You are on page 1of 16

Available online at www.sciencedirect.

com

Computers & Education 51 (2008) 939954


www.elsevier.com/locate/compedu

The eect of scaolding students context-generating


cognitive activity in technology-enhanced case-based learning
Stavros N. Demetriadis a,*, Pantelis M. Papadopoulos a,
Ioannis G. Stamelos a, Frank Fischer b
a
b

Aristotle University of Thessaloniki, PO Box 114, 54124 Thessaloniki, Greece


Ludwig Maximilians Universitat, Leopoldstrasse 13, 80802 Munchen, Germany

Received 4 April 2007; received in revised form 17 September 2007; accepted 23 September 2007

Abstract
This study investigates the hypothesis that students learning and problem-solving performance in ill-structured
domains can be improved, if elaborative question prompts are used to activate students context-generating cognitive processes, during case study. Two groups of students used a web-based learning environment to criss-cross and study casebased material in the software project management domain. The experimental group was additionally prompted to consistently answer a set of questions based on a model of context-generating processes, meant to engage students in deeper
processing of information presented in cases. Students were also classied as having either complex or simple EB prole (based on their epistemological beliefs record), thereby establishing a 2 2 factorial design. Results indicated that scaffolding treatment had a signicant main eect on students performance, with the experimental group performing better in
both domain knowledge acquisition and knowledge transfer post-test items. There is also tentative indication that EB prole and scaolding treatment interact, with complex-EB learners beneting most from the scaolded condition. Overall,
the study provides evidence that it is possible to improve individual learning in a technology environment for case-based
learning, by implementing appropriate questioning strategies that trigger students to activate their context-generating cognitive processes, while studying the contextually rich material of cases.
 2007 Elsevier Ltd. All rights reserved.
Keywords: Interactive learning environments; Teaching/learning strategies; Multimedia/hypermedia systems; Pedagogical issues

1. Introduction
The premise of this study is that question prompts that activate students context-generating cognitive processes (Kokinov, 1995) can eciently scaold them to process the contextual information presented in cases
and improve their performance in ill-structured problem-solving, where contextual issues play a crucial role
(Jonassen, 1997). The eectiveness of questioning strategies to scaold learning has been highlighted in several
*

Corresponding author. Tel.: +30 2310997902; fax: +30 2310998419.


E-mail address: sdemetri@csd.auth.gr (S.N. Demetriadis).

0360-1315/$ - see front matter  2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compedu.2007.09.012

940

S.N. Demetriadis et al. / Computers & Education 51 (2008) 939954

situations. In reviewing the literature Ge (2001) concludes that scaolding strategies such as coaching through
prompts (e.g. King, 1992; Scardamalia, Bereiter, & Steinbach, 1984) and guiding students to self-generate
questions (e.g. King, 1991; Rosenshine, Meister, & Chapman, 1996) may promote comprehension, monitor
cognitive thinking, and facilitate general problem-solving and reective thinking. Researchers expect that
these strategies should be successful in supporting also ill-structured problem-solving tasks (Ge & Land,
2004), although it has been pointed out that few studies have focused on the eectiveness of scaolding in this
kind of problems (Ge & Land, 2003). Available studies conrm these expectations, showing that scaolding
students may signicantly improve their skills in ill-structured problem-solving as well (e.g. Davis & Linn,
2000; Ge & Land, 2003; Lin & Lehman, 1999).
This study provides additional evidence on the role of question prompts as students cognitive scaolds
when learning in ill-structured domains. The focus is on case-based learning (CBL) in technology-enhanced
environments and on scaolding the problem representation task. A specic questioning strategy is implemented in a web-based environment for CBL during learners criss-crossing activity prescribed by cognitive
exibility theory (CFT) (Spiro & Jehng, 1990). The aim of the questioning strategy is to guide students attention on important aspects of the case-based material thus facilitating the problem representation process. The
study explores how this questioning intervention can improve students conceptual knowledge of the domain
and also knowledge transfer in novel problem-solving situations. According to theory, learning from cases in
an ill-structured domain can be facilitated, when students are guided to study thematic instantiations and
focus on the exhibited variability across various cases, thereby constructing exible knowledge schemata
(e.g. Spiro, Feltovich, Jacobson, & Coulson, 1992). However, it has not been investigated if additional scaffolding interventions, such as question prompts, can further enhance learning. The results of our study inform
designers of technology-enhanced learning environments (TELEs) on the value of question-based scaolding
for students when criss-crossing contextually rich case-based material.
An additional issue explored is whether students epistemological beliefs aect their learning in the above
condition. It has been argued that students cognitive engagement in instructional situations may be aected
by their epistemological beliefs (DeBacker & Crowson, 2006). A previous study (Jacobson, Maouri, Mishra, &
Kolar, 1996) has illustrated that students who conceptualize knowledge as a complex structure that requires
personal involvement and increased eort, are in better position to benet from learning in a complex and
demanding learning situation. In this study we speculate that learning, by criss-crossing cases and being scaffolded by question prompts, can become a demanding learning task for students and may not be equally well
accepted by them, depending on their epistemological beliefs.
In the following, we present the theoretical underpinnings of the study, the research design and results,
along with discussion on emerging implications for the design of TELEs.
1.1. Question prompts as students cognitive scaolds
Question prompts are sets of static questions, used to guide and scaold the learning process oering both
cognitive and metacognitive support to students (Ge, 2001). They usually appear in the form of procedural,
elaboration or reection prompts and they have consistently proven to be of value in diverse situations. For
example, McDaniel and Donelly (1996) found that adding why . . . questions to textual information
resulted in greater factual and inference learning. King (1992) showed how elaboration prompts of generic
format like How are . . . and . . . alike? What are the strengths and weakness of . . .?, How does . . .
aect . . .? can signicantly facilitate knowledge building by eliciting explanatory responses and high level
thinking elaboration. Self-generating questions can also be benecial for learning. A review by Rosenshine
et al. (1996) concludes that training learners to generate questions improved their overall comprehension of
textual materials. The authors suggest that generic questions may have been more eective because they
promote deeper understanding, initiate recall of background knowledge, require integration of prior knowledge and provide more direction for processing (p. 200). Question prompts have been used in technologyenhanced learning environments to help direct students towards learning-appropriate goals such as focusing
student attention, modeling the kinds of questions students should be learning to ask, and helping make
their thinking visible and thus an object for reection (Azevedo & Hadwin, 2005; Azevedo, Cromley, & Seibert, 2004; Hmelo & Day, 1999).

