Anda di halaman 1dari 12

Diagnosis of Learning Styles Based on Active/Reflective

Dimension of Felder and Silverman’s Learning Style

Model in a Learning Management System

Ömer Şimşek, Nilüfer Atman, Mustafa Murat İnceoğlu, and Yüksel Deniz Arikan

Ege University, Department of Computer and Instructional Technologies

35100 Bornova, Izmir, Turkey,,,

Abstract. Learner centered education is important both in point of face to face

and Web based learning. Due to this importance, diagnosis of learning styles of
students in web based or web enhanced educational settings is important as
well. This paper presents prediction of learning styles by means of monitoring
learner interface interactions. A mathematics course executed on a learning
management system (Moodle) was monitored and learning styles of the learners
were analyzed in point of active/reflective dimension of Felder and Silverman
Learning Styles Model. The data from learner actions were analyzed through
literature based automatic student modeling. The results from Index of Learning
Styles and predicted learning styles were compared. For active/reflective di-
mension 79.6% precision was achieved.

Keywords: Learning styles, Felder and Silverman’s Index of Learning Styles,

Moodle, Web-enhanced learning.

1 Introduction
Individuals have different backgrounds, motivation and preferences in their own
learning processes. Web-based systems that ignore these differences have difficulty in
meeting learners’ needs effectively [1]. Therefore, when designing instructional mate-
rial, it is important to accommodate elements that reflect individual differences in
learning. One of these elements is learning styles [2]. Learning styles of learners must
be determined for adapting instructional material that best suits students learning
styles. There are many ways of detecting learners learning style and mostly it is per-
formed by questionnaires or scales. However, using questionnaires for determining
learning styles has some disadvantages such as not all the students are motivated to
fill out questionnaire. As a result, mismatches between real behavior and question-
naire answers could exist. To overcome these problems, instead of allocating time to
filling out questionnaires, student models could be constructed. An alternative ap-
proach to collect the information pertinent to a student model is to track the student’s
behavior and responses and then make inferences about general domain competence,
cognitive traits, and learning styles. The challenge of this approach is to identify and

D. Taniar et al. (Eds.): ICCSA 2010, Part II, LNCS 6017, pp. 544–555, 2010.
© Springer-Verlag Berlin Heidelberg 2010
Diagnosis of Learning Styles Based on Active/Reflective Dimension of Felder 545

collect sufficient information to make reliable and useful inferences [3]. The aim of
this paper is to automatically detect learning styles of learners through analyzing their
behaviors in an LMS course.

2 Related Work
Detecting automatically learning style requires reviewing two main contexts such as
learning styles and adaptive systems.

2.1 Learning Styles

There are many models about learning styles in literature such as Kolb [4], Dunn &
Dunn [5], Honey & Mumford [6], Myers-Briggs [7]. This study is based on Felder
and Silverman’s Learning Styles Model because of its applicability to e-learning and
compatibility to the principles of interactive learning systems design [8].
Students learn in many ways ⎯ by seeing and hearing; reflecting and acting; rea-
soning logically and intuitively; memorizing and visualizing and drawing analogies
and building mathematical models; steadily and in fits and starts [9]. The ways in
which an individual characteristically acquires, retains, and retrieves information are
collectively termed the individual’s learning style [10]. In 1988 R. Felder and L.
Silverman proposed a learning style model that classifies five dimensions of learning

Table 1. Dimensions of Felder and Silverman Learning Style Model

Dimensions of Felder and Silverman Learning

Style Model
Sensory Perception
Visual Input
Active Process
Sequential Understanding

Sensory/intuitive dimension specified the type of the information that students pre-
fer to perceive. The dimension of visual/auditory (lately amended as visual/verbal) is
refer to trough which sensory channel is external information most effectively per-
ceived; active/reflective represent how does the student process the information and
finally sequential/global shows how does the student progress towards understanding
[9].Lately, inductive/deductive was excluded from the model.

• sensing learners (concrete, practical, oriented toward facts and procedures)

or intuitive learners (conceptual, innovative, oriented toward theories and
546 Ö. Şimşek et al.

• visual learners (prefer visual representations of presented material--pictures,

diagrams, flow charts) or verbal learners (prefer written and spoken explana-
• active learners (learn by trying things out, working with others) or reflective
learners (learn by thinking things through, working alone);
• sequential learners (linear, orderly, learn in small incremental steps) or
global learners (holistic, systems thinkers, learn in large leaps) [11].
Although there are four dimensions in this model; ‘process’ dimension is investigated
because this dimension’s characteristics are supported in proposed LMS course. Due
to having various collaborative and individual web activities, the study focused on
active/reflective dimension of Felder and Silverman learning style model. The promo-
tion of individual activities such as watching videos, viewing pdf documents, answer-
ing questionnaires, attempting quizzes and collaborative activities such as forum
discussions are the reasons of choosing this dimension. Also, In Garcia et al. [12], the
study results showed the low prediction achievement for active/reflective dimension
because of the lack use of communication tools such as forums, chat, mail, etc. In this
study, it is aimed at giving more detailed data for active/reflective dimension.
In process dimension, individuals prefer to process in two ways actively— through
engagement in physical activity or discussion, or reflectively— through introspection
[9]. Active learners learn well in situations that enable them to do something physical
and reflective learners learn well in situations that provide them with opportunities to
think about the information being presented [10]. Active learners work well in groups;
reflective learners work better by themselves or with at most one other person. Active
learners tend to be experimentalists; reflective learners tend to be theoreticians [9].
The Index of Learning Styles [ILS] was created in 1991 by Richard M. Felder, and
Barbara A. Soloman, then it was installed on the World Wide Web in 1996. ILS is a
44 item questionnaire and there are eleven questions per four dimensions [13]. The
ILS scales are bipolar, with mutually exclusive answers to items, i.e. either (a) or (b).
Because there is an odd number of items on each scale, if items are scored as +1 and –
1, respectively, the total score on a scale from –11 to +11 shows an emerging prefer-
ence for the given modality [14]. If the score on a scale is 1 between 3, it indicates
there is a balanced preference on the two dimensions of that scale. If the score on a
scale is 5-7, it could be said there is a moderate preference for one dimension of the
scale and will learn more easily in a teaching environment which favors that dimen-
sion. If the score on a scale is 9-11, there is a strong preference for one dimension of
the scale and may have real difficulty learning in an environment which does not
support that preference [13].

2.2 Student Modeling

In literature, there are adaptive educational hypermedia systems considering learning

styles such as CS383 [15], MANIC [16], MASPLANG [17], AHA! [18], TANGOW
[19]. Adaptive hypermedia systems build a model of the goals, preferences and
knowledge of each individual user, and use this model throughout the interaction with
the user, in order to adapt to the needs of that user [20]. The student modeling module
performs two main functions: first, initializes the student model when a new student
Diagnosis of Learning Styles Based on Active/Reflective Dimension of Felder 547

