-
Persuasive apps are becoming increasingly popular, with researchers and practitioners
making more efforts to find effective ways to persuade people to act in beneficial ways.
In the e-commerce domain, for example, vendors like Amazon and eBay are looking
for better ways to persuade online shoppers to buy more of their products and services.
In the health domain, e.g., physical activity, eating, smoking, etc., designers of
interventions are looking for better ways to help people adopt a healthier lifestyle [
1
]. In
recent years, research has shown that personalizing products and services can lead to
higher persuasion and user satisfaction [
2
]. Further, as cited in [
2
], research in the field
of psychology has shown that two apparently similar individuals, who share similar
characteristics, such as gender, age, culture, etc., may behave differently in the same
situation due to differences in personality [
3
]. However, in the persuasive technology
domain, a field which is still growing, limited studies have investigated the
relationships between personality traits and influence strategies [
4
] for the purpose of designing
more effective behavior change apps. Alkış and Temizel [
4
], who have worked on this
previously, focused on Turkish population only. This may not generalize to other
countries or cultures. This makes it important for further research to be conducted. In this
paper, we investigate the influence of personality traits on Cialdini’s six principles of
persuasion [
5
], focusing on the Canadian population. We conducted an online survey
among 216 Canadian participants to determine the links between the Big Five
personality traits [
6
] and Cialdini’s persuasive strategies [
5
]. We chose Cialdini’s principles
because they are universally established persuasive strategies, which have found wide
acceptance in persuasive technology research [
7–9
] and wide application in the fields
of advertising and marketing [
10
] owing to their strong theoretical underpinning. On
the other hand, we chose the Big Five due to its wide acceptance, application and
empirical validity [
2
]. The Big Five is known as a comprehensive model, which is
applicable across individuals and cultures [
11
]. Our results reveal that Conscientiousness,
Agreeableness and Openness are the most consistent predictors of Cialdini’s persuasive
strategies. Our findings provides designers better insight in designing more effective
persuasive apps for behavior change.
2
Related Work
Limited research has shown the link between personality and Cialdini’s persuasive
strategies [
4
]. Halko and Kientz [
13
] conducted a study in the health domain in this
area. Using correlation analysis, they found that: 1) the more conscientious a person is,
the less likely s/he is to use social applications which employ Cooperation and
Competition strategies; 2) the more agreeable a person is the more effective would
Reinforcement strategy be; and 3) the more open a person is to experience, the more likely for
him or her to welcome Extrinsic, Intrinsic and Negative Reinforcement strategies. With
respect to personality and Cialdini’s principles of persuasion, to the best of our
knowledge, only two prior studies (Alkış and Temizel [
4
] and Sofia et al. [
15
]) have
been conducted. Alkış and Temizel found that Agreeableness, Conscientiousness and
Openness are the strongest predictors of Cialdini’s six principles of persuasion. Sofia
et al. [
15
] also found the same personality traits, including Neuroticism, as the strongest
predictors of the six persuasive principles. However, both studies focused on
collectivist cultures (Greece and Turkey) and used convenience samples. Our study differs from
these prior studies in three ways. First, our focus is on an individualist culture (Canada),
which is independent and concerned about personal goals and aspirations, as opposed
to the collectivist culture which is group-dependent and concerned about collective
goals and aspirations [
16
]. Second, our demographic is more heterogeneous, as we used
Amazon Mechanical Turk (AMT), in addition to email and the University of
Saskatchewan website, for the recruitment of participants, unlike the previous studies, which
recruited students only from their respective universities. Third, we aim to investigate
the generalizability of the prior findings by Alkış and Temizel [
4
]—whose study is
more similar to ours—from one demographic to another. Specifically, we investigate
whether the results based on the Turkish population can be replicated among the
Canadian population as well as the differences that exist between both demographics.
Method
In this section, we present our research question and hypotheses, the measurement
instruments, and the demographