Accepted Manuscript

An Application of Adaptive Games-Based Learning based on Learning Style to Teach

SQL

Mario Soflano, Thomas M. Connolly, Thomas Hainey

PII: S0360-1315(15)00093-7
DOI: 10.1016/j.compedu.2015.03.015
Reference: CAE 2823

To appear in: Computers & Education

Received Date: 16 July 2014

Revised Date: 21 March 2015
Accepted Date: 24 March 2015

Please cite this article as: Soflano M., Connolly T.M. & Hainey T., An Application of Adaptive Games-
Based Learning based on Learning Style to Teach SQL, Computers & Education (2015), doi: 10.1016/
j.compedu.2015.03.015.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Abstract: The fact that each student has a different way of learning and processing information has long been
recognised by educationalists. In the classroom, the benefits derived from delivering learning content in ways that
match the student’s learning style have also been identified. As new modes of delivery of learning content such
as computer-assisted learning systems (e.g. eLearning) have become increasingly popular, research into these
has also identified the benefits of tailoring learning content to learning styles. However, in games-based learning
(GBL), the adaptation based on learning style to enhance the educational experience has not been well
researched. For the purpose of this research, a game with three game modes has been developed: 1) non-
adaptivity mode; 2) a mode that customises the game according to the student’s learning style identified by using
a learning style questionnaire; and 3) a mode that has an in-game adaptive system that dynamically and
continuously adapts its content according to the student’s interactions in the game.

PT
This paper discusses the term adaptivity in a GBL context and presents the results of an experimental study
investigating the differences in learning effectiveness of the different game modes compared to a paper-based
learning. The study was performed with 120 Higher Education students learning the database language SQL
(Structured Query Language). The results show that the game developed, regardless of mode, produced better

RI
learning outcomes than those who learned from a textbook while adaptive GBL was better in terms of allowing
learners to complete the tasks faster than the other two game versions.

Keywords: Games-Based Learning, Adaptivity, Learning Style, SQL, NeverWinter Nights, RPG

SC
1 Introduction
The fact that each student has a different way of learning and processing information has long been

U
recognised by educationalists (Kolb, 1984; Felder & Brent, 2005). In the classroom, the benefits
derived from delivering learning content in ways that match the student’s learning style have also
AN
been identified (Smith & Renzulli, 1984; Price, 2004). As new modes of delivery of learning content
such as computer-assisted learning systems (e.g. eLearning) have become increasingly popular,
research into these has also identified the benefits of tailoring learning content to learning styles
(Miller, 2005).
M

However, as Connolly and Stansfield (2006) have suggested, eLearning simply replicates the
traditional education system (classroom style) and may be overly focussed on method of delivery, i.e.
delivering materials over the web rather than on actual teaching and learning, and indeed motivating
and engaging the students in the learning process. In contrast, games, particularly video games,
D

appear to be able to engage people over extensive periods of time and also motivate them to re-play
the game repeatedly until they have mastered it (Kirriemuir & McFarlane, 2004). Therefore, some
TE

educationalists (for example, Prensky, 2006) have considered games to be a potential platform in
supporting student learning and have turned their attention to what is now called games-based
learning (GBL).
While many GBL applications have been developed in the last two decades, there remains a lack of
EP

empirical evidence to support the use of GBL for learning purposes (Connolly et al., 2012). Given that
there appears to be genuine advantages for learning outcomes to be derived from the adaptation of
teaching materials to learning styles in the classroom and remotely through eLearning, it may also be
possible that GBL applications that are adapted to the individual’s learning style would improve
C

learning outcomes.
Kirriemuir and McFarlane (2004) have suggested that games, unlike classroom learning or eLearning,
AC

provide a different type of engagement as they demand constant interaction and generate a ‘flow’ that
could assist in engaging students. It is therefore possible for students to adopt different leaning styles
in GBL than they adopt in other learning settings.
In the next section, adaptivity is discussed followed by a short review of learning style theories and a
review of previous empirical work in adaptive GBL particularly those based on learning styles. Section
3 presents the game that has been created for this research study and Section 4 discusses the
research methodology used to evaluate the game. Section 5 discusses the evaluation of the adaptive
GBL application. The evaluation includes analysis of the difference in learning effectiveness and
completion times between experimental groups. Section 6 provides conclusions from the study and
discusses future research directions.
 ACCEPTED MANUSCRIPT
2 Previous Research

2.1 Definition of Adaptivity and Adaptability

In computing, there are two types of adaptation process: adaptability and adaptivity (Jameson, 2003).
Adaptability refers to the ability of the user to ‘adapt’ to the system by explicitly customising the
system according to their preferences (Bontcheva, 2002). On the other hand, adaptivity, which is
usually used in the context of a user-adaptive system, refers to the ability of the system to identify the
user’s preferences or characteristics and customise the system accordingly; that is, the user implicitly
influences the adaptation process (Mulwa et al., 2010).

PT
In modern computer systems, adaptability is usually implemented by providing options that allow the
user to customise the system according to their preferences. For example, in eLearning a student can
choose a font size and font style associated with the learning materials. Conversely, adaptivity does
not explicitly require input from the users and it is usually hidden from them. The users simply see the

RI
result of the customisation process provided by the system.
Although both types can exist in a computer system, each type has differences in terms of its usage.
Nowadays, adaptability through user customisation is widely adopted and exists in many computer

SC
systems. Whilst it is certainly useful in some circumstances, adaptability requires direct manipulation
from the user that can result in an increase in the user’s cognitive load, especially if there are many
options the user needs to choose from (Oppermann, Rashev & Kinshuk, 1997). Cognitive load theory
(Sweller, 1994) suggests that learning may be enhanced if instructional materials are designed to
reduce the levels of extraneous cognitive load.

U
Conversely, adaptivity can capture the interactions between the user and the computer system and
AN
the adaptivity analyses the historical interactions before making an automatic adjustment. Adaptivity is
considered to be less intrusive compared to adaptability as it does not require the user to make any
changes and, as a result, the interaction between the user and the system can be maintained. This is
useful especially in a system that has a considerable amount of elements that the user would need to
manipulate. In addition, adaptivity can be used for a behavioural pattern analysis of the user’s
M

interaction with the system. Such an analysis may be used to create different user experiences. The
disadvantage of adaptivity is that the user does not have direct control in customising the system.
According to Ismailovic et al. (2012), adaptivity involves players monitoring, characterisation and
D

assessment and adaptive intervention. The purpose of adaptivity is to support, encourage and
motivate the player by considering the player’s strength and weaknesses (Budiharto et al., 2013;
TE

Ming-Chun & Shyan-Ming, 2013; Gasimov et al., 2010). Adaptivity has also been used to make
games more challenging, engaging and fun as well as to increase replayability (Li et al., 2014; Ibanez
and Delgado-Mata, 2011; Sweetser and Wyeth, 2005). However, as noted in the comprehensive
survey by Karpinskyj et al (2014), there is still a lack of empirical evidence related to the effect of
adaptivity in player experience. Doukianou et al (2014) has also noted the limited empirical evidence
EP

in games-based learning particularly to assess human attributes and behaviours.

For this particular research, adaptivity is used to refer to the system’s ability to automatically
customise certain elements of the system based on a series of the student’s interactions with the
C

system.
According to Charles et al. (2005) and Melis and Monthienvichienchai (2004) adaptation can be
AC

incorporated into games through:

1) A player’s character: all actions undertaken by the character have implications; for example, if the
character is wounded, the movement of the character is slower.
2) Non-player character (NPC): the player can access this feedback by ‘talking’ to the NPC. Besides
providing feedback, the conversation itself may be used to alter the story based on the selections
the player has made in the conversation.
3) The game environment: adaptation through the game environment can be categorised into:
customisation, contextualisation and personalisation. In the context of GBL, customisation is
related to the functionalities of the learning system; for example, when the student reaches a
particular level, a new control will appear. Contextualisation means that the content that is going
to be delivered is adaptable according to the student’s performance, learning history or response
to certain missions. Personalisation relates more to the student’s preference, for example, the font
size.
 ACCEPTED MANUSCRIPT
4) Feedback/scaffolding: can be used to inform the student about their status in the game and to
help the student to achieve the game and educational objectives (Jackson, Krajcik & Soloway,
1998).
In this paper, adaptivity was implemented through the conversation system where the adaptive
system analyses the preferences of the players and adapts the type of presentation on the next
conversation between the player and NPC. The adaptivity mechanism of this research is discussed
further in Section 3.3

2.2 Learning Styles

While many learning styles have been proposed, one of the learning styles that has been used in

PT
many computer-assisted learning applications with its reliability and validity being confirmed is the
Felder-Silverman learning style. The Felder-Silverman learning style model (1988) was developed by
Richard Felder and Linda Silverman. The model consists of four main elements as shown in Table 2
(Sharda, 2007; Graf, Viola, Leo & Kinshuk, 2007) with two of the main elements replicating those

RI
selected by Kolb and MBTI (perception and processing). As shown in Table 1, each main element
consists of two sub-elements that represent the element a person may have, for example a person
may prefer to use visual or verbal materials (Graf et al., 2007).

