Content uploaded by Fidelia Orji
Author content
All content in this area was uploaded by Fidelia Orji on Jul 10, 2021
Content may be subject to copyright.
Personalization of Persuasive Technology in Higher Education
Fidelia A. Orji1, Kiemute Oyibo1, Rita Orji2, Jim Greer,1 and Julita Vassileva1
Department of Computer Science
1University of Saskatchewan, Saskatoon, Canada, and 2Dalhousie University, Halifax, Canada
fidelia.orji@usask.ca, kiemute.oyibo@usask.ca, rita.orji@dal.ca, jim.greer@usask.ca
and jiv@cs.usask.ca
ABSTRACT
The success of persuasive systems in changing people’s attitudes
and behaviours has been established in various domains.
Specifically, research has shown that personalized persuasive
technology is more effective at achieving the desired goal than the
one-size-fits-all approach. However, in the education domain, there
are limited studies on the personalization of persuasive strategies to
students. To advance persuasive technology research in this area,
we investigated the susceptibility of undergraduate students (n =
243) to four commonly employed persuasive strategies (Reward,
Competition, Social Comparison, and Social Learning) in PT
design. We aim to use our findings to provide design guidelines for
personalizing persuasive systems in education. These four strategies
were chosen because research on persuasion has established their
effectiveness in changing behaviour and/or attitude. The results of
our analysis reveal that students are more likely to be susceptible to
Reward, followed by Competition and Social Comparison (both of
which come in the second place) and Social Learning (the least
persuasive). Moreover, there is no gender difference in the
persuasiveness of the strategies. Hence, in choosing persuasive
strategies to motivate students’ learning and success in the
education domain, among the strategies we investigated, Reward
should be given priority, followed by Competition and Social
Comparison, while Social Learning should be least favoured.
KEYWORDS
Persuasive Technology, Persuasive system design, Persuasion
Profile, Persuasive Strategies, Personalization, Persuasion in
Education.
ACM Reference format:
Fidelia Orji, Kiemute Oyibo, Rita Orji, Jim Greer, and Julita
Vassileva. 2019. Personalization of Persuasive Technology in
Higher Education. In Proceedings of the 27th ACM Conference
on User Modelling, Adaptation and Personalization, June 9–12,
2019, Larnaca, Cyprus, 4 pages.
https://doi.org/10.1145/3320435.332047
1 Introduction
The effectiveness of persuasive technology (PT) in motivating
people to achieve certain goals has been established in various
domains. Specifically, PT is an interactive system that is designed to
change users’ behaviour, attitude, and opinions about an issue
without using coercion or deception [5]. The ubiquity of
technological devices and applications in recent years has changed
the way we live our lives and do things drastically. For instance,
students attend lectures with their smartphones, tablets and laptops
and use them to read, record, type or search for information in real
time. Even while at home or on the move (e.g., on the bus), students
constantly interact with their personal devices, thereby opening up
new opportunities for these devices to be leveraged in education.
Specifically, Christy and Fox [2], and Filippou et al. [4] have shown
that PTs can be applied in education to help students improve their
academic performance.
In the last few years, researchers of PTs have identified
persuasive strategies, which are being employed in bringing about
behaviour change in the various domain [5, 7]. However, research
has shown that personalized PTs that are tailored to users’
susceptibilities are more likely to be effective in achieving
behaviour or attitude change [6, 8]. Thus, in the education domain,
there is a need for researchers to investigate the susceptibility of
students to the commonly used persuasive strategies in literature to
provide design guidelines for personalizing PTs for the education
domain.
