Available online at [Link].

com

ScienceDirect
Cognitive Systems Research 64 (2020) 134–145
[Link]/locate/cogsys

The theory of learning styles applied to distance learning

Action editor: Tarek Richard Besold
Roberto D. Costa a,b,⇑, Gustavo F. Souza a, Ricardo A.M. Valentim b,c, Thales B. Castro c
a
IFRN – Federal Institute of Education, Science and Technology of Rio Grande do Norte, Natal, RN, Brazil
b
UFRN – Federal University of Rio Grande do Norte – Program in Electrical and Computer Engineering, Natal, RN, Brazil
c
UFRN – Federal University of Rio Grande do Norte – Laboratory of Technological Innovation in Health, Natal, RN, Brazil

Received 13 November 2019; received in revised form 28 April 2020; accepted 16 August 2020
Available online 6 September 2020

Abstract

Distance Education (DE) associated with the use of Virtual Learning Environments (VLE) as interaction tools between the student
and the educator has become a large research niche spread around the world. Techniques to improve learning effectiveness in VLEs seek
to find relation-ships connections between pedagogical advances and educational technological resources available in VLEs. In this con-
text, this work sought to associate the theory of Learning Styles with the behavior of the student of Distance Education, observing their
interaction with the Virtual Learning Environment and trying to associate them with their learning style identified by the CHAEA ques-
tionnaire. For this purpose, the CHAEA questionnaire was applied to a group of distance learning students and the correlation between
their interactions with VLE and their learning styles was verified. The results show that there is no correlation between these elements,
which may show, in fact, the lack of coherence between these theories. Seeking a model for data, a linear regression was applied and the
results were, then, confirmed.
Ó 2020 Elsevier B.V. All rights reserved.

Keywords: Distance learning; Learning styles; Behavior patterns; Distance education

1. Introduction (Brazil, 2005) in its article 80 of the National Education

Guidelines and Framework Law [Lei de Diretrizes e Bases
The growth in the offer of courses mediated by technol- da Educação Nacional] (LDB) of December 20th, 1996,
ogy and offered through Virtual Learning Environments and regulated by decree No. 5622 of December 19th,
(VLEs) has been allowing the significant increase in the 2005, as:
records of the databases of the institutions in which they
“Educational modality in which the didactic-pedagogical
are generated through interactions between agents involved
mediation in teaching and learning processes occurs through
in the teaching and learning process in courses provided as
the use of information and communication media and tech-
Distance Education (DE).
nologies (TIC, in Portuguese), with students and teachers
The term “Distance Education“ or simply the acronym
developing educational activities in different places or times.”
“DE” is defined by the Ministry of Education and Culture
Bozkurt (2019) defines Distance Education as any learn-
⇑ Corresponding author at: IFRN – Federal Institute of Education,
ing activities within formal, informal, and non-formal
Science and Technology of Rio Grande do Norte, Natal, RN, Brazil.
domains that are facilitated by information and
E-mail address: [Link]@[Link] (R.D. Costa). communication technologies to lessen distance, both

[Link]
1389-0417/Ó 2020 Elsevier B.V. All rights reserved.
 R.D. Costa et al. / Cognitive Systems Research 64 (2020) 134–145 135

physically and psychologically, and to increase interactivity this article’s theme; in section four, we present the methods
and communication among learners, learning sources and used in this research; in section five, we bring the results
facilitators. and discussions of this work; and, finally, in section six,
Bacich, Neto, and Trevisani (2015) states that the educa- we bring the final considerations.
tional processes in Distance Education are mediated by
digital resources, in which the students use technological 2. Theoretical framework
resources to establish a new connection with knowledge,
defining their routine and pace of studying. With Distance 2.1. Learning Styles
Learning, this study routine, managed by the student and
mediated by VLEs, will provide the student with tailored The Learning Styles may be considered a series of dis-
studies, thus improving the learning process. tinct behaviors gathered in a single definition that, accord-
In order to better carry out the teaching–learning process on ing to the concept presented by Alonso, Gallego, and
an individual basis in education, researchers as James Keefe Honey (2002), based on studies by Keefe (1979), describes
(Keefe, 1979), David Kolb (Kolb, 1984), Catalina Alonso, the Learning Styles as cognitive, affective, and physiologi-
Domingos Gallego, Peter Honey (Alonso, Gallego, & cal traits used as fairly stable indicators for how students
Honey, 2002), Alan Mumford (Honey & Mumford, 1986), perceive, interact with, and respond to their learning
among others, have been making effort to study each student’s environments.
individual differences and different types of learning, in search According to Kolb (1984), the Learning Styles may be
for behavior patterns and regularities that may indicate the stu- defined as personal ways to process information, feelings,
dents’ Learning Styles and, then, support the development of and behaviors in learning situations. For Messick (2009),
more effective pedagogical actions. on his turn, Learning Styles may be seen as different ways
Learning Styles regard how each person assimilates and a student can learn.
processes the information received, i.e., the way the person According to Alonso, Gallego, and Honey (2002),
learns something. Each individual develops a different way everyone uses different learning styles, although one is usu-
of learning, which is why it is so important that the educa- ally predominant. This leads to the confirmation that all
tor knows his/her students and himself/herself, which humans develop Learning Styles, some more predominant
allows him/her to diversify his/her way of teaching in class- than others, but what it is necessary is to identify them and
rooms and work on different learning styles. use them in learning.
Among the different ways of learning we may highlight
those students who prefer working on an individual basis, 2.2. Adaptive Learning
thinking by themselves and developing their own reflection
abilities, and those who prefer working collectively, The concept of adaptive learning surfaces in the ‘70s
through group activities, promoting interaction and the with the emergence of intelligent tuition, created through
relationship between them. In short, these are only two dif- artificial intelligence (AI) studies (Brusilovsky & Peylo,
ferent forms of behavior indicating different preferences 2003). It relies on the assumption that the intelligent system
and thus, different Learning Styles. created will be able to adapt itself to the student’s learning
The behavior patterns of the students may be deter- process, thus resulting in a better and more effective learn-
mined through their interaction with the educational envi- ing experience.
ronment, either in the traditional classroom or in the VLE. According to Brusilovsky and Peylo (2003), adaptive
In the latter, behavior is determined through their interac- learning is an educational method that uses, as a strategy
tion with resources and activities proposed by the profes- to promote interactions of the teaching and learning pro-
sor. All this access to the virtual environment by the cess consistent with each student’s specific needs, techno-
student is recorded in the VLE’s database. By checking logical means in which the teaching content is
such database, it is possible to keep track of his/her inter- broadcasted and adapted according to the student’s learn-
actions and to infer behavior patterns. ing needs, based on responses to the exercises, assignments,
Once the behavior of the students in the VLE being used and experiences presented.
and the predominant Learning Style of each student are Also, according to Brusilovsky and Peylo (2003), adap-
known, this work’s question is raised, which is to know tive systems are those seeking to be different for different
whether there is a direct connection between the student’s students, considering the information accumulated in indi-
Learning Style and his/her behavior in the VLE when vidual models in which the participants interact.
he/she participates in courses offered as EAD, that is, The main adaptive technologies used by web and hyper-
whether the interactions of the student with the resources media systems are:
and activities in the VLE have any connection with
his/her predominant Learning Style.  Adaptive content selection takes place when the user
In this regard, this article is organized as follows: In sec- searches for information and the system adaptably fil-
tion two, we present the theoretical framework used in this ters and gives priority to the most important items in
study; in section three, we describe some works related to the search.
136 R.D. Costa et al. / Cognitive Systems Research 64 (2020) 134–145