SC
Table 1: Elements of Felder-Silverman’s learning style
Main Elements Sub-Elements A Sub-Elements B
Perception Sensing: prefer to use existing ways Intuitive: prefer to try new ways,

U
than exploring new ways, prefer to prefer to learn abstract material, not
learn concrete materials, careful with careful with details
AN
details
Input Visual: prefer to learn materials Verbal: prefer to learn materials by
presented in pictures, chart or text or audio, having difficulty with
diagrams visual style
M

Processing Active: learning by doing, social Reflective: learning by thinking it

oriented through, impersonal oriented
D

Organisation Sequential: building up from specific Global: (learning from a general

knowledge until a more general knowledge into a more specific
knowledge is attained knowledge
TE

The Perception element consists of two sub-elements: sensing and intuitive. Sensing students prefer
to learn facts and concrete learning materials, whilst intuitive students prefer abstract materials with
EP

general principles rather than concrete materials. The Input element relates to the presentation of
materials preferred by students. Visual students prefer to learn using pictures, diagrams, graphics,
etc., whilst verbal students prefer to listen or read the learning materials. Verbal students tend to be
able to remember written or spoken text or a sentence. For example, in an advertisement, some
C

people remember the slogan, whilst others remember what the advertisement looks like.
In the Processing element, active students prefer to learn by doing and enjoy discussion with other
AC

people, whilst reflective students prefer to reflect on the learning materials alone (learning by thinking
it through). Active students can be categorised as extroverted and reflective students can be
categorised as introverted. In the Organisation of materials, sequential students learn in a step-by-
step/linear sequence, following a logical sequence, whilst global students absorb learning materials
randomly and when they have learned enough, they will understand the whole picture. Global
students can also be categorised as holistic students and sequential students as serial students whilst
sensing - sequential students have a tendency to be convergent and intuitive and global students
have a tendency to be divergent students.
Amongst the theories, the Felder-Silverman learning style has been identified as having a greater
number of benefits compared to other learning style theories. The Felder-Silverman learning style
model has been widely used in eLearning and GBL and its reliability and validity have been tested
(Kuljis & Liu, 2005; Litzinger et al., 2005; Cook, 2005; Livesay et al., 2002; Van Zwanenberg,
Wilkinson & Anderson, 2000; Zywno, 2003). When compared to other learning style models, the
 ACCEPTED MANUSCRIPT
Felder-Silverman model represents elements from most models, which indicates the generalisability
of the model (Limongelli et al., 2009).
However, not all elements of learning style theory can be adopted in GBL. According to Becker (2007)
one of the main characteristic of games, regardless of the genre, is their interactivity, which requires
players to actively interact with the game, and indicates that the ‘reflection’ element of the Felder-
Silverman learning style may not be relevant to GBL. This statement is also supported by Boyle,
Connolly and Hainey (2011) who indicated that games provided an active, experiential, situated and
problem-based learning environment.

2.3 Adaptive GBL Applications - Empirical Evidence

PT
A systematic literature search was conducted on electronic databases including: Association for
Computing Machinery (ACM), Cambridge Journals Online, Institute of Electrical and Electronic
Engineering (IEEE), Index to theses, IGI Global, Ingenta Connect, Science Direct, Springer Link,
Wiley Online Library, Extended Academic ASAP, Simulation and Gaming and Emerald. The following

RI
search terms were used:
(adaptive or adaptivity or personalisation or personalization or “learning style”) and (“serious
games” OR “games-based learning”)

SC
The search returned 1334 papers, however, after detailed analysis only 27 papers were identified as
being relevant to adaptive GBL applications with empirical evidence and these are summarised in
Table 2 while only 2 papers are related to adaptive GBL based on learning style.

U
Table 2: Existing Empirical Evidence in Adaptive GBL
Authors Area Adaptivity
AN
Felicia & Pitt (2007, 2008, Learning style and personality traits (The
Maths
2009) Big-5 model).
Conati & Zhou (2002), Cognitive theory of emotions (joy, distress,
Maths
Conati & Zhao (2004) pride, shame, admiration and reproach).
M

Peirce, Conlan & Wade Number of interactions between the

Physics of optics
(2008) student and certain elements of the game.
Learning style based on Dunn and Dunn’s
Lynch et al. (2008) Food safety
D

theory.
Hwang et al. (2012) Plantation Felder-Silverman (sequential - global).
Lee & Ko (2011) Logic programming Successful rate of the code execution.
TE

Changes occurring in the environment

Yongyuth et al. (2010) Agriculture
caused by the student’s actions.
Number of mistakes the student has
Demmel et al. (2011) Languages
made.
EP

Smeddinck, Siegel & Exercise for Parkinson’s Speed and Accuracy

Herrlich (2013) Disease Patients

Rodriguez-Cerezo et al. Introductory Computer Number of mistakes the student has

C

(2013) Language made.

For young children with Cognitive and affective state
AC

Bernardini, Porayska- Autism Spectrum

Pomsta & Smith (2013)
Sampayo-Vargas et al.
(2013)
Kerfoot & Baker (2012)
Hertzog et al. (2014)
Conditions (ASC)
practice
Spanish cognates The performance of the student
Number of mistakes the student has
Urology
made.
Water Management / Changes occurring in the environment
Irigation caused by the player’s actions.

Lester et al. (2014) Microbiology Paths / actions chosen by the player

 ACCEPTED MANUSCRIPT
Wilson et al (2009); Siegler
Numbers The performance of the student
& Ramani (2011)

Praet & Desoete (2014) Arithmetic The performance of the student

Kujala et al. (2010) Reading (Finnish) The performance of the student

Berns et al. (2013) Language (German) The performance of the student

PT
Tavassolian et al. (2014) Time-balancing The performance of the student

Exergame, Chinese

RI
Zhang et al (2012) The performance of the student
culture, Olympics

Tsai et al (2012) Language (Chinese) Facial expression

SC
Tong & Chignell (2014) Cognitive Assessment The performance of the student

U
Amongst the papers found, only two research papers contain empirical evidence related to learning
styles: Lynch et al. (2008) and Hwang et al. (2012). Lynch et al. (2008) developed Ootle-U to teach
AN
aspects of food safety. The game is adapted to the student’s learning style, based on Dunn and
Dunn’s theory: motivation/learning enjoyment (M); persistence towards task completion (P); sense of
responsibility (R); structure (S); alone versus peer (AP); auditory (A); visual (V); tactile (T); and
kinesthetic (K). The authors suggested that M, P, R and S were related to emotional preferences
M

while A, V, T and K could be categorised as perceptual preferences. Their experiment involved 217
participants to measure the learning effectiveness of the game. The effectiveness was measured by
comparing the pre-test and post-test scores. The experiment also assessed the impact of learning
D

style on learning effectiveness and the correlation between the learning achievement and the game.
The results showed that the participants’ knowledge in food safety was improved although the results
indicated that the improvement was not significant because the students had prior knowledge of the
TE

subject matter. With respect to learning style, the experiment indicated that a greater number of
participants preferred to mix their learning methods rather than consistently use a single learning
method. The authors suggested that perhaps the participants, who were considered as digital natives,
interacted with various different media in their everyday life.
EP

Hwang et al. (2012) developed an adaptive GBL application based on learning style to teach about
plantation. Their evaluation of the game involved 46 participants and they investigated whether
learning in a game that could match the student’s learning style was better than a game without any
consideration of the student’s learning style in terms of learning effectiveness and motivation. The
C

experiment also evaluated the easiness and usefulness of the game. A pre-test/post-test
experimental design was used with an equal number of participants in the control group (who played
AC

the game without any adaptivity) and the experimental group (who played the game with adaptivity
based on the learning style identified before the game). The learning style adopted in this research
was the ‘sequential/global’ element of the Felder-Silverman learning style, identified using their
learning style questionnaire, while a test sheet was used in the pre-test and post-test to measure the
knowledge of the participant on the subject matter. The results showed that the adaptive version had
a significantly higher learning effectiveness.
From the empirical evidence generated from the adaptive GBL that has been discussed above,
adaptivity has the potential for improving learning effectiveness in GBL. Moreover, it has also shown:
• The current trend in adaptive GBL research is the investigation of the effect of theories used as
the basis for the adaptivity. Such theories may be personality theories as researched by Felicia
and Pitt (2009) or learning styles as used by Lynch et al. (2008) and Hwang et al. (2012).
 ACCEPTED MANUSCRIPT
• The majority of research in adaptive GBL uses a pre-test/post-test experimental design. This is
mostly used when measuring learning effectiveness as reflected in the differences between pre-
test and post-test either within or between control and experimental groups. The analysis
methods adopted in the studies are mainly quantitative.
• The adaptive systems mostly use log files or a database to collect information about interactions
between the games and the students.

3 Learning-style-based Adaptive GBL

For the purpose of this research, an adaptive GBL application based on learning styles was

PT
developed. The game was intended to teach the basics of the database programming language SQL
(Structured Query Language) while the learning style adopted in this game was the Felder-Silverman
learning style model, particularly the presentation elements (picture-text). The Felder-Silverman
learning style model has been widely used in eLearning and its reliability and validity have been

RI
tested. When compared to other learning style models, the Felder-Silverman model represents
elements from most models, which indicates the generalisability of the model. The selected genre of
the game developed for this research was role-playing games (RPG) using the NeverWinter Nights 2
engine.

SC
In this study, there were three modes of the same game designed and developed:
• A non-adaptive mode of the game. This mode treats all students the same and takes no account
of the student’s learning style.

U
• An out-of-game adaptive mode. The characteristics of a student are identified by means of the
Felder-Silverman learning style questionnaire completed in advance of playing the game and the
AN
gameplay is then customised according to the student’s learning style.
• An in-game adaptive mode. In this mode, the student’s characteristics are identified during the
gameplay. As it is possible for the student to change learning style in the course of the game, the
M

game has an adaptive system that can automatically customise the game in real-time according
to the student’s current learning style. The student’s current learning style is identified by
analysing historical interactions between the student and the game.
D

The difference between the modes concerns the nature of the adaptive approach adopted while the
rest of the game elements such as storyline, game environment, controls and game interface, are
identical. The adaptive approach itself was implemented through the presentation of the learning
TE

materials presented by the conversation system of the game.

3.1 SQL
In this research, a module on Structured Query Language (SQL) was selected as the basis of
EP

research. SQL was developed as the standard language to allow an individual to define and
manipulate data stored in a relational database. SQL is used in many modern Database Management
Systems (DBMSs) available today and the language itself is descriptive and has a similar structure to
English. Therefore, SQL has been identified as one of the essential programming skills and it is
C

normally taught as part of computing/IT courses.

