Narrative-Based Interactive Learning for Scam Prevention: Rich Within
Reach
Weile Tu, Bryan Zl Lim, Victor Wd Ong, Juay Hee Tan, Ashe Xy Lee, Peisen Xu and Anand Bhojan
Department of Computer Science, National University of Singapore, Singapore, Singapore
Keywords:
Gamification, Interactive Learning, Experiential Design, Digital Literacy, Scam Prevention, Fraud Awareness.
Abstract:
This paper introduces Rich Within Reach, a narrative-driven, decision-based, educational game designed to
improve scam identification skills using realistic and interactive scenarios. The game leverages engaging nar-
ratives and gameplay mechanics to help players identify phishing, scareware, and invoice scams in email and
SMS contexts. Analyzing player performance metrics, the study uncovers improved abilities in scam detection,
particularly for phishing emails, while highlighting persistent challenges in SMS-based scam identification,
especially for invoice scams. These findings underscore the potential of targeted, gamified interventions to
strengthen digital literacy and fraud awareness. By integrating experiential learning principles, Rich Within
Reach not only equips users with practical scam prevention skills, but also makes a case for the use of interac-
tive learning to address modern cybersecurity challenges. The paper concludes with insights into the game’s
design and implications for broader educational applications.
1 INTRODUCTION
In an increasingly digital world, people are often vul-
nerable to sophisticated scams. In recent years, Sin-
gapore has witnessed a significant surge in scam-
related activities, with the number of reported scam
cases up 46.8% from 31,728 cases in 2022 to 46,563
cases in 2023 (Chua, 2024), underscoring the escalat-
ing challenge of scam prevention. Traditional educa-
tional campaigns have laid the groundwork for public
awareness, with the Singapore Police Force (SPF) at
the forefront of interactive anti-scam initiatives such
as the Anti-Scam Center established in 2019 (Na-
tional Crime Prevention Council Singapore, 2024).
However, the dynamic and evolving nature of
scams necessitates more engaging and adaptive learn-
ing methodologies to educate the public. Interactive
learning is a potent strategy to equip individuals with
the skills and knowledge to identify and thwart scam
attempts using immersive and participatory methods.
Academic research further underscores the efficacy
of interactive learning in scam prevention, as this al-
lows participants to explore the psychological mecha-
nisms that make individuals susceptible to scams and
comparatively evaluate various interventions, includ-
ing interactive educational tools, to mitigate such vul-
nerabilities (Tay and Teiw, 2023).
The National Crime Prevention Council (NCPC)
has also developed interactive web games designed
to test users’ decision-making skills in life-like chal-
lenges, in collaboration with other major institutions
such as banks, telecommunication providers, and
government agencies (Dass, 2021). These initiatives
highlight the pivotal role of interactive learning in
scam prevention, demonstrating its effectiveness and
accessibility to the broader population.
Rich Within Reach (RWR) takes a narrative-based
approach to interactive scam education, with the aim
of equipping players with scam identification and pre-
vention skills by role-playing as an entrepreneur. A
player is required to distinguish legitimate business
and personal opportunities from potential scams on
email and SMS mediums. Compared to traditional
education delivered unilaterally, RWR educates play-
ers on scams using engaging, real-world game scenar-
ios.
In summary, our contributions are two-fold:
1. The design and development of RWR, an interac-
tive, game-based artefact to scam prevention that
offers players a safe and immersive environment
to learn to identify and respond to scams effec-
tively.
2. The analysis of player metrics (inputs and re-
sponses to real-time game events), which reveals
the potential of interactive learning as an innova-
tive tool for educating users on scam prevention.
Tu, W., Lim, B. Z., Ong, V. W., Tan, J. H., Lee, A. X., Xu, P. and Bhojan, A.
Narrative-Based Interactive Learning for Scam Prevention: Rich Within Reach.
