Design-Based ;Hoadley, 2002, in press).

One noteworthy aspect of

:his story i s that the design-based research paradigm
aermitted not only evaluative research and hypothesis
Research Methods and formation to take place, but also allowed careful
refinement and testing of a new psychological theory,
Theory Building: the theory of socially relevant representations (SRRs)for
learning (Hoadley, 1999a; Hoadley & Kirby, 2004).
A Case Study of In the sections below, I describe some of the initial
design goals, early designs of the predecessor to the
Research with SpeakEasy software and associated educational
activities, and the theories used to construct these
SpeakEasy designs.* Then, I describe h o w early studies led to
development and refinement of the theory of SRRs.
Finally, I discuss how later DBR studies allowed testing
of the theory and delineate some of the implications of
Christo her M. Hoadley
7
The Pennsy vania State University
this research trajectory for DBRM in general.

SpeakEasy and Its Predecessor, the

Multimedia Forum Kiosk
In 1992, the Web was still an idea more than a
Design-based research methods are an emerging reality. Multimedia was novel and expanding onto the
research paradigm that blends empirical research with computing scene. Educational CD-ROMs were
design and implementation. This article discusses how becoming more important, and video online was no
a design-based research trajectory can help not only
longer restricted to specialized videodisc or high-end
suggest and refine theories, but also permit their
falsification. In a series of design-based research studies workstations for the first time. Though the Web was
on an online discussion tool (called SpeakEasy), the basically unknown, the Internet was not; e-mail and
author shows how a theory of socially relevant Usenet newsgroups were beginning to ascend i n
representations for learning was formed, applied, popularity not only with academic computer scientists,
refined, and ultimately tested. but also with college students, businesspeople, and
home users through services like Prodigy and AOL. At
,As described in the introductory article in this special that time, Sherry Hsi, Christina Schwarz, and I began
issue, design-based research methods (DBRM) i n work on an online multimedia discussion tool called
education are an exciting, productive set of research the Multimedia Forum Kiosk. We had a simple core
methods that blend a desion stance with empirical design goal: To create a tool that would allow
research in context to advance our theoretical collaborative knowledge building while taking
knowledge of learning, while simultaneously producing advantage of multimedia. We wanted something that
valued educational outcomes for learners in real would be "better than Netnews" (our unofficial mantra)
settings 'Design-Based Research Collective, 2003; by supporting higher quality, more coherent discourse
Kelly, 2003). In this article, 1 describe the development through a more intuitive graphical user interface. We
of an online discussion tool for learning, called also identified as one of our use-case scenarios the
SpeakEasy {Hoadlev, Hsi, & Berman, 1995a), and collaborative qualitative analysis or discussion of video
discuss how a long trajectory of design-based research data. Implicitly, our dissatisfaction w i t h Usenet
not only allowed theory to inform design, but also newsgroups hinted at a tacit goal we shared: building a
enabled clesiqn to inform psychological theory system that w o u l d promote greater social
connectedness rather than disconnection.
Our initial system, built i n Hypercard, drew on
some examples we were familiar with, namely a video
bulletin board that Sherry had helped develop at the
Apple Multimedia Lab (for a much later write-up of this
design's evolution, see Bellamy, Woolsey, & Kerns,
Christopher M. Hoadley designs, builds, and studies ways for
technolop to enhance collaboration and learning. He is 1995), and Scardamalia and Bereiter's CSILE (Com-
currently Assistant Proiessor of Instructional Systems and of
Information Sciences and Technology at Penn State
University. He was previously president of the International *More information about the design of the SpeakEasy and its
Society for the Learning Sciences. He co-rounded and leads predecessor, the Multimedia Forum Kiosk, is available
the Spencer-tunded Design-Based Research Collective (e- elsewhere (Hoadley, 1999a, 2002, in press; Hoadley & Hsi,
mail: EdTechDBRCbS'toplie.~iet). 1993; Hoadley et a/., 1995a).