AC

SQL has two major components (Connolly & Begg, 2014):

• Data Definition Language (DDL): related to the structural design of the database, such as:
CREATE TABLE, ALTER TABLE and DROP TABLE.
• Data Manipulation Language (DML): related to data manipulation such as retrieval, update, delete
and insert.
DDL is normally used during the database design process while DML is used to manage the data flow
from and into the database including insert, update and delete data and more importantly the learner
also can specify which data to be retrieved according to his/her needs. The learning outcomes for the
game to be developed in this research are as follows:
• How to write an SQL command.
• How to retrieve data from the database using the SELECT statement.
 ACCEPTED MANUSCRIPT
• How to build SQL statements that:
- use the WHERE clause to retrieve rows that satisfy a simple condition;
- use the WHERE clause to retrieve rows that satisfy multiple conditions;
- sort query results using ORDER BY;
- use the aggregate functions of SQL;
- use of the DISTINCT keyword to eliminate duplicates rows from the result.
This will allow students to construct some basic forms of SQL retrieval syntax (SQL SELECT
statements), such as:

PT
SELECT * FROM user_tbl
SELECT * FROM user_tbl WHERE city = ‘Glasgow’

RI
SELECT firstname, lastname FROM user_tbl WHERE age > 27 AND city = ‘Glasgow’
SELECT firstname, lastname FROM user_tbl ORDER BY age DESC
SELECT COUNT(*) FROM user_tbl

SC
SELECT MIN (age) FROM user_tbl
SELECT MAX (age) FROM user_tbl

U
SELECT AVG (age) FROM user_tbl
SELECT DISTINCT (city) FROM user_tbl
AN
In a database and SQL course, a case study is often used that requires learners to analyse a
particular task and then create an SQL statement according to the needs identified. In this research,
this case study will be implemented through the game story. The game element is what is expected to
motivate students to learn SQL and also to give them the opportunity to test their knowledge through
M

the game missions. The process of completing the game involves a learner in appropriately applying
the material delivered to create the correct SQL query if they are to make progress through the game.
D

3.2 Gameplay
TE

The game was developed to achieve the above learning outcomes in a 3D Role-Playing-Game (RPG)
created through the Aurora toolset from NeverWinter Nights 2. In this game, learners control an
avatar that they can personalise at the beginning of the game. The avatar can be controlled by point-
and-click using a mouse and also by using keyboard shortcuts. The same method also controls the
EP

game’s 3D camera.

3.3 Game Story

C

When designing an educational game it is necessary to consider how to teach the learning materials
without losing the ‘fun’ part of the game so learners can remain motivated to learn the materials. One
AC

way of maintaining motivation is to blend the materials and the game story in such a way that the
learning materials are part of the challenge of the game (Marsh, Li, Klopfer, Chuang, Osterweil &
Haas, 2011). It was decided to base the game on the story of a criminal investigator, a role to be
adopted by the learner, who undertakes quests for information to arrest criminal masterminds, using
warrants the learner gains based on the evidence and information collected. During the course of the
game, the learner is also required to collect data and evidence by talking to Non-Player Characters
(NPCs) before a warrant to arrest the criminals can be issued. Figure 1, 2 and 3 show the sequence
of conversation on the first mission.
 ACCEPTED MANUSCRIPT

PT
RI
SC
Figure 1: Conversation Sequence - First

U
AN
M
D
TE
EP

Figure 2: Conversation Sequence - Second

C
AC
 ACCEPTED MANUSCRIPT

PT
RI
SC
Figure 3: Conversation Sequence - Third

The data and evidence collected are automatically stored in a fictional database while the

U
conversation system allows the player to choose the learning style, in this case the type of
presentation, preferred. In the non-adaptive mode of the game, the game is passively recording the
AN
type of presentation chosen by the player. In the in-game adaptive mode, the game's adaptive system
automatically sets the default type of presentation according to the type selected the most by the
player. On each conversation, the player is also provided with an option to change the type of
presentation and such changes are recorded and analysed by the adaptive system for the next
conversation.
M

The game is designed to require the learner to retrieve the required data and evidence from the
fictional database by using appropriate SELECT statements in order to get the warrant. The game
structures learning by increasing the level of complexity of the SQL queries that the learner must
D

construct as the game progresses. When the player makes a mistake when constructing the
statements, the player can retry it and also will be given a chance to choose their preferred learning
TE

style by selecting their preferred presentation. As the process of real-life crime investigation parallels
to some extent the process of investigation reflected in the game story, this element may help the
learner to understand how the learning content is applied in real life.
The game contains three main missions and a number of side missions. For each mission, the learner
EP

learns different forms of SELECT statements specified above. To help the learner complete the tasks,
there are instructions specific to each task and the learner can choose how these instructions are
presented, either as text or picture/diagram, to reflect their learning style. By completing these tasks,
the learner can learn how to create and use SELECT statements by entering the appropriate
C

statement into a textbox to solve a given task. The learner is given six chances to create the SQL.
Once the learner has successfully created SQL queries in response to the task requirements, the
AC

warrant to arrest the key criminals will be issued, which will allow the learner to arrest these criminals.
If the learner fails to create the correct SQL after the six chances, the learner will still get the warrant
so the learner is not blocked from progressing. When the key criminal on a mission is arrested, the
learner will receive a reward and positive feedback from NPCs, such as the chief inspector who
usually gives the missions, and then the game story will proceed to the next mission.
As for the detail of the story, at the beginning of the game the learner can create and personalise his
or her own avatar. Once the avatar is created, the learner will appear in the police headquarters as a
new recruit. In the first mission, the learner is required to investigate the case of a missing classified
document, which later connects to further missions investigating cases of corruption, potential
rebellion and an assassination plot. For the purposes of gathering data and evidence, students will
have to travel to particular areas of the game world to talk to NPCs and sometimes they will encounter
enemies. The game also contains fighting scenes that require a simple point-and-click using the
mouse. The storyline is designed in such a way that the learner can easily find out what to do by
using the ‘Journal’, which provides details of the missions including the relevant locations for the
 ACCEPTED MANUSCRIPT
missions, and where to go by using a map that pinpoints important characters and locations.
Provision of the ‘Journal’ and the map are to help the learner focus on solving the missions, which are
related to the learning objectives, rather than be distracted by the mechanics of the game with related
problems such as getting lost when trying to find the relevant places or wasting time in finding NPCs.
Figure 4 and 5 provide some screenshots from the game.

PT
RI
U SC
AN
Figure 4: Screenshot of an area – Central Plain
M
D
TE
C EP
AC

Figure 5: Screenshot of the map, inventory and biodata notebook

4 Research Questions and Methodology

4.1 Research Questions

Research Question: How do GBL and adaptive GBL compare to traditional (paper-based) learning?
 ACCEPTED MANUSCRIPT
The main purpose of this research question is to evaluate whether the use of a game with adaptivity
based on a student’s in-game learning style can improve that student’s learning compared to the use
of one without any use of adaptivity, or a game with adaptivity based on a student’s learning style as
identified from a learning style questionnaire and non-game/text-based learning. Thus, this
experiment comprises four independent groups:
• A paper-based group that require the participants in the group to read the SQL materials from a
textbook (control group);
• A ‘non-adaptive game’ group that will play the game in non-adaptivity mode;
• An ‘out-of-game adaptive game’ group that will have their learning style identified by a learning-
style questionnaire before playing the game and each participant in the group will play the game

PT
that is customised to his or her previously identified learning style;
• An ‘in-game adaptivity game’ group that will play the game in the in-game-adaptivity mode. This
mode allows the game to automatically identify and adapt to a student’s learning style based on
his/her interaction with the learning content.

RI
The learning outcomes of each group were then analysed and compared.

SC
4.2 Methodology
In this experiment, 120 university students with no knowledge of SQL voluntarily participated in the
study. Only the participants who scored 0 in the pre-test SQL test (Appendix 2) were selected in this
study. The participants were 83 undergraduate students (69.2%) and 37 postgraduate students

U
(30.8%) with 43 students from the University of the West of Scotland (35.8%), 46 students from
Heriot-Watt University (38.3%), 19 students from Napier University (15.8%), 9 students from
AN
University of Edinburgh (7.5%), 2 students from Glasgow University (1.7%) and 1 student from
Strathclyde University (0.8%). The participants were also selected from various programmes. There
were 78 participants (65%) from computing programmes, 1 student from accounting (0.8%), 3
students from business (2.5%), 10 students from engineering (8.3%), 1 student from finance (0.8%), 3
students from languages (2.5%), 1 student from law (0.8%), 7 students from life science (5.8%), 9
M

students from management (7.5%), 2 students from mathematics (1.8%), 1 student from medicine
(0.8%), 1 student from nursing (0.8%), 1 student from psychology (0.8%) and 2 students from social
science programmes (1.8%). For the purposes of analysis, the participants were categorised into two
D

groups, computing and non-computing students, because the number of participants in each of the
non-computing programmes were too low to be analysed individually.
TE

The participants were recruited from the universities by visiting campuses, classes, libraries and
student accommodation. Particularly at UWS, student mailing lists and posters were used. The
participants were recruited voluntarily without any incentive offered. When approached, the
participants were asked if they would like to participate in the experiment and the process of the
experiment was described. The participants were told the experiment was to investigate how the
EP

subject (SQL) is learnt and that this experiment would involve tests to measure the learning
effectiveness. There was no time limit for the experiment, however, the approximate duration of this
experiment was explained at this point. The treatment on each group described on section 5.1 was
also explained. Once the participants agreed to voluntarily participate in the experiment, the
C

participants were then randomly allocated to a group. For those who were allocated to the game
groups, the experiment was conducted by using the researcher's laptops while the paper-based group
AC

participants were given a textbook. The textbook used was Database Systems: A Practical Approach
to Design, Implementation and Management by Connolly and Begg (2004) and the highlighted pages
were the basic SELECT, conditional SELECT and aggregate functions (page 117 to 133) although the
participants had the freedom to read any sections they wished. The textbook contains a case study
including SQL examples and solutions. Besides presented examples and solutions relevant to the
case study implemented in the game story, the game also presented the same SQL theory as the
textbook.
Before the experiment, the participants were asked to answer the Felder-Silverman learning style
questionnaire (Appendix 1) to identify their individual learning style. During the game, the interaction
between the participants and the game were recorded automatically to identify the participant’s
learning style during the game. At the end of the game, the participants were asked to answer an SQL
post-test (Appendix 3). The Felder-Silverman questionnaire, SQL pre-test and post-test were all
paper-based. The SQL pre-test used different case study than the post-test and the case study in
game is different than the textbook. The game recorded interactions between the participants and the
 ACCEPTED MANUSCRIPT
learning materials and the records are to be used to analyse in more details on the learning style in
GBL.
There were three experimental groups that represented each mode of the game and one control
group in which the students would learn SQL by using a textbook. Each participant was allocated
randomly to one of the four groups. In this section, the groups will be referred to as: the paper-based
group, the non-adaptive game group, the out-of-game learning style (LS) group and the in-game
adaptive group. In the paper-based group, the participants were given a textbook about SQL while
participants on the other groups learnt SQL through a game with a mode of adaptivity described
previously.
The participants were required to answer the SQL tests and those aspects of the game relating to the

