Butch why they are takingthe course using virtual reality. Butch replied that hewas excited to be in this experimental course andthat he would be putting this course on his resumebecause he thought that it would help him get a jobin business. By the end of the course, Sundance cameto understand the value of what they were doing,stating that this was one of the most challenging andrewarding courses he had taken. Additionally, andmore importantly, both students learned astronomyas evidenced in class presentations, reflection papers,class discussions, and on the final examination.

DISCUSSION

The VSS course was designed to support stu-dents as they constructed concrete artifacts and, inthe process, rich understandings of astronomical phe-nomena. Specifically, our data collection focused onthe following issues: (1) the role of the teacher inthis participatory environment; (2) the particular dy-namics that formed in each group; (3) the modelingprocess; (4) the resources used, specifically student-developed inscriptions; and (5) the role of technol-ogy, and whether learning the technology interferedwith learning astronomy.

Role of Teacher. In constructivist learning envi-ronments, the teacher’s role is reconfigured, fromdidactic caretaker and keeper of knowledge to a facil-itator of the knowledge construction process, direct-ing students down profitable paths, modeling an en-gaged mind, problem-solving with students, andproviding a rich context with needed resources (Sav-ery and Duffy, 1996). Most interactions between theinstructors and the students were Socratic in natureand were centered on the students’ model or themodeling process. When the instructor observed stu-dents struggling with a difficult astronomical concepthe would ask the students the question: ‘‘What doesyour model say?’’ If the students could not answerthis question by manipulating their model, the in-structor would ask the students what needed to beadded to their model to answer the question. Thisdialogue would lead the students to modify theirmodels until they could answer the questions (includ-ing more formal Gedanken experiments) by studyingtheir model. Most importantly, these questions al-lowed students (and instructors) to determine gaps,

and to aid students in improving their models (andtheir understandings) without students losing owner-ship over their models and the learning process.

In general, the instructors maintained this roleof facilitator. However, the instructors did presentseveral ‘‘mini’’ or what the CTGV (1993) called ‘‘just-in-time’’ lectures throughout the course. These‘‘mini-lectures’’ were delivered when students ap-peared confused or frustrated, or needed to under-stand a particular astronomy concept to continuetheir work. Although these lectures occurred mostoften in response to individual group concerns, occa-sionally when the instructors would observe similardifficulties across multiple groups they would presenta just-in-time lecture to the entire class. Anotherteaching technique that the instructor used was tohave ‘‘stand-and-deliver’’ sessions in which one stu-dent would attempt to describe challenging conceptsto the rest of the class. If the student failed to demon-strate a satisfactory understanding, another ‘‘stand-and-deliver’’ session would be held the next classsession, allowing students time to research the issueand formulate a response to the instructor’s ques-tions.

Group Dynamics. There appeared to be verydifferent group dynamics both between groups andamong groups over time. The most obvious differ-ence was between the groups composed of two stu-dents and groups of three students. Groups of twostudents tended to work collaboratively over the en-tire semester. In contrast, in the groups of three, onemember of each team appeared to be central to themodeling process. The leaders of each team submit-ted the best papers and generally had more thoughtbehind their arguments and comments than studentswho played a secondary role in the development ofthe projects. Further, students had the greatest de-gree of expertise with the content that they wereresponsible for modeling. While increasing depth ofunderstanding is important, steps need to be made tointegrate student responsibilities so they learn about‘‘all’’ aspects of the group’s model. We are currentlyaddressing this limitation by placing greater emphasison the follow-up activity in which one student fromeach group is held responsible for explaining hergroup’s model to members from the other groups.

The most notable similarity among groups wasin terms of the groups’ transition from reactive partic-ipation to active participation. Initially, group mem-bers simply did what was next, spending little timeplanning out tasks and individual responsibilities. Asa result, first projects were ill-conceived and lacked

VSS Project 21

the sophistication of the second project. In contrast,in Project Two, teams spent much time planning,gathering resources, and dividing up responsibilities,resulting in a higher quality project.