To advance research in this area, this paper investigates the
responsiveness of students to four commonly used persuasive
strategies in PTs [9] which have been shown to be effective in
encouraging users to change specific attitude or behaviour. The
strategies include Reward, Competition, Social Comparison, and
Social Learning. We conducted a study among 243 undergraduate
students using a validated tool called persuadability inventory (PI)
developed by Busch et al. [1]. Specifically, we adapted four PI
scales to reflect the education domain. The results of our analysis
reveal that, overall, students are more likely to be persuaded by
Reward, followed by Competition and Social Comparison, and
Social Learning (the least persuasive). Specifically, there was no
significant difference between the perceived persuasiveness of
Competition and that of Social Comparison. Moreover, males and
females are equally susceptible to all strategies. Thus, in choosing
persuasive strategies to motivate students’ learning and success in
higher education, among the strategies we investigated, Reward
should be given priority, followed by Competition and Social
Permission to make digital or hard copies of all or part of this work for personal or
classroom use is granted without fee provided that copies are not made or
distributed for profit or commercial advantage and that copies bear this notice and
the full citation on the first page. C opyrights for components of this work owned by
others than ACM must be honored. Abstracting with credit is pe rmitted. To copy
otherwise, or republish, to post on servers or to redistribute to lists, requires prior
specific permission and/or a fee. Request permissions from Permissions@acm.org.
UMAP ’19, June 9–12, 2019, Larnaca, Cyprus.
© 2019 Association of Computing Machinery.
ACM ISBN 978-1-4503-6021-0/19/06...$15.00.
DOI: http://dx.doi.org/10.1145/3320435.332047
Comparison, while Social Learning should be given the least
priority.
Our main contributions in the education domain are as follows.
First, we show that the susceptibilities of students with different
persuasive strategies are not equal; rather, they differ. Thus, to
improve the effectiveness of PTs for educations, PTs built with
students most preferred strategy should be personalized them.
Second, we show that students, studying in an individualist
culture, are most likely to be susceptible to Reward and least
likely to be susceptible to Social Learning. Third, we show that
both males and female are equally susceptible to Reward,
Competition, Social Comparison, and Social Learning. Hence, in
personalizing PTs developed with the four strategies, students’
susceptibility to the strategies could be used, rather than gender.
Based on our literature search, our study is the first to present these
findings in the context of PT in education.
2 Study Design and Methods
Our study aims to examine students’ susceptibility to four
commonly used persuasive strategies in PT design. Precisely, it
aims to answer the following research questions:
1. Which of the four strategies (Reward, Competition, Social
Comparison, and Social Learning) are students most susceptible to?
2. Are there gender differences in the perceived persuasiveness of
the strategies?
2.1 Measurement Instruments and Data
Collection
To measure students’ susceptibility to the four persuasive strategies
with respect to learning, we adapted the PI developed by Busch et al.
[1] to the education domain. Our instruments comprised 6 items for
Reward, 5 items for Competition; 6 items for Social Comparison,
and 5 items for Social Learning. Participants rated a 9-point Likert
scale ranging from “1 = Strongly Disagree” to “9 = Strongly Agree.”
Participants in this study were first-year undergraduate students
in a university, who were at least 16 years old. Elimination of bias
in the questions ordering was done by using page randomization
functionality provided by fluidsurvey.usask.ca, which varies the
order in which the questions were presented to each student. A total
of 276 responses were received. After filtering out incomplete
responses, 243 valid responses (76 males and 167 females) were
retained for further analysis.
2.2 Data Analysis
We performed Shapiro-Wilk normality test on our data to determine
the type of analysis to carry out. The test showed that our entire data
is not normally distributed as two of the strategies failed the test (p
< 0.05). As a result, we chose to carry out non-parametric analyses.
To evaluate the reliability of the respective constructs measuring the
four persuasive strategies, we used McDonald’s omega (ω) test [3]
in R’s “psych” package. All of the four strategies passed the
reliability test: ω > 0.7.
Moreover, to analyze our data, we used the non-parametric
Analysis of Variance (ANOVA), which was based on the Aligned
Rank Transformation for Non-Parametric Factor Analysis proposed
by [12]. Specifically, we used the “ARTool” package in R. We
began our analysis by computing the overall and gender-based
mean rating of each of the four strategies. Thereafter, we performed
Repeated Measure ANOVA (RM-ANOVA) to determine whether
there are main effects of gender and strategy and/or interaction
between both factors with respect to participants’ susceptibility to
the four persuasive strategies. Finally, we carried out within-group
and between-group analyses using the pair-wise contrast function in
the ARTool package and Kruskal-Wallis rank sum test, respectively.