PT
questionnaire. The SQL tests were conducted twice: in both the pre-test and post-test. In the pre-test,
the SQL test was used to identify the participants’ knowledge of SQL. Those who had no knowledge
of SQL were selected for the experiment to ensure the participants had the same initial knowledge of
SQL. By doing so, the authors tried to reduce the possibilities of recruiting participants who may have

RI
learnt SQL before as they may be easily recall what they have learned before considering the forms
of SQL are similar and the materials in this experiment are normally taught together. The SQL post-
test was used to test the students’ knowledge of SQL after learning SQL through the game or after

SC
studying the prescribed textbook. The SQL pre-test had similar questions to the post-test with 11
questions in total. The maximum points achievable for the SQL pre-test and post-test was 16. The
analysis of the learning effectiveness to be conducted was a non-parametric test according to the
Levene statistic = 8.280 (p<0.001), while the analysis for the completion time was also a non-
parametric test according to the Levene statistic = 3.879 (p<0.024). A Levene statistic with p<0.05

U
means the homogeneity cannot be assumed. In this paper, the confidence value used was 95%
(p<0.05).
AN
5 Results
This section describes the results of the experiment starting from differences in learning effectiveness
M

followed by the completion time analysis between the experimental groups. Before discussing the
results, it is noted that the tests were marked by the researcher and by an independent marker, an
SQL expert. A reliability of the marking of the tests was conducted using the Cronbach Alpha test
(Kinnear & Gray, 2008). The test showed that the Cronbach Alpha value of the reliability in marking
D

was 0.990, which means the post-test scores marked by the researcher were consistent with the
scores marked by the independent rater.
TE

5.1 Learning Effectiveness

The learning effectiveness between the paper-based group and the games group is compared to test
the following hypothesis:
EP

• H0A: There is no difference in learning effectiveness between the paper-based group and the
game groups.
• H1A: There is a difference in learning effectiveness between the paper-based group and the game
C

groups.
The learning effectiveness of the paper-based SQL learning for 30 university students had a mean of
AC

2.6 (SD=3.8065). The mean score of the learning effectiveness from the non-adaptivity game group,
which consisted of 30 students, was 12.517 (SD = 4.690). The learning effectiveness of the out-of-
game LS group, which consisted of 30 students, had a mean score of 12.35 (SD = 4.065). The
learning effectiveness of the in-game adaptive group, which also consisted of 30 students, had a
mean score of 14.042 (SD = 2.354).
When compared to any mode of the game, a significant difference in learning effectiveness between
the groups was found. By using Mann-Whitney U tests, the learning effectiveness for the non-
adaptivity game group was found to be significantly higher when compared to the paper-based
learning group (Z = -5.903, p<0.001). The learning effectiveness of the paper-based learning group
was also found to be significantly lower to the out-of-game LS game group (Z = -5.995, p<0.001) and
the in-game adaptive group (Z = -6.495, p<0.001) and so H0A is rejected.
The analysis of the differences in learning effectiveness between the game groups has three
hypotheses to investigate how differences between the non-adaptivity game group and the other
 ACCEPTED MANUSCRIPT
game groups and between the out-of game learning LS group and the in-game adaptive group. The
first hypothesis is:
• H0B: There is no difference in learning effectiveness between the non-adaptivity game group and
the out-of-game LS group.
• H1B: There is a difference in learning effectiveness between the non-adaptivity game group and
the out-of-game LS group.
A Mann-Whitney U test showed there was no significant difference between the non-adaptivity group
and the out-of-game LS group (Z = -0.872, p<0.389) and so H0B is accepted.
The second hypothesis is:

PT
• H0C: There is no difference in learning effectiveness between the non-adaptivity game group and
the in-game adaptive group.
•

RI
H1C: There is a difference in learning effectiveness between the non-adaptivity game group and
the in-game adaptive group.
A similar result was found when comparing the learning effectiveness of the non-adaptivity group to

SC
the game with the in-game adaptivity (Z = -0.378, p<0.711) and so H0C is accepted.
The third hypothesis is:
• H0D: There is no difference in learning effectiveness between the out-of game learning LS group
and the in-game adaptive group.
•

U
H1D: There is a difference in learning effectiveness between the out-of game learning LS group
AN
and the in-game adaptive group.
Similar to the result of the first and second hypotheses, a Mann-Whitney U test showed there was no
significant difference between the out-of game learning LS group and the in-game adaptive group (Z =
-1.546, p<0.124) and so H0D is accepted.
M

These tests showed that the learning effectiveness for the modes of the game were significantly
higher than those for paper-based learning. A Kruskal-Wallis test was then performed using to
compare all these groups (paper-based, game without adaptivity, game with out-of-game LS and
D

2
game with in-game adaptivity) and was found to be significant beyond the 0.01 level (χ = 58.153,
p<0.01) indicating there is a significant difference between the means of learning effectiveness
between groups. A summary of the results is presented in Table 3.
TE

Table 3: Comparison of Learning Effectiveness

Groups Learning Effectiveness between

EP

Groups
Paper-based Group and Non-adaptive Group Z = -5.903, p<0.001 (Non-adaptive
Group > Paper-based Group)
C

Paper-based Group and Out-of-game LS Group Z = -5.995, p<0.001 (Out-of-game LS

AC

Group >Paper-based Group)

Paper-based Group and In-game Adaptive Group Z = -6.495, p<0.001 (In-game
Adaptive Group>Paper-based Group)
Non-adaptive Group and Out-of-game LS Group Z = -0.872, p<0.389

Non-adaptive Group and In-game Adaptive Group Z = -0.378, p<0.711

Out-of-game LS Group and In-game Adaptive Group Z = -1.546, p<0.124

χ = 58.153, p<0.01
2
All Groups
 ACCEPTED MANUSCRIPT
5.2 Differences in Learning Effectiveness between Learning Styles
This section investigates the differences in the learning effectiveness for those students with different
learning styles (picture/text) as reflected by the SQL post-test. In the paper-based group, there were 8
participants (27%) with a text-based learning style and the mean for learning effectiveness was 1.5
(SD = 2.712), while 22 participants (73%) had a picture-based learning style and the mean for their
learning effectiveness was 3.0 (SD = 4.114). In the non-adaptivity group, there were 13 participants
(43%) with a text-based learning style and the mean of the learning effectiveness was 13.481 (SD =
3.209), while 17 participants (57%) had a picture-based learning style and the mean of their learning
effectiveness was 11.779 (SD = 5.551). In the out-of-game LS group, there were 13 participants
(43%) with a text-based learning style and the mean for learning effectiveness was 10.788 (SD =

PT
4.377), while 17 participants (57%) had a picture-based learning style and the mean of their learning
effectiveness was 13.544 (SD = 3.477). In the in-game adaptive group, there were 14 participants
(47%) with a text-based learning style and the mean for the learning effectiveness was 13.411 (SD =
3.001), while 16 participants (53%) had a picture-based learning style and the mean of their learning

RI
effectiveness was 14.594 (SD = 1.486).
When comparing the learning effectiveness of participants with different learning styles in all groups, a
Mann-Whitney U test indicated no difference in learning effectiveness between those participants who

SC
had a text-based learning style and those who had a picture-based learning style (Z = -0.011,
p<0.993). The same result was also found when comparing the learning effectiveness of participants
with different learning styles across the game groups (Z = -1.477, p<0.142). The result suggests the
participants with text-based learning style did not have a significantly higher learning effectiveness

U
than the participants with picture-based learning style and vice versa.

5.3 Differences in Learning Effectiveness between Programmes

AN
This section investigates the differences in the learning effectiveness reflected by the SQL post-test
for those on different programmes (computing/non-computing). In the paper-based group, there were
4 participants (13%) from non-computing programmes and the mean for learning effectiveness was
M

4.875 (SD = 1.702), while there were 26 participants (87%) from computing programmes and the
mean for their learning effectiveness was 2.250 (SD = 3.938). In the non-adaptivity group, there were
6 participants (20%) from non-computing programmes and the mean for learning effectiveness was
11.583 (SD = 4.104), while there were 24 participants (80%) from computing programmes and the
D

mean for their learning effectiveness was 12.750 (SD = 4.877). In the out-of-game LS group, there
were 19 participants (63%) from non-computing programmes and the mean for learning effectiveness
TE

was 11.382 (SD = 4.61), while there were 11 participants (37%) from computing programmes and the
mean for their learning effectiveness was 14.203 (SD = 2.192). In the in-game adaptivity group, there
were 13 participants (43%) from non-computing programmes and the mean for learning effectiveness
was 13.019 (SD = 2.803), while there were 17 participants (53%) from computing programmes and
the mean for their learning effectiveness was 14.824 (SD = 1.627).
EP

When comparing the learning effectiveness of participants from different programmes in all groups, a
Mann-Whitney U test indicated no difference in learning effectiveness between participants from non-
computing programmes and participants from computing programmes (Z = -0.274, p<0.787).
C

However, a Mann-Whitney U test indicated the learning effectiveness of participants from computing
programmes across the game groups was significantly higher than the learning effectiveness for
those on non-computing programmes (Z = -2.797, p<0.006) potentially because participants from
AC

computing programme are more familiar with game environment which allows them to effectively
adapt to the learning process throughout the game.
 ACCEPTED MANUSCRIPT
5.4 Differences in Learning Effectiveness between Education Levels
This section investigates the differences in learning effectiveness, as reflected by the SQL post-test,
for the different education levels (undergraduate/postgraduate). In the paper-based group, there were
26 undergraduate participants (87%) and the mean for learning effectiveness was 1.942 (SD =
3.5618), while there were 4 postgraduate participants (13%) and the mean for their learning
effectiveness was 6.875 (SD = 2.4958). In the non-adaptivity group, there were 19 undergraduate
participants (63%) and the mean of learning effectiveness was 10.934 (SD = 5.281) while there were
11 postgraduate participants (37%) and the mean for their learning effectiveness was 15.250 (SD =
0.783). In the out-of-game LS group, there were 18 undergraduate participants (60%) and the mean
for the learning effectiveness was 13.222 (SD = 3.659), while there were 12 postgraduate participants

PT
(40%) and the mean for their learning effectiveness was 11.042 (SD = 4.445). In the in-game
adaptivity group, there were 20 undergraduate participants (67%) and the mean of the learning
effectiveness was 13.925 (SD = 2.755) while there were 10 postgraduate participants (33%) and the
mean of their learning effectiveness was 14.275 (SD = 1.315).

RI
When comparing the learning effectiveness of participants with different education levels in all groups,
a Mann-Whitney U test indicated that the learning effectiveness of postgraduate participants was
significantly higher than undergraduate participants (Z = -2.138, p<0.033). However, a Mann-Whitney

SC
U test indicated there was no significant difference in learning effectiveness between undergraduate
and postgraduate participants across the game groups (Z = -0.068, p<0.949). This finding might
suggest that the learning from a textbook required a more experienced learner than is required by a
game.

U
5.5 Differences in Learning Effectiveness between Genders
AN
This section investigates the differences in learning effectiveness reflected by the SQL post-test
according to gender (male/female). In the paper-based group, there were 5 female participants (17%)
and the mean for learning effectiveness was 4.2 (SD = 4.0094), while there were 25 male participants
(83%) and the mean for their learning effectiveness was 2.280 (SD = 3.7669). In the non-adaptivity
M

group, there were 7 female participants (23%) and the mean for learning effectiveness was 12.857
(SD = 3.660) while there were 23 male participants (77%) and the mean for their learning
effectiveness was 12.413 (SD = 5.029). In the out-of-game LS group, there were 8 female participants
(27%) and the mean for learning effectiveness was 11.094 (SD = 4.698), while there were 22 male
D

participants (73%) and the mean for their learning effectiveness was 12.807 (SD = 3.827). In the in-
game adaptive group, there were 8 females participants (27%) and the mean for learning
TE

effectiveness was 12.875 (SD = 1.512), while there were 22 male participants (73%) and the mean for
their learning effectiveness was 14.466 (SD = 2.486).
When comparing the learning effectiveness of participants based on gender, a Mann-Whitney U test
indicated no difference in learning effectiveness between male and female participants (Z = -0.528,
EP

p<0.600). However, a Mann-Whitney U test indicated the learning effectiveness of male participants
across the game groups was significantly higher than the learning effectiveness for female
participants (Z = -2.115, p<0.035), suggesting that for this type of game (RPG) and/or subject matter
may not be the most appropriate for female learners.
C

5.6 Completion Times

AC

In this section, the completion time of the game groups are analysed. The completion time of paper-
based learning group was not included because the completion time of the group was not directly
comparable to completion time of the game groups. The completion time is the time required for the
player to complete the game. The completion times were recorded automatically during the gameplay.
This analysis includes a comparison with the learning style and other characteristics associated with
the player (gender, education level, programmes of study), as well as its correlation with the learning
effectiveness. The mean completion time when learning through the game without adaptivity was
94.67 minutes (SD = 12.893), while the mean completion time for the out-of-game LS group was
90.67 minutes (SD = 14.284) and the mean completion time for the in-game adaptive group was
84.91 minutes (SD = 8.174).
The analysis of the differences in completion times between the game groups has three hypotheses
to investigate how differences between the non-adaptivity game group and the other game groups
 ACCEPTED MANUSCRIPT
and between the out-of game learning LS group and the in-game adaptive group. The first hypothesis
is:
• H0E: There is no difference in completion times between the non-adaptivity game group and the
out-of-game LS group.
• H1E: There is a difference in completion times between the non-adaptivity game group and the
out-of-game LS group.
When comparing the completion times between the non-adaptive group and the out-of-game LS
group, a Mann-Whitney U test showed no significant difference in completion times between these
two variants (Z = -1.339, p<0.184) and so H0E is accepted.

PT
The second hypothesis is:
• H0F: There is no difference in completion times between the non-adaptivity game group and the
in-game adaptive group.

RI
• H1F: There is a difference in completion times between the non-adaptivity game group and the in-
game adaptive group.

SC
A Mann-Whitney U test showed the completion times for the in-game adaptive group was shorter than
for the non-adaptive group (Z = -2.814, p<0.005) and so H0F is rejected.
The third hypothesis is:
• H0G: There is no difference in completion times between the out-of game learning LS group and

U
the in-game adaptive group.
•
AN
H1G: There is a difference in completion times between the out-of game learning LS group and the
in-game adaptive group.
A Mann-Whitney U test showed no significant difference in completion times between the out-of-game
LS group and the in-game adaptive group (Z = -0.955, p<0.345) and so H0G is accepted.
M

To conduct an overall analysis between the groups, the completion times for all the groups were
compared and analysed using a Kruskal-Wallis test. The test also showed that there was a significant
difference in completion times between groups with the in-game adaptive game requiring the shortest
D

amount of time to complete, followed by the out-of-game LS game, and the game without adaptivity
2
(χ = 7.056, p<0.03). This finding might suggest that dynamically adapting the content is beneficial in
supporting the student’s learning, allowing the student to learn more quickly. A summary of the results
TE

is presented in Table 4.
Table 4: Comparison of Completion Times
Groups Completion Time between Groups
EP

Non-adaptive Group and Out-of-game LS Group Z = -1.339, p<0.184

Non-adaptive Group and In-game Adaptive Group Z = -2.814, p<0.005 (In-game Adaptive
C

Group<Non-adaptive Group)
AC

Out-of-game LS Group and In-game Adaptive Group Z = -0.955, p<0.345

χ = 7.056, p<0.03
2
All Game Groups

5.7 Correlation between Completion Times and Learning Effectiveness

The bivariate analysis using the Pearson correlation test (Kinnear & Gray, 2008) shows that the
correlation between the completion times and learning effectiveness in the non-adaptive group is
weak (r(30) = 0.349, p<0.059). The correlation is also weak in the out-of-game LS group (r(30) =
0.349, p<0.059). Similar to that shown for the previous groups, the correlation between learning
effectiveness and the completion time in the in-game adaptive group is weak (r(30) = 0.182, p<0.336).
Overall, the results show completion time does not have any effect on learning effectiveness.
However, it is also shown the In-game Adaptive Group has a higher correlation between completion
 ACCEPTED MANUSCRIPT
time and learning effectiveness compared to other group. Perhaps by using a larger sample size, the
significance in the correlation may be confirmed.

5.8 Differences in Completion Times between Programmes

In the non-adaptivity group, the mean of the completion times for the participants from non-computing
programmes was 98.67 (SD = 15.883), while from computing programmes the mean was 93.67 (SD =
12.229). In the out-of-game LS group, the mean of the completion times for the participants from non-
computing programmes was 93.11 (SD = 12.875), while from computing programmes the mean was
86.45 (SD = 16.201). In the in-game adaptive group, the mean of the completion times for the
participants from non-computing programmes was 84.62 (SD = 7.985), while from computing

PT
programmes the mean for the completion times was 87.59 (SD = 7.977).
To investigate the differences in completion times between the different programmes for each group,
a Mann-Whitney U test was used. This test showed that there was no significant difference in

RI
completion times for participants from the different programmes in the non-adaptive group (Z = -
0.104, p<0.929), or in the out-of-game LS group (Z = -1.509, p<0.136) and in the in-game adaptive
group (Z = -0.154, p<0.893). The result suggests the participants’ from computing did not have a
significantly faster completion time than the non-computing participants and vice versa.

SC
5.9 Differences in Completion Times between Education Levels
In the non-adaptivity group, the mean of the completion times for the undergraduate participants was
91.95 (SD = 11.345), while for the postgraduate participants it was 99.36 (SD = 14.562). In the out-of-

U
game LS group, the mean of the completion times for the undergraduate participants was 89.61 (SD =
13.682), while for the postgraduate participants it was 92.25 (SD = 15.621). In the in-game adaptive
AN
group, the mean of the completion times for the undergraduate participants was 84.30 (SD = 8.163),
while for the postgraduate participants it was 90.30 (SD = 6.165). To investigate the difference in
completion times on the basis of level of education for each group, a Mann-Whitney U test was used.
This test showed no significant difference in completion time for participants of different education
M

levels in the non-adaptive game group (Z = -1.142, p<0.263), in the out-of-game LS group level (Z = -
0.382, p<0.715) and in the in-game adaptive group (Z = -1.052, p<0.311). The result suggests the
postgraduate participants did not have a significantly faster completion time than the undergraduate
participants and vice versa.
D

5.10 Differences in Completion Times between Learning Styles

TE

In the non-adaptivity group, the mean of the completion times for participants with a picture-based
learning style was 94.06 (SD = 14.289), while for those with a text-based learning style it was 95.46
(SD = 11.326). In the out-of-game LS group, the mean of the completion times for participants with a
picture-based learning style was 91.12 (SD = 13.402), while for participants with a text-based learning
EP

style it was 90.08 (SD = 15.903). In the in-game adaptive group, the mean of the completion time for
participants with a picture-based learning style was 88.19 (SD = 8.534), while for those with a text-
based learning style it was 84.14 (SD = 6.982).
C

To investigate the difference in completion times between learning styles for each group, a Mann-
Whitney U test was used. This test showed no significant difference in completion time for those
participants with different learning styles in the non-adaptive game group (Z = -0.042, p<0.975), in the
AC

out-of-game LS group (Z = -0.231, p<0.829), or in the in-game adaptive group (Z = -1.172, p<0.254).
The result suggests the participants with text-based learning style did not have a significantly faster
completion time than the participants with picture-based learning style and vice versa.