S.N. Demetriadis et al. / Computers & Education 51 (2008) 939954

941

Ge and Land (2004), however, argue that little has been done to implement scaolding methods in ill-structured domains and that much of what has been written on scaolding ill-structured problem-solving has not
been empirically validated (p. 6). Ill-structured problems are highly contextualized problems with vagueness
and ambiguity dominating some or all aspects of the problem. They are typically encountered in professional
contexts (real-world problems) and the cognitive skills necessary for their solution are dierent from those
required for well-structured problem-solving (Jonassen, 1997), including extensive problem representation,
active construction of the problem space, justication, monitoring and evaluating (Sinnott, 1989; Voss & Post,
1988).
Certain studies have focused on the way that question prompts can scaold learners engaging in ill-structured problem-solving activities. Ge and Land (2003) provide evidence that the students who received prompts
in four fundamental problem-solving processes (problem representation, solution formulation, justication,
and monitoring/evaluation), signicantly outperformed those without any prompting. Lin and Lehman
(1999) report that justication prompts facilitated students in knowledge transfer to a contextually dissimilar
problem situation by directing their attention to understanding when, why, and how to employ experiment
design principles and strategies. Davis and Linn (2000) investigated how scaolding in the form of reection
prompts, can help students be autonomous integrators of their knowledge. They conclude that self-monitoring
prompts encourage students to reect on their own understanding, thereby demonstrate an integrated understanding of the relevant science.
1.2. Scaolding students in case-based learning
Although case-based learning (CBL) is a well-established instructional method, when practicing CBL, two
instructionally challenging issues need always careful interventions (Feltovich, Spiro, Coulson, & Feltovich,
1996):
(a) First, how to help students avoid misconceptions by not oversimplifying the material. Students need to
work through several cases to develop deeper domain-specic knowledge (such as domain concepts,
rules, principles), for it is only when many cases have been acquired, that domain knowledge can be
migrated out of the cases to form a general model of the domain (Kolodner, 1993, p. 131).
(b) Second, the knowledge transfer problem, or how to support students apply their knowledge to novel
problem situations, which may signicantly dier from those encountered in the instructional setting.
Kolodner (1993, p. 138) argues that people . . . do not always remember the right cases on which to
base their reasoning and argumentation. If knowledge is to be successfully transferred then students
need to be supported in their ability to recall relevant cases for reasoning in novel problem-solving
situations.
Indeed, the problem representation process in ill-structured problem-solving requires that solvers recall and
reect on a large amount of related information from memory (Voss & Post, 1988) and actively construct a
problem space that contains (a) the possible states of the problem, (b) the problem operators, and (c) the problem constraints. In doing so, solvers need to rely upon exible cognitive schemata (Spiro & Jehng, 1990) that
entail ecient conceptual indexing of relevant cases.
Advocates of cognitive exibility theory (CFT) have long suggested that students should be guided to
criss-cross the informational landscape shaped by dierent cases, in order to develop the kind of exible
learning necessary in ill-structured domains (e.g. Spiro & Jehng, 1990). Flexible learning denotes students
ability to construct non-rigid cognitive schemata that can be appropriately restructured to improve performance in diverse problem conditions. The criss-crossing method is expected to facilitate cognitive exibility,
by guiding students to view domain structural themes from various viewpoints and in various situations
(Spiro et al., 1992). A theme is considered as a more abstractly represented domain key issue, frequently
appearing in cases but with its case-specic instantiations exhibiting considerable variability depending on
case context. For example, in the domain of software project management a frequently discussed theme is
changing requirements (during project development). By considering the instantiations of this specic
theme across various cases (possibly also linked to other relevant themes), students are expected to attain

942

S.N. Demetriadis et al. / Computers & Education 51 (2008) 939954

exibility of learning, emerging not from the repetitiveness of similar information (rigid schema) but from the
context-dependent variability studied across theme instantiations (exible schema).
At this point we emphasize that to fully reap the benets of criss-crossing an instructor has to engage
students in some form of context-oriented analysis. Simply speaking, context-oriented analysis means
that students reect on the eect that various contextual constraints have on the problem-solving process
presented in a case, within the perspective of a specic theme or across various themes. To support students
context-oriented analysis in technology-supported environments, Spiro and Jehng (1990) propose that
themes include appropriate thematic commentary to illustrate how the generic theme is tailored to apply
to that particular context and also how a particular instantiation of a theme is tailored to instantiations
elsewhere in the learning material (p. 178). Their suggested approach essentially presents to students the
outcomes of a pre-conducted analysis and has certain shortcomings: rst, it is not always quite clear what
the commentary should include. Second, it assigns to the learner a passive role instead of triggering active
processing of the material. Third, if case material is to be further updated (as it can be easily done in technology-supported environments) it is not possible for the instructor to foresee all possible commentaries pertinent to future theme instantiations. Finally, there is by no means indication that simply providing
additional thematic commentary could advance learners metacognitive skills relevant to learning in ill-structured domains.
We argue that in technology-supported environments for case-based learning students should be scaolded
by appropriate question prompts to reect on important contextual issues in each theme instantiation they
study while criss-crossing the material. The eect of such questioning on student learning has not been studied
so far, but based on the encouraging research evidence on the eectiveness of question prompts we expect
that learning can be improved, when using a TELE that prompts students in conducting context-oriented
analysis. Furthermore, we believe that such TELEs will be of signicant value in the instruction of large number of students, where limited instructors availability may severely reduce other forms of studentteacher
interactions that could help students reect on the context of the case-based material.
1.3. Question prompts for supporting context-oriented analysis
We follow Ge and Land (2003) guideline that questions should be elicited from content-domain experts.
Since, however, our objective is not to construct domain-specic, but rather method-specic prompts, we stipulate that the appropriate question prompts should trigger those cognitive processes that are relevant to generating the context of a situation. Kokinov (1999) argues that elements of contextual information available to
a reasoner are induced from at least three cognitive processes (characterized as context-generating processes): perception, memory recall and reasoning. Perception-induced context refers to contextual information
available through perception of the environment. Memory-induced context refers to elements which are
recalled from memory and older representations, which are reactivated. Finally, reasoning-induced context
refers to representations that are derived through reasoning process (for example while setting goals, dening
strategy, etc.). The above model implies that a questioning scheme for context-oriented analysis should provide (at least) three question prompts:
(a) A prompt, relevant to the perception process (observe prompt), asking learners to identify important
case-specic information (for example events, decisions, etc.) and their eect on the situation.
(b) A prompt, relevant to memory recall process (recall prompt), asking learners to link information from
step (a) to similar/relevant information encountered in other cases.
(c) A prompt, relevant to reasoning process (conclude prompt), asking learners to do some reasoning
based also on results from previous steps, preferably reaching useful conclusions regarding the professional practice in the eld.
Implementing this scaolding approach we developed eCASE, a web-based environment for supporting
case-based learning (Papadopoulos, Demetriadis, Stamelos, & Tsoukalas, 2007). eCASE was used in this
study to engage students in case-based learning oering also scaolding question prompts. In eCASE, an
instructor can exibly accomplish three major tasks:

S.N. Demetriadis et al. / Computers & Education 51 (2008) 939954

943

(1) Upload advice-cases to the system: An advice-case is a full case presenting some useful experience in
the eld (Jonassen & Hernandez-Serrano, 2002). Advice-cases are organized in smaller parts (caseframes) each part presenting a domain theme, that is, some meaningful and self-contained aspect
of the whole case. For example, an advice-case in eCASE could possibly be organized in three
case-frames, under the titles The role of end-users, Changing requirements and Executive support and commitment which are considered as important themes in the software project management
domain.
(2) Prepare scenarios for students and assign to them paths for criss-crossing: A scenario is a problem-case
anchoring student learning in realistic and complex problem situations in the eld. After presenting
the problem, a scenario poses to students some critical open-ended questions, engaging them in decision-taking processes, as if they were eld professionals. To support students in taking informed decisions the instructor can dene paths to guide them in their navigation through advice-cases. A
path is a succession of case-frames that the instructor considers relevant to the scenario and suggests
to students as reading material. For example, when preparing a scenario where the theme changing
requirements is of importance, the instructor can built a path linking all case-frames, in various
advice-cases, presenting this theme. Thus, a scenario can be accompanied by one or more such paths
(depending on its learning objectives) and this feature implements CFT prescriptions for thematic
criss-crossing.
(3) Scaold students in context-oriented analysis: The instructor can activate the presentation of scaolding
questions to students during criss-crossing. The system allows free input of question text; however it also
oers guidelines on how to construct questions based on the aforementioned observerecallconclude
model. The instructor decides on the fade-out level of the scaold, ranging from persistent (students
have to provide answers to the scaolding questions each time they study a new case-frame in a path) to
optional (question prompts are presented but there is no requirement for students to provide answers).
The instructor has also the option of providing feedback to students responses tailored to the content of
each case-frame.

1.4. Inuence of students epistemological beliefs on learning


This study considers also the eect of students epistemological beliefs (EB) on learning, an aspect of learning that has not been adequately explored in relation to cognitive scaolding. Learners, in general, develop
their own private system of epistemological beliefs, that is, beliefs about how learning works and how human
mind develops understanding of real-world situations. Schommer (1994) summarizes that epistemological
beliefs aect learning both directly and indirectly and their eects are often mediated by other aspects of cognition. Furthermore, she identies ve major epistemological dimensions: certainty of knowledge, ranging
from knowledge is absolute to knowledge is tentative; the structure of knowledge, ranging from knowledge is
organized as isolated bits and pieces to knowledge is organized as highly interwoven concepts; the source of
knowledge, ranging from knowledge is handed down by authority to knowledge is derived through reason, the
control of knowledge acquisition, ranging from the ability to learn is xed at birth to the ability to learn can be
changed; and the speed of the knowledge acquisition, ranging from knowledge is acquired quickly or not-at-all
to knowledge is acquired gradually (Schommer, 1994, p. 28).
Jacobson et al. (1996) have shown how students who are more open to knowledge and learning complexities (students with complex-EB) benet more from complex hypermedia educational environments relatively to those with simple-EB. More recently, DeBacker and Crowson (2006) conclude that
epistemological beliefs is an important variable for understanding learners treatment of knowledge in instructional settings and the nature of their cognitive engagement. It seems therefore that EB constitute a kind of
learners internal value system, which although possibly inaccurate (it can be in contradiction with current scientic knowledge) nevertheless it may have important inuence on the way that students manage their own
learning, by regulating the activation of students cognitive potential. We hypothesize that learning in a TELE,
where the criss-crossing technique along with question prompting are implemented, can increase the extraneous load (Sweller, 1999) imposed on students by the specic instructional design. This, along with the

944

S.N. Demetriadis et al. / Computers & Education 51 (2008) 939954

intrinsic load due to the complexity of ill-structured domain, might be counter-motivational for students
with simpler EB prole. Although dicult to measure epistemological beliefs (Schommer, 1994), we devised
an instrument for assessing students EB prole and relate it to their learning outcomes.
2. Overview of the study
2.1. Goal of the study
The main goal of the study is to provide evidence on the eciency of a questioning strategy for activating
students context-generating cognitive processes, implemented in a TELE for supporting case-based instruction. Furthermore, the study explores the inuence of students EB prole on the learning outcomes.
2.2. Domain of instruction
The domain of instruction was software project management, a domain of considerable complexity and
need for knowledge transfer in job-related situations. Software project management was chosen because it
is hard to teach and learning relies largely on past experiences and project successes and failures. Diculties
in this domain stem from the fact that software processes are not well-dened, their product is intangible and
often hard to measure, and large software projects are dierent in various ways from other projects (Sommerville, 2004). In addition, many aspects of software management are not adequately formalized and involve
subjective quantication, e.g. risk prioritization. As a consequence, software managers recall and use their
knowledge about projects they have managed (or are aware of) in the past, and base their decisions on management patterns and anti-patterns. It is worth mentioning that this eld has been ranked rst among 40 computer science topics whose instruction needs to be intensied in academia because of demands in professional
context (Kitchenham, Budgen, Brereton, & Woodall, 2005).
2.3. Research questions and hypotheses
Research questions focused on whether the systematic use of the question prompts can aect (a) the acquisition of conceptual domain knowledge and (b) the ability of knowledge transfer in novel problem situations.
The issue is not trivial since it is not clear to what extend students in the non-scaolded group are also activating those processes without being triggered by an external representation (prompts). It was also investigated (c) whether students EB prole was related to their post-test performance. The three null hypotheses
tested in this study were:
 H01 (conceptual): Students in both experimental and control group perform the same in a test on acquisition of ill-structured domain conceptual knowledge.
 H02 (transfer): Students in both experimental and control group perform the same when dealing with a
novel problem situation.
 H03 (EB eect): Learning outcomes are not aected by students EB prole.

3. Method
3.1. Participants
Thirty-two Computer Science students (17 females) in their 3rd (out of 4) year of studies volunteered to
participate in the study. Students were already enrolled in one of two laboratory classes from the beginning
of the semester. After conducting pre-tests the two classes were proven to be comparable regarding students
domain prior knowledge and EB prole distribution (see further below). For practical reasons, therefore, we
decided to keep the two classes intact and assign each of them to a dierent treatment condition (experimental
group: 16 students, 10 females, control group: 16 students, 7 females). A bonus grade for the laboratory course