DOI: 10.5220/0013436100003932
Paper published under CC license (CC BY-NC-ND 4.0)
In Proceedings of the 17th International Conference on Computer Supported Education (CSEDU 2025) - Volume 1, pages 581-589
ISBN: 978-989-758-746-7; ISSN: 2184-5026
Proceedings Copyright © 2025 by SCITEPRESS – Science and Technology Publications, Lda.
581

2 LITERATURE REVIEW
Interactive games have been used for scam educa-
tion to educate players on recognising and preventing
online scams. For example, Scam Busters (Money-
Sense, 2021) is an interactive web game developed by
MoneySense to help younger audiences understand
specific types of scams such as fraud and phishing.
Players assist a character named Liam and his fam-
ily to identify potential scams and highlight clues to
avoid them. However, Scam Busters targets younger
audiences and exclude adult or elderly audiences who
are more often targets of scams. The categories are
also limited, and do not encompass modern, more
complex scams such as ransomware, scareware, and
impersonation scams.
Another initiative is ScamSpace (ScamSpace,
2024), which is a platform designed to help users
avoid scams on social media platforms like TikTok. It
offers a gamified approach to learning about scam de-
tection, where a player progresses through a module-
based learning tree and navigates a simulated interac-
tive TikTok feed, differentiating between scams and
legitimate content. RWR takes a broader approach,
presenting scams on multiple common media formats
such emails and SMSes as opposed to solely being fo-
cused on specific social media platforms. This multi-
modal engagement allows players to develop nuanced
scam detection skills that adapt to various contexts.
RWR is also differentiated from other initiatives
by weaving scam prevention skills into an engaging
narrative-driven game. Unlike traditional educational
tools that deliver content unilaterally, RWR provides
dynamic, time-sensitive scenarios that mimic real-
world scams, and challenges players to identify and
avoid them. By blending experiential learning and en-
tertainment, RWR makes scam prevention education
active, memorable, and enjoyable, while fostering a
players practical scam-identification skills.
3 GAME DESIGN AND
DEVELOPMENT
RWR is a single-person role-playing game (RPG)
where players are required to distinguish between
scams and legitimate opportunities in order to
progress. The game was designed to equip players
with real-world skills in a simulated, risk-free envi-
ronment. This section will cover the design and de-
velopment of RWR with an emphasis on creating an
interactive, narrative-driven environment.
Figure 1: Scenario development in tutorial for Macs upon
game start.
3.1 Narrative and Scenario
Development
RWR starts by introducing the Macs, an entrepreneur
(Figure 1) who is presented with numerous “invest-
ment opportunities”. Some are legitimate and will
make Macs wealthier while others are scams that will
drain Macs of his hard-earned wealth. The player’s
objective is to grow Macs’ wealth to $100,000 from a
starting capital of $25,000 by identifying and avoid-
ing scams while responding to legitimate opportu-
nities within a time limit. Scams range from im-
personation, phishing, bogus investments and ran-
somware that mimic common scam tactics in Singa-
pore documented in recent studies (Singapore Police
Force, 2024). These scenarios allow players to ex-
perience the subtleties of real scams in a safe setting,
while learning transferrable skills to recognize similar
scams in their personal lives.
3.2 User Interface and Experience
(UI/UX) Design
RWR was designed to simulate real-life devices and
contexts. Interfaces were designed to resemble real-
world interfaces, using similar themes, fonts and art
styles. A sketch-inspired style was used for a more ca-
sual feel compared to traditional modes of education.
Interfaces were also intentionally kept minimal and
user-friendly for clarity, allowing a player to focus
on the core gameplay - reading and making decisions
on text-based content. Interactive objects are marked
with a simple orange outline. Game directions are
provided through an interactive tutorial that is loaded
upon game start (Figure 2). Overall, the game was de-
signed to be minimal and clear, to increase accessibil-
ity for different audiences. This is supported by pre-
vious research indicating that well designed UI/UX in
educational games can significantly enhance knowl-
CSEDU 2025 - 17th International Conference on Computer Supported Education
582

edge retention by reducing cognitive load and increas-
ing engagement (AnNing et al., 2024).
Figure 2: User interface design in tutorial for Macs upon
game start.