 Adaptive browsing takes place when the system controls For Nunes (1994), Distance Education may be consid-
the URLs while the user browses in order to provide ered a set of tools that provide assistance to a great amount
adaptive support to the browsing process. of students, regardless of their geographic location, with
high quality, as it does not compromise the content or
Finally, adaptive presentation takes place when the the assistance method.
system presents content adaptable to the user’s style when In Moore and Kearsley (2011), Distance Education is
he/she visits a specific page. defined as a teaching and learning methodology facilitated
by technologies, in the process of which the agents involved
2.3. Educational Data Mining are physically or temporally separated.
According to Belloni (1999), Distance Education is inten-
The Portuguese term Mineração de Dados Educacionais sely associated with the terms production and qualification,
came from the English term “Educational Data Mining” as educational institutions that work with Distance Education
(EDM). It is a research area mainly focused on the devel- invest more and more on virtual learning platforms for Dis-
opment of methods for exploring datasets collected in the tance Education due to its flexibility, practicality, low cost
databases of Educational Environments. (Baker, in comparison with classroom education, and technological
Carvalho, & Isotani, 2011). environments quite proper for and that facilitate the teach-
This exploration technique is widely used by researchers ing–learning process and professor training.
in the area of Computing in Education, more specifically, Barros (2003) argues, still, that Distance Education, in
Artificial Intelligence Applied to Education, in order to addition to being a teaching and learning process mediated
investigate scientific questions about the educational area, by technologies, has other characteristics to be considered.
especially in Distance Education. Among them, attention is called to time and space differ-
Currently, there are still other definitions for the term ences and the need to develop habits for self-learning.
“Educational Data Mining (EDM)”. The international Accordingly, Distance Education may be seen as a
society for Educational Data Mining defines the term as teaching methodology that allows offering a quality educa-
“an emerging discipline, concerned with developing meth- tion that overcomes time and space and that is constantly
ods for exploring the unique types of data that come from updating in light of new emerging technologies, at the same
educational settings, and using those methods to better allowing incorporation of these new technologies to the
understand students, and the setting which they learn in” teaching and learning process.
(McGaw, Peterson, and Baker, 2010)
Romero and Ventura (2010) define EDM as “the imple- 2.5. Virtual learning environment
mentation of Data Mining (DM) techniques for specific
types of datasets arising from educational environments According to Silva (2011), a Virtual Learning Environ-
in order to address important educational issues”. ment (VLE) is software with tools for creation, tuition,
The type of data analyzed in EDM is more diverse than and management of activities that are usually found in
that observed in data traditionally used in classic courses.
approaches of Data Mining. At the same time, this diver- These environments are different in several aspects: the
sity in the data represents a potential for implementing language in which they were developed, the tools they offer,
resources essential for supporting improvements in Educa- the functionalities for interaction between the participants
tion (Romero & Ventura, 2010). in the learning process. Some worth-mentioning examples
of environments are: Moodle (2016), AulaNet (2012),
2.4. Distance Education TelEduc (2012), among others.
These VLEs are full of tools with pedagogical resources
In light of the requirements of the market, Distance and activities through which it is possible to provide several
Education arises as a new educational model, in order to functionalities for interaction and interactivity between the
support the propagation of knowledge in a more agile participants, which can be classified as synchronous and
way, facilitating access to learning. asynchronous.
There is a broad spectrum of definitions of the term Synchronous communication tools are those where the
“Distance Education”. In Brazil (2005), Distance Educa- participants are connected to the environment at the same
tion, established in Art. 80 of the “Lei de Diretrizes e Bases time in a real-time interaction. Asynchronous communica-
da Educação Nacional” of December 20th, 1996, and reg- tion tools, on the other hand, are those that do not require
ulated by decree-law No. 5622 of December 19th, 2005, is the participants’ physical presence, and they are able to
characterized as an educational modality that seeks to communicate without the limitations imposed by time.
overcome space and time restrictions through the pedagog-
ical implementation of information and communication 3. Related works
media and technologies and that, without excluding class-
room activities, is organized with peculiar methods, man- The first work highlighted is Dorça (2012), which uses
agement, and evaluation. pedagogical concepts of Learning Styles combined with
 R.D. Costa et al. / Cognitive Systems Research 64 (2020) 134–145 137