S.N. Demetriadis et al. / Computers & Education 51 (2008) 939954

945

was awarded to each student that completed all the phases of the study. Students were domain novices and
they had never before been engaged in case-based learning.
3.2. Design
In the study we established a 2 2 factorial design with scaolding treatment and students EB prole as
independent variables (Table 1). Students were classied as having either a simple or complex EB prole
based on their EB score.
3.3. Procedure
Students proceeded through the study in four distinct phases: pre-test, familiarization phase, study phase
and post-test. In the pre-test session, students were asked to complete: a demographic questionnaire, a prior
domain knowledge instrument and an instrument for recording students epistemological beliefs (EB instrument). During familiarization phase, students of both groups were instructed to login to the eCASE environment (whenever and from wherever they wanted) and work on a relatively simple scenario prepared for them,
including two short paths. In general, students had to read the material in paths and based on that to provide
answers to the scenario open-ended questions. No feedback was oered to them and no scaolding question
prompts were presented while studying the case-frames in paths. Students were allowed a week interval to
complete the activity and the objective of this phase was to become aware of the scenario-paths learning unit
and familiarize themselves with eCASE user-interface navigation tools.
Immediately after ending the familiarization phase, the control group continued with the study phase (it
lasted 1 week), closing with a post-test. The experimental group started its study phase (which also lasted 1
week) right after the control group completed its post-test. This shift in the study period of the experimental
group was necessary, to avoid any biases due to possible communication between students of the two groups,
since the study phase took place in natural setting.
3.4. Pre-testing and EB instrument
During pre-test the demographic questionnaire recorded some general information (sex, age, and study
semester), students internet user prole (type, location, frequency and purpose of connection) and digital document reading preference (reading from the screen vs. reading from a print-out). The prior domain knowledge
instrument included a set of eight open-ended question items relevant to domain conceptual knowledge. Students answers to this instrument were assessed using a four-step scale (0 = Rejected, 1 = Rejected-major revisions, 2 = Accepted-minor revisions, 3 = Accepted).
For identifying students EB prole we designed an EB instrument implementing a method reported in
other similar studies (Jacobson et al., 1996): the instrument included 10 paired statements, one positively
and one negatively worded, each pair referring to one of the epistemological dimensions suggested by Schommer (1994), that is, certainty, structure, source, control and speed of knowledge acquisition (see Appendix A
for sample statements). Students were asked to indicate their level of agreement (using a ve-step Likert scale,
from 1 = Strongly Disagree to 5 = Strongly Agree). Scores in negatively worded statements were afterwards
inverted and a mean score was calculated for each epistemological dimension. Higher values were indicators of
a students attitude in favor of more open, exible, constructive and complex perspective of knowledge. The
mean of these scores across the ve epistemological dimensions was students nal EB record, varying from
Table 1
A 2 2 factorial design of the study with scaolding and EB prole (simplecomplex) as independent variables
EB prole

Scaolding treatment
Scaolded (experimental)
Non-scaolded (control)

Simple-EB

Complex-EB

SiS (n = 8)
SiNS (n = 8)

CoS (n = 8)
CoNS (n = 8)

946

S.N. Demetriadis et al. / Computers & Education 51 (2008) 939954

1 (a rather simple EB prole) to 5 (a complex EB prole). The EB mean for all students was
M = 3.66 (n = 32, SD = 0.44). Students with EB record above the median were classied as complex-EB
students (n = 16, M = 4.03, SD = 0.31, range= [3.70, 4.90]) and the rest of them as simple EB students
(n = 16, M = 3.30, SD = 0.17, range = [2.90, 3.60]). After administering the EB questionnaire we discovered
that students were already equally distributed to the two conditions, in terms of their EB prole (both groups
had eight complex-EB and eight simple-EB students).
The reliability of the EB instrument was only moderate (Cronbachs alpha = .51) and at the lower end of
ranges reported in other studies on measuring student EB prole (from .63 to .85 (Schommer, Calvert, Gariglietti, & Bajaj, 1997) and from .51 to .78 (Schommer, 1993)). This indicates low correlation of the items but
should not necessarily be considered as a problem, since the instrument was designed to measure students
beliefs on ve dierent epistemological dimensions supposed to be independent (at least in principle).
3.5. Study conditions in control group
Control group was given three new scenarios accompanied by ve paths (totally) to study. The scenarios
were text-based presentations of problem cases relevant to the development of software projects and the scenario questions focused on managerial decisions that students should take in order to deal with the problem
situations. These new scenarios were more complex and demanding compared to that in the familiarization
phase addressing more themes of the domain. Consequently, the accompanying paths included more caseframes from advice-cases. These advice-cases were selected and adapted from authentic software management
cases reported in the literature (e.g. Ewusi-Mensah, 2003, see Appendix B for an excerpt).
Students studied individually and they had to read each scenario (in any preferred order) and then navigate
through respective paths to study the suggested advice-case material. This information presented how managerial decisions in relevant past authentic software development projects were proven to be either inecient
(causing, for example, project delays, personnel churn or cost overruns), or skillful, thus leading to a successful
project completion (for example, correct matching of skill proles in groups of co-workers). After navigating
through all paths of a scenario, the system allowed students to upload their scenario answers and see instructors feedback, which was pre-entered to the system and was the same for all students. Thus, they were given
the opportunity of self-assessing the quality of their response. Students were also supplied with an attitude
instrument to ll in during study phase.
3.6. Study conditions in experimental group
Students in experimental group studied exactly the same material in the same way that control group did,
except from one dierence. Each time they navigated to a new case-frame in a path the observerecallconclude scaolding questions appeared and students had to submit answers to these questions (Fig. 1).
Although minor dierences existed in the exact format of the question text in the various case-frames, in general they were stated as follows: (a) What concrete events (decisions, etc.) imply possible problems during
project development? (b) In what other case do you recall having encountered similar problems in project
development? (c) What are some useful implications for the successful development of a project?
The recall question was additionally providing a case archive link for students to navigate and review
relevant case-frames in already studied advice-cases. The above questions are considered as triggering the students context-generating cognitive processes in the sense that they specically prompt learner to activate the
perception process (focus on concrete events), the memory recall process (recall relevant information from
other studied cases) and also the reasoning process (suggest useful implications for domain problem-solving).
The scaolding fade-out level for the study group was set to no fade-out condition. This means that the
question prompts were persistently presented to students each time they navigated to a new case-frame.
Answers for the scaolding questions had to be submitted in order for the system to consider nalized the
study of each case-frame and eventually allow students to upload their answers to scenario questions. The
requirement for answering the scaolding questions was imposed to avoid what sometimes is reported as a
problem with question prompts, namely the failure of students to engage in deeper processing due to supercial engagement with question prompts (Greene & Land, 2000).

S.N. Demetriadis et al. / Computers & Education 51 (2008) 939954

947

Fig. 1. The implementation of scaolding question prompts in the eCASE environment. Students submit their answers using this web
form, which appears when clicking the Whats your opinion? button available in every case-frame of a path.

Feedback was provided to students after submitting their answers, however it was not an extensive correctanswer modeling feedback (as in the scenario questions), but instead coaching oriented, simply providing some
incomplete hints just to make students understand whether they were on the right track. It should be emphasized that feedback did not present any additional information to the students in experimental condition as
compared to those in the control condition. Furthermore, the recall question did not provide any feedback
at all but only a general suggestion to review already studied relevant case-frames. Feedback information was
also pre-entered to the system and was the same for all students in the experimental group. After submitting
their answers in all case-frames of the scenario paths, the system allowed students to upload their answers to
the scenario questions and that was the end of the study in the specic scenario. Students in the experimental
condition were also supplied with the same attitude instrument expanded with one more section referring to
the design of the scaold.
3.7. Post-test instruments
Post-tests included two instruments: an attitude and a learning eectiveness instrument. The rst, as already
mentioned, was handed to students during the study phase and it included 12 statements (8 for the control
group) for assessing the design of various user-interface features and the learning experience as a whole. It also
focused on recording students opinions and likes/dislikes regarding the learning experience. Students denoted
their degree of agreement using a ve-step Likert scale (1 = Strongly Disagree to 5 = Strongly Agree). This
instrument also allowed them to freely provide comments and suggestions for improving the environment.
The learning eectiveness instrument was answered by all students during their post-test session. It comprised two sections focusing on (a) acquired domain-specic conceptual knowledge, and (b) students potential