3.3 Gameplay Mechanics
RWR employs time-bound, decision-making tasks to
simulate real-world scams. The time element is
added to add gameplay pressure and simulate real-
life, where an individual is presented with scams
while occupied with other tasks. Players are thus re-
quired to quickly analyze in-game prompts to distin-
guish between legitimate opportunities or fraudulent
scams and progres in the game (Figure 3).
Figure 3: A fraudulent email example of an invoice scam.
This mechanic aligns with active learning princi-
ples, where engagement in decision-making boosts
memory retention and real-world application (Laine
and Lindberg, 2020). Progress is measured using an
in-game money system that rewards players for cor-
rectly identifying legitimate opportunities and penal-
izes players for ignoring legitimate opportunities or
responding to fraudulent scams. This also encour-
ages re-playability and enhances learning through a
frequent feedback loop.
3.4 Content Generation
A mix of manual writing and various algorithms is
used for content generation.
3.4.1 Newspaper Content
The Newspaper interface presents static content
which a player can access through the main scene to
learn about common scams types that show up in the
game. This is manually written from research con-
ducted by the team.
3.4.2 Email & SMS Challenges
Email & SMS challenges are randomised using a gen-
erative algorithm that randomly selects and mutates
content from a repository of template emails/SMSes
generated using GPT-4 and manually vetted by the
authors.
Each email is broken into descriptive fields of
“sender”, “title”, “body”, “date”, “attachment” (op-
tional), and functional fields of “isScam”, “hackable”,
“daysLeft” etc. which allows for both content and
logic to be generated based on a single email object in
the repository, and also allows for field-specific muta-
tions to be made by the email generation algorithm.
Mutations include randomly creating typographical
errors (e.g. modifying an email’s local part before the
symbol, modifying a legitimate sender’s name that
closely resembles the original name) The algorithm
also tags each email object generated with a category
(e.g. phishingScam), which allows the team to imple-
ment game and separate database logic to track and
analyze a player’s response to different scam types.
Similarly, each SMS is also composed of descrip-
tive and game logic fields such as “sender”, “recipi-
ent”, “body”, “isScam” etc. Likewise, this allows for
field-specific mutations to be applied, and game-logic
to be created from an SMS object in the repository.
Lastly, the frequency of email & SMS generation
is also randomized using an exponential probability
function which increases the likelihood of challenges
as the game day progresses. The reward/penalty of an
challenge response is randomly determined at “run-
time”, i.e. when a player makes a decision.
3.5 Iterative Development and Testing
RWR was developed iteratively across three phases -
Alpha, Beta and Gold. Each iteration was tested by
the authors’ coursemates. The feedback was used to
improve narrative, game mechanics and overall game-
play flow. A weekly development blog was also main-
tained. Iterative development has been shown to en-
hance user satisfaction in educational games, particu-
larly when feedback is used to adjust game complex-
ity and usability (Viudes-Carbonell et al., 2021).
Narrative-Based Interactive Learning for Scam Prevention: Rich Within Reach
583

3.6 Education Framework
The game’s design is grounded in experiential learn-
ing theory, which posits that individuals learn best
by transforming theoretical knowledge into practical
skills, thereby enhancing the likelihood of real-world
application (Kolb, 1984). The narrative-driven design
of RWR maps seamlessly onto Kolb’s (Kolb, 1984)
Experiential Learning Cycle. Players are introduced
to realistic scam scenarios (Concrete Experience), en-
couraged to reflect on their choices through directed
feedback (Reflective Observation), equipped with ab-
stract principles for scam detection via in-game con-
tent (Abstract Conceptualization), and challenged to
apply these strategies in progressively complex situ-
ations (Active Experimentation). This alignment en-
sures that the learning experience is not only engag-
ing but also theoretically grounded, promoting deeper
knowledge retention and real-world applicability. Re-
search has further shown the efficacy of experiential
learning in educational settings, with a study suggest-
ing that experiential learning activities have signifi-
cantly improved student’s ability to apply theoreti-
cal concepts in practice (Sarah Yardley and Dornan,
2012). Similarly, a meta-analysis has demonstrated
that interactive learning methods, such as simula-
tions and role-playing, effectively enhance learners’
critical thinking and problem-solving skills (Rogers,
2015). In the context of scam prevention, interactive
learning tools have been shown to increase aware-
ness and preparedness. With academic literature on
game design framework revealing that mobile games
designed to teach phishing awareness led to a signifi-
cant improvement in users’ ability to identify fraudu-
lent emails (Arachchilage and Love, 2013).