5.11 Differences in Completion Times between Genders

In the non-adaptivity group, the mean of the completion time for female participants was 95.00 (SD =
10.599), while for male participants the mean was 94.57 (SD = 13.727). In the out-of-game LS group,
the mean for the completion time of female participants was 93.12 (SD = 15.887), while for male
participants it was 89.77 (SD = 13.945). In the in-game adaptive group, the mean of the completion
time of female participants was 84.12 (SD = 7.039), while for male participants it was 87.09 (SD =
8.309).
To investigate the difference in completion times for each group according to gender, a Mann-Whitney
U test was used. This test showed no significant difference in completion times for participants of
 ACCEPTED MANUSCRIPT
different genders in the non-adaptive game group (Z = -0.025, p<0.991), in the out-of-game LS group
(Z = -0.470, p<0.654) or in the in-game adaptive group (Z = -0.485, p<0.647). The result suggests the
male participants did not have a significantly faster completion time than the female participants and
vice versa.

6 Conclusions and Future Directions

Adapting the learning content according to the student’s learning style has been found to have a
positive impact on learning. To investigate whether the same benefits apply to GBL, an experiment
was conducted with research question as follows: How do GBL and adaptive GBL compare to
traditional (paper-based) learning? The experiment investigates the differences in learning

PT
effectiveness between different adaptive modes of a game and more traditional (paper-based)
learning and the difference in the completion times of the game groups.
The first conclusion is GBL and both types of adaptive GBL applications can be used to teach SQL

RI
regardless of the learner’s education level, gender, programme of study or learning style as identified
by the learning style questionnaire. Although the participants had no prior knowledge of SQL, they
were motivated to try to learn SQL when playing the games as they observed SQL to be part of the
game and understood that SQL was required to complete the game.

SC
Second, the research found that all forms of GBL (without adaptivity and both types of adaptive GBL
applications) produced higher learning outcomes compared to paper-based learning of SQL. In
general, this finding is in line with the empirical evidence and the systematic literature review by
Hwang et al. (2012), Conati and Zhao (2004), Felicia and Pitt (2009), Hainey (2010) and Connolly,

U
Boyle, MacArthur, Hainey and Boyle (2012) and the empirical evidence produced by adaptive GBL
identified in Section 2.3. It demonstrates the potential of GBL for use as a platform to support the
AN
teaching SQL.
Third, although, the in-game adaptive group has the highest means of learning effectiveness
compared to other game groups as described in Section 6.1, the difference were not significant.
However, the completion time of the in-game adaptive group is found to be significantly lower than the
M

other game groups which indicate that adaptivity has the potential to significantly shortened
completion time. Felicia and Pitt (2009) also found a reduction in completion times between their
experimental adaptive GBL group and the paper-based control group.
D

The results show the out-of game LS group has shorter completion times than the non-adaptive game
group which may be because there was only one type of learning material presented. However, it has
TE

a lower learning effectiveness than the non-adaptive game group which indicates the player may have
different preferences during the gameplay and it may be beneficial to allow the player to change their
preferences during the gameplay. These findings confirm the importance of adapting the game
elements during the gameplay, in this case the learning materials, based on the characteristics of the
player. The in-game adaptivity has the advantage that it monitors and adapts the game elements
EP

accordingly, which resulted in only relevant learning materials at the time are presented. By filtering
the learning materials, it helps to shorten the player’s completion time and reduce the process and
cognitive load of the player when learning through the game.
C

Fourth, when comparing the learning effectiveness of each group, the results show: 1) in general,
there is no difference in learning effectiveness between learning styles; 2) participants from computing
programmes have significantly higher learning effectiveness compared to participants from non-
AC

computing programmes in all game groups; 3) postgraduate participants have significantly higher
learning effectiveness compared to undergraduate participants in all groups; and 4) male participants
have a significantly higher learning effectiveness compared to female participants in all game groups.
In terms of completion time, the results show: 1) participants in the in-game Adaptive group require
shorter time to complete the game followed by the out-of-game group and non-adaptive group; 2)
there is no correlation between completion time and learning effectiveness; 3) there is no significant
difference in completion time between genders, learning styles, programmes of study and education
levels.
The results from this research have suggested that GBL, particularly adaptive GBL, can be used to
teach a subject such as SQL. However, as suggested by Hays (2005) and Hainey (2010), it is not
possible to generalise from the findings produced here relating to teaching SQL as learning content
via GBL to other contents and other methods of delivery. Another limitation is that the findings may be
specific to the integration of these particular learning materials and the mechanics of the game. For
 ACCEPTED MANUSCRIPT
example, the game developed for the purposes of this research has a conversation system
customised to accommodate the teaching of SQL so if this same game were to be used to teach a
different subject, the conversation system would need to be adapted to meet the needs of the new
subject. Such modifications in the game’s specification could give rise to different learning outcomes.
The result of the analysis involving sub-groups such as female/male, undergraduate/postgraduate
and computing/non-computing may require further investigation due to the imbalance between the
number of participants on each sub-group.
For future research, the adaptivity may be improved in future research particularly to address multiple
learning styles in a game and to create more complex adaptivity in various elements of the game and
the results are expected to contribute to empirical evidence of the beneficial effects of GBL and

PT
adaptive GBL in particular.

References

RI
Becker, K. (2007). Digital Game-Based Learning Once Removed: Teaching Teachers. British Journal
of Educational Technology, 38(3), 478-488.
Berns, A., Gonzalez-Pardo, A. & Camacho, D. (2013). Game-like Language Learning in 3D Virtual

SC
Environments. Computers and Education, 60, 210-220
Bernardini, S., Porayska-Pomsta, K., & Smith, T. J. (2013). ECHOES: An Intelligent serious game for
fostering social communication in children with autism. Journal of Information Sciences, 264, 41-60.
Bontcheva, K. (2002). Adaptivity, Adaptability, and Reading Behaviour: Some Results from the

U
Evaluation of a Dynamic Hypertext System, AH’2002, 69-78.
Boyle, E.A., Connolly, T.M. & Hainey, T. (2011). The Role of Psychology in Understanding the Impact
AN
of Computer Games, Entertainment Computing, 2(2), 69-74.
Budiharto, W., Rachmawati, R. N., Ricky, M.Y., Chyntia, P. B. R. (2013). The Psychological Aspects
and Implementation of Adaptive Games for Mobile Application. International Joint Conference on
Awareness Science and Technology and Ubi-Media Computing (iCAST-UMEDIA). IEEE
M

Charles, D., McNeill, M., McAlister, M., Black, M., Moore, A., Stringer, K., Kerr, A., & Kucklich, J.
(2005). Player-centred game design: Player modelling and adaptive digital games. Proceedings of
DiGRA 2005 Conference: Changing Views – Worlds in Play, 285-298.
D

Conati, C., & Zhao, X. (2004). Building and Evaluating an Intelligent Pedagogical Agent to Improve
the Effectivenesss of an Education Game, International Conference on Intelligent User Interfaces,
IUI'04, Funchal, Madeira, Portugal, 6-13.
TE

Conati, C., & Zhou, X. (2002). Modelling Students’ Emotions from Cognitive Appraisal in Educational
th
Games. Proceedings of the 6 International Conference on Intelligent Tutoring Systems, London, UK,
Springer-Verlag, ACM
EP

Connolly, T.M. & Begg, C.E. (2004). Database Systems: A Practical Approach to Design,
th
Implementation and Management, 4 edition, Addison Wesley Longman.
Connolly, T.M. & Begg, C.E. (2014). Database Systems: A Practical Approach to Design,
th
Implementation and Management, 6 edition, Addison Wesley Longman.
C

Connolly, T. M., Boyle, E. A., MacArthur, E., Hainey, T., & Boyle, J. M. (2012). A systematic literature
review of empirical evidence on computer games and serious games. Computers & Education, 59,
AC

661–686.
Connolly, T.M., & Stansfield, M.H. (2006). Enhancing eLearning: Using Computer Games to Teach
Requirements Collection and Analysis. Second Symposium of the WG HCI & UE of the Austrian
Computer Society, 23 November 2006, Vienna, Austria.
Cook, D. A. (2005). Reliability and Validity of Scores from the Index Of Learning Styles. Journal of
Academic Medicine, 80(10), 97-101
Demmel, R. B., Kohler, B., Krusche, S., & Schubert, L. (2011). The serious game: wemakewords.
th
Proceedings of the 10 SIGPLAN symposium on New ideas, new paradigms, and reflections on
programming and software, New York, USA, ACM, 109-110
Doukianou, S., Petridis, P. & Dunwell, I. (2014). A Usability Evaluation of Game-based Approaches
th
Assessing Risk and Delayed Gratification. Proceedings of the 6 International Conference on Games
and Virtual Worlds for Serious Applications (VS-GAMES), IEEE, 1-8
Felder, R. M., & Brent, R. (2005). Understanding Student Differences. Journal of Engineering
Education, 94(1), 57-72.
 ACCEPTED MANUSCRIPT
Felicia, P., & Pitt, I.J. (2007). Evaluating the Effect of Personalities on the Design of Educational
Games. Proceedings of the ECGBL Conference, Paisley, Scotland
Felicia, P., & Pitt, I.J. (2008). Personalising Educational Games to Students' Learning Styles.
Proceedings of the International Technology in Education and Development Conference, Valencia,
Spain
Felicia, P. & Pitt, I.J. (2009). Profiling Users in Educational Games. In Games-based Learning
Advancement for Multisensory Human Computer Interfaces: Techniques and Effective Practices (Eds:
T.M. Connolly, M.H. Stansfield, E Boyle), Idea-Group, Hershey.
Gasimov, A., Tan, C. H., Phang, C. W., & Sutanto, J. (2010). Visiting mobile application development:
what, how and where. Proceedings of the International Conference on Mobile Business and Global