42 EDUCATIONAL TECHNOLOGY/january-February 2005

puter-Supported Intentional Learning Environments) two more theories were significant influences. We
software (Scardamalia, Bereiter, McLean, Swallow, & came across work on small-group interaction that
Woodruff, 1989). We admired the Apple project's use framed the types of discursive moves typically taken in
of the brand-new QuickTime multimedia technology, collaboration; this supported and extended our
although it depended on an expensive, specialized semantic labels (Bales, 1969). We also followed Pea's
hardware setup to allow video capture to make more specific theories of the mechanisms for learning
postings. We also were influenced by Scardamalia and through collaborative discussion (Pea, 1993a, 19930).
Bereiter's use of scaffolding to improve the quality of This theory helped us focus on process support, instead
discussion in their tool. Taking both strategies to heart, of merely focusing on usability of the system; we
we created a kiosk-based bulletin board system. Unlike emphasized developing social supports that worked in
Netnews, which was subject-based, our tool was conjunction with the software to produce certain kinds
topically based, with all participation centered on a of interaction that constituted "productive discourse."
question or discussion point posed by a topic author. As is typical, these theories did not completely
The Multimedia Forum Kiosk system had two main specify what types of design decisions we should
interfaces. The opinion area provided a welcome to the undertake, either in the interface or in the surrounding
topic; and, in addition to presenting the topic author's activities. Should we permit anonymity? What types of
motivating question, it supplied a representation of all group size would best help learning? What types of
of the participants in that topic's discussion and their questions, topics, or adult moderation might best
points of view (revisable over time, but only one engender "productive discussion"? None of these
opinion per person), showing a community gestalt. The questions could be answered from first principles with
discussion area was more similar to what we now are the theories at hand, but we drew on our own and
accustomed to as threaded discussion. It provided a expert intuitions (including expert designers at Apple,
chance for people to post comments and respond to Tal igent, Interval Research, and elsewhere) for
one another, with a graphical tree structure (rather than guidance and used iterative refinement to develop
a outline-based view, this was truly tree-like). Similar to workable educational interventions using prototypes of
CSILEfs "think types," each comment subtitled with a the tool.
semantic label (such as "and," "but," "or," "?" and so At this point, we used both formative evaluation and
on) that identified how the comment related to prior iterative refinement just to get something working in
ones. practice. We tried our tool in a wide variety of
In our system, we used multimedia in various contexts, including graduate courses in engineering, an
innovative ways. Topics could include short video clips informal lounge in an education department, science
as a conversation starter. Topics, opinions, and museums, a self-paced study center, and (in what
comments were individually authored; and picture would eventually become our primary research
icons were used to highlight this authorship. Two context) a middle school physics classroom. In this
design decisions that were byproducts of our context, we ended up supporting an inquiry-based
technological constraints were the fact that the system science curriculum through students' discussing topics
was a same-place, different-time system (due to limited that asked them to explain science phenomena,
networks for sharing large files); and that users were presented through multimedia (Bell, Davis, & Linn,
permitted to log in anonymously. Because of the photo 1995; Hoadley & Bell, 1996). We eventually arrived at
icons and the relative difficulty of obtaining and a very stable set of activities based on this particular
digitizing images of users, the kiosk had some guest learning context.
accounts to allow use of the system by people who had Some of our early findings centered on the
not been previously photographed and set up with importance of the context of use. We began with basic
accounts. usability studies-our software fit the HCI "ten minute
rule" (meaning that novices could walk up and use the
system without any training, uncovering all its features
Designing from Theory and in ten minutes or less). But usability is only one
Our Early Evaluation Work component of the success of such a system, and we
Initially our designs were based upon theories of began to understand how important the activities and
collaboration and learning. Our emphasis on context of use were for us. For example, when we used
collaboration came from Vygotskian notions of co- the system with engineering students to help assess
construction and collaboration (Newman, Griffin, & learning climate issues, we discovered through
Cole, 1989; Wertsch, 1979, 1985), but more comparison of different courses that physical access to
specifically we used ideas about knowledge-building the system was important, but more important was the
communities (Scardamalia & Bereiter, 1991, 1993) and degree to which the course instructor showed evidence
communities of practice (Lave & Wenger, 1991) to of incorporating feedback provided by students. One
guide our vision of successful use of the system. Later, particularly interesting example involved a non-native

EDUCATIONAL TECHNOLOGYfjanuary-February 2005

speaker of English; this student reported that the MFK description). We also used a variety of informal data
made a big difference in his own learning climate, sources, such as our own experiences in the context as
because-unlike in face-to-face interaction in the participant observers or the input of expert designers,
class-he could compose and edit contributions to and used these to help guide not only the design, but
class discussions on his own time (Hsi & Hoadley, also the more formal research activities such as
1995). We later demonstrated with the middle-school treatment design and data collection. Constructing
science students that identity, gender, anonymity, and theories, design hypotheses, and sometimes more
online participation played an important role not only fragile understandings of how our treatments played out
in students' participation in the discussion, but also in in context were the primary goals. Falsification was an
their learning (Hsi & Hoadley, 1997). important part of our work; in some cases, we were
As the next stage of our design-based research and able to falsify null hypotheses, and in many others, we
building on this work, prior to the introduction of the were able to falsify incomplete or incorrect under-
first Netscape browser, we developed the second ever standings of the learning taking place.
Web-based threaded discussion tool,* called SpeakEasy Through our design work, we stumbled on an area
(Hoadley eta/., 1995a). SpeakEasy retained most of the that appeared to be central to our success-how
Multimedia Forum Kiosk interface; but, because it was students construed the social space they were entering
Web-based, this application no longer depended on when they participated in online discussion. Intuitively,
limited access to a single kiosk (different-place, we felt that there was a great deal of social
different-time communication, instead of same-place, connectedness and motivation wrapped up in our tool,
different-time communication). Our tree-based layout in part due to the nature of the representations in the
was a casualty of the limits of HTML and was replaced interface. We began to develop a theory of how social
by the indented outline-style interface. The opinion cues, such as the face icons and our semantic labels,
area and discussion area remained. might be influencing student cognition and might
--
facilitate sensemaking and learning (Hoadley, 1998;
Hoadley & Hsi, 1996; Hoadley, Hsi, & Berman, 1995b;
Building a Theory of Hsi & Hoadley, 1994). As a design strategy, we began
Socially Relevant Representations explicitly manipulating and studying various social
How did we use this design-based research to representations i n the interface, such as identity (as
investigate theory? While there is debate about what conveyed through the face icons) (Hsi, 1997; Hsi &
precisely are the characteristics of design-based Hoadley, 1997). We also began hypothesizing
research, I believe our example to be relatively mechanisms by which social cues might aid learning.
prototypical. Our work involved a theory-driven, One method we used to aid in development of this
iterative design of an artifact and associated activities theory was to look longitudinally over our design
that constituted a learning environment over a long trajectory. As soon as we began to suspect that social
period of time (eight years, to be exact). We tested our cues were facilitating participation and learning, we
intervention in a variety of learning contexts; each began to manipulate our design accordingly. In some
iteration was implemented and evaluated in some cases, this took the form of following success; we tried
authentic learning setting. Our design work was in embedding additional social cues in our SpeakEasy
some cases connected to intuitive or informal interventions, such as linking to descriptive personal
knowledge, but more often was the result of embedded homepages, adding short usernames to the comment
investigations over the course of the development icons, and moving some of what had been placed in
trajectory. We altered our intervention on a daily basis seed comments into introductory videos of actors
if needed to support its success in context, but we also voicing those seed comments.
created planned comparisons between successive Embedded within our design-based research
versions of the intervention, or between versions trajectory was a chance to run traditional treatment-
administered simultaneously to different populations in control studies, in which some participants received
a similar or the same setting. Our data collection was one version of the intervention and others received
principled, but wide ranging; we collected some data another. We could sometimes compare an improved
to help orient us to the phenomena at hand, but also version against baseline data (one semester of students
collected targeted data related to the planned compared against another), but in addition we could
comparisons (experiments) embedded in the work, and randomly assign students to conditions, some of which
gathered still other data to help elaborate or explore were designed to gather information by introducing
poorly understood aspects of the phenomena (rich changes we thought might be worse, to test our
budding theories. For instance, we compared students
in discussions who were forced to make all comments
*We were beaten to the punch by Hypernews by a few anonymously with students who were forced to identify
months. themselves on all comments.