logs on the system for the first time, second, updates the student model based on the
student’s interaction with the system [21].
The majority of adaptive systems focusing on learning styles are using a collabora-
tive student modeling approach by asking students to fill out a questionnaire for de-
tecting their learning styles. While collaborative student modeling requires students to
explicitly provide some information about their preferences and needs, an automatic
student modeling approach is based on the concept of looking at what students are
really doing in a course and inferring their preferences and needs from their behavior
and actions in the course [22].
One of the recent researches that automatically diagnosing learning styles, belongs
to Cha et al. [23]. In their research, they detected learning styles with data-driven
approach based on user interface behaviors. System was based on Felder and
Silverman Learning Styles Model and four dimensions of model were investigated.
The interface of the system was designed with Macromedia Flash and interface be-
haviors of 70 participants were monitored. First, the learning styles of learners were
detected then interface was adapted respecting learning styles.
Another research, detecting learning styles with fully automatic student modeling
belongs to Garcia et al. [12]. They used data-driven approach and evaluated Bayesian
networks at detecting the learning styles of a student. The Bayesian network models
different aspects of a student behavior while he/she works with the system. Then, it
infers his/her learning styles according to the modeled behaviors [12]. Forum, chat, e-
mail, reading materials (concrete or abstract), exam revisions, exercises, answer
changes, exam results, etc., were investigated as behavior patterns. Also the system
was based on Felder and Silverman’s Learning Styles Model and three dimensions
were investigated, namely; perception, process and understanding. 50 students were
used to train Bayesian networks and system was tested by 27 students. Using the
evaluation method which was developed by themselves, they obtained a precision of
77% in the perception dimension, 63% in the understanding dimension, and 58% in
the processing dimension.
There is also an alternative way for fully automatic detection which was developed
by Graf namely; literature based approach [22]. Differently from data-driven ap-
proach, literature based approach, is to use the behavior of students in order to get
hints about their learning style preferences and then apply a simple rule-based method
to calculate learning styles from the number of matching hints. This approach is simi-
lar to the method used for calculating learning styles in the ILS questionnaire and has
the advantage to be generic and applicable for data gathered from any course due to
the fact that Felder and Silverman Learning Style Model is developed for learning in
general [22]. In the research, 75 students attended ‘Object Oriented Modeling’ course
for seven weeks and student behaviors analyzed in a Learning Management System,
with the help of literature based approach. Sensory/intuitive, active/reflective and
sequential/global dimensions of Felder and Silverman’s Learning Style Model were
investigated. In this study, first behavior patterns which were frequently used in LMS
were determined and thresholds for each patterns propounded with the help of ex-
periments and reviewing literature. Data from student behaviors compared with
548 Ö. Şimşek et al.

thresholds and calculate matching hints for detecting learning styles. For evaluating
how close the predicted values to ILS, the formula was developed by Garcia et al.
[12] was used, and 79.33% prediction achievement was found for active/reflective,
77.33% for sensory/intuitive and 77.33% for sequential/global dimension.
Atman et al. [1] used literature-based approach for detecting learning styles in a
web-based course. In the study, a web based education system proposed and each
module is labeled for their corresponding learning style dimension. Each module is
labeled such as Visual_Active, Visual_Reflective, Verbal_Active and Ver-
bal_Reflective. This makes analyzing process faster and transportable to other dimen-
sions. To evaluate system effectiveness, learners filled Index of Learning Styles Ques-
tionnaire at the beginning of the course. Scores of predicted learning styles and ILS
scores are compared by using the formula developed by Garcia et al. [12]. 17 college
students’ behaviors are analyzed with literature based approach and results show that
the prediction achievement is 83.15% for active/reflective dimension.

3 Automatic Detection of Learning Styles With Literature-Based


In this study, behaviors of 27 freshmen enrolled in Mathematics course in department

of computer education and instructional technologies were analyzed with literature-
based approach and learning styles of learners automatically detected. The course
which carried out in the spring semester of 2008-2009 was enhanced with web activi-
ties. The course took 14 weeks and during the course “Derivative” topic was chosen.
Derivative is one of the important topics in calculus and students have difficulty in
understand it. Therefore the face to face course was enhanced with an LMS including
resources and activities online. In this course, interactions with forums (10), content
including videos (37) and PDFs(6), questionnaires(5), quizzes(3) and user profile
viewing were investigated.

Fig. 1. Screenshot of the course

Diagnosis of Learning Styles Based on Active/Reflective Dimension of Felder 549

The forum discussions were mostly about the topic “derivatives” and also included
announcements about the course, social dialogs between students. All of the video
contents had screen captured video lectures of derivative subject; without the lec-
turer’s view but the voice with digital pen moves on whiteboard software. The PDF
content included exam solutions, exam dates and exam scores.
Also forum postings and the time spent on forum discussions are considered. Fo-
rum and user profile actions are one of the frequently used tools in this course so data
from these behaviors are monitored and analyzed. There are five questionnaires in this
course such as personal information, views about derivative subject, video lesson
assessment questionnaire, assessment of derivative subject and views about the
teacher in the videos.
To evaluate student performance three short exams were used and time used for
each exam and the total number of performed question for each exam are considered
and determined as behaviors. The tendencies of students to view other student’s pro-
files (classmates) were also investigated with regard to their clicking on a profile and
the spent time on the activity.