Modeling Process. The practice of modeling oc-curred in two stages, an enactment stage and a visual-ization stage. The first stage in the students’ practiceof modeling was the enactment stage. For example,students begin with questions like ‘‘When does aneclipse occur?’’ In a traditional course, the lecturewould contain a definition and an assortment of two-dimensional drawings. The focus would probably beon the difference between lunar and solar eclipses.In the VSS course, there was no lecture. Studentsstarted by collecting resources and planning themodel they would build to answer the question. Fol-lowing the collection of resources and planning, theybegan the construction of their virtual models. Stu-dents enacted the facts and concepts about the Earth,Moon, and Sun they collected. Facts such as the 23-degree tilt of the Earth, the 24-hour day, or the 13-degree motion of the Moon against the backgroundstars were enacted when students constructed theirVR models.

The second stage is the visualization stage. Oncethe virtual model was built, students used the modelto answer their initial question(s). However, this re-quires significant planning and thought by the stu-dents. Students engaged in systematic observation oftheir models. They created ‘‘viewpoints’’ from whichto ‘‘see’’ their models at appropriate times. In science,knowing when and where to look is sometimes themost important aspect of an investigation. Studentsattached viewpoints to objects (say, the Moon look-ing back at the Earth to first detect an eclipse) tocreate relative perspectives. They then were able tomove to other ‘‘global’’ viewpoints to see the relativepositions of the Earth, Moon, and Sun.

Finally, students used visualization techniquesto view important relationships. To answer questionsabout an eclipse, students in the VSS course used thevisualization technique of constructing transparentdisks to visualize the Moon’s orbital plane and theplane of the ecliptic. The intersections of these twoplanes form the ‘‘line of nodes.’’ When the Earth,Moon, and Sun are all on the line of nodes, an eclipseoccurs. This became readily apparent as students vis-ualized these planes. The concept of ‘‘line of nodes’’is often left out of introductory astronomy coursesbecause the dynamic, three-dimensional nature ofthe concept is so difficult for students to understandat a meaningful level. Another difficult concept for

students to grasp is the vast variety of scales of sizein astronomy (e.g., the fact that the distance of theMoon from the Earth is 60 Earth radii—in otherwords, the Moon is far away from the Earth; or thatthe distance of the Earth from the Sun is another400 times larger than the distance of the Moon fromthe Earth). This is difficult to communicate througha textbook or even a carefully planned lecture usingpictures and slides; however, as students enactedthese distance into their virtual models they wereable to immediately experience these distances byzooming in and out of the appropriate scales. Thepower of the modeling process in which students con-structed ‘‘their’’ models was evident in one student’spost interview:

I definitely saw this as much better than a lecture-based course. Many times during a lecture-basedcourse I may get lost or might not be that attentive.This is impossible with this project since it is soindependent and I have to do it myself. The self-motivation and independence makes it vastly moreinteresting than the mandatory listening to a profes-sor speak his mind . . . I have been in both A105and A115 astronomy classes. And by far, I haveretained more information in this A100 intro classthan both two classes combined.

Resources and Inscriptions. The primary re-sources used by students were their textbooks,the on-line syllabus, the World Wide Web, theCosmoWorlds tutorial, each other, and the instructor.With respect to the textbooks, the instructor stated,‘‘In all my years teaching this course, I have neverseen books that were so marked up and dog-eared.’’Initially, students relied heavily on their textbooks,on-line supports (e.g., tutorials, the syllabus) and theinstructor. However, during the course they becamemore reliant on the other members of their owngroup and even went to other groups as a resource.Some of the most exciting resources were the inscrip-tions developed by the students themselves.