Bloom’s Taxonomy to statistically classify students in one Style, may provide a better explanation on the individual
or more combinations of Learning Styles. In order to do differences in learning. In other words, the Learning Style
so, reinforcement machine learning techniques were used. theory by itself cannot measure the best way for the student
The system created by Dorça (2012) uses probabilistic to learn, since it, by itself, has no solid (consistent) grounds
techniques to infer the Learning Style or set of Learning to do so.
Styles in which the student is inserted. In order to do so, The abovementioned statements were the result of a crit-
the student needs to take a performance appraisal (activi- ical analysis about problems with the Learning Style the-
ties) in which he/she is evaluated and classified through ory. Three points were discussed, namely: the lack of a
cognitive concepts in one of the six (6) classes of Bloom’s clear and explanatory structure about Learning Styles;
Taxonomy that indicate the evolution of the student’s Learning Style-measurement issues; and the inconsistency
learning. In the end, the system recommends Learning to connect the Learning Style to the students’ behaviors
Objects (Resources and Activities) of the Virtual Environ- when fulfilling assignments.
ment used, based on the stochastic selection of combina- Regarding the lack of a clean and explanatory structure
tions of identified Learning Styles. addressing the Learning Style, An and Carr (2017) empha-
In parallel, this work uses a system called “Adaptive and size that, for a theory to be well consistent it must properly
Intelligent Systems for Distance Learning (SAIE)”, in explain the common processes and the mechanisms causing
order to identify the students’ Learning Styles. Therefore, the student to have preferences regarding Leaning Style.
it does not follow the Learning Style identification way, However, instead, Learning Style theories tend to define
namely, by identifying the student’s behavior, directly lists of the student’s Learning Style preferences without
determined in the educational environment in which the explaining which of the cognitive, motivational, and
student interacts, accessing the resources and activities pro- personality-related mechanisms support these preferences.
posed by the professor in the course in effect. The individuals simply have these characteristics and it is
Another work worth highlighting is the one carried out not possible explain the process of development defining
by Silva, Ferreira, and Pimentel (2016), called “Adaptive whether or not the individual belongs to a specific Learning
Presentation for Learning Objects based on Learning Style rather than another.
Styles.” Such study presents a proposal for adapting the Among the problems assessed we may highlight the lack
Learning Objects (LO) to be created by the professor- of solid theoretical grounds, lack of researches supporting
author according to the student’s Learning Style. For that, and providing grounds for the Learning Style theory, low
they used Questionnaire “Index of Learning Styles (ILS)”, reliability and validity of the instruments for analysis of
which is based on the Felder and Silverman Learning Style the theory, and inconsistency when establishing a connec-
Model (FSLSM) (Felder & Silverman, 1988), and related tion between the student’s Learning Style and his/her
to the presentation characteristics for Learning Objects. behavior while performing assignments.
They listed the characteristics of the resources forming Yet another problem described is the lack of interaction
part of a LO and classified them in each one of the types between the research subjects of the Learning Styles men-
of Learning Styles according to Felder and Silverman tioned by Sternberg (2001), which is a result of the exis-
(1988) classification, based on the principles of Mayer’s tence of several theories on the subject, and each one of
(2005) “Cognitive Theory of Multimedia Learning”, as them has its own distinct conceptual structure, thus con-
“The principles of this theory help to prevent misuse of verting the studies on the subject into something more
resources in several formats, which may distract and dis- abstract. There is also the fact that Learning Style research-
courage the student using this type of resource and lead ers do not consider other factors, such as the cognition the-
to an unsuccessful learning process.” ory, as a major factor regarding the student’s learning
This was the base for establishing the parameters and process.
attributes required to define a LO model based on Felder According An and Carr (2017) Learning Style theories
and Silverman (1988) Learning Styles. Then, the resources have critical problems when it comes to the form of assess-
that the professor-author will use in Distance Learning ment and definition of the Learning Style of individuals,
courses depend on this classification of LO according to for they “rankings”, thus implying a false dichotomy.
the classification of Learning Styles developed by Felder Another problem is that the detection of Learning Styles
and Silverman (1988). is essentially made through self-report questionnaires that
We also highlight two other works in which Learning may be unsuitable in terms of being conclusive about
Style theories accepted in the academic setting were unsuc- behaviors or skill levels, thus defining a low reliability in
cessful. The first we can mention is the work carried out by terms of effectiveness for the assessment instruments
An and Carr (2017). In their work, entitled “Learning adopted for Learning Styles.
styles theory fails to explain learning and achievement: In a second work brought for analysis, author Kirschner
Recommendations for alternative approaches”, An and (2017, through a work entitled “Stop propagating the
Carr (2017) reviewed the issues identified in the Learning learning styles myth”, states that there is a great difference
Style theory and then presented some alternative between the way someone rather learn a particular subject
approaches that, associated with the concept of Learning and what really leads such individual to effective and
138 R.D. Costa et al. / Cognitive Systems Research 64 (2020) 134–145