The following were implemented to achieve the
learning objectives:
1. Concise, Static Newspaper Content. In the
Newspaper of the Main Scene, the team has in-
cluded extremely concise information about dif-
ferent types of scams. By focusing only on: (i)
What are these scams; (ii) How they manifest; and
(iii) How to avoid them; Players can read about
scams and refer to this for gameplay aid.
2. Scam Detection as Core Gameplay. The core
gameplay mechanic replicates real-life scams, al-
lowing players to train their “scam detection” in-
stincts; this lets them respond to actual scams bet-
ter. The game is also intentionally fast-paced, to
replicate real life conditions - where a player is
often presented with scams in their hectic lives.
Lastly, the game also simulates for a player what
happens in reality should they fall for a scam -
they lose more money than they’ve earned.
3. Directed, Instantaneous Feedback. The game
also provides directed, informative feedback im-
mediately after the player makes a decision, i.e.
if a player falls for a scam, the game explains
to the player in the post-decision prompt about
why the Email or SMS was a scam and how they
could have detected it. This directed feedback al-
lows players to be more aware of the scams they
are likely to fall for, and how they can detect and
avoid it in the future.
4 EVALUATION FRAMEWORK
The evaluation framework assesses RWR’s effective-
ness in educating players on scam awareness and pre-
vention strategies. This section outlines the key met-
rics and the methodology used to gauge the educa-
tional impact of the game.
4.1 Evaluation Metrics
To measure RWR’s impact quantitatively, our team
has developed 45 key metrics (shown in Table 1l) that
aligns with both educational and interactive learning
goals in three broad categories:
1. Scam Detection Skills, which focuses on evaluat-
ing players’ abilities to recognize and respond to
scams. These include the accuracy of scam iden-
tification, frequency of legitimate emails correctly
identified, the different platforms (e-mail or SMS)
being engaged, and the total number of emails and
SMS sent within an in-game day;
2. Time Element, which analyzes the time taken for
player to respond to an email or SMS event gener-
ated, number of in-game days that the player has
spent within the game (as retention rate), and time
taken to respond correctly. These provide insights
into how time constraints influence learning re-
tention and decision-making speed, which com-
plements with the Scam Detection Skills category
metrics and is essential in real-life scam preven-
tion where quick decisions are needed;
3. Overall Gameplay, which evaluates the broader
aspects of player engagement by tallying whether
the player has been able to achieve the goal objec-
tive easily with the UI/UX design and scenario de-
velopment. These metrics ensure that the game re-
mains user-friendly and engaging, as a high score
indicates that the game has provided an effective
and enjoyable learning experience for players.
1
Available at https://metaverse.comp.nus.edu.sg/projects/rwr.
CSEDU 2025 - 17th International Conference on Computer Supported Education
584

Table 1: Excerpt of part of the 45 metrics suite collected
into MongoDB. Full list of variables available in Table A1
of the appendix
1
.
GameDay TotalEmails-
Received
CorrectEmails-
Identified
ScamEmails-
Identified
MissedEmails TotalPhoneSMS-
Received
CorrectPhone-
SMSIdentified
ScamPhoneSMS-
Identified
MissedPhone-
SMS
Figure 4: Example of the actual data collected for some of
the metrics.