3 Results
In this section, we present participants mean ratings of the strategies,
the interaction and main effects of gender and strategy, the between-
and within-group differences.
3.1 Overall mean rating of the perceived
persuasiveness of the strategies
Figure 1 and Figure 2 show the overall and gender-based mean
ratings of the strategies, respectively. Overall, and regardless of
gender, all of the strategies are perceived as persuasive, as each
strategy has an average rating that is greater than or equal to
(approximately) the neutral score of 5. However, more analysis
needs to be carried out to determine the main effect of and
interaction between gender and strategy.
Figure 1: Overall mean ratings of persuasive strategies
Figure 2: Gender-based mean ratings of persuasive strategies
3.2 Interaction and Main Effect of Gender and
Strategy
Our RM-ANOVA showed that there is no interaction between
gender and strategy (F3, 960 = 2.25, p = 0.080621). However, there
is a main effect of strategy (F3, 960 = 147.88, p < 0.001), but no
main effect of gender (F1, 960 = 1.85, p = 0.173746). Specifically,
the lack of a main effect of genders means that males and females
do not significantly differ (p > 0.05). This was confirmed by a
further between-group analysis, which we did not show in this
paper for the sake of brevity.
3.3 Within-Group Analysis: Pairwise Comparison
of Strategies
Given that there is a main effect of strategy, we decided to
conduct a posthoc within-group analysis also known as pairwise
comparisons between the strategies (see Table 1). The results
show that each pair of strategies significantly differ (p < 0.001),
except for Competition and Social Comparison pair, where there
is no significant difference between their mean values (p > 0.05).
Thus, as shown in Figure 1, Reward (7.49) is the most persuasive
strategy, followed by Competition (5.51) and Social Comparison
(5.51), both of which occupy the second position. Moreover,
Social Learning (4.98) turns out to be the least persuasive. Thus,
beginning with the most to the least persuasive strategy, the
overall persuasion profile for our target university students is
Reward, Competition, Social Comparison, and Social Learning.
Table 1: Overall pairwise comparison of strategies
Contrasts
Estimate
SE
df
t.ratio
p.value
Reward - Competition
343.95
22.19
960
15.50
< .0001
Reward - Social
Comparison
343.19
22.19
960
15.47
< .0001
Reward - Social
Learning
430.80
22.19
960
19.41
< .0001
Competition - Social
Comparison
- 0.76
22.19
960
-0.03
1.000
Competition - Social
Learning
86.85
22.19
960
3.91
0.0006
Social Comparison -
Social Learning
87.61
22.19
960
3.91
0.0005
4 Discussions
This study aims at uncovering the susceptibility of university
students to four commonly used persuasive strategies (Reward,
Competition, Social Comparison, and Social Learning) in PT design.
Our results show that there is a main effect of strategy on the overall
perceived persuasiveness of the strategies. This means that some
strategies are perceived as more persuasive than others. We discuss
the implication of our findings.
With respect to our first research question, the results of our
analysis (see Figure 1 and Table 1) reveal that students are more
likely to be susceptible to Reward than the other three strategies.
This means that Reward strategy has the potential of being the most
effective in motivating students to learn or work harder compared to
the Competition, Social Comparison, and Social Learning strategies.
This finding is consistent with that of Oyibo et al. [9], in which the
susceptibility of individuals to these strategies is studied in a general
context. This suggests that their finding, which we replicated, may
cut across domains. This means that, with respect to the four
persuasive strategies we investigated, which are drawn from the PI
[1], Reward is likely to be the most persuasive, irrespective of the
domain being used as a case study. An example question in the
Reward scale is, “It is important to me that my efforts in courses
are rewarded with good grades.” One possible reason why Reward
tends to be the most persuasive in the education domain is that the
overall learning outcome for most students, apart from gaining
knowledge, is to earn good grades, which are a form of reward for
their hard work. Unlike knowledge, which might be immeasurable
in the meantime and remote (manifesting when students are out of
school and are gainfully employed), grades and other rewards (such
as prizes and recognition) are more or less tangible and immediate.