In our VSS course, students constructed inscrip-tions (tables, formulas, charts) after prolonged dis-cussion concerning astronomical data found in theirtextbook, the WWW, or told to them by the instruc-tors. Student-developed inscriptions were not widelyused until the second project in which students wereconfronted with the modeling of significant astro-nomical data and concepts. For example, beforeDonatello’s group began the construction of theirEarth-Moon-Sun model, the group spent two classperiods researching their textbook and the WWWfor the necessary astronomical data needed to createa realistic model. They tabulated their findings in

22 Barab et al.

a single inscription (table of numbers representingastronomical sizes and distances, and list of stepsnecessary to the modeling process). This inscriptionserved the group as a memory device (keeping trackof what parts of the model were incomplete), planner(timeline), task management check (which groupmember was responsible for which portion of themodel), and final check of their model (similarity oftheir model to the real solar system). During theconstruction of their model, group members continu-ally referred back to the inscription to verify theirmodel’s properties, which facilitated each groupmember’s ownership and understanding of the over-all model. These inscriptions stimulated discussion,and were instantiated into the model constructionprocess once students recognized the value of sharingtheir ideas and planning their strategy for con-structing their models.

Technology vs. Astronomy. Initially, studentsspent a great deal of time learning the software tooland, at times, its lack of usability and intuitivenesscreated frustration. This was evident in the student’sabove statement regarding whether the goal of thecourse is to learn astronomy or the technology. How-ever, on other occasions, the technology promptedstudents to engage in debates with their peers andinstructors concerning astronomical phenomenaand concepts.

For example, a particularly challenging issue wasmodeling the necessary conditions for the occurrenceof eclipses. The students could not model an eclipsewith complete accuracy and maintain relative scalebetween the Earth, Sun and Moon due to limitationsin the software. Eclipses are possible because theMoon and the Sun appear to be about the sameangular size in the sky when viewed from the Earth.However, if the students created a model that cor-rectly represented this fact, they could not observean eclipse due to the large distances between theobjects in their model. At this point, students wereforced to decide what concepts their model woulddemonstrate (correct scale or conditions for aneclipse), acknowledging that in the process theywould lose some realism. In struggling with this issue,students developed a richer understanding of eclipsesand gained an appreciation of one purpose of mod-els—simplify the situation under study through simu-lation, or imitation to elucidate key concepts. Com-menting on the importance of the modeling process,one student stated:

I don’t think I would have understood this processas well without the models because the five-degree

tilt was just a number in my other classes. I did notunderstand the need for the 5-degree tilt until wesaw that eclipses were happening every month!

IMPLICATIONS

Given the power of current technologies, educa-tors must find ways of integrating 3-D technologiesinto learning environments to improve and deepenstudents’ understanding of the content, to empowerstudents with the tools of scientists, and to do so ina fashion that is financially viable and sustainable. Inlab-based studies, VR technology has been shownto have great potential for learning and instruction(Dede et al., in press). However, even with the con-stant downward spiral of cost/perform function forthe past 20 years, these technologies are prohibitivelyexpensive to learning institutions. We need effectiveinstructional models that take advantage of the learn-ing potential and enabling opportunities for doingscience, and doing so in a cost-effective manner thatwill make feasible the use of these innovative technol-ogies. A curriculum that requires every student tohave $20,000 head-mounted displays for nine hoursa week and four technicians supporting the labora-tory will have minimal impact on education in gen-eral. The challenge is to address questions of feasibleintegration. The VSS course discussed here movedin this direction by using low-end Silicon Graphicsmachines to create a window-on-the-world and mov-ing from a lecture to a project-based course. We areteaching the course on regular personal computers(less than $2000 a machine) using VRML freeware.

The VSS Project, at its core, involves the replace-ment of lecture-based instruction as the central in-structional activity in an introductory astronomycourse with students constructing virtual solar systemmodels. In the VSS, from day one, students are con-structing models of the solar system. Or more pre-cisely, they constructed models of the solar system inan inquiry process to frame and answer fundamentalastronomy questions. Central to the design of thecourse was our pedagogical commitment, which in-volved moving away from lectures and towards im-mersing students within participatory learning envi-ronments. Although there is a growing theoreticalbase from which to derive principles for the designof technology-based, participatory learning environ-ments, it is essential that we provide an empiricalbase to ground this perspective; that is, we need tocontinue to examine the learning that is actually oc-

VSS Project 23

curring within these contexts. To this end, we havedeveloped our research agenda as a series of designexperiments (Brown, 1992) in which we engineer var-ious design modules that are introduced as curricularconstraints and that offer new learning opportunitiesfor our students. The interactions related to thesemodules are then captured using video cameras, in-terviews, and document analysis so that we can tracethe impact of the module. The findings from thisstudy can provide educators with an understandingof this process.