At the end of every game day, all evaluation met-
rics for that game day are collected and stored in an
online MongoDB Atlas NoSQL database, preparing
the dataset for further processing and analysis (Fig-
ure 4). By implementing real-time data collection
alongside conventional survey methods, this measure
improves both the volume and accuracy of data gath-
ered for analysis. This layered approach enables
a more consistent and quantitative understanding of
player performance, reducing biases which may arise
from conventional survey methodologies.
4.2 Evaluation Methodology
The learning outcomes were evaluated using quantita-
tive in-game metrics and pre- and post-game surveys,
to accurately assess users’ before and after under-
standing of scams using specific scenario-based quiz
questions (Figure 5).
Figure 5: Pre-survey and post-survey quiz question on
phishing scams.
The pre- and post-surveys included technical
questions, such as identifying the warning signs of
ransomware and recognizing the deceptive nature of
scareware and incorporated scenario-based questions
where players were asked to respond to situations like
receiving an email about a suspicious login attempt
on their account, testing their ability to identify the
appropriate next steps.
The surveys also gauged players’ self-reported un-
derstanding of different scams before and after play-
ing the game, giving insight to whether players each
feel that the game has ”subjective helped” them un-
derstand scams better.
In combination, the surveys provided both objec-
tive (e.g. most effective educational elements in the
game) and subjective (self-reported learning ”gains”)
learning outcomes to gauge the effectiveness of RWR
as a reliable interactive learning platform for scam
awareness and prevention.
5 RESULTS
The results demonstrate that RWR has objectively
and subjectively improved players’ understanding and
ability to identify scams. This section summarizes
key findings from empirical data collected through in-
game metrics, pre- and post-game surveys, and qual-
itative feedback, along with limitations and poten-
tial areas of improvements for future research on the
use of interactive learning. 21 pre-survey responses
and 12 post-survey responses for RWR and 74 player
records were used to analyse player engagement and
learning outcomes.
5.1 Pre-Game Survey Analysis
In the pre-survey, majority of respondents (47.6%) re-
ported a very high level of understanding of common
scam types, rating their knowledge as “very well” (5
out of 5-point scale). This response, along with 33.3%
of respondents rating their understanding “well” (4
out of 5-point scale), reflects that the participanst are
relatively confidence with their foundational aware-
ness of scams prior to playing RWR (Figure 6).
Figure 6: Pre-survey results for level of understanding of
scams.
Prior to playing RWR, majority of respondents
have encountered phishing (81%), fake invoices
Narrative-Based Interactive Learning for Scam Prevention: Rich Within Reach
585

(66.7%), scareware (61.9%), and impersonation
(61.9%) scams (Figure 7). This shows that most play-
ers start with a certain level of real-world scam aware-
ness. The game’s design, therefore, builds on this
familiarity by training players to respond faster and
more accurately to scams they encounter.
Figure 7: Pre-survey results for common scams encoun-
tered.
Figure 8: Pre-survey score distribution of scenario-based
questions.
The pre-survey also assessed respondents’ foun-
dational knowledge of scams through a set of 15
multiple-choice questions, shown in Figure 7. The
results showed a mean score of 10.86 and a median
score of 11, suggesting that respondents generally
possessed a good understanding of scams (Figure 8)
before playing RWR.
5.2 In-game Survey Analysis
5.2.1 Scam Identifications by In-Game Days
To evaluate the progression in players’ ability to cor-
rectly identify email scams across different days, we
analyzed the distribution of the percentage of cor-
rectly identified emails for Day 1 and Day 2. Figure 9
illustrates this distribution, with separate histograms
for Day 1 and 2.
The distribution of correct email identification
percentages on Day 1 shows a wide spread, with
prominent peaks around 20%, 40%, and 60%.
The distribution indicates varied performance among
players, with clusters of lower (20%) and moderate
(40–60%) correct identifications. This spread sug-
gests that players on Day 1 had mixed success in iden-
tifying email scams accurately, with a concentration
around moderate success rates. The mean and me-
Figure 9: In-game survey results for correct emails identi-
fied.
Figure 10: In-game survey results for correct SMS identi-
fied.
dian for Day 1 are 49.73% and 50%, respectively, in-
dicating a relatively symmetrical distribution centered
around the midpoint.