According to Oyibo et al. [9], “Reward has the tendency to provide
immediate reinforcement and present users something to work for
since it is often difficult to visualize the short-term benefit of most
behaviour” (p. 40). As a result, students, irrespective of gender, are
more likely to be motivated by Reward strategy than the other
strategies. Another example question in the Reward scale is,
“External rewards motivate me in learning.”
Moreover, Competition and Social Comparison, which turned
out to be the second most persuasive strategies, may also be
influenced by students’ susceptibility to Reward. An example
question in the Competition scale is, “I push myself hard when I
am in competition with other students in a course.” Similarly, an
example question in the Social Comparison scale is, “I like
comparing my academic performance against other students'
performance in a course.” As we know, the tendency for people to
be competitive (which necessitates social comparison) is very much
likely to be influenced by their natural drive for reward. This is
confirmed by the work of Oyibo et al. [11]. The authors found that
irrespective of gender [9] or culture [10] the more individuals are
susceptible to Reward, the more susceptible they will be to
Competition and Social Comparison. In other words, as discussed
by Oyibo and Vassileva [11], Reward, Competition, and Social
Comparison as persuasive strategies “can co-exist and are
compatible in a given application targeted at a population that is
motivated by any of the three persuasive strategies” (p. 288). Thus,
we recommend that in PT for education, if a designer wants to
implement a persuasive system in a personal (non-social) context,
Reward should be the first port of call. One simple way of
implementing Reward may be by simply and directly mapping
grades to corresponding virtual rewards. In this case, non-point-
based rewards such as badges may be better or more effective, as
the grades of students are already point-based. On the other hand, if
the designer wants to implement a persuasive strategy in a social
context, Competition and/or Social Comparison should be the first
port of call. As we have pointed out, these two strategies, which can
co-exist in a given PT app, can be realized using a leaderboard,
which embodies Competition and Social Comparison
simultaneously.
With respect to our second research question on gender
differences, the results of our RM-ANOVA show that gender has
no significant influence on the perceived persuasiveness of the
four investigated strategies. This means that personalizing PTs
built on the four strategies using gender might not likely improve
their effectiveness in motivating students for active learning
activities. Finally, among the four commonly employed
persuasive strategies we investigated, Social Learning should be
the least favoured when deciding on a persuasive strategy to be
implemented in a persuasive application for education. One
possible explanation of why Social Learning turned out to be the
least persuasive in the context of education is privacy and
confidentiality. Many students, especially those from
individualist cultures, would like to keep their performance (e.g.,
grades) private and confidential. Another possible explanation, in
the context of learning, why Social Learning turned out to be the
least persuasive may be the fact that the student population we
investigated lived and studied in an individualist culture
(Canada), where people are expected to be independent and self-
reliant. Thus, unlike Competition, for example, which is an
intrinsic motivation that cuts across cultures, Social Learning
turned out to be the least persuasive, indicating that students’
learning in this culture is more likely to be influenced by intrinsic
motivation than the observation of and/or learning from others.
An example question in the Social Learning scale is, “Before
making academic decisions, I ask for advice from my peers or
others who know better.”
5 Limitations
The main limitation of our study is that the perceived
persuasiveness of the strategies was measured using a validated
persuasive tool (based on subjective self-report) as against using
a real-life (more objective) persuasive system. Therefore, the
actual persuasiveness of the strategies may differ when measured
in an actual persuasive application for education.
6 Conclusion
This paper presents the results of the susceptibility of 243 university
students to four commonly used persuasive strategies: Reward,
Competition, Social Comparison, and Social Learning. The results
show that irrespective of gender, students are most likely to be
susceptible to Reward, followed by Competition and Social
Comparison (both of which are equally persuasive). Moreover, both
genders are least likely to be susceptible to Social Learning. Thus,
among the four persuasive strategies we investigated, the Reward
strategy should be given the highest priority in PT design targeted at
promoting learning, especially in a personal context. Moreover, in a
social context, Competition and Social Comparison should be given
a higher priority than Social Learning in the PT design for education.