efficient learning. He also states that the preference for a Hence, Kirschner (2017) concluded that, if the purpose
particular way of studying is not a learning style, as the is to define instructions that take into account the differ-
so-called Learning Styles are based on types and classify ences between students, it is better to assess the students’
individuals in different groups. cognitive skills rather than their learning styles, because
In the abovementioned work, Kirschner (2017) describes cognitive skills are better indicators for identifying how
the concept of Learning Style in education as an “urban one can learn more efficiently. Additionally, these cognitive
legend”, as, according to the essence of his criticism, the skills must be objectively measured in an ordinal scale
concepts of Learning Style do not show any sustainable instead of by subjective self-reports that are used to desig-
support for the following hypotheses: the student actually nate the person’s type based on one or more arbitrary cri-
has a certain predominant Learning Style; the student teria. Conclusively, Kirschner (2017) states that, at this
acknowledges that this predominant Learning Style is his moment, there is no evidence base suitable for justifying
own, personal style, and; comprehensive learning depends the incorporation of learning style assessment into overall
primarily on ideal instructions according to the student’s educational practice. Thus, limited educational resources,
dominant Learning Style. as in the case of Learning Styles, would be better used
The author also mentions another work by Kirschner & when associated with other educational practices becoming
van Merrienboer (2013), according to which Learning evident in the education scenario. (Coffield et al., 2004;
Styles barely classify learners, as most of the styles defined Pashler, McDaniel, Rohrer, & Bjork, 2009; Rohrer &
are based on types since the student is not assigned to a Pashler, 2012).
style based on a set of scores in different levels, but instead Based on the works presented, we may verify that there
he/she is classified into a specific group that is often non- is still much to study on the theory of Learning Styles, for,
exclusive. This is the first problem with this classification, when it comes to the way a human being learns, there are
as such individuals may not be grouped in specific and dis- many variances involving the process, not only the stu-
tinct groups, as the majority of the differences between dent’s behavior in learning situations, but also physiologic,
individuals, at any level, is gradual and not nominal. physical, and mental conditions, once all these variances
In this regard, Kirschner (2017) states that those proposing influence human behavior in a way that, at a given
the use of learning styles tend to disregard such fact and to use moment, one may belong to a certain group of Learning
arbitrary criteria, such as a median or a mean in a certain scale, Styles and, at another moment, the same person belongs
in order to designate a person to a specific style trying to label a to a different one.
student under a Learning Style. In Coffield, Moseley, Hall, and
Ecclestone (2004), the author managed to identify 30 different 4. Methodology
dichotomic Learning Styles in the literature, proving that this
great amount of dichotomies identified in the literature is some- The approach method adopted was the hypothetical-
thing that causes great confusion in terms of concepts about deductive by Prodanov and Ernani (2013), as it is a prob-
Learning Styles. lem situation that must be investigated through hypotheses
Other problems referred to in the work of Stahl (1999) proposed in order to answer the questions that led to
are the inconsistencies and low reliability on the tests to addressing the topic raised in this scientific work.
assess acknowledgement of the learning styles which are This method begins with a problem that shall be pre-
not being used. The most used method is self-report, in sented clearly to make it easier to create a theory and to
which the student denies or confirms the questions pre- identify other instruments that may assist the researchers
sented through questionnaires. However, Veenman, Prins, during the development of their work. Next, the research-
and Verheij (2003) found that this method is highly doubt- ers observe the behavior of their theory. In fact, this is
ful, as the students are not capable and/or are not willing to the phase of tests, where the purpose of the research will
report sincerely what they do or what they think they do. be evaluated. Eventually, the following phase creates a
Additionally, according to Knoll, Otani, Skeel, and Van hypothesis which shall express.
Horn (2017) self-reporting method is a bad predicting fac- Shall express what was noted in the previous phaseThe
tor to define how people learn more efficiently, as in many question researched in this work was the existence or inex-
cases what individuals prefer is not, by definition, what is istence of a direct connection between a student’s Learning
best for them to learn. In other words, doubt arises as to Style and his/her behavior in Virtual Learning Environ-
whether or not the students actually know what is best ments when interacting with the resources and activities
for them. available in the Educational Environment. Will we be able
Clark (1982) performed some meta-analyses and found to identify the Learning Styles of the students of an online
that students who claimed to prefer a particular way of class through their behavior when interacting with the tools
learning, usually, did not learn better or even had an infe- in the platform?
rior performance when their preferences were applied, and In order to answer this question, the research process
he verified, thus, that a teaching method that matches the was divided in four (4) stages. In the first stage, we applied
way of learning preferred by the student is unproductive the simplified Learning Styles identification questionnaire
or even prejudicial to the learning process. CHAEA 32, which corresponds to the work of the
 R.D. Costa et al. / Cognitive Systems Research 64 (2020) 134–145 139

researchers Vega and Patino (2013), who aimed at adapting 4.2. Identification of Learning Styles
the CHAEA 80 questionnaire proposed by Alonso,
Gallego, and Honey (2002), reducing the number of ques- In order to identify the Learning Style of each student,
tions from 80 to 32, without sacrificing the quality of the we tabulated the answers given in the questionnaire exhib-
data collected. The second stage was the analysis of the ited in Table 2 and gauged them according to the list
data of the students who replied to the questionnaire iden- described in Table 1. At the end, we counted how many
tifying the predominant Learning Styles of each student. In answers to each option (1; 2; 3; and 4) were chosen by each
the third stage, we consulted the VLE’s database in order student, as shown in Table 3.
to collect data about the students’ behavior, using the inter- After we calculated the students’ score for each Learn-
actions with the tools of the VLE as reference. In the fourth ing Style. In order to do so we added the whole score of
stage, we assessed the correlations between the Learning the questions regarding each style, then we obtained the
Styles and the behavior of the students. values of the questions related to the Activist, Theorist,
Reflector, and Pragmatist styles. Such as, for example the
4.1. Application of the questionnaire Activist Score of student 1 (Table 3) in which there was a
final result of 22 points (68.75%).
The list of questions from the original CHAEA 80 ques- Observing Table 3, we can verify that Student 1 shows
tionnaire can be found in Table 1, with their numbers dis- 86.11% of characteristics of the pragmatist Learning Style,
placed on the left side with respect to the simplified 85.71% of the reflector, 78.12% of the theorist, and 68.75%
CHAEA 32 questionnaire, where the numbers are placed of the activist. Thus we can conclude, based on this
on the right side, in which we may exemplify that question method, that Student 1 has a predominance of the pragma-
03 of CHAEA 80 questionnaire corresponds to question 02 tist Learning Style.
of the simplified CHAEA 32 questionnaire – 03|02 and At the conclusion of this stage, we verified that there
question 18 of CHAEA 80 corresponds to question 05 of were students with very similar score in more than one type
CHAEA 32 questionnaire – 18|05 and so on. of Learning Style, thus we used another technique consist-
The simplified CHAEA 32 questionnaire introduced in ing of categorizing the score of the type of Learning Style
Table 2 was initially used in a higher educational institu- into 5 categories. For each type of Learning Style it was
tion with 650 students. After removing inconsistent data, assigned a category (1–5) corresponding to the interval
i.e., the incomplete filling or fail to answer all questions, between the lowest and highest score found in the samples
we reached a total final number of 598 students. of the students for each type of Learning Style. This tech-
According to Miranda and Morais (2008), the question- nique was used by Honey and Mumford (1986) and fol-
naire exhibited in Table 2 was intended to identify the pre- lowed by Alonso and Gallego (2000), in which a
dominant learning style of each participant, forecasting correspondent score is defined for each type of style based
their attitudes in light of the learning process. So that the on the scores of the sample of students who answered the
answers to the questions in the questionnaire would be questionnaire.
more humanized, a likert scale (Almeida and Freire, The categorizations of the ranges in the Learning Styles
2017) was used; unlike closed YES and NO questions, were defined with the percentage of 10% for samples with
the likert scale enable us to measure and know the intervie- the highest scores, classified as “Very High” level, prefer-
wee’s level of agreement with the proposed statements. ably; the next 20% integrate the samples classified as
The likert scale is very useful for providing greater “High” level; the next 40% are classified as “Moderate”
details of the interviewee’s opinion, so that the answers level; the next 20% as “Low” level; and, finally, the last
are not leveled only in the extremities, but can also express 10% are classified as “Very Low” level of predominance,
intermediate positions by offering four options to be con- as shown in Tables 4 and 5.
sidered: 1 – “totally disagree”; 2 – “disagree”; 3 – “agree”; Upon analysis of data of 598 students, we had five cat-
4 – “totally agree”, enabling the interviewees to define their egories of Learning Styles, in which we can highlight the
answer in a more flexible and undoubted way. students who scored in the interval of 26–32 points as being

Table 1
List of both questionnaires: CHAEA 80 and CHAEA 32 – Simplified.
Style Number Of Questions List Of Questions Of Both Questionnaires – (Original = 80|32 = Simplified)
ACTIVIST 8 03|02 27|09 35|13 37|14 43|17 51|19 74|30 77|31
REFLECTOR 7 18|05 19|06 31|10 32|11 63|25 69|27 70|28
THEORIST 8 11|03 17|04 21|07 33|12 54|22 64|26 71|29 78|32
PRAGMATIST 9 01|01 24|08 38|15 40|16 47|18 52|20 53|21 56|23 57|24
140 R.D. Costa et al. / Cognitive Systems Research 64 (2020) 134–145

Table 2
Simplified CHAEA with 32 questions adapted to Portuguese.
No. Items Levels
1 I am famous for saying what is on my mind clearly and shamelessly. 1 2 3 4
2 I often behave without considering the consequences. 1 2 3 4
3 I am adept in self-discipline, at following a certain order, for example, in food diet, studies, and physical exercises, etc. 1 2 3 4
4 I prefer structure to disorder. 1 2 3 4
5 My concern is in carefully interpreting the information available before reaching a conclusion. 1 2 3 4
6 Before taking actions, I carefully analyze the advantages and disadvantages. 1 2 3 4
7 I often try to be consistent with my principles according to criteria and values. 1 2 3 4
8 I prefer realistic and concrete people than idealistic. 1 2 3 4
9 Most of the times I openly express my feelings. 1 2 3 4
10 I am very cautious about drawing conclusions. 1 2 3 4
11 I would rather count with the greatest amount of information I can get, that is, the more, the better. 1 2 3 4
12 I tend to be perfectionist. 1 2 3 4
13 I enjoy living spontaneously and not have to plan everything in advance. 1 2 3 4
14 I feel very uncomfortable with very analytical people. 1 2 3 4
15 I often evaluate people’s ideas for their practical value. 1 2 3 4
16 In meetings I support practical and realistic ideas. 1 2 3 4
17 I have new and spontaneous ideas in discussion groups. 1 2 3 4
18 I often find myself in better and more practical ways of doing things. 1 2 3 4
19 I look for new experiences on a daily basis. 1 2 3 4
20 When I hear about a new idea or approach I immediately try to find concrete applications. 1 2 3 4
21 I believe we should focus on the central idea of things as quickly as possible. 1 2 3 4
22 I always strive for conclusions and clear ideas. 1 2 3 4
23 I get inpatient when people offer me irrelevant or incoherent explanations. 1 2 3 4
24 I always check in advance if everything works as they should. 1 2 3 4
25 I always consider several alternatives before making a decision. 1 2 3 4
26 I frequent try to anticipate the future. 1 2 3 4
27 I often thing deeply about topics and issues. 1 2 3 4
28 Conscious work fills me with satisfaction and pride. 1 2 3 4
29 Considering the developments, I try to discover the principles and theories behind them. 1 2 3 4
30 I am frequently the life and soul of the parties. 1 2 3 4
31 I often get carried away by my intuition. 1 2 3 4
32 If I am part of a working group, I try to get people to follow a plan and a methodology. 1 2 3 4

very likely to have the activist Learning Style as predomi- style, finishing the identification of the students’ Learning
nant, as shown in Table 4. Style.
Based on this classification, we developed a new filter, in Fig. 1 shows the number of students and their respective
which we separated the records in categories that showed in predominant learning styles. There is a higher concentration
all styles only scores 5 (Very Low), 4 (Low), or 3 (Moderate), in the reflector style, i.e., the database is very unbalanced.
or the combination between them. Table 5 shows the situa- This imbalance between the classes may be attributed to sev-
tion of student 5, who got low scores for two out of four eral factor, one of which is the fact that, for these learning
Learning Styles and moderate scores for the other two, styles, the questionnaire has questions that are apparently
and who was, then, classified as having an Undefined Learn- harder for the students to answer negatively, such as, for
ing Style, which corresponds to students who got moderate, example, question 5, “Do I worry about carefully interpret-
low, and very low scores in all Learning Style categories. ing the information available before coming to a conclu-
Notice that, according to Table 5, only the students who sion?”, as well as question 6, “Before I do something, do I
had a Learning Style between categories 1 (Very High) or 2 carefully analyze the advantages and inconveniences?”,
(High) remain in the classification of Learning Styles. and question 10, “Am I careful when it comes to drawing
Another issue was the identification of students who conclusions?”. Denying these questions seems to mean that
belonged to more than one class of Learning Style. As to the student is not very careful in decision making, which is
solve that, the percentages described in Table 3 were an unwanted characteristic for a student.
checked in order to identify the highest score among those In Table 6, we present the number of responses to the
presenting conflicting categorization in Table 5. So, in case questions related to the reflector learning style, which are
of conflicting categorization, we can use the example of 1 – Strongly disagree, 2 – Disagree, 3 – Agree, and
Student 3, who got High scores for both Reflector and 4 – Strongly agree, in the Almeida and Freire (2017)
Pragmatist styles, then the percentages regarding these “Likert” scale. It is possible to note that there is a high
two styles were observed in Table 3, where we can see concentration of responses under the options 3 and 4
50% for reflector style, and 44.44% for the pragmatist; thus, (92.50%), which shows the students’ difficulty in answering
student 3 was defined as having a predominant reflector negatively.
 R.D. Costa et al. / Cognitive Systems Research 64 (2020) 134–145 141