44 EDUCATIONAL TECHNOLOGY/January-February 2005

 More than simply analyzing the dependent variables were able to determine that some students
in these cases, we collected rich descriptive data about approximately half) use social cues i n navigating
how the students participated with the system and used SpeakEasy, that some (but not all) students encode and
these to further inform our theory. By trying out various ,emember these cues in their understanding of the topic
representations, collecting data from not only af discussion, and that on average providing students
participants but from the literature on interfaces, and with social cues led to better learning outcomes in
iterating, we improved our understanding of how a SpeakEasy discussions (Hoadley, 1999a, 1999b;
particular treatment corresponded to a theoretical Hoadley & Linn, 2000).
entity. Elsewhere (Hoadley, 2002) I describe how our
notion of anonymity changed as a result of this kind of Theory Building and Testing
iteration-initially we saw anonymity as evidence of Our design-based research methods helped us build
social inclusiveness because it improved participants' theory in four ways: induction, orientation, constraints,
perceptions of social safety, but later we realized that and manipulation.
social safety was independent of the use of the
anonymity function and that in fact this feature Induction. Induction i s perhaps the most obvious
threatened inclusivity. form of research, yet i s often overlooked. We collected
Sometimes design-based research gave us a chance a great deal of data on the use of our tool in various
to refine the ways we operationalized our theory, contexts. While some have argued that over-collection
helping to increase the alignment between our of data may be seen as a weakness of design-based
theoretical notions (in this case, "representations of research (Dede, 2004), i t also permits hypothesis
social cues in interfaces") and how those notions were generation from a large body of knowledge and is
embodied i n the research setting and their effects especially effective when deduction from known first
measured. Sandoval has termed this type of exploration principles is not possible. Additionally, because the
the testing of "embodied hypotheses" (Sandoval, 2002, design-based researcher has intimate knowledge of the
in press) We had a lot of exploring to do in terms of research setting, the researcher can draw educated
what constituted a more or less socially relevant guesses about which explanations for the observed
representation. In one very promising early study, we phenomena might be most parsimonious. It is difficult
compared the standard version of SpeakEasy that was to apply "Occam's razor" from afar.
designed to "feel like a conversation" with an interface
that used similar screen layouts, but that was altered to Orientation. Learning in context i s a complex sys-
feel more like posting to a database. We used visual tem, with multitudinous interacting factors. Which
icons that were not faces to represent contributors, and matter most for success? This is an important design
changed the discussion area from a discursive question that poses challenges for research. By focusing
organization (semantic labels and a response-oriented on the design of the learning environment and the
thread structure) to a topical organization (labels such improvement of that design over time, the design-based
as "pro" and "con" and a conceptually oriented thread research process is forced to converge quickly on what
structure). Our results were somewhat surprising; works through refinement and hill-climbing within a
students participated equivalently in both formats (same design space. The net result is to help orient the
number of comments, same comment quality), but had researcher quickly to those factors that are most design-
different outcomes. On the positive side, more students relevant in a given context with a given set of tacit
changed their ideas based on the activity in the socially design assumptions.
relevant condition, but frequently they changed their
ideas for the worse! By doing post hoc analysis of the Constraints. An infinite number of theories can be
comments themselves, we were able to trace this back generated to explain any finite set of data. However,
to some of the particular seed comments we provided. much as sampling may be used to test a model created
This led to continued refinement of our notion of with one subset of data by attempting prediction on
mechanisms for socially relevant representations (SRRs) another subset of the same data, design-based research
to support learning and also led to changes in the allows theories to be falsified by systematic additional
activity structure that were irrelevant to our ideas about analysis of existing data. Comprehensive data
how SRRs worked, but that were necessary (but not collection in design-based research permits not only
sufficient) for learning and conceptual progress. this type of model validation based on subsampling,
The research trajectory culminated in a series of but also allows additional post hoc analysis in which
studies that suggested that SRRs can enhance learning models, hypotheses, and in some cases theories can be
outcomes, but indicated there are significant individual put up for falsification by making predictions about
differences in how these representations are used. data already collected and then examining that data.
Through a combination of interviews, surveys related to Since creation of rich, contextualized interventions is
social orientation, and learning and recall tasks, we enormously expensive, and orientation to relevant