With respect to learning astronomy, it was ourinitial contention that astronomy education mustmake a profound transition from an emphasis ondelivering content through large-class lectures to afocus on supporting students as they construct con-crete artifacts and (in the process) build rich under-standings. In terms of construction of meaning, stu-dents did not memorize astronomical concepts as self-contained entities or as facts to be regurgitated on atest. Rather, understandings emerged as ephipheno-mena (or ‘‘residue’’, Heibert et al., 1996) arisingwithin their participation in model building activities.Concepts (e.g., plane of the ecliptic, relative scale)were embodied within practices in which their mean-ing and value were actualized, not simply realized(Barab et al., 1998). For example, the concept ofrelative size was not a separate activity that studentsdeveloping a solar system were expected to study;rather, it was a way of defining the constraints of theirvirtual worlds. In this way, concepts were relegated toconceptual tools that were embodied within practiceand stood in sharp contrast to their more frequenttreatment as abstract, disembodied facts introducedthrough didactic lectures (Bransford et al., 1989;Roupp et al., 1993). The concrete instantiation ofstudents’ understandings into VR artifacts facilitatedthe development of grounded understandings, not asseparate concepts stored in the learner’s brain but asdistributed descriptions that were situated across andthrough their experiences (Barab et al., 1999; Pea,1993).

In this article, we have described and criticallyexamined a course that engaged students learningastronomy through the building of VR worlds. Ourfindings indicate that this initial course was a success-ful innovation in which the learning of astronomyoccurred through the construction of VR models. Aswe continue to do research and analyze the data, wewill gain a richer understanding of the potential ofthese participatory, technology-rich contexts for sup-porting learning. We challenge ourselves and our col-

leagues to continue this exploration so that we mayground theoretical conjectures regarding the poten-tial of these contexts in empirical findings.

ACKNOWLEDGMENTS

This research was supported in part by a ProffittGrant from the Research and University GraduateSchool at Indiana University, and a Seed Grant fromthe Center for Innovative Learning Technologies. Inaddition, we would like to thank the Virtual Reality/Virtual Environments group for their support in us-ing the Automatic Virtual Environment (CAVE).We would also like to thank Tom Keating for hisvaluable suggestions on this manuscript.

REFERENCES

Allen, B. S., and Otto, R. G. (1996). Media as lived environments:The ecological psychology of Educational technology. In:Jonassen, D. (Ed.), The Handbook of Research for Educa-tional Communications and Technology, Simon and SchusterMacmillan, New York, NY, pp. 199–226.

Barab, S. A. (1999). Ecologizing instruction through integratedunits. Middle School Journal 30: 21–28.

Barab, S. A., Cherkes-Julkowski, M., Swenson, R., Garrett, S.,Shaw, R. E., and Young, M. F. (1999). Principles of self-organization: Ecologizing the learner-facilitator system. TheJournal of the Learning Sciences 8(3&4): 349–390.

Barab, S. A., and Duffy, T. (in press). From practice fields tocommunities of practice. Jonassen, D., and Land, S. (Eds.),Theoretical Foundations of Learning Environments, LawrenceErlbaum Associates.

Barab, S. A., Hay, K. E., Barnett, M. G., and Squire, K. (1998,May). Constructing Knowledge and Virtual Worlds: Knowl-edge Diffusion in Future Camp 97. Presented at the annualmeeting of the American Educational Research Association,San Diego, California.

Barab, S. A., Hay, K., and Duffy, T. (1998). Grounded Construc-tions and How Technology Can Help. Technology Trends43(2): 15–23.

Baxter, J. (1991). A constructivist approach to astronomy in theNational Curriculum. Physics Education 26: 38–45.