Table 3
Learning Style of each student.
Questions/Students Student 01’s Answer Student 02’s Answer Student 03’s Answer Student 04’s Answer Student 05’s Answer
Question 1 – Pragmatist type 4 1 3 4 4
Question 2 – Activist type 2 4 3 1 2
Question 3 – Theorist type 3 1 2 1 2
Question 4 – Theorist type 4 3 1 4 3
Question 5 – Reflector type 3 3 2 4 3
Question 6 – Reflector type 3 4 2 4 4
Question 7 – Theorist type 3 4 1 4 4
Question 8 – Pragmatist type 4 4 1 4 4
Question 9 – Activist type 3 2 3 4 2
Question 10 – Reflector type 4 3 2 4 3
Question 11 – Reflector type 4 3 2 4 4
Question 12 – Theorist type 3 3 3 4 2
Question 13 – Activist type 3 3 2 4 2
Question 14 – Activist type 4 3 2 4 4
Question 15 – Pragmatist type 3 3 2 4 3
Question 16 – Pragmatist type 4 3 2 4 3
Question 17 – Activist type 3 3 3 4 3
Question 18 – Pragmatist type 3 3 2 4 3
Question 19 – Activist type 3 3 1 4 3
Question 20 – Pragmatist type 3 2 2 4 3
Question 21 – Pragmatist type 4 3 1 0 3
Question 22 – Theorist type 4 3 1 4 3
Question 23 – Pragmatist type 3 3 1 4 4
Question 24 – Pragmatist type 3 2 2 4 4
Question 25 – Reflector type 3 3 1 4 4
Question 26 – Theorist type 2 3 4 4 2
Question 27 – Reflector type 3 3 3 0 3
Question 28 – Reflector type 4 3 2 4 4
Question 29 – Theorist type 3 3 3 4 3
Question 30 – Activist type 2 3 4 4 1
Question 31 – Activist type 2 3 3 4 2
Question 32 – Theorist type 3 3 3 4 3
Total Answers 32 32 32 30 32
Activist Score 22 24 21 29 19
Reflector Score 24 22 14 24 25
Theorist Score 25 23 18 29 22
Pragmatist Score 31 24 16 32 31
Activist Percentage 68.75% 75.0% 65.62% 90.62% 59.38%
Reflector Percentage 85.71% 78.57% 50.0% 85.71% 89.29%
Theorist Percentage 78.12% 71.88% 56.25% 90.62% 68.75%
Pragmatist Percentage 86.11% 66.67% 44.44% 88.89% 86.11%

Table 4
Categorization of Learning Styles – 5 categories.
Learning Styles Activist Theorist Reflector Pragmatist
Maximum Score 32 32 33 36
Minimum Score 8 8 7 9
1. Very High – 10% 26. . ... . .0.32 31. . ... . .32 27. . ... . .33 32. . ... . .36
2. High – 20% 23. . ... . .0.25 27. . ... . .30 24. . ... . .26 29. . ... . .31
3. Moderate – 40% 18. . ... . .0.22 22. . ... . .0.26 20. . ... . .0.23 23. . ... . .28
4. Low – 20% 15. . ... . .0.17 19. . ... . .0.21 16. . ... . .19 21. . ... . .22
5. Very Low – 10% 08. . ... . .14 08. . ... . .18 07. . ... . .15 09. . ... . .20

4.3. Identifying the behaviors access and interaction history of the students in the courses
in Distance Education and the VLE was stored. The pur-
This stage aimed at identifying the behaviors of the stu- pose was to collect information of their behavior in the
dents who were part of the selected classes that participated platform.
in the survey. To obtain the behavior variables, we carried out several
In order to do so, we had to access some Tables in the queries in the Structured Query Language (SQL) code in
Virtual Learning Environment database, where the entire the database of the VLE. Fig. 2 shows an example of one
142 R.D. Costa et al. / Cognitive Systems Research 64 (2020) 134–145

Table 5
An example of the classification of students.
Student Activist Score Theorist Score Reflector Score Pragmatist Score Student’s Learning Style
Student 1 Moderate High Very High High Reflector
Student 2 Moderate Very High High High Theorist
Student 3 Moderate Moderate High High Reflector-Pragmatist
Student 4 Moderate High High Moderate Theorist-Reflector
Student 5 Moderate Moderate Low Low Undefined

Fig. 1. Distribution of the Learning styles of the 598 students

(Number  Styles).

of the queries carried out to check the student’s response Fig. 2. Example of the SQL code – Time to Submit an Assignment.
time when submitting an assignment.
Every query carried out showed a value corresponding
to the behavior variable of an action of each student According to the results presented in Fig. 3, we noticed
according to his/her interaction with the VLE, such as, that, since all variables show low linear correlation with the
for example, the number of messages sent by the student Learning Styles, these variables are not good indicators of
during the course, which is shown in Table 7. learning styles. The strongest correlation was found in the
Once we knew how each student behaves or how they variable “number of accesses to page (V1)”, value of which
reacted to the professors’ encouragement through the was 0.073.
resources and activities available in the VLE, we were Fig. 4 shows a point chart listing the variables “number
able to discover the individual behavior pattern of all of accesses to page” and “number of accesses to chat” for
students. each of the five categories of learning styles. It is possible to
The variables identified: the number of accesses per note that there is no evidence of correlation among the
course, messages sent, and accesses to the resources: folder, groups, i.e., there are no apparent patterns.
page, file, and URL; we also identified the number of Despite these evidences, a linear regression between
accesses to task, questionnaire, and forum activities. these variables and learning styles was made. The regres-
sion model containing all behavior variables showed only
4.4. Analysis of correlation two variables (amount of access to books and amount of
access to files) as significant, at the level of 5%, however
Fig. 3 shows the value of Pearson’s linear correlation the determination coefficient was very low (R2 = 0.033).
coefficient (Morettin & Bussab, 2017) between the behavior This result again shows that there is no correlation between
variables and the Learning Styles of the students. behavior variables and learning styles.

Table 6
An example of the classification of students.
Questions No answer 1 – Strongly disagree 2 – Disagree 3 – Agree 4 – Strongly agree
05 2 3 30 275 288
06 1 7 44 272 274
10 0 7 56 363 172
11 1 4 25 283 285
25 2 6 47 399 144
27 2 9 62 298 227
28 2 8 11 251 326
Total 10 44 275 2141 1716
Percentage 0.24 1.05 6.57 51.15 40.99
 R.D. Costa et al. / Cognitive Systems Research 64 (2020) 134–145 143