Finally, to improve the effectiveness of PTs designed with the four
strategies in improving students’ learning, personalization could be
done based on students most preferred strategy. In future work, we
intend to investigate the effectiveness of these strategies in a real-
life persuasive system to determine the generalizability of our
findings to an actual application setting.
ACKNOWLEDGMENTS
This work is supported by the NSERC Discovery Grant of the fifth
author.
REFERENCES
[1] Marc Busch, Johann Schrammel, and Manfred Tscheligi. 2013.
Personalized persuasive technology - Development and validation of
scales for measuring persuadability. In International Conference on
Persuasive Technology. Springer, Berlin, Heidelberg, 33–38.
https://doi.org/10.1007/978-3-642-37157-8_6
[2] Katheryn R. Christy and Jesse Fox. 2014. Leaderboards in a virtual
classroom: A test of stereotype threat and social comparison
explanations for women’s math performance. Computers and
Education 78: 66–77. https://doi.org/10.1016/j.compedu.2014.05.005
[3] Thomas J. Dunn, Thom Baguley, and Vivienne Brunsden. 2014. From
alpha to omega: A practical solution to the pervasive problem of
internal consistency estimation. British Journal of Psychology 105, 3:
399–412. https://doi.org/10.1111/bjop.12046
[4] Justin Filippou, Christopher Cheong, and France Cheong. 2016.
Modelling the impact of study behaviours on academic performance to
inform the design of a persuasive system. Information and
Management 53, 7: 892–903. https://doi.org/10.1016/j.im.2016.05.002
[5] B J Fogg. 2002. Persuasive Technology: Using Computers to Change
What We Think and Do. Morgan Kaufmann Publishers. Retrieved
December 18, 2017, from
http://www.elsevierdirect.com/product.jsp?isbn=9781558606432
[6] Maurits Kaptein, Boris De Ruyter, Panos Markopoulos, and Emile
Aarts. 2012. Adaptive Persuasive Systems: A Study of Tailored
Persuasive Text Messages to Reduce Snacking. ACM Transactions on
Interactive Intelligent Systems 2, 2: 1–25.
https://doi.org/10.1145/2209310.2209313
[7] Harri Oinas-Kukkonen and M Harjumaa. 2009. Persuasive systems
design: key issues, process model, and system features.
Communications of the Association for Information Systems, vol. 24,
no. 28, pp. 485–5009.
[8] Rita Orji, Regan L. Mandryk, and Julita Vassileva. 2017. Improving
the Efficacy of Games for Change Using Personalization Models.
ACM Transactions on Computer-Human Interaction 24, 5: 1–22.
https://doi.org/10.1145/3119929
[9] Kiemute Oyibo, Rita Orji, and Julita Vassileva. 2017. Investigation of
the persuasiveness of social influence in persuasive technology and the
effect of age and gender. International Workshop on Persuasive
Technology 1833: 32–44. https://doi.org/10.1145/3099023.3099071
[10] Kiemute Oyibo, Rita Orji, and Julita Vassileva. 2017. Investigation of
the Social Predictors of Competitive Behavior and the Moderating
Effect of Culture. In Adjunct Publication of the 25th Conference on
User Modeling, Adaptation and Personalization - UMAP ’17, 419–
424. https://doi.org/10.1145/3099023.3099113
[11] Kiemute Oyibo and Julita Vassileva. 2017. Investigation of social
predictors of competitive behaviour in persuasive technology. In
International Conference on Persuasive Technology. Springer, Cham,
279–291. https://doi.org/10.1007/978-3-319-55134-0_22
[12] Jacob O. Wobbrock, Leah Findlater, Darren Gergle, and James J.
Higgins. 2011. The aligned rank transform for nonparametric factorial
analyses using only anova procedures. In Proceedings of the 2011
annual conference on Human factors in computing systems - CHI ’11,
143. https://doi.org/10.1145/1978942.1978963