EDUCATIONAL TECHNOLOGY/January-February 2005

factors in these interventions requires intimacy with the more general theories of psycholoe;~or human-
research setting, this type of post hoc analysis and/or machine interaction (Carroll & Rosson, 1992)], and the
theory constraint through prior data can be invaluable context might all be incompletely understood,
for testing ideas in an authentic way. Often, the choice faced by a researcher is whether it
is more valuable to do messy research on big,
Manipulation. Manipulation is the heart of experi- complicated questions using design-based research, or
mentation. In traditional controlled laboratory whether i t i s more valuable to perform tightly
experimentation, we manipulate the world in precisely controlled studies while limiting oneself to questions
controlled ways to test our theories' ability to predict and interventions that can be tightly controlled. Or, put
the outcomes. But what can one do when theory another way, "The contingency of artificial phenomena
incompletely predicts 3n outcome, or when variables has always created doubts as to whether they tall
cannot be controlled? Some w o u l d argue that properly within the compass of science. Sometimes
inferencine must not take place in these circumstances, these doubts are directed at the ...difficulty of
but Dewey would argue otherwise: "The conjunction of disentangling prescription from description. This seems
problematic and determinate characters i n nature to me not to be the real difficulty. The genuine problem
renders every existence, as well as every idea and is to show how empirical propositions can be made at
human act, an experiment in fact, even though not in all about systems that, given different circumstances,
design. To be intelligently experimental i s but to be might be quite other than they are" (Simon, 1969).
conscious of this intersection of natural conditions so as Design-based research can be one way to build
to profit by it instead of being at its mercy" (Dewey, theoretical, empirical propositions about learning with
1925). technology. 0
The very nature of design (as opposed to problem-
solving or deduction) is reasoning under uncertainty, in
underconstrained problem spaces, where the outcomes
of actions cannot be fully specified. In design-based References
research, w e can take advantage of this type of
reasoning, by incrementally increasing our under- Bales, R. F. (1969). Personality and interpersonal behavior.
standing of a particular designed intervention i n a New York: Holt, Rinehart, and Winston.
particular context over time. We poke it, prod it, and Bell, P., Davis, E. A., & Linn, M. C. (1995). The knowledge
integration environment: Theory and design. In S.
continuously monitor the results. Much as calculus
Goldman & J. Greeno (Eds.), Computer supported
suggests approximating the change in a function's
output by applying smaller and smaller changes to its
collaborative learning '95 (pp. 14-2.1). Mahwah. NJ: .
Lawrence Erlbaum Associates.
input and taking the limit, we add incremental "deltas" Bellamy, R., Woolsey, K., & Kerns, C. (1995). D e s i g n
to our interventions in design-based research and view experiments with media-rich messaging. Paper presented
the results. This provides not only a means to get to a at the Annual Meeting of the American Educational
particular outcome (i.e., formative evaluation), but also Research Association, San Francisco.
provides a way to systematically explore the design Carroll, J. M., & Rosson, M. B. (1992). Getting around the
space through hill climbing. As the relationship task-artifact cycle: How to make claims and design by
between interventions and outputs becomes better scenario. Transactions o n Information Systems, 70(2),
understood in one context, it provides the opportunity 181-212.
Dede, C. (2004). If design-based research i s the answer, what
to attempt generalization to other contexts.
is the question? Journal of the Learning Sciences, 73(1),
105-1 14.
Conclusion Design-Based Research Collective. (2003). Design-based
Design-based research boils down to trying to research: An emerging paradigm for educational inquiry.
understand the world by trying to change it. This article Educational Researcher, 32(1), 5-8.
focuses on some of the theory-building aspects of one Dewey, J. (1925). Volume I: 1925. In J. A. Boydston (Ed.),
trajectory of design-based research. Certainly, there are John Dewey, the later works (Electronic ed.). Carbondale,
many ways to test theories, and many more ways to IL: Southern Illinois University Press.
build or propose them. Design-based research i s only Hoadley, C. (1998). Shaping social interactions for
one way to do these. It i s not recommended in knowledge integration through technology. In B. K.
situations where other, less demanding methods (such Nichols, A. C. Kemp, & D. Jackson (Eds.), 71st NARSJ
Annual Meeting (p. 166). San Diego: National Association
as good guesswork and a simple experiment) will do.
for Research in Science Teaching.
However, DBR can be useful for creating and testing
Hoadley, C. (1 999a). Scaffolding scientific discussion using
theories in situations, such as complex educational socially relevant representations in networked multimedia.
technology innovations, where little is fixed: the Unpublished Ph.D. Dissertation, University of California,
interventions, the theories [both what Reigeluth terms Berkeley.
instructional-design theories (Reigeluth, 1999) and Hoadley, C. (1999b). Social text: Learning in online peer