948

S.N. Demetriadis et al. / Computers & Education 51 (2008) 939954

for knowledge transfer in novel problem situation. The rst section included three domain conceptual knowledge questions (see Appendix C for sample items). The answers to these questions were not to be found as such
in the study material but rather to be constructed by combining parts of information presented in various
occasions in the case material. The second section presented a dialogue-formatted scenario a completely different format as compared to the narrative format of the scenarios and the advice-cases presented during study
phase, to avoid student answering based on supercial clues (see Appendix D for an excerpt). In this scenario,
various stakeholders (company CEO, CFO, clients, technicians, etc.) were discussing managerial issues of an
ongoing software project in an everyday professional context. Students were asked to identify elements in the
scenario that might be indicators of inecient management and suggest resourceful alternatives. This is considered as a far transfer task since for accomplishing it students were required to have developed a deep
enough understanding of domain problem situations and could not simply apply step by step a learned
procedure.
Students post-test answer sheets of both conditions were mixed together to avoid any biases and were
assessed by two independent raters. A 010 scale was used and raters followed predened instructions on
how to assess each specic item. Eventually each student received two scores: a score for answering
domain-specic conceptual knowledge questions (conceptual score), and a score for the new scenario analysis (transfer score). These scores were calculated as the mean value of the respective conceptual and
transfer scores provided by the two raters. The deviation between scores from dierent raters was not to
exceed the 20% level (two grades in the assessment scale) else raters had to discuss the issue and reach a consensus (however, this was never required).
4. Results
4.1. Pre-test
The majority of students had PSTN (dial-up) internet connection and logged-in the eCASE system from
home. They were experienced internet users, using internet services every day for purposes such as email communication, information seeking, chatting, entertainment, and downloading. Most of them (65%) preferred to
read instructional material from a print-out rather than directly from screen. The above traits were common
for students in both conditions. Pre-test results indicated that students were domain novices scoring very low
(control group M = .56 (out of 3), SD = .41; experimental group M = .82, SD = .47; F(1, 28) = .61, p > .05,
for EB prole and F(1, 28) = 2.73, p > .05 for scaolding treatment). The two conditions, therefore, were comparable regarding students prior knowledge.
4.2. Post-test
Attitude instrument: Students responses in the attitude instrument armed that they could easily accomplish all routine tasks in the eCASE system (navigate through the environment, select appropriate material
and upload their answers). Table 2 presents detailed results in some interesting items of the attitude
instrument.
For all statistical analyses a level of signicance at .05 was chosen. Applying two-way ANOVA yielded signicant main eect for the EB prole in item 5 (F(1, 28) = 6.81, p < .05) and for scaolded treatment in item 6
(F(1, 28) = 7.51; p < .05).
Learning eectiveness instrument: Table 3 presents the two groups post-test results in the conceptual and
transfer dimension. These were the dependent measures concerning students learning outcomes.
Applying two-way ANOVA revealed a signicant main eect for the scaolding treatment regarding both
dependent measures (conceptual: F(1, 28) = 4.55; p = .042, ES = .78; transfer F(1, 28) = 5.93; p = .021,
ES = .89). Cohens d formula was used for calculating eect sizes. Neither the main eect for the EB prole
(conceptual F(1, 28) = 2.02, p = .166; transfer F(1, 28) = .83, p = .37) nor the interaction between the two independent variables reached statistical signicance (conceptual: F(1, 28) = 1.14, p = .295; transfer:
F(1, 28) = 1.48, p = .233). Post hoc t-tests revealed the following: (a) scaolded complex-EB students performed better than non-scaolded complex-EB, and the dierence was signicant for both dependent mea-

S.N. Demetriadis et al. / Computers & Education 51 (2008) 939954

949

Table 2
Students responses in some items of the attitude instrument
Group

SiS

CoS

Scaolded (SiS + CoS)

Non-scaolded (SiNS + CoNS)

16
3.38 (1.15)

na

2. Scaolding questions were tiresome and I would prefer them not to exist
n
8
8
16
M (SD)
3.13 (1.55)
3.75 (0.89)
3.44 (1.26)

na

1. Scaolding questions helped me understand the case material


n
8
8
M (SD)
3.50 (1.31)
3.25 (1.04)

3. If answering the scaolding questions was not mandatory, I would have tried to answer the scenario questions without actually reading the
path material
n
8
8
16
M (SD)
1.50 (1.07)
2,00 (1.31)
1.75 (1.18)
na
Group

Scaolded (SiS + CoS)

Non-scaolded (SiNS + CoNS)

4. Studying in eCASE was an overall positive learning experience


n
16
16
M (SD)
4.38 (0.50)
4.19 (0.83)

Simple-EB (SiS + SiNS)

Complex-EB (CoS + CoNS)

16
4.31 (0.48)

16
4.25 (0.86)

5. Studying in eCASE was complicated for me


n
16
16
M (SD)
2.25 (1.00)
1.50 (0.52)

16
1.88 (0.96)

16
1.88 (0.81)

6. How much time did you spend studying in eCASE (estimated hours)?
n
16
16
M (SD)
5.78 (3.24)
6.47 (2.63)

16
7.41 (3.05)

16
4.84 (2.20)

Table 3
Post-test results in learning eectiveness instrument
Simple-EB

Complex-EB

SD

Conceptual
Scaolded
Non-scaolded

8.08
7.70

(0.52)
(1.29)

Total

7.89

Transfer
Scaolded
Non-scaolded
Total

Total

SD

8
8

8.95
7.83

(0.65)
(1.25)

(0.97)

16

8.39

8.75
8.25

(1.28)
(1.16)

8
8

8.50

(1.21)

16

SD

8
8

8.52
7.77

(0.72)
(1.23)

16
16

(1.13)

16

8.14

(1.06)

32

8.88
7.38

(0.64)
(1.40)

8
8

8.81
7.81

(0.98)
(1.32)

16
16

8.13

(1.31)

16

8.31

(1.25)

32

sures (conceptual: t(14) = 2.24, p = .042; transfer: t(14) = 2.74, p = .016). (b) Scaolded complex-EB performed better than scaolded simple-EB in the conceptual measure (t(14) = 2.95, p = .011). The above may
suggest some interaction between student EB prole and the scaolding treatment. However, considering
the transfer measure the dierence between scaolded complex-EB and scaolded simple-EB was not significant (t(14) = .25, p = .809).
5. Discussion
5.1. Hypotheses testing
In the light of the above results the rst two null hypotheses of the study (concerning the eect of the scaffolding treatment) are rejected and alternatives are proposed. The third hypothesis referring to the eect of the
EB prole on learning outcomes is however tentative (see Table 4).