3.1 Investigated Patterns

Six features are investigated in this study. Each pattern valued as ‘-’ for reflective and
‘+’ for active dimensions according to characteristics of the pattern. For determining
the patterns, first the literature reviewed also new patterns such as patterns dealing
with user profile is introduced in this study. In Content objects, both video and PDF,
learners are expected to listen or read and learned by thinking through so the value of
these patterns are marked as ‘-’. If one of the students spends more time or visit more
than thresholds, it could be said, there is an evidence for reflective learning. While
Content features have reflective properties, forum posting valued as ‘+’ because ac-
tive learners tend to retain and understand information best by doing, discussing or
applying it or explaining it to others actively. On the other hand reflective learners
learn well in situations that provide them with opportunities to think about the infor-
mation being presented so it is expected to spend more time in forums for reflective
learners. In this study, one of the frequently used features is user profile view, it is
investigated as well. Visiting user profile is expected to increase, while the time spent
on user profile view decrease for active learners. Performed quiz and questionnaire
questions are considered as active characteristics and the time spent on quiz or ques-
tionnaires regarded as reflective. Table 2 shows the investigated patterns, patterns
descriptions and relevant thresholds for each pattern.
After determining patterns for dimensions, relevant thresholds for each pattern
must be defined to analyze learner behaviors systematically. The data that comes from
the learner behavior is compared with these thresholds so it gives hints about learning
styles of the learner. Thresholds are used as evaluation criteria for data coming from
learner behaviors. Most of the thresholds values are determined with the help of re-
viewing literature. Thresholds for content objects are defined assuming that these
objects are required to read in order to understand the topic; a value of 75% and 100%
of the available content objects is recommended [22]. Visiting forums 7 to 14 per
week and posting 1-10 is considered normal values [22], [24]. For user profile, no
recommendations were given. It can be assumed 50% and 75% visit frequency of user
550 Ö. Şimşek et al.

profile view and 25% and 50% for time on user profile view. Thresholds for per-
formed quiz and questionnaire questions are set to 25% and 75% based on the as-
sumptions of Garcia et al. [12].

Table 2. Thresholds for Patterns

Behavior Pattern Pattern Description Thresholds

percentage of visited video content

Content Content(video)_visit (-)(based on the number of available 75% - 100%
Video content objects
percentage of time spent on video
Content(video)_stay (-) content (based on average time) 50% - 75%
percentage of visited PDF content
Content Content(PDF)_visit (-) objects (based on the number of available 75%- 100%
PDF content objects
percentage of time spent on PDF content
Content(PDF)_stay (-) (based on average time)
50% - 75%
Number of postings in the forum (per
Forum_post (+) week)
Fo- Percentage of division of average staying
rum_discussion_stay/vis time to visit frequency in a forum 50% - 75%
it (-) discussion.
Percentage of visited user profiles (based
User Profile
Userview_visit (+) on average value)
50% - 75%
Percentage of division time spent on
Userview_stay (-) user profiles to visit number (based on 25% - 50%
average value)
percentage of performed quiz questions
Quiz Quiz_visit (+) (based on the total amount of available 25% - 75%
percentage of time spent on quiz (based
Quiz_stay (-) on a predefined expected value) 50% - 75%

percentage of performed questionnaire

Questionnaire_visit (+) questions (based on the total amount of 25% - 75%
available questions)
Questionnaire percentage of time spent on questionnaire
(based on a predefined expected value)
Questionnaire_stay (-) 50% - 75%