46 EDUCATIONAL TECHNOLOGY/january-February 2005

 . ,
Processing Conference, Jackson Hole, Wyoming. zone: Working for cognitive change i n school. New York:
Hoadley, C. (2002). Creating context: Design-based research Cambridge University Press.
in creating and understanding CSCL. In C. Stahl (Ed.), "ea, R. D. (1993a). Learning scientific concepts through
Computer support for collaborative learning 2002 (pp. material and social activities: Conversational analysis
453-462). Mahwah, NJ: Lawrence Erlbaum Associates. meets conceptual change. Educational Psychologist, 28131,
Hoadley, C. (in press). Methodological alignment in design- 265-277.
based research. Educational Psychologist. Pea, R. D . (1993b). Seeing what we b u i l d together:
Hoadley, C., & Bell, P. (1996, Sept.). Web for your head: The Distributed multimedia learning environments for
design of digital resources to enhance lifelong learning. D- transtormative communications. journal o f the Learning
Ub Magazine. Sciences, 3 ( 3 ) ,285-299.
Hoadley, C., & Hsi, 5. (1993). A multimedia interface for Reigeluth, C. M. (1 999). What is instructional-design theory
knowledge building and collaborative learning. In Adjunct and how is it changing? In C. M. Reioeluth iEd.i,
Proceedings of the International Computer Human Instructional-design theories and models: A new paradign7
Interaction Conference (InterCHI) '93 (pp. 103-1 04). o f instructional theory (Vol. II, pp. 5-29!. ~Mahwah,Nj:
Amsterdam, The Netherlands: ACM Press. Lawrence Erlbaum Associates.
Hoadley, C., & Hsi, S. (1996, April). Towards a theory o f Sandoval, W. A. (2002, April). Learning from designs:
collaborative networking i n the science classroon7. Paper Learning environments as embodied hypotheses. Paper
presented at the Annual Meeting o f the American presented at the Annual Meeting of the American
Educational Research Association, New York. Educational Research Association, New Orleans.
Hoadley, C., Hsi, S., & Berman, B. P. (1995a). The Sandoval, W. A. (in press). Developing learning theory by
Multimedia Forum Kiosk and SpeakEasy. i n P. Zeilweger refining conjectures embodied i n educational designs.
(Ed.), Proceedings o f the Third ACM International Educational Psychologist.
Conference on Multimedia (pp. 363-364). San Francisco: Scardamalia, M., & Bereiter, C. (1991). Higher levels of
ACM Press. agency for children in knowledge building: A challenge
Hoadley, C., Hsi, S., & Berman, B. P. (1995b, April). for the design of new knowledge media. Journal o f the
N e t w o r k e d multimedia f o r c o m m u n i c a t i o n a n d Learning Sciences, 7 (1 ), 37-68.
collaboration. Paper presented at the Annual Meeting of Scardamalia, M., & Bereiter, C. (1993). Computer support tor