950

S.N. Demetriadis et al. / Computers & Education 51 (2008) 939954

Table 4
Hypotheses testing
Null hypothesis

Result

Alternative

H01 (conceptual): Students in both experimental


and control group perform the same in a test
on acquisition of ill-structured domain
conceptual knowledge
H02 (transfer): Students in both experimental
and control group perform the same when
dealing with a novel problem situation

Rejected on the basis of


ANOVA analysis:
F(3, 28) = 4.55; p = .042,
ES = .78
Rejected on the basis of
ANOVA analysis:
F(3, 28) = 5.93; p = .021,
ES = .89
Tentative. Post hoc t-test oer a
basis for rejection but further
investigation is required

HA1 (conceptual): Students who study in a


scaolded CBL condition perform better in a
test on acquisition of ill-structured domain
conceptual knowledge
HA2 (transfer): Students who study in a
scaolded CBL condition perform better in a
novel problem situation

H03 (EB eect): Learning outcomes are not


aected by students EB prole

Students in the scaolded group achieved signicantly higher scores in both dependent measures, with high
eect sizes. Checking their answers to scaolding questions in the system database conrmed that all 16 students of the group did actually try to provide full answers and hence their study condition was indeed aected
by the question prompts.
This outcome is a clear indication that explicitly asking scaolding questions to activate students contextgenerating cognitive processes can have positive eect on learning from cases, and the assumption that students will equally well activate these processes, without prompting, is not strong. It is worth re-emphasizing
here that the same information was presented to both groups (the feedback presented to the experimental
group when answering the scaolding questions was very limited and did not present any additional information). Furthermore, students in both conditions were oered a common questioning activity (answering scenario questions) to engage them in active information processing. However, the superior performance of
the experimental group indicates that the impact of question prompts during the study of advice-cases was
signicant, resulting in deeper domain knowledge understanding and potential for knowledge transfer in novel
problem situations.
Why is that so? Research on the role of scaolding has revealed that various types of scaolds can help
learners improve their learning both at cognitive and metacognitive level (Azevedo & Hadwin, 2005). More
specically, Davis and Linn (2000) emphasize that those students who are not prompted to articulate their
ideas, miss opportunities to integrate their knowledge. We argue that students in the experimental group processed information and integrated it in their cognitive schemata more eciently while articulating their understanding in the form of answers to the question prompts. This allowed them to recall relevant information
more easily, when facing the post-test questions on conceptual and transfer tests.
However, students improved performance in the transfer test needs explanation that goes further beyond
the simple recall of information. In general, higher performance in far transfer tasks is an indication that students have developed more abstract understanding, which provides a bridge between contextually dissimilar
situations (Salomon & Perkins, 1989) supporting students to solve new problems in a exible way (Brown &
Campione, 1994). The ndings of this study suggest that prompting students to connect between cases (the
memory recall prompt) and articulate their conclusions (the conclude prompt) helped them to develop
a deeper (more abstract) internal representation of the principles underlying the domain, thus performing better in a dierent context in the far transfer test. This outcome also supports an alternative option to the suggestion by Spiro and Jehng (1990) that thematic commentary should be presented to students while crisscrossing case-based information. Students can be prompted to reect on the issues that the thematic commentary would otherwise present to them and this cognitive activity can be highly benecial for them when trying
to develop domain deeper understanding and transfer knowledge to novel situations.
Overall, the present study provides substantial evidence that computer-based question prompts can eectively
scaold learners in the fundamental process of problem representation (Voss & Post, 1988) in ill-structured
domains, activating their context-generating cognitive processes and guiding them to reect on the eect that
various contextual constraints have on problem-solving. Other researchers (Ge & Land, 2003) have also
reported on the eectiveness of prompts for problem representation but without using computer-based scaolds.

S.N. Demetriadis et al. / Computers & Education 51 (2008) 939954

951

Interestingly enough the post hoc t-test analysis provided evidence implying an EB eect on student learning. This result is consistent with ndings from other studies documenting the impact of epistemological beliefs
on learning (Jacobson et al., 1996; Schommer, 1993, 1994). The impact of the scaold appears to be more signicant for the complex-EB students. Scaolded complex-EB students perform better than non-scaolded students of the same EB prole (in both dependent measures) and they also perform better than scaolded
simple-EB students (but only in the conceptual measure). The improved performance of this subgroup can
be explained assuming that complex-EB students were more responsive to the increased workload induced
by the scaolding intervention and therefore beneted more from the scaolded study condition. Previous
studies exploring also the EB eect provided similar indication that, in general, students with more complex
epistemological beliefs learn more eectively in complex hypertext/hypermedia environments (Jacobson et al.,
1996; Jacobdon & Spiro, 1995). However, it is not clear why improved performance was not detected for the
transfer measure in comparison to scaolded simple-EB students. Therefore, although the present analysis has
yielded interesting results regarding the EB eect, the small sample size and the moderate reliability of the EB
instrument make us suggest further investigation.
5.2. Student attitudes
Students responses to the attitude instrument indicated that their impression of the supportive role of the
scaold was positive but rather weak (item 1 in Table 2). It also appears that consistently answering scaolding questions was tiresome for them (item 2). In the current study, we applied the no fade out condition as a
guarantee that students will use the scaolding questions mindfully (Ge & Land, 2003). However, an improved
design could simply remind the learners of the question prompts after rst practicing it a few times. Clark and
Mayer (2003) suggest a similar approach for teaching self-explanation techniques, stating that periodically
reminding the learner to self-explain should be sucient (p. 191).
Students in both conditions value the overall learning experience in eCASE as positive (item 4) and not
complicated (item5). However, the analysis of item 5 (studying in eCASE was complicated for me) for
the scaolded group revealed that simple-EB students agreed with this statement signicantly stronger in comparison to the complex-EB subgroup. This, we believe, is a positive indication that the EB instrument has successfully detected a certain level of students epistemological dierences. Finally, students were asked to make
a rough estimation of the time they spent during study phase in the eCASE environment (item 6). The dierence between the two groups is signicant with the scaolded group spending more on-task time, obviously
due to the reection induced by the scaold and the subsequent writing process. Unfortunately our design
did not enable the separate estimation of each time interval to attain a clearer picture of their contribution
to the increased student workload.
5.3. Study limitations
This study employed a formal experimentation design integrating also aspects of a more naturalistic eld
approach, such as the fact that students were free to self-organize their study during a week time interval. Such
a prolonged study period was necessary in order to allow the domain complexity to emerge through the several
available scenarios and advise-cases, something not feasible in a typical experimental laboratory session lasting few hours. Certainly a point of concern is the way that the two groups were formed. Students had selected
their preferred group in the beginning of the semester and we had very good practical reasons to keep the two
laboratory classes intact as experimental and study group (one such reason being that students were proved to
be already equally distributed by means of their EB prole). Employing pre-formed classes as study groups
has also been reported elsewhere (Ge, 2001), however we acknowledge that an improved design should randomly assign individual students to study groups.
Another issue of concern is the small sample size. Studies with group sizes comparable to that of the present
study often appear in the literature (for example, Azevedo et al., 2004; Jeong & Joung, 2007; Nam & SmithJackson, 2007; Zydney, 2004). However, in case of small sample size, it is important to validate the application
of parametric tests by investigating the homogeneity of variance and the normality of the sample distribution
(Ge, 2001; Nam & Smith-Jackson, 2007). In the present study Levene test was applied to test the homogeneity