Bednar, A. K., Cunningham, D., Duffy, T. M., and Perry, D. J.(1992). Theory into practice: How do we link? In Duffy, T.,and Jonassen, D. (Eds.), Constructivism and the Technologyof Instruction, Erlbaum, Hillsdale, New Jersey, pp. 17–34.

Blumenfeld, P. C., Marx, R. W., Soloway, E., and Krajcik, J. (1996).Learning with peers: From small group cooperation to collab-orative communities. Educational Researcher 25(8): 37–40.

Bransford, J. D., Franks, J. J., Vye, N. J., and Sherwood, R. D.(1989). New approaches to instruction: Because wisdom can’tbe told. In Vosinadou, S., and Ortony A. (Eds.), Similarityand Analogical Reasoning, Cambridge University Press, Mas-sachusetts, pp. 470–497.

Brown, A. L. (1992). Design experiments: Theoretical and method-ological challenges in creating complex interventions in class-room settings. The Journal of the Learning Sciences 2:141–178.

Byrne, C. M. (1996). Water on tap: The use of virtual reality as

24 Barab et al.

an educational tool. Ph.D. Dissertation. University of Wash-ington, Seattle, Washington.

Carr, K. M. (1997). A Constructivist Approach to Reflective Judge-ment and Science Literacy in Introductory College ScienceInstruction. Paper presented at the Northern Rocky MountainEducational Research Association.

Cognition and Technology Group at Vanderbilt. (1993). Anchoredinstruction and situated cognition revisited. Educational Tech-nology 33: 52–70.

Confrey, J., and Doerr, H. M. (1994). Student modelers. InteractiveLearning Environments 4(3): 199–222.

Cordova, D. I., and Lepper, M. R. (1996). Intrinsic motivation andthe process of learning: Beneficial effects of contextualization,personalization, and choice. Journal of Educational Psychol-ogy 88: 715–730.

Dede, C., Salzman, M. C., Loftin, R. B., and Sprague, D. (in press).Multisensory immersion as a modeling environment for learn-ing complex scientific concepts. Roverts, N., Feurzeig, W.,and Hunter, B. (Eds.), Computer Modeling and Simulationin Science Education, Springer-Verlag, New York, NY.

Dewey, J. (1963). Experience & Education, Collier MacMillan,New York, NY (Original work published in 1938).

Edwards, L. D. (1995). The design and analysis of a mathematicalmicroworld. Journal of Educational Computing Research12: 77–94.

Gay, E. (1994, Winter). Is virtual reality a good teaching tool?Virtual Reality Special Report, 51–59.

Gilbert, S. W. (1991). Model building and a definition of science.Journal of Research in Science Teaching 28(1): 73–79.

Guba, E. G., and Lincoln, Y. S. (1983). Epistemological and meth-odological bases of naturalistic inquiry. In Madaus, G. F.,Scriven, M. S., and Stufflebeam D. L. (Eds.) Evaluation Mod-els: Viewpoints on Educational and Human Services Evalua-tion, Kluwer-Nijhoff Publishing, Boston, Massachusetts, pp.311–334.

Hay, K. E., and Barab, S. A. (1998, May). Building Worlds: Toolsof Virtual Practice. Presented at the annual meeting of theAmerican Educational Research Association, San Diego, Cal-ifornia.

Hay, K. E., Johnson, H., Barab, S. A., and Barnett, M. G. (inpress). The next best thing: Virtual reality in the astronomyclassroom. To appear in Mercury.

Heibert, J., Carpenter, T. P., Fennema, E., Fuson, K., Human, P.,Murray, H., Olivier, A., and Wearne, D. (1996). Problemsolving as a basis for reform in curriculum and instruction: Thecase of mathematics. Educational Researcher 25(4): 12–21.

Jackson, S. L., Stratford, S. J., Krajcik, and Soloway, E. (1994).Making dynamic modeling accessible to precollege sciencestudents. Interactive Learning Environments 4(3): 233–257.