Table 7
Variables of the students’ behavior in the interaction with the VLE.
Behavior variables Students
Student 1 Student 2 Student 3 Student 4 Student 5 Student 6 Student 7 Student 8
No. of accesses_page 0 4 10 0 6 0 7 0
no. of accesses_folder 2 13 7 0 12 13 8 2
no. of accesses_file 34 87 10 17 1 146 80 24
no. of accesses_url 9 2 0 6 6 1 3 0
no. of accesses_course 220 392 133 35 148 372 417 97
No. of messages sent 35 33 30 8 22 13 11 32
no. of accesses_assignment 26 82 6 11 41 57 85 22
no. of accesses_forum 13 18 42 6 72 1 60 37
no. of accesses_questionnaire 101 429 5 2 2 1171 658 2
Learning_style Undefined Activist Undefined Activist Reflector Reflector Reflector Pragmatist

define exactly, through the questionnaire, the extent to

which he/she learns a given topic or concept. The way we
learn is complex and still not clear to us, and this may lead
the student to give answers other than what he/she actually
learns.
Behavior descriptors: In this work, we used 14 behavior
Fig. 3. Correlation between the behavior variables and the Learning variables. All of them refer to the number of accesses made
Styles. by the students to resources offered by the professor in the
courses taken during the semester. It is possible that these
variables are not actually able to describe the student’s
behavior so that we can identify them as descriptors of
Learning style. In this context, the term behavior follows
the definition given by Psychology that, according to
Cabral (1996), behavior is the set of procedures or reac-
tions that someone has before the environment that sur-
rounds him/her under certain circumstances. Thus, it is
possible to note that behavior is a very extensive concept,
and it may not be completely characterized by a set of vari-
ables. Skinner (2003) states that:
Fig. 4. Dispersion diagram between two behavior variables.

“Behavior is a difficult subject, not because it is inaccessi-

5. Analysis of results
ble, but because it is extremely complex. As it is a process,
rather than a thing, it cannot be easily immobilized for
There are many possibilities to explain the lack of corre-
observation.”
lation between behavior variables and learning styles. Here
we point out three of them, the ones we deem more rele- Thus, the fact that behavior is so complex to character-
vant: answers to the questionnaire, behavior descriptors, ize contributes to a specific set of variables not being able
and definition of learning styles. to “capture” all dimensions associated with behavior.
Answers to the questionnaire: students applied to ques- Definition of learning styles: In this work, we used the
tionnaire CHAEA 32 through the VLE, in which the stu- concept defined by Alonso et al. (2012), which Classifies
dent attends to the Distance Education course he/she Learning Styles as cognitive, affective, and physiological
contracted, i.e., the student is alone in front of the com- traits used as fairly stable indicators for how students per-
puter and free to answer as he/she will, not necessarily ceive, interact with, and respond to their learning environ-
assuming the responsibility of being true when answering. ments. The authors consider the existence of four learning
This suspicion takes place due to how questionnaire styles: activist, reflector, pragmatist, and theorist. How-
CHAEA presents some questions we deem to generate a ever, labeling the way someone learns in four categories
high degree of resistance by the students to answering neg- does not seem to be enough to capture the complexity
atively, like the questions presented in Table 4. and variety of ways of thinking and acquiring knowledge
In addition, the questionnaire aims at identifying the (learning).
student’s learning style based on his/her answers, i.e., a According to the author Kirschner (2017), there is a
self-report, but the student may not be mature enough to huge difference between the way someone prefers to learn
144 R.D. Costa et al. / Cognitive Systems Research 64 (2020) 134–145

and what actually leads to an effective learning, as prefer- References

ence for a study method is not a learning style. Another
point to be highlighted is that the Learning Styles theory Almeida, L. S., & Freire, T. (2017). Metodologia da Investigação em
classifies people nominally, where the student tends to have Psicologia da Educação (5rd ed.). Braga-PT: Psiquilı́brios.
Alonso, C. M., & Gallego, D. J. (2000). Aprendizaje y Ordenador. Madrid-
a preference for one or other style, rather than gradually, ES: Dykinson.
where the student tends to have preference for one style Alonso, C. M., Gallego, D. J., & Honey, P. (2012). Los Estilos de
and, during the educational process, they develop prefer- Aprendizaje: Procedimientos de Diagnóstico y Mejora (8rd ed.). Bilbao-
ences for other learning styles. ES: Editorial Mensajero.
In this perspective, we still cannot affirm that students An, D., & Carr, M. (2017). Learning styles theory fails to explain learning
and achievement: Recommendations for alternative approaches. Per-
who behave in a certain way in the educational environ- sonality and Individual Differences, 116, 410–416. [Link]
ment belong to a specific style, because these behaviors 10.1016/[Link].2017.04.050.
listed as typical may be redefined if the student simply suf- Aulanet, 1998. EduWeb-Aulanet. Disponı́vel em [Link]
fers from an external influence such as an infirmity or a trip [Link]/public/papers/[Link]. Acesso em 16 de Abr
2018.
during the educational process.
Bacich, L., Neto, A. T., & Trevisani, F. de M., 2015. Ensino Hı́brido:
Another point to be highlighted is that identifying the Personalização e Tecnologia na Educação. Penso Editora, Porto
Learning Style through the students’ self-report, obtained Alegre-RS.
by filling in the questionnaire used to get to know their pre- Baker, R. S. J., Carvalho, A. M. J., & Isotani, S. (2011). Mineração de
dominant style, was labeled as a bad predictor to identify Dados Educacionais: Oportunidades para o Brasil. Revista Brasileira
the best way for students to learn, because not always their de Informática na Educação, 19, Número 2.
Barros, D. M. V. (2003). Educação a distância e o universo do trabalho.
preferences are necessarily what is best for their learning. Bauru-SP: EDUSC.
Knoll et al. (2017) conclude that the learning styles are Belloni, M. L. (1999). Educação a distância. Campinas-SP: Autores
associated with subjective aspects of learning, but not with Associados.
objective aspects of learning. Bozkurt, A. (2019). From Distance Education to Open and Distance
Learning: A Holistic Evaluation of History, Definitions, and Theories.
In S. Sisman-Ugur & G. Kurubacak (Eds.), Handbook of Research on
6. Conclusions Learning in the Age of Transhumanism (pp. 252–273). Hershey, PA:
IGI Global.
The Learning Styles are elements that identify and Brazil, 2005. Decreto 5.622 de19 de dezembro de 2005 regulamenta o Art.
80 da Lei no 9.394, de 20 de dezembro de 1996. Disponı́vel em:
define the way an individual learns. Under the educational
[Link]
scope, such information could be used to facilitate stu- Acesso em: 16 de Abr 2018.
dent’s learning, changing education planning, especially Brusilovsky, P., & Peylo, C. (2003). Adaptive and intelligent web-based
within the current context of Distance Education, which educational systems. International Journal of Artificial Intelligence in
is developed in Virtual Learning Environments. The first Education, 13, 159–172.
Cabral, A. (1996). Dicionário técnico de psicologia (14rd ed.). São Paulo-
condition required, and extremely necessary, is to find a
SP: Editora Cultrix.
satisfactory correlation between the Learning Styles and Coffield, F., Moseley, D., Hall, E., & Ecclestone, K. (2004). Learning
the behavior patterns during learning. styles and pedagogy in post-16 learning: A systematic and critical review.
When searching for this correlation based on the group London-UK: Learning & Skills Research Centre.
of students that voluntarily replied questionnaire Clark, R. E. (1982). Antagonism between achievement and enjoyment in
ATI studies. Educational Psychologist, 17(2), 92–101. [Link]
CHEA32, we observed that there is no correlation between
10.1080/00461528209529247.
these styles and the behavior variables used in the study. Dorça, F. A. (2012). Uma Abordagem Estocástica Baseada em Aprendiza-
This lack of correlation suggests that some assumptions gem por Reforço para Modelagem Automática e Dinâmica de Estilos de
may not be supported, as the veracity of the information Aprendizagem de Estudantes em Sistemas Adaptativos e Inteligentes
provided by the students in the questionnaire or the model para Educação a Distância Tese (Doutorado). Universidade Federal de
Uberlândia-MG.
of learning styles being used itself. Another possibility that
Felder, R. M., & Silverman, L. K. (1988). Learning and teaching styles in
may be researched is the set of behavior variables that may engineering education. Journal of Engineering Education, 78, 674–681.
be identified to effectively define the behavior. As a future Honey, P., & Mumford, A. (1986). The Manual of Learning Styles.
work, it is suggested to insert variables of duration and per- London-UK: McGraw Hill.
formance of students in the activities in the VLE and check Keefe, J. W. (1979). Learning style: An overview, In NASSP’s Student
learning styles: Diagnosing and prescribing programs. VA: National
if there is a greater correlation with these variables.
Association of Secondary School Principals, Reston-US (pp. 1–17).
VA: National Association of Secondary School Principals, Reston-US.
Declaration of Competing Interest Kirschner, P. A., & Van Merrieenboer, J. J. G. (2013). Do learners really
know best? Urban legends in education. Educational Psychologist, 48,
169–183. [Link]
The authors declare that they have no known competing Kirschner, P. A. (2017). Stop propagating the learning styles myth.
financial interests or personal relationships that could have Computers & Education, 106(5), 166–171. [Link]
appeared to influence the work reported in this paper. [Link].2016.12.006.
 R.D. Costa et al. / Cognitive Systems Research 64 (2020) 134–145 145