952

S.N. Demetriadis et al. / Computers & Education 51 (2008) 939954

of variance and the ShapiroWilk test to investigate whether the distribution satised the normality criterion.
Applying both tests in all pair-wise situations resulted in non-signicant outcome (p > 0.05) indicating that the
samples had homogenous variances and normal distribution of data. The above validate our choice to apply
and report the results of parametric tests such as ANOVA and t-test. However, the statistics for the EB eect
on student learning are based on even smaller group size, thus limiting the generalisability of this result. In line
with other researchers who suggest replication of study when the sample is small (for example, Ge & Land,
2003; Ge, Chen, & Davis, 2005), we have chosen to report this nding as an interesting, though tentative element, so that future studies with larger samples can further explore this issue and provide more solid statistical
analysis.
5.4. Implications for the design of TELEs
The results of this study suggest that designers of TELEs for learning in ill-structured domains may consider as a viable option the integration of context-oriented generic questioning prompts, based on models of
cognitive activity which is necessary for the processing of contextually rich material. Broadening our perspective we suggest that developers of TELEs need to consider methods for the ecient and exible integration of
questioning components in their products, in the light of research that consistently highlights the benecial
eects of questioning (and self-questioning) strategies on students learning. Although questions for the learner are routinely included in instructional software, this is mostly done in an ad hoc way, hardwired to the
system and tailored to specic content, without really supporting students metacognitive awareness, development of self-questioning skills and instructors exible intervention. Questioning schemes like the one used in
this study, can be further formalized and be given some appropriate computerized format, thus enabling a
more intelligent and ecient manipulation by the system. We believe that such scaolding tools can become
fundamental software components that would allow instructors to exibly adapt all functional aspects of a
questioning strategy they wish to implement in the technology-enhanced learning environment.
In sum, this study provided clear evidence that a questioning strategy implemented in a technology-based
environment can eciently scaold students during thematic criss-crossing of case-based material. Furthermore, it oered indication (although somehow tentative) that student EB prole aects learning in a situation
where the cognitive load of case-based learning is further increased by the activity induced by scaolding.
Acknowledgements
Dierent parts of this study have been funded by Kaleidoscope Network of Excellence (contract No
507838) and Greek-German interstate program IKYDA06.
Appendix A. EB instrument sample (paired items on structure dimension)
(a1) Positively worded
Everything man knows is eventually somehow connected to all other issues and inuences them. Knowledge is like a huge network where one may set out from one node and navigate to various others, realizing
that knowledge on an issue may inuence the knowledge we possess on any other.
(a2) Negatively worded
Human knowledge seems to me that it is essentially a vast number of unrelated things that we know on
various subjects. Nothing relates, for example, gardening to astrophysics. So, knowledge looks more like a list
of disparate issues.
Appendix B. Example of studied material
The following is an excerpt from a case in the software project management domain (Ewusi-Mensah, 2003,
p. 94). This and other material of similar complexity, adapted from the international literature, was presented
to students in case-frames.
The collapse of Phar-Mur Luc and the contract with UHC Consortium

S.N. Demetriadis et al. / Computers & Education 51 (2008) 939954

953

In May 1993, following the collapse of Phar-Mur Luc, a Midwestern pharmacy chain responsible for the
15% of FoxMeyers revenue stream, its [FoxMeyers] top management felt added pressure to make up for
the lost business and ensure its survival. As a result, FoxMeyer aggressively pursued University Health System
(UHC) Consortium, a national network of major teaching hospitals, by underbidding the contract, expecting to
make up for the smaller prot margin with the increase in volume they felt was assured by the Delta project. The
successful acquisition of the UHS contract, however, set in motion a stream of actions the necessitated changes
in the original requirements for SAP R/3 implementation. For example, because UHS hospitals are scattered
across the United States, it required the opening of six new warehouses in the West. In addition, the implementation schedule had to be shortened to get parts of the SAP nancial software running three months sooner
than planned to accommodate the requirements of the UHS contract. FoxMeyers own CIO admitted that
the new contract [i.e., UHS] increased the diculties of the already-ambitious plan. This fact is echoed by
Jesitus, who reported that the SAP contract on the project indicated that the new contract introduced a significant change in FoxMeyers business. It added to the problems of Delta, because the focus of the project changed drastically from the original specications to ensure that the UHS consortiums needs could be satised.
Appendix C. Post-test domain knowledge sample question (conceptual)
As a software project manager you have to decide on recruiting partners for your project team. (a) What
criteria do you think are important in selecting the members of the team? (b) What should you pursuit (or
avoid) in order for the team members to eciently work together during project development?
Appendix D. Excerpt from post-test scenario (transfer test)
[John] As I told you, I am convinced that this a system will help tune up the company performance. Smith
asked me to recommend someone who is well aware of the company functioning, to discuss details. I
want you to work on this with him and keep me informed.
[Anne] Thatll be ne, but I think Peter will be of considerable help here. He knows details on all technical
issues much better than I do and he is responsible for the storehouse.
[John] Peter? No! Id rather leave him out of this! He was one of my fathers rst employees and I really
dont think he knows so much about technology as you think he does. I prefer you to take over, that
you are younger in the job. Peter and the others in the store will be informed when the rst components of the system will be ready.
References
Azevedo, R., Cromley, J. G., & Seibert, D. (2004). Does adaptive scaolding facilitate students ability to regulate their learning with
hypermedia?. Contemporary Educational Psychology 29, 344370.
Azevedo, R., & Hadwin, A. F. (2005). Scaolding self-regulated learning and metacognition Implications for the design of computerbased scaolds. Instructional Science, 33, 367379.
Brown, A. L., & Campione, J. C. (1994). Guided discovery in a community of learners. In K. McGilly (Ed.), Classroom lessons: Integrating
cognitive theory and classroom practice (pp. 229270). Cambridge, MA: MIT Press.
Clark, R. C., & Mayer, R. E. (2003). e-Learning and the science of instruction. San Francisco: Pfeier.
Davis, E. A., & Linn, M. (2000). Scaolding students knowledge integration: Prompts for reection in KIE. International Journal of
Science Education, 22(8), 819837.
DeBacker, T. K., & Crowson, H. M. (2006). Inuences on cognitive engagement: Epistemological beliefs and need for closure. British
Journal of Educational Psychology, 76, 535551.
Ewusi-Mensah, K. (2003). Software development failures. Cambridge, MA: The MIT Press.
Feltovich, P. J., Spiro, R. J., Coulson, R. L., & Feltovich, J. (1996). Collaboration within and among minds: Mastering complexity,
individuality and in groups. In T. Koschmann (Ed.), CSCL: Theory and practice of an emerging paradigm (pp. 2544). Mahwah, NJ:
Lawrence Erlbaum Associates.
Ge, X. (2001). Scaolding students problem-solving processes on an ill-structured task using question prompts and peer interactions.
Unpublished doctoral thesis. Retrieved, July 7, 2006, from http://etda.libraries.psu.edu/theses/approved/WorldWideIndex/ETD-75/
index.html.
Ge, X., Chen, C.-H., & Davis, K. A. (2005). Scaolding novice instructional designers problem-solving processes using question-prompts
in a web-based learning environment. Journal of Educational Computing Research, 33(2), 219248.