Johnson, D. W., and Johnson, R. T. (1990). Cooperative learningand achievement. In Sharan, S. (Ed.), Cooperative learning:Theory and research, Praeger, New York, pp. 23–37.

Johnson, D. W., and Johnson, R. T. (1994). Learning together. InSharan S. (Ed.), Handbook of Cooperative Learning Methods,Greenwood Press, Connecticut, pp. 51–65.

Jonassen, D. H. (1991). Evaluating constructivist learning. Educa-tional Technology 31: 28–33.

Jonassen, D. H. (1996) Computers in the Classroom: Mindtoolsfor Critical Thinking, Prentice-Hall, Englewood Cliffs, NewJersey.

Keating, T., Barnett. M., and Barab, S. A. (1999, May). StudentLearning Through Building Virtual Models. Presented at theAnnual Meeting of the American Educational Research Asso-ciation, Montreal, Candada.

Koschmann, T. (1996). CSCL: Theory and Practice of an EmergingParadigm (Edited Volume). Erlbaum, Mawah, New Jersey.

Land, S. M., and Hannafin, M. J. (1996). A conceptual frameworkfor the development of theories-in-action with open-ended

learning environments. Educational Technology Research andDevelopment 44: 37–53.

Lave, J., and Wenger, E. (1991). Situated Learning: LegitimatePeripheral Participation. Cambridge University Press, NewYork.

Lehrer, R., Horvath, J., and Schauble, L. (1994). Developingmodel-based reasoning. Interactive Learning Environments4(3): 219–231.

Lincoln, Y. S., and Guba, E. G. (1986). But is it rigorous? Trustwor-thiness and authenticity in naturalistic evaluation. New Direc-tions for Program Evaluation 30: 73–84.

McCellan, H. (1996). Virtual realities. In Jonassen, D. (Ed.) Hand-book of Research for Educational Communications andTechnology, Kluwer-Nijhoff Publishing, Boston, Massachu-setts, pp. 457–487.

Nastasi, B., and Clements, D. H. (1991). Research on cooperativelearning: Implications for practice. School Psychology Review20: 110–131.

National Science Foundation, Division of Undergraduate Educa-tion (1998). ‘‘Shaping the future: New expectations for under-graduate education’’ [On-line]. Available: http://www.ehr.nsf.gov/EHR/DUE/documents/review/96139/start.htm

Newman, D., Griffin, P., and Cole, M. (1989). The ConstructionZone, Cambridge University Press, New York.

Olson, S. (1998). Stargazing. Teacher Magazine Feb: 25–28.Osberg, K. M., Winn, W., Rose, H., Hollander, A., and Hoffman,

H. (1997, May). ‘‘The Effect of Having Grade Seven StudentsConstruct Virtual Environments on their Comprehension ofScience.’’ In Proceedings of the Annual Meeting of the Ameri-can Educational Research Association. Chicago, Illinois.

Papert, S. (1991). Situating constructionism. In Harel, I., and Pa-pert, S. (Eds.), Constructionism: Research Reports and Essays,1985–1990, Ablex, Norwood, New Jersey, pp. 1–11.

Pea, R. (1993). Practices of distributed intelligence and designsfor education. In Salomon, G. (Ed.), Distributed Cognitions:Psychological and Educational Considerations, CambridgeUniversity Press, UK, pp. 47–87.

Pfundt, H., and Duit, R. (1991). Students’ Alternative Frameworksand Science Education. Bibliography. 3rd. Kiel Univ. (WestGermany). Institut fuer die Paedagogik der. Naturwis-senschaften.

Pyramid Film and Video. (1988). A private universe. An insightfullesson on how we learn [Flier] Author, Santa Monica, Cali-fornia.

Roth, W.-M. (1996). Knowledge diffusion in a grade 4–5 classroomduring a unit of civil engineering: An analysis of a classroomcommunity in terms of its changing resources and practices.Cognition and Instruction 14: 170–220.

Roth, W.-M. (1998). Designing communities, Kluwer AcademicPublishers, Dordrecht.