PT
Mobility Roundtable (ICMBGMR), 74-81.
Graf, S., Viola, S. R., Leo, T., & Kinshuk. (2007). In-Depth Analysis of the Felder-Silverman Learning
Style Dimensions. Journal of Research on Technology in Education, 40(1), 79-93
Hertzog, T., Poussin, J. C., Tangara, B., Kouriba, I. & Jamin, J. Y. (2014). A Role Playing Game to

RI
Address Future Water Management Issues in a Large Irrigated System: Experience from Mali.
Agricultural Water Management, 137, 1-14
Hwang, G. J., Sung, H. Y., Hung, C. M., Huang, I., & Tsai, C. C. (2012). Development of a

SC
Personalized Educational Computer Game Based on Students’ Learning Styles. Special Issue on
Personalized Learning, 60(4), Springer, 623-638
Ibanez, J, & Delgado-Mata, C. (2011). Adaptive Two Players Videogames. Expert Systems with
Applications, 38, 9157-9163

U
Ismailovic, D., Haladjian, J., Kohler, B., Pagano, D., & Brugge, B. (2012). Adaptive Serious Game
Development. 2nd International Workshop on Games and Software Engineering (GAS). IEEE, 23 - 26
AN
Jackson, S. L., Krajcik, J. & Soloway, E. (1998). The Design of Learner-adaptable Scaffolding in
Interactive Learning Environments. Proceedings of ACM CHI 1998 Conference on Human Factors in
Computing Systems, 197-194.
Jameson, A. (2003). Adaptive Interfaces and Agents. In Human–Computer Interface Handbook. (Eds:
M

Jacko, J. A., Sears, A.). Lawrence Erlbaum, Mahwah, 305–330

Karpinskyj, S., Zambetta, F. & Cavedon, L. (2014). Video Game Personalisation Techniques: A
Comprehensive Survey. Entertainment Computing, 5, 211-218
D

Kerfoot, B. P. & Baker, H. (2012). An Online Spaced-Education Game to Teach and Assess
Residents: A Multi-Institutional Prospective Trial, Journal of the American College of Surgeons,
214(3):367-73
TE

Kinnear, P. R., & Gray, C. D. (2008). SPSS 16 Made Simple. Psychology Press. New York.
Kirriemuir, J., & McFarlane, A. (2004). Literature Review in Games and Learning. Report 8, NESTA,
Futurelab, Bristol.
EP

Kolb, D. (1984). Experiential Learning. Englewood Cliffs, New Jersey. Prentice-Hall Inc.
Kujala, J, V., Richardson, U. & Lyytinen, H. (2010). Estimation and Visualization of Confusability
Matrices from Adaptive Measurement Data. Mathematical Psychology, 54, 196-207
Kuljis, J., & Liu, F. (2005). A Comparison of Learning Style Theories on the Suitability for elearning.
C

Proceedings of the 2005 ASTED International Conference on Web Technologies, Applications, and
Services, 191-197
AC

Lee, M. J., & Ko, A. J. (2011). Personifying Programming Tool Feedback Improves Novice
Programmers’ Learning. Conference on International Computing Education Research (ICER),
Providence, Rhode Island, USA, ACM, 109-116
Lester, J., Lobene, E., Mott, B. & Rowe, J. (2014). Serious Games with GIFT: Instructional Strategies,
Game Design, and Natural Language in the Generalized Intelligent Framework for Tutoring. In Design
Recommendations for Adaptive Intelligent Tutoring Systems: Volume 2 – Instructional Management
(Eds: Sottilare, R., Graesser, A., Hu, X., and Goldberg, B.), U.S. Army Research Laboratory, Orlando,
Florida, 205-215
Li, Y. N., Yao, C., Li, D. J. & Zhang, K. (2014). Adaptive Difficulty Scales for Parkour Games. Visual
Languages and Computing, 25, 868-878
Litzinger, T. A., Lee, S. H., Wise, J. C., & Felder, R. M. (2005). A Study of the Reliability and Validity
of the Felder-Soloman Index of Learning Styles. Proceeding of the 2005 American Society for
Engineering Education Annual Conference and Exposition, American Society for Engineering
Education
 ACCEPTED MANUSCRIPT
Livesay, G.A., Dee, K.C., Nauman, E.A., & Hites, Jr., L.S. (2002). Engineering Student Learning
Styles: A Statistical Analysis Using Felder’s Index of Learning Styles. 2002 ASEE Conference and
Exposition, Montreal, Quebec, June 2002.
Limongelli, C., Sciarrone, F., Temperini, M., & Vaste, G. (2009). Adaptive Learning with the LS-Plan
System: A Field Evaluation. IEEE Transactions on Learning Technologies, 2(3), 203-215
Lynch, R. A., Steen, M. D., Pritchard, T. J., Buzzell, P. R., & Pintauro, S. J. (2008). Delivering Food
Safety Education to Middle School Students Using a Web-Based, Interactive, Multimedia, Computer
Program. Journal of Food Science Education, 7, Wiley, 35–42
Marsh, T., Li, Z. N., Klopfer, E., Change, X., Osterweil, S., & Haas, J. (2011). Fun and Learning:
Blending Design and Development Dimensions in Serious Games through Narrative and Characters.

PT
In Serious Games and Edutainment Applications (Eds: Ma, M., Oikonomou, A., Jain, L. C.), 273-288
Melis, E., & Monthienvichienchai, R. (2004). They call it learning style but it's so much more.
Proceedings of World Conference on e-learning in Corporate, Government, Healthcare, and Higher
Education 2004, Chesapeake, VA, 1383-1390

RI
Miller, L. M. (2005). Using learning styles to evaluate computer-based instruction. Computers in
Human Behavior, 21, 287-306
Ming-Chun, C., & Shyan-Ming, Y. (2013). The psychological aspects and implementation of adaptive

SC
games for mobile application. International Joint Conference on Awareness Science and Technology
and Ubi-Media Computing (iCAST-UMEDIA), 163 - 169
Mulwa, C., Lawless, S., Sharp, M., Arnedillo-Sanchez, I., & Wade, V. (2010). Adaptive Educational
Hypermedia Systems in Technology Enhanced Learning: A Literature Review. Proceedings of the

U
2010 ACM Conference on Information Technology Education, ACM
Oppermann, R., Rashev, R., & Kinshuk. (1997). Adaptability and Adaptivity in Learning Systems. in
AN
nd
Knowledge Transfer, (Eds: Behrooz, A), pAce, London, 2 ed, 173-179.
Peirce, N., Conlan, O., & Wade, V. (2008). Adaptive Educational Games: Providing Non-invasive
Personalised Learning Experiences. Second IEEE International Conference on Digital Games and
Intelligent Toys Based Education; IEEE, 28-35
M

Praet, M. & Desoete, A. (2014). Enhancing Young Children's Arithmetic Skills Through Non-intensive,
Computerised Kindergarten Interventions: A Randomised Controlled Study. Teaching and Teacher
Education, 39, 56-65
Prensky, M. (2006). Don’t Bother Me, Mom, I'm Learning!: How Computer and Video Games Are
D

Preparing Your Kids for 21st Century Success and How You Can Help. St Paul, Paragon House.
Price, L. (2004). Individual differences in learning: Cognitive control, cognitive style, and learning
TE

style. Educational Psychology, 24, 681-698.

Rodriguez-Cerezo, D., Sarasa-Cabezuelo, A., Gomez-Albarran, M., & Sierra, J. L. (2014). Serious
games in tertiary education: A case study concerning the comprehension of basic concepts on
computer language implementation courses. Computers in Human Behavior, 31, 558-570
EP

Sampayo-Vargas, S., Cope, C. J., He, Zhen., Byrne, G.J. (2013). The effectiveness of Adaptive
Difficulty Adjustments on Student Motivation and Learning in an Educational Computer Game,
Computers and Education, 69, 452-462
Sharda, N. (2007). Authoring Educational Multimedia Content Using Learning Styles and Story Telling
C

Principles. EMME’07, Augsburg, Bavaria, Germany, 93-102

Siegler, R. S., Ramani, G. B. (2011). Improving Low-Income Childrens' Number Sense. Space, Time
AC

and Number in the Brain, 343-354

Smeddinck, J., Siegel, S., & Herrlich, M. (2013). Adaptive Difficulty in Exergames for Parkinson’s
Disease Patients. Proceedings of Graphics Interface Conference, 141-148.
Smith, L. H., & Renzulli, J. S. (1984). Learning Style Preferences: A Practical Approach for Classroom
Teachers. Theory into Practice, 23, 44-50.
Sweetser, P. & Wyeth, P. (2005). Gameflow: a Model For Evaluating Player Enjoyment in Games.
Computers and Entertainment, 3(3), 3
Sweller, J. (1994).Cognitive Load Theory, Learning Difficulty and Instructional Design. Learning and
Instructions, 4(4), 295-312
Tavassolian, A., Stanley, K. G. & Gutwin, C. (2014). The Effect of Temporal Adaptation Granularity
and Game Genre on the Time-Balancing Abilities of Adaptive Time-Varying Minigames.
Entertainment Computing, 5, 43-54
 ACCEPTED MANUSCRIPT
Tong, T. & Chignell, M. (2014). Developing a Serious Game for Cognitive Assessment: Choosing
Settings and Measuring Performance. Proceedings of the Second International Symposium of
Chinese CHI, ACM, 70-79
Tsai, T. W., Lo, H. Y. & Chen, K. S. (2012). An Affective Computing Approach to Develop the Game-
based Adaptive Learning Material for the Elementary Students. Proceedings of the Joint International
Conference on Human-Centered Computer Environments, ACM, 8-13
Van Zwanenberg, N., Wilkinson, L. J., & Anderson, A. (2000). Felder and Silverman's Index of
Learning Styles and Honey and Mumford's Learning Styles Questionnaire: How do they compare and
do they predict academic performance? Journal of Educational Psychology, 20(3), 365-380.
Wilson, A. J., Dehaene, S., Dubois, O., Fayol, M. (2009). Effects of an Adaptive Game Intervention on

PT
Accessing Number Sense in Low-Socioeconomic-Status Kindergarten Children, Mind Brain
Education, 3 (4), 224–234
Yongyuth, P., Prada, R., Nakasone, A., Kawtrakul, A., & Prendinger, H. (2010). AgriVillage: 3D Multi-
Language Internet Game for Fostering Agriculture Environmental Awareness. Proceeding of the

RI
International Conference on Management of Emergent Digital Ecosystems, NY, ACM, 145-152.
Zhang, M., Xu, M., Han, L., Liu, Y., Lv, P. & He, G. (2012). Virtual Network Marathon with Immersion,
Scientificalness, Competitiveness, Adaptability and Learning. Computers and Graphics, 36, 185-192

SC
Zywno, M. S. (2003). A contribution of validation of score meaning for Felder-Soloman’s Index of
Learning Styles. Proceedings of the 2003 Annual ASEE Conference. Washington, DC: ASEE

U
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT
Appendix 1. Felder-Silverman Learning Style Questionnaire
1. I understand something better after I

(a) try it out.