It is recommended in Garcia et al. [12], the time spent on content objects, quiz and
questionnaires are assumed as 50% and 75%.
When analyzing visiting frequencies in literature-based student modeling ap-
proach, total number of content is considered. The score of visiting frequency of
learner proportioned to total number of the respecting content. This percentage com-
pared with the thresholds. On the other hand, when analyzing the time spent on a
feature, predefined values are used. In features like “Questionnaire” and “Quiz”, is
determined, but some problems came up while determining expected values for user
profile view. For this reason, average time was used as criteria in this study. Up
Diagnosis of Learning Styles Based on Active/Reflective Dimension of Felder 551

and down values are excluded while averaging process to prevent affecting results
negatively. For example, total video staying time of a learner is rated to a average
value, which was obtained of values excluded top and low values, gives us a percent-
age. Then, this percentage is compared with thresholds.
The next step is to compare data coming from learner and thresholds and calculate
hints to detect learning styles of the learner. For example if the percentage of number
of total performed questions to all of the questions is between thresholds, this gives us
there is balanced evidence. If the values are less than threshold, there is a weak pref-
erence for active and strong evidence for reflective learning. If the percentage of the
learner is higher than threshold, it could be said there is a strong evidence for active
and weak evidence for reflective learner.

3.2 Method of Evaluation

Evidences for active learning is marked as “3”, for balanced learning “2” is marked
and finally “1” is pointed for reflective learning. The average of total hints ranged
between 1 to 3 and these results are normalized 1 to 0; 1 for active and 0 for reflective
learning. 0.25 and 0.75 were used as thresholds [22]. Later the results from Index of
learning Styles are mapped into 3 – item scale and compared with the predicted val-
ues to see prediction precision.

Fig. 2. Overview of evaluation process of automatic detection [1]

Comparison of the values is done with the formula developed by Garcia et. al [12].
In this formula, if predicted learning styles and ILS values is equal, formula returns 1.
If one the values is balanced and the other is a preferred learning style of the two
poles of that dimension, function then returns 0.5. Finally if each of two values differs
from each other, the function returns 0. This formula is performed for each student
and divided to the number of learners.
552 Ö. Şimşek et al.

3.3 Study Group

The study group consists of 19 male and 8 female students and 23 of them have their
own computers. The prior knowledge of group members about the subject is 23 low, 4
moderate and 0 advanced. In acquiring information about their prior knowledge, a
pretest was used. In this context, describing qualification of the group members re-
garding to students views may help understanding learning styles of students while
discussing the results. Besides, 9 of the students have not taken any online enhanced

Table 3. Characteristics of the Study Group

Frequency of Frequency of Frequency of

Computer use Internet use
using Internet controlling mail- sending mail in
skills skills
in a week box in a day a day
Strongly insuffi- Strongly insuffi-
Never (0) Never (1) Don't send mail (4)
cient(0) cient (0)
Insufficient (2) Insufficient (2) 0-1 hour (4) Once (9) 1 mail (5)
More than 2 mails
Moderate (5) Moderate (4) 1-3 hours (4) Twice (2)
Sufficient (10) Sufficient (10) 3-5 hours (1) No opinion (9) No opinion (16)
Strongly sufficient Strongly suffi-
Above 5 hours (11)
(5) cient (4)
No opinion (7) No opinion (7) No opinion (7)

The table shows study groups’ computer and Internet use skills and frequencies of
using Internet in a week, controlling mailbox and sending mail in a day. The table
shows us the members of group have sufficient computer use skills, sufficient Internet
use skills, and spend time using Internet, mostly controlling mailbox once a day and
sending very few mail in a day. Table 3 summarizes the characteristics of the study
group. Although there are group members who have no opinion about the variables, it
is assumed that skills and opportunities of students show normal distribution.

4 Results and Discussion

In this study, learner behaviors in an LMS course are analyzed through the help of
literature-based approach. Felder and Silverman Learning Styles Model is chosen and
active/reflective dimension of this model is investigated. 27 college students’ behav-
iors analyzing results show that the precision is 79.63% for process dimension.
Comparison of the ILS questionnaire result and diagnosed learning style is shown in
Table 4. Bold italic parts, illustrate mismatches.
As seen in Table 4, most of the students have balanced (Neutral: NEU) learning
styles so that literature-based approach detected learning styles balanced too. Table 5
shows the compare results of prediction achievement for this study and the other
Diagnosis of Learning Styles Based on Active/Reflective Dimension of Felder 553