The distribution for Day 2 shows a noticeable
shift, with higher concentrations in the 40–60% range,
indicating an improvement in players’ performance in
correctly identifying emails. The clustering around
these percentages suggests that more players were
able to recognize email scams more accurately by
Day 2, indicating a potential learning effect or fa-
miliarity with the game mechanics and scam cues.
The mean and median for Day 2 were calculated as
66.75% and 63.16%, respectively, both values higher
than those for Day 1. The increase in both mean and
median points to an overall improvement in correct
email identification skills.
The findings demonstrate an improvement in play-
ers’ ability to identify email scams. The concentration
of Day 2 results around the higher percentages, along
with increased mean and median values, suggests that
players were able to more accurately identify email
scams over time; this reflects learning and/or adapta-
tion, where repeated gameplay or feedback may have
enhanced players’ detection skills.
CSEDU 2025 - 17th International Conference on Computer Supported Education
586

In addition to email identification, we analyzed
players’ performance in identifying SMS scams
across both days. Figure 10 provides a distribution
of the percentage of correct SMS identifications for
Day 1 and Day 2.
The distribution of correct SMS identification per-
centages on Day 1 is highly variable, with peaks
around 20% and 60%. The clustering at 20% indi-
cates that a significant portion of players had difficulty
in accurately identifying SMS scams, suggesting that
they may have struggled with recognizing scam in-
dicators in SMS messages. However, there are also
players with higher correct identification rates, up to
60%, indicating some players were relatively success-
ful despite the general challenges.
The distribution for Day 2 shows a concentra-
tion around the same lower and moderate percentages,
with fewer instances above 60%. Despite the slight
presence of players who achieved moderate success,
the distribution reveals a lack of significant improve-
ment from Day 1, as reflected in the overlapping clus-
ters at lower ranges. This consistency across days
implies that players’ ability to identify SMS scams
may have reached a plateau, or that SMS scams were
inherently more challenging to identify compared to
email scams. Both the mean and median for SMS
identification across days remained relatively indif-
ferent, further supporting the observation of minimal
progression in SMS identification skills.
The analysis of SMS identification results indi-
cates limited improvement between Day 1 and Day
2. Unlike email identification, where players demon-
strated noticeable progress, the SMS identification
data reveals continued challenges, with little indica-
tion of a learning effect. This finding suggests that
identifying SMS scams requires more nuanced under-
standing or targeted support, as players did not show
the same level of adaptation observed in email identi-
fication.
Notably, the majority of real-time records were
concentrated in the early stages of gameplay, with
27 entries from the first in-game day (17 entries on
emails, and 10 on SMSes) and 11 from the second
(7 on emails and 4 on SMSes). This distribution in-
dicates a natural decline of player count as the game
progresses. The players are also more willing to en-
gage with the email contents as compared to SMSes.
We mention this in Section 5.5.
These findings suggest that players may inherently
find email scam cues more discernible, whereas SMS
scams require more nuanced interpretation, which
may not be readily learned through simple repetition.
5.2.2 Scam Identification by Type
The in-game metrics also examines the frequency
of incorrect identifications by scam category in both
email and SMS formats, (Figure 11 and Figure 12).
In email challenges, Phishing (10 errors), Scare-
ware (12 errors), and Impersonation scams (10 er-
rors) were the most challenging for players, reflecting
the deceptive nature of these scams. The high error
rates suggest that players found it difficult to distin-
guish these scams from legitimate emails. In contrast,
Invoice scams were incorrectly identified only once,
suggesting they were more easily recognized, likely
due to distinct indicators relating to financial docu-
mentation.
Figure 11: In-game survey results for common e-mail scam
mistakes by players.
Figure 12: In-game survey results for common SMS scam
mistakes by players.