(b) think it through.

2. I would rather be considered

(a) realistic.

(b) innovative.

PT
3. When I think about what I did yesterday, I am most likely to get

(a) a picture.

RI
(b) words.
4. I tend to

(a) understand details of a subject but may be fuzzy about its overall structure.

SC
(b) understand the overall structure but may be fuzzy about details.
5. When I am learning something new, it helps me to

(a) talk about it.

(b) think about it.

U
AN
6. If I were a teacher, I would rather teach a course

(a) that deals with facts and real life situations.

(b) that deals with ideas and theories.

M

7. I prefer to get new information in

(a) pictures, diagrams, graphs, or maps.

D

(b) written directions or verbal information.

8. Once I understand
TE

(a) all the parts, I understand the whole thing.

(b) the whole thing, I see how the parts fit.

9. In a study group working on difficult material, I am more likely to
EP

(a) jump in and contribute ideas.

(b) sit back and listen.

10. I find it easier
C

(a) to learn facts.

AC

(b) to learn concepts.

11. In a book with lots of pictures and charts, I am likely to

(a) look over the pictures and charts carefully.

(b) focus on the written text.

12. When I solve math problems

(a) I usually work my way to the solutions one step at a time.

(b) I often just see the solutions but then have to struggle to figure out the steps to get
to them.
 ACCEPTED MANUSCRIPT
13. In classes I have taken

(a) I have usually gotten to know many of the students.

(b) I have rarely gotten to know many of the students.

14. In reading nonfiction, I prefer

(a) something that teaches me new facts or tells me how to do something.

(b) something that gives me new ideas to think about.

15. I like teachers

PT
(a) who put a lot of diagrams on the board.

(b) who spend a lot of time explaining.

16. When I'm analyzing a story or a novel

RI
(a) I think of the incidents and try to put them together to figure out the themes.

(b) I just know what the themes are when I finish reading and then I have to go back

SC
and find the incidents that demonstrate them.
17. When I start a homework problem, I am more likely to

(a) start working on the solution immediately.

U
(b) try to fully understand the problem first.
18. I prefer the idea of
AN
(a) certainty.

(b) theory.
19. I remember best
M

(a) what I see.

(b) what I hear.

D

20. It is more important to me that an instructor

(a) lay out the material in clear sequential steps.

TE

(b) give me an overall picture and relate the material to other subjects.
21. I prefer to study
EP

(a) in a study group.

(b) alone.
22. I am more likely to be considered
C

(a) careful about the details of my work.

(b) creative about how to do my work.

AC

23. When I get directions to a new place, I prefer

(a) a map.

(b) written instructions.

24. I learn

(a) at a fairly regular pace. If I study hard, I'll "get it."

(b) in fits and starts. I'll be totally confused and then suddenly it all "clicks."
25. I would rather first

(a) try things out.

(b) think about how I'm going to do it.

 ACCEPTED MANUSCRIPT
26. When I am reading for enjoyment, I like writers to

(a) clearly say what they mean.

(b) say things in creative, interesting ways.

27. When I see a diagram or sketch in class, I am most likely to remember

(a) the picture.

(b) what the instructor said about it.

28. When considering a body of information, I am more likely to

PT
(a) focus on details and miss the big picture.

(b) try to understand the big picture before getting into the details.
29. I more easily remember

RI
(a) something I have done.

(b) something I have thought a lot about.

SC
30. When I have to perform a task, I prefer to

(a) master one way of doing it.

(b) come up with new ways of doing it.

U
31. When someone is showing me data, I prefer
AN
(a) charts or graphs.

(b) text summarizing the results.

32. When writing a paper, I am more likely to
M

(a) work on (think about or write) the beginning of the paper and progress forward.

(b) work on (think about or write) different parts of the paper and then order them.
33. When I have to work on a group project, I first want to
D

(a) have "group brainstorming" where everyone contributes ideas.

TE

(b) brainstorm individually and then come together as a group to compare ideas.
34. I consider it higher praise to call someone

(a) sensible.
EP

(b) imaginative.
35. When I meet people at a party, I am more likely to remember

(a) what they looked like.

C

(b) what they said about themselves.

36. When I am learning a new subject, I prefer to
AC

(a) stay focused on that subject, learning as much about it as I can.

(b) try to make connections between that subject and related subjects.
37. I am more likely to be considered

(a) outgoing.

(b) reserved.
38. I prefer courses that emphasize

(a) concrete material (facts, data).

(b) abstract material (concepts, theories).

 ACCEPTED MANUSCRIPT
39. For entertainment, I would rather

(a) watch television.

(b) read a book.

40. Some teachers start their lectures with an outline of what they will cover. Such outlines are

(a) somewhat helpful to me.

(b) very helpful to me.

41. The idea of doing homework in groups, with one grade for the entire group,

PT
(a) appeals to me.

(b) does not appeal to me.

42. When I am doing long calculations,

RI
(a) I tend to repeat all my steps and check my work carefully.

(b) I find checking my work tiresome and have to force myself to do it.

SC
43. I tend to picture places I have been

(a) easily and fairly accurately.

(b) with difficulty and without much detail.

U
44. When solving problems in a group, I would be more likely to
AN
(a) think of the steps in the solution process.

(b) think of possible consequences or applications of the solution in a wide range of

areas.
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT
Appendix 2. Pre-test SQL questions
MyCompany Co. has a database called MyCompanyDB and it contains information about staffs. The
information about staffs is stored in a table called staff_tbl with columns as followed: firstname,
lastname, address, gender, age and contractremaining. The contractremaining represent in years.
Here is the table including the data:

Firstname lastname address gender age contractremaining

John Doe Market Street Male 33 3

PT
Diane Keaton Waverley Road Female 32 1

RI
John Cusack Wester Hailes Male 26 2

Richard Black Cowgate Male 29 4

SC
Carlos Wilson Haymarket Male 36 2

Dawn Leigh Leith Street Female 34 2

U
Helen Ho Morningside Female 40 1
AN
You are required to retrieve the information using SQL based on the following conditions:
M

1. Retrieve all staff information

2. Retrieve only first name and last name of the staffs
3. Retrieve all information about the staff who lives at Haymarket
D

4. Retrieve all information of Diane Keaton

TE

5. Retrieve first name, last name and address of the male staffs who has contracts remaining of
less than 3 years
6. Retrieve the total number of records in the table
EP

7. Retrieve the lowest number in age in the table

8. Retrieve the highest number in age in the table
9. Retrieve the average of age in the gable
C

10. Retrieve the distinct values in the field contractremaining

11. Retrieve all information and sort the results descending to the field age
AC

Your answer (in SQL statement):

1.
2.
3.
4.
5.
6.
7.
8.
 ACCEPTED MANUSCRIPT
Appendix 3. Post-test SQL questions
MyRealEstate Company Co. has a database called estatesDB and it contains information about
properties handled by the company. The information about properties is stored in a table called
properties_tbl with columns as followed: propertyname, city, size, price, numberbedroom and
typeofproperty. The size represents in square feet. Here is the table including the data:

propertyname city size price numberbedroom typeofproperty

SummerSun Glasgow 1200 300000 3 Villa

PT
Alesto Glasgow 700 200000 3 Flat

RI
Alpaydin Edinburgh 1100 500000 3 Villa

Kittle Glasgow 500 130000 2 Flat

SC
LeBlond Edinburgh 800 300000 2 Flat

Cotrell Glasgow 400 75000 1 Flat

U
Oram Aberdeen 875 220000 4 House
AN
You are required to retrieve the information using SQL based on the following conditions:
M

1. Retrieve all properties information

2. Retrieve only property name and city of the properties
3. Retrieve all information about the property called LeBlond
D

4. Retrieve all information of Flats in Glasgow

TE

5. Retrieve property name, city and price of properties which have size bigger than 600 and
have more than 2 bedroom
6. Retrieve the total number of records in the table
EP

7. Retrieve the lowest number in price in the table

8. Retrieve the highest number in price in the table
9. Retrieve the average of price in the table
C

10. Retrieve the distinct values in the field city

11. Retrieve all information and sort the results descending to the field numberbedroom
AC

Your answer (in SQL statement):

1.
2.
3.
4.
5.
6.
7.
8.
 ACCEPTED MANUSCRIPT
Acknowledgements
This work has been co-funded by the EU Lifelong Learning Programme under contract 519057-LLP-
1-2011-1-UK-KA3-KA3NW (Ed2.0Work – European Network for the integration of Web2.0 in
education and work) and as part of the Games and Learning Alliance (GaLA) Network of Excellence
on ‘serious games’ funded by the European Union in FP7 – IST ICT, Technology Enhanced Learning
(see http://www.galanoe.eu).

PT
RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT
Highlights:
- Systematic review of the literature presenting empirical evidence on adaptive GBL
- Evaluation of an adaptive GBL RPG game based on the Felder-Silverman learning style
- Learning effectiveness analysis on paper-based learning and learning through games
- Results show the game groups improve learning effectiveness
- Results show in-game adaptive game group has shortest completion times

PT
RI
U SC
AN
M
D
TE
C EP
AC