Roth, W.-M., and Bowen, G. M. (1995). Knowing and interacting:A study of culture, practices, and resources in a grade 8 open-inquiry science classroom guided by a cognitive apprentice-ship metaphor. Cognition and Instruction 13: 73–128.

Roth, W. M., and McGinn, M. K. (1998a). Inscriptions: Towarda theory of representing as social practice. Review of Educa-tional Research 68(1): 35–59.

Roth, W. M., and McGinn, M. K. (1998b). Knowing, researching,and reporting science education: lessons from science andtechnology studies. Journal of Research in Science Teaching35(2): 213–235.

Ruopp, R., Gal, S., Drayton, B., and Pfister, M. (1993). LabNet:Toward a Community of Practice, Lawrence Erlbaum Associ-ates, Hillsdale, New Jersey.

Rutherford, F. J., and Ahlgren, A. (1990). Science for all Ameri-cans, Oxford University Press, New York.

Sabelli, N. (1994). On using technology for understandings science.Interactive Learning Environments 4(3): 195–198.

Savery, J., and Duffy, T. (1996). Problem based learning. An in-

VSS Project 25

structional model and its constructionist framework. In Wil-son, B. (Ed.) Constructivist Learning Environments: CaseStudies in Instructional Design, Educational Technology Pub-lications, New Jersey, pp. 135–148.

Scardamalia, M., and Bereiter, C. (1993). Computer support forknowledge-building communities. The Journal of The Learn-ing Sciences 3: 265–283.

Scriven, M. S. (1983). Evaluation methodologies. In Madaus,G. F., Scriven., M. S., and Stufflebeam, D. L. (Eds.) Evalu-ation Models: Viewpoints on Educational and Human Ser-vices Evaluation, Kluwer-Nijhoff Publishing, Massachusetts,pp. 229–260.

Sfard, A. (1998). On two metaphors for learning and the dangersof choosing just one. Educational Researcher 27: 4–13.

Sharan, S. (1994). Cooperative learning and science. In Sharan,S. (Ed.), Handbook of Cooperative Learning Methods, Green-wood Press, Connecticut, pp. 227–244.

Slavin, R. E. (1995). Cooperative Learning: Theory, Research, andPractice, Allyn and Bacon, Needham Heights, Massachusetts.

Solomon, J. (1983). Learning about energy: how pupils think intwo domains. European Journal of Science Education 5: 49–59.

Stake, R. E. (1983). Program evaluation, particularly responsiveevaluation. In Madaus, G. F., Scriven., M. S., and Stufflebeam,D. L. (Eds.), Evaluation Models: Viewpoints on Educationaland Human Services Evaluation, Kluwer-Nijhoff Publishing,Massachusetts, pp. 287–310.

Stratford, S. J. (1997). A review of computer-based research inprecollege science classrooms. The Journal of Computers inMathematics and Science Teaching 16: 3–23.

Stratford, S. J., Krajcik, J., and Soloway, E. (1998). Secondarystudents’ dynamic modeling processes: Analyzing, reasoningabout, synthesizing, and testing models of stream ecosystems.Journal of Science Education and Technology 7(3): 215–234.

Tobin, K., Espinet, M., Byrd, S. E., and Adams, D. (1988). Alterna-tive perspectives of effective science teaching. Science Educa-tion 72: 433–451.

Vygotsky, L. (1978). Mind in Society: The Development of HigherPsychological Processes, Harvard University Press, Cam-bridge, Massachusetts.

Wandersee, J. H., Mitzes, J. J., and Novak, J. D. (1994). Researchon alternative conceptions in science. In Gabel, D. L. (Ed.),Handbook on Science Teaching and Learning, Macmillan Pub-lishing Co., New York.

White, B. Y., and Frederikson, J. R. (1998). Inquiry, modeling,and metacognition: Making science accessible to all students.Cognition and Instruction 16(1): 3–118.

Whitehead, A. N. (1929). The Aims of Education and Other Essays,MacMillan, New York.

Winn, W. (1995). The Virtual Reality Roving Vehicle Project.Technological Horizons in Education Journal 23(5): pp.70–75.