Knoll, A. R., Otani, H., Skeel, R. L., & Van Horn, K. R. (2017). Learning Rohrer, D., & Pashler, H. (2012). Learning styles: Where’s the evidence?
style, judgements of learning, and learning of verbal and visual Medical Education, 46, 630–635. [Link]
information. British Journal of Psychology., 108, 544–563. [Link] 2923.2012.04273.x.
org/10.1111/bjop.12214. Romero, C., & Ventura, S. (2010). Educational data mining: A review of
Kolb, D. A. (1984). Self-scoring inventory and interpretation booklet the state of the art. IEEE Transactions On Systems, Man, and
(Revised Edition). Boston-US: Hay McBer. Cybernetics, Part C, 40, 601–618. [Link]
Mayer, R. et al. (2005). The Cambridge handbook of multimedia learning. TSMCC.2010.2053532.
New York-US: Cambridge University Press. Silva, Z., Ferreira, L., & Pimentel, A. (2016). Modelo de apresentação
Messick, S. (2009). The nature of cognitive styles: Problems and promise adaptativa de objeto de aprendizagem baseada em estilos de apren-
in educational practice. Educational Psychologist, 19, 59–74. https:// dizagem. Brazilian Symposium on Computers in Education (XXVII
[Link]/10.1080/00461528409529283. Simpósio Brasileiro de Informática na Educação -SBIE) (volume 27,
Miranda, L., & Morais, C. (2008). Estilos de Aprendizagem: O pp. pag 717). . [Link]
questionário CHAEA adaptado para lı́ngua portuguesa. Revista de Silva, R. S. (2011). Moodle para autores e tutores. 2ª Ed. rev. e ampl.,
Estilos de Aprendizagem, no 1(01), 66–87. Novatec Editora, São Paulo-SP.
Moodle (2016). Modular object-oriented dynamic learning environment. Skinner, B. F. (2003). Ciência e comportamento humano. São Paulo-SP:
Disponı́vel em [Link] Acesso em: Abr. de 2018. Martins Editora.
Moore, M. G., & Kearsley, G. (2011). Distance education: A systems view Stahl, S. A. (1999). Different strokes for different folks? A critique of
of online learning (3rd ed.). Belmont-US: Wadsworth Cengage learning styles. American Educator, 23(3), 27–31.
Learning. Sternberg, R. J. (2001). Epilogue: Anothermysterious affair at styles. In R.
Morettin, P. A., & Bussab, W. O. (2017). Estatı́stica básica (9rd ed.). São J. Sternberg & L. Zhang (Eds.), Perspectives on thinking, learning, and
Paulo-SP: Editora Saraiva. cognitive styles (pp. 249–252). NJ: Erlbaum, Mahwah-US. [Link]
Nunes, I. B. (1994). Noções de educação a distância. Revista de Educação org/10.4324/9781410605986.
a Distância, 7–25. Teleduc, 2018. TelEduc-Ambiente e-learning. Disponı́vel em: [Link]
Pashler, H., McDaniel, M., Rohrer, D., & Bjork, R. (2009). Learning [Link], acesso em: 15 de dez. 2018.
styles: Concepts and evidence. Psychological Science in the Public Veenman, M. V. J., Prins, F. J., & Verheij, J. (2003). Learning styles: Self-
Interest, 9, 105–119. [Link] reports versus thinking-aloud measures. British Journal of Educational
Peterson, P. L., Baker, E., & McGaw, B. (2010). International encyclopedia Psychology, 73, 357–372. [Link]
of education. Amsterdam: Elsevier Ltd. 000709903322275885.
Prodanov, C. C., & Ernani, C. F. (2013). Metodologia do trabalho Vega, M. C., & Patino, M. C. (2013). CHAEA 32 simplificada: Propuesta
cientı́fico: Métodos e técnicas da pesquisa e do trabalho acadêmico – 2. basad em Análisis Multivariantes Trabajo de Fin de Master.. España:
ed. – Editora Feevale, Novo Hamburgo-RS. Departamento de Estadı́stica. Universidad de Salamanca.