954

S.N. Demetriadis et al. / Computers & Education 51 (2008) 939954

Ge, X., & Land, S. M. (2003). Scaolding students problem-solving processes in an ill-structured task using question prompts and peer
interactions. Educational Technology Research & Development, 51(1), 2138.
Ge, X., & Land, S. M. (2004). A conceptual framework for scaolding ill-structured problem-solving processes-using question prompts
and peer interactions. Educational Technology Research & Development, 52(2), 522.
Greene, B. A., & Land, S. M. (2000). A qualitative analysis of scaolding use in a resource-based learning environment involving the
World Wide Web. Journal of Educational Computing Research, 23(2), 151179.
Hmelo, C., & Day, R. (1999). Contextualized questioning to scaold learning from simulations. Computers & Education, 32(2), 151164.
Jacobdon, M. J., & Spiro, R. J. (1995). Hypertext learning environments, cognitive exibility and the transfer of complex knowledge: An
empirical investigation. Journal of Educational Computing Research, 12(5), 301333.
Jacobson, M. J., Maouri, C., Mishra, P., & Kolar, C. (1996). Learning with hypertext learning environments: Theory, design, and
research. Journal of Educational Multimedia and Hypermedia, 5(3/4), 239281.
Jeong, A., & Joung, S. (2007). Scaolding collaborative argumentation in asynchronous discussions with message constraints and message
labels. Computers & Education, 48(3), 427445.
Jonassen, D. H. (1997). Instructional design models for well-structured and ill-structured problem-solving learning outcomes. Educational
Technology Research & Development, 45(1), 6594.
Jonassen, D. H., & Hernandez-Serrano, J. (2002). Case-based reasoning and instructional design: Using stories to support problem
solving. Educational Technology Research & Development, 50(2), 6577.
King, A. (1991). Eects of training in strategic questioning on childrens problem solving performance. Journal of Educational Psychology,
83(3), 307317.
King, A. (1992). Facilitating elaborative learning through guided student-generated questioning. Educational Psychologist, 27(1), 111126.
Kitchenham, B., Budgen, D., Brereton, P., & Woodall, P. (2005). An investigation of software engineering curricula. Journal of Systems
and Software, 74(3), 325335.
Kokinov, B. (1995). A dynamic approach to context modeling. In P. Brezillon & S. Abu-Hakima (Eds.), Proceedings of the IJCAI-95
workshop on modeling context in knowledge representation and reasoning. LAFORIA 95/11.
Kokinov, B. (1999). Dynamics and automaticity of context: A cognitive modeling approach. In P. Bouquet, L. Serani, P. Brezillon, M.
Benerecetti, & F. Castellani (Eds.), Modeling and using context. Lecture notes in articial intelligence (vol. 1688). Berlin: Springer.
Kolodner, J. (1993). Case-based reasoning. San Mateo, CA: Morgan Kaufman Publishers Inc..
Lin, X., & Lehman, J. D. (1999). Supporting learning of variable control in a computer-based biology environment: Eects of prompting
college students to reect on their own thinking. Journal of Research in Science Teaching, 3(7), 837858.
McDaniel, M. A., & Donelly, C. M. (1996). Learning with analogy and elaborative interrogation. Journal of Educational Psychology, 88,
508519.
Nam, C. S., & Smith-Jackson, T. L. (2007). Web-based learning environment: A theory-based design process for development and
evaluation. Journal of Information Technology Education, 6.
Papadopoulos, P., Demetriadis, S., Stamelos, I., & Tsoukalas, I. (2007). Online case-based learning: Design and preliminary evaluation of
the eCASE environment. In Kinshuk et al. (Eds.), Proceedings of IEEE 6th International Conference on Advanced Learning
Technologies (ICALT 2006) (pp. 751755). Los Alamitos, CA: IEEE.
Rosenshine, B., Meister, C., & Chapman, S. (1996). Teaching students to generate questions: A review of the intervention studies. Review
of Educational Research, 66(2), 181221.
Salomon, G., & Perkins, D. N. (1989). Rocky roads to transfer: Rethinking mechanisms of a neglected phenomenon. Educational
Psychologist, 24, 113142.
Scardamalia, M., Bereiter, C., & Steinbach, R. (1984). Teachability of reective processes in written composition. Cognitive Science, 8,
173190.
Schommer, M. (1993). Epistemological development and academic performance among secondary students. Journal of Educational
Psychology, 85(3), 406411.
Schommer, M. (1994). An emerging conceptualization of epistemological beliefs and their role in learning. In R. Garner & P. A. Alexander
(Eds.), Beliefs about text and instruction with text (pp. 2540). Hillsdale, NJ: Erlbaum.
Schommer, M., Calvert, C., Gariglietti, G., & Bajaj, A. (1997). The development of epistemological beliefs among secondary students: A
longitudinal study. Journal of Educational Psychology, 89(1), 3740.
Sinnott, J. D. (1989). A model for solution of ill-structured problems: Implications for everyday and abstract problem solving. In J. D.
Sinott (Ed.), Everyday problem solving: Theory and application (pp. 7299). New York: Praeger.
Sommerville, I. (2004). Software engineering (7th ed.). Addison-Wesley.
Spiro, R. J., Feltovich, P. J., Jacobson, M. J., & Coulson, R. L. (1992). Cognitive exibility, constructivism, and hypertext: Random access
instruction for advanced knowledge acquisition in ill-structured domains. In T. Duy & D. Jonassen (Eds.), Constructivism and the
technology of instruction (pp. 5775). Hillsdale, NJ: Erlbaum.
Spiro, R. J., & Jehng, J. (1990). Cognitive exibility and hypertext: Theory and technology for the non-linear and multidimensional
traversal of complex subject matter. In D. Nix & R. Spiro (Eds.), Cognition, education, and multimedia. Hillsdale NJ: Ehrlbaum.
Sweller, J. (1999). Instructional design in technical areas. Camberwell, Australia: ACE Press.
Voss, J. F., & Post, T. A. (1988). On the solving of ill-structured problems. In M. H. Chi, R. Glaser, & M. J. Farr (Eds.), The nature of
expertise (pp. 261285). Hillsdale, NJ: Lawrence Erlbaum Associates.
Zydney, J. M. (2004). Students dening complex problems through questioning: The eectiveness of scaolding in a multimedia program.
In Paper presented at the annual meeting of the eastern educational research association clearwater.

You might also like