As for SMSes, Invoice scams had the highest er-
ror rate (16 errors), significantly higher than in emails,
indicating that players struggled to recognize finan-
cial scams in SMS form. Scareware (10 errors) and
Personal scams (8 errors) also had high failure rates,
similar to email results, suggesting consistent diffi-
culty across formats. Phishing (6 errors) was some-
what easier for players to identify in SMS, while Im-
personation and Ransomware scams had lower er-
ror rates (3 and 1 errors, respectively).
Narrative-Based Interactive Learning for Scam Prevention: Rich Within Reach
587

Players found Phishing, Scareware, and Imper-
sonation scams challenging across formats, while
SMS-specific challenges were prominent in Invoice
scams. These findings therefore suggest a need for
targeted guidance in recognizing scams that exploit
different mediums. Enhancing in-game cues for com-
plex scam types, particularly in SMS format, may im-
prove players’ scam detection skills and overall learn-
ing outcomes.
5.3 Post-Game Survey Analysis
In the post-survey, a majority of respondents (66.7%)
reported a very high level of understanding of com-
mon scam types, rating their knowledge as “very
well” (5 out of 5 on a 5-point scale). Additionally,
25% of respondents in the post-survey rated their
understanding as “well” (4 out of 5). This reflects
an improvement of distribution from the pre-survey,
indicating enhanced awareness and confidence (Fig-
ure 13). These results suggest that RWR has con-
tributed meaningfully to deepening players’ founda-
tional knowledge of scams, with a clear increase in
participants reporting a high or very high level of un-
derstanding after gameplay.
Figure 13: Post-survey results for level of understanding of
scams.
Furthermore, reflecting on their knowledge be-
fore playing the game, a majority of respondents
(50%) admitted that their foundational understand-
ing of scams had been only moderate (3 out of 5 on
a 5-point scale) (Figure 14). This retrospective as-
sessment, contrasted with the pre-survey data, where
the majority were confident about their understanding
(47.6% rated as ”very well”, or 5 out of 5), suggests
that RWR provided meaningful learning experiences
that helped participants’ increase the scam awareness.
The shift in awareness underscores the game’s effec-
tiveness in enhancing players’ confidence and under-
standing of common scam types.
The post-survey also evaluated respondents’ foun-
dational knowledge of scams using the same set of fif-
teen multiple-choice questions. The results revealed a
mean score of 11.17 and a median score of 12, show-
ing a slight improvement from the pre-survey mean
Figure 14: Post-survey results for level of understanding
retrospectively.
Figure 15: Post-survey score distribution of scenario-based
questions.
of 10.86 and median of 11 (Figure 15). This increase
suggests that participants not only retained their base-
line knowledge but also enhanced their understanding
of scams and their mechanisms after playing RWR.
5.4 Observational Findings and
Hypothesis
The main hypothesis was that playing RWR would
objectively improve players’ ability to discern scams
from legitimate communications. This was tested by
comparing players’ accuracy in scam identification
over two Game Days. The in-game metrics support
this - players showed a slight improvement in re-
sponse accuracy (Correct responses to emails/SMS-es
out of total responses) on Game Day 2 versus Game
Day 1. Likewise, pre- and post-survey data supported
these findings, where respondents’ minimum mean
and median scores increased after gameplay, which
indicates player learning and improvement in scam
identification. As such, RWR did improve players’
abiity to discern scams.
The second hypothesis proposes that players will
better understand common scam types after playing
RWR. Compared to the previous hypothesis which
was observed objectively, this hypothesis is supported
by subjective observations - respondents expressed an
improvement in their self-assessed understanding of
common scam types post-game, and indicate that they
earlier overestimated their understanding of scams.
Both hypotheses are supported by the in-game
metrics and surveys, suggesting that RWR is effective
in educating players on scams and scam identifica-
tion.
CSEDU 2025 - 17th International Conference on Computer Supported Education
588

5.5 Areas of Improvements
An initial hypothesis was that playing the RWR
would train players to respond faster to scams they
encounter. However, player metrics do not show
a marked improvement in response times between
Game Day 1 versus Game Day 2. From the extremely
short response times, players could also have reacted
impulsively rather than thoughtfully. Further devel-
opment is required to increase engagement, and more
data is required for analysis.