Table 4. Comparisons of ILS Results to Predicted Results

No ILS Predicted No ILS Predicted

Result Result Result Result

Table 5. Comparison Results

Modeling Web- Study Participants Precision for

Approach based/LM active/reflective
Data-driven Web-based Garcia et al. 27 62.50%
Literature- LMS Graf [22] 75 79.33%
Literature- Web-based Atman et al. 17 83.13%
based [1]
Literature- LMS This study 27 79.63%

The table indicates that literature based approach has better precision prediction
than the data-driven approach. In this study, the results have close values to other
The number of members in the study group, the bandwidth opportunity, the prior
knowledge about the subject may affect students’ tendencies to stay in a online course
activity. In addition to these limitations; lack of information about the students who
have no opinion for computer (7) and Internet (7) use skills and frequency of using
Internet in a week (7) are limitations of the study as well. Also, the 9 students who
have not taken any online courses should be taken into consideration while predicting
their learning styles with help of their stay time and click frequencies in a web
554 Ö. Şimşek et al.

activity. Because, a student who have never used an online course may spend more
time to understand how a forum discussion works, how to attempt a quiz. So, the
analysis results can assume active learner as balanced or as reflective.
Another limitation is the subject of the courses. The investigated course of this is
study is Mathematics, in Atman et al.[1] learner behaviors of an English course was
analyzed, while in Graf [22], “Object Oriented Modeling”, and in Garcia et al. [12],
“Artificial Intelligence” were chosen subject of the courses. Although there are many
limitation variables for predicting learning style, the results show a high precision for
active/reflective dimension.

5 Conclusion
Diagnosing learning styles of students automatically through analyzing learner behav-
iors in an LMS course is important from pedagogical aspect. The study based on
Felder and Silverman Learning Styles Model and process dimension is investigated
through the help of literature-based approach. In this way, the data of learning styles
of learners can be provided to adaptive systems. Predicted scores and the ILS scores
are compared.
While results demonstrated promising outcomes, also it is not expected to achieve
100% prediction with analyzing staying time and visiting frequency of students. Real
life behaviors and web behaviors can differ in sometime. It is said, there could be dif-
ferences with questionnaire items which are consist of real life behaviors and actions
(i.e. clicking, staying time) and actions in a web environment. To cope with these prob-
lems, it is more suitable to compare learning styles with a psychometric tool which was
developed to evaluate especially web actions of learners. With this type of measure-
ment instrument could be presented, when a learner first time logged in to LMS. By
this way, course administrators and course instructors can monitor learning styles of
learners and this information can help them to plan and organize instructional activities
more effectively. In addition to improving an online learning style scale, it is important
to determine ones learning style while spending time and clicking on a web activity, it
is important to know computer and Internet use skills and prior knowledge. These
levels affects students’ tendency to join Web activities, as well.

1. Atman, N., Inceoğlu, M.M., Aslan, B.G.: Learning Styles Diagnosis Based On Learner
Behaviors in Web Based Learning. In: Gervasi, O., Taniar, D., Murgante, B., Laganà, A.,
Mun, Y., Gavrilova, M.L. (eds.) Computational Science and Its Applications – ICCSA
2009. LNCS, vol. 5593, pp. 900–909. Springer, Heidelberg (2009)
2. Bajraktarevic, N., Hall, W., Fullick, P.: Incorporating Learning Styles in Hypermedia En-
vironment: Empirical Evaluation. In: de Bra, P., Davis, H.C., Kay, J., Schraefel, M. (eds.)
Proceedings of the Workshop on Adaptive Hypermedia and Adaptive Web-Based Sys-
tems, pp. 41–52. Eindhoven University, Nottingham (2003)
3. Graf, S., Lin, T.: Kinshuk: The Relationship Between Learning Styles and Cognitive
Traits. Computers in Human Behavior 24, 122–137 (2008)
4. Kolb, A.Y., Kolb, D.A.: The Kolb Learning Style Inventory - Version 3.1, Technical
Specification. Hay Group, Boston (2005)
Diagnosis of Learning Styles Based on Active/Reflective Dimension of Felder 555