the American Educational Research Association, San knowledge-building communities. Journal o f the Learning
Francisco. Sciences, 3(3), 265-283.
Hoadley, C., & Kirby, j. (2004). Socially relevant Scardamalia, M., Bereiter, C., McLean, R. S., Swallow, I., &
representations in interfaces for learning. In Y. B. Katai, W. Woodruff, E. (1 989). Computer-supported intentional
A. Sandoval, N. Envedy, A. Scott Nixon, & F. Herrera learning environments. journal o f Educational Computing
(Eds.), International Conference o f the Learning Sciences Research, 6(1), 55-68.
(ICLS) 2004 (pp. 262-269). Mahwah, NJ: Lawrence Simon, H . A. (1969). The sciences o r the artificial.
Erlbaum Associates. Cambridge, MA: MIT Press.
Hoadley, C., & Linn, M. C. (2000). Teaching science through Wertsch, I . (1985). Vvgotsky and the social formation of
on-line, peer discussions: SpeakEasy i n the Knowledge mind. Cambridge, MA: Harvard University Press.
Integration Environment. International Journal of Science Wertsch, J. V. (1979). From social interaction to higher
Education, 22(8), 839-858. psychological processes: A clarification and application of
Hsi, S. H . (1997). Facilitating knowledge integration in Vygotsky's theory. Human Development, 2211 ), 1-22.
science through electronic discussion: The Multimedia
Forum Kiosk. Unpublished Ph.D. dissertation, University
of California, Berkeley.
Hsi, S., & Hoadley, C. (1994, April). An interactive
multimedia kiosk as a tool for collaborative discourse,
reflection, and assessment. Paper presented at the Annual Features on Web Site
Meeting of the American Educational Research
Association, New Orleans. Visitors to the Web Site maintained for this magazine
Hsi, S., & Hoadley, C. (1995). Assessing curricular innovation
will find the following features:
in engineering: Using the Multimedia Forum Kiosk. Paper
presented at the Annual Meeting o f the American
See all of these features at: BooksToRead.com/etp
Educational Research Association, San Francisco.
Hsi, S., & Hoadley, C. (1997). Productive discussion in
science: Gender equity through electronic discourse. Sample Articles. At least two recently published
Journal o f Science Education and Technology, 1 0(1), 23- articles from this magazine are always available at
36. the site.
Kelly, A. E. (2003). Research as design. Educational Contributing Editors. The complete list of our
Researcher, 32(1), 3-5. regular contributors is available at the site.
Lave, I., & Wenger, E. (1991 ). Situated learning: Legitimate Author Guidelines. Prospective authors of articles
peripheral participation. New York: Cambridge University for the magazine are encouraged to read these
Press.
guidelines.

EDUCATIONAL TECHNOLOGY/January-February 2005 47