Due to time constraints, a major limitation is the
relatively small dataset for analysis - only 27 entries
were recorded for Day 1 and 11 for Day 2. Some
players also only engaged with one medium (SMS or
email), distorting the data collected. More player test-
ing is required to bolster the findings.
RWR can also be developed to include progres-
sive difficulty depending on the player’s progress and
response times. This can improve retention rate by
posing more challenging scenarios to well-informed
players, and reduce impulsive decision-making by
posing longer prose to players who react impulsively
(short response time + incorrect answer). Data col-
lected will be more reflective of a player’s learning
progress.
A more well-developed narrative with a branching
storyline dependent on the player’s decisions (akin to
”personality tests” or ”determine your fate” style of
games can also be implemented to engage a player
better, and in turn increase retention and reinforce
learning over time.
Lastly, more comprehensive in-game feedback
can be provided to a player to reinforce learning, re-
tain players, and make players feel a better sense of
enrichment. For example, in-game metrics can be
analysed and shown to a player at the end of ev-
ery Game Day - their improvement in response time,
scams most vulnerable to, etc.
6 CONCLUSIONS
RWR is a significant step towards the field of interac-
tive learning for scam prevention, by contributing to
the growing body of research on the intersection of
gamification and cybersecurity education. RWR also
demonstrates the tangible impact that thoughtfully de-
signed educational games can have on societal chal-
lenges. Through its engaging narrative, realistic sce-
narios, and decision-based gameplay, the game equips
players with the practical skills needed to recognize
and respond well to various scams. Empirical stud-
ies reaffirm the success of RWR in enhancing scam
awareness and its response, which demonstrates the
overall effectiveness of interactive learning as a plat-
form to raise public awareness on scams. Our team
hopes that RWR serves as a prelude to help liaise with
related authorities to increase the impact of combat-
ing the proliferation of fraudulent scams.
REFERENCES
AnNing, Ahmad, M., and Ibrahim, H. (2024). User-
centered mobile app design for education: Enhanc-
ing engagement and learning outcomes. ESP Interna-
tional Journal of Advancements in Science & Technol-
ogy, 2(1).
Arachchilage, N. A. G. and Love, S. (2013). A game design
framework for avoiding phishing attacks. Computers
in Human Behavior, 29(3):706–714.
Chua, N. (2024). Scam victims in s’pore lost $651.8m in
2023, with record high of over 46,000 cases reported.
Dass, D. (2021). Police life: The anti-scam centre: A col-
laborative approach against scams.
Kolb, D. (1984). Experiential Learning: Experience As
The Source Of Learning And Development, volume 1.
Prentice Hall.
Laine, T. H. and Lindberg, R. S. N. (2020). Designing
engaging games for education: A systematic litera-
ture review on game motivators and design princi-
ples. IEEE Transactions on Learning Technologies,
13(4):804–821.
MoneySense (2021). Scam busters.
National Crime Prevention Council Singapore (2024).
Xiam the scams.
Rogers, D. T. (2015). Further validation of the learning al-
liance inventory: The roles of working alliance, rap-
port, and immediacy in student learning. Teaching of
Psychology, 42(1):19–25.
Sarah Yardley, P. W. T. and Dornan, T. (2012). Experiential
learning: Transforming theory into practice. Medical
Teacher, 34(2):161–164. PMID: 22288996.
ScamSpace (2024). Your child’s safety on tiktok starts here.
Singapore Police Force (2024). Mid-year scams and cyber-
crime brief 2024.
Tay, S. and Teiw, Y. K. (2023). Phishing for a job? investi-
gating scam victimisation and interventions in singa-
pore.
Viudes-Carbonell, S. J., Gallego-Dur
´
an, F. J., Llorens-
Largo, F., and Molina-Carmona, R. (2021). Towards
an iterative design for serious games. Sustainability,
13(6).
Narrative-Based Interactive Learning for Scam Prevention: Rich Within Reach
589