5. Dunn, R., Dunn, K., Price, G.E.: Learning Style Inventory, Lawrence, KS. Price Systems
6. Honey, P., Mumford, A.: The Learning Styles Helper‘s Guide. Peter Honey Publications
Ltd., Maidenhead (2006)
7. Myers, I.B., McCaulley, M.H.: Manual: A Guide to the Development and Use of the
Myers-Briggs Type Indicator. Consulting Psychologists Press, Palo Alto (1998)
8. Kuljis, J., Liu, F.: A Comparison of Learning Style Theories on the Suitability for Elearn-
ing. In: Hamza, M.H. (ed.) Proceedings of the Iasted Conference on Web Technologies,
Applications, and Services, pp. 191–197. ACTA Press (2005)
9. Felder, R., Silverman, L.: Learning and Teaching Styles. Journal of Engineering Educa-
tion 94(1), 674–681 (1988)
10. Felder, R., Henriques, E.R.: Learning and Teaching Styles In Foreign and Second Lan-
guage Education. Foreign Language Annals 28(1), 21–31 (1995)
11. Felder, R.M.: Matters of Style. ASEE Prism 6(4), 18–23 (1996)
12. García, P., Amandi, A., Schiaffino, S., Campo, M.: Evaluating Bayesian Networks Precision
for Detecting Students Learning Styles. Computers &Education 49(3), 794–808 (1995)
13. Felder, R.M., Soloman, B.A.: Index of Learning Styles Questionnaire (October 2009),
14. Zywno, M.S.: A Contribution to Validation Score Meaning for Felder-Soloman‘s Index of
Learning Styles. In: Proceedings of the 2003 American Society for Engineering Education
Annual Conference & Exposition (2003)
15. Carver, C.A., Howard, R.A., Lane, W.D.: Addressing Different Learning Styles through
Course Hypermedia. IEEE Transactions on Education 42(1), 33–38 (1999)
16. Stern, M.K., Steinberg, J., Lee, H.I., Padhye, J., Kurose, J.: Manic: Multimedia Asynchro-
nous Networked Individualized Courseware. In: Proceedings of the World Conference on
Educational Multimedia/Hypermedia and World Conference on Educational Telecommu-
nications (Ed-Media/Ed-Telecom), Calgary, Canada, pp. 1002–1007 (1997)
17. Peña, C.-I., Marzo, J.-L., de la Rosa, J.-L.: Intelligent Agents in a Teaching and Learning
Environment on the Web. In: Petrushin, V., Kommers, P., Kinshuk, G.I. (eds.) Proceedings
of the International Conference on Advanced Learning Technologies, pp. 21–27. IEEE
Learning Technology Task Force, Palmerston North (2002)
18. Stash, N., Cristea, A., de Bra, P.: Authoring of Learning Styles in Adaptive Hypermedia:
Problems and Solutions. In: Proceedings of the International World Wide Web Confer-
ence, pp. 114–123. ACM Press, New York (2004)
19. Carro, R.M., Pulido, E., Rodriguez, P.: Tangow: A Model for Internet-Based Learning. In-
ternational Journal of Continuing Engineering Education and Lifelong Learning 11(1/2),
25–34 (2001)
20. Brusilovsky, P.: Adaptive Hypermedia. User Modeling and User-Adapted Interaction 11,
87–110 (2001)
21. Nwana, H.: User Modeling and user adapted interaction in an intelligent tutoring system.
User Modeling and User-Adapted Interaction 1(1), 1–32 (1991)
22. Graf, S.: Adaptivity in Learning Management Systems Focusing On Learning Styles. Un-
published Ph. D. Thesis (2007)
23. Cha, H.J., Kim, Y.S., Park, S.H., Yoon, T.B., Jung, Y.M., Lee, J.-H.: Learning Style Diag-
nosis Based on User Interface Behavior for the Customization of Learning Interfaces in an
Intelligent Tutoring System. In: Ikeda, M., Ashley, K.D., Chan, W. (eds.) ITS 2006.
LNCS, vol. 4053, pp. 513–524. Springer, Heidelberg (2006)
24. Rovai, A.P., Barnum, K.T.: On-Line Course Effectiveness: An Analysis of Student Inter-
actions and Perceptions of Learning. Journal of Distance Education 18(1), 57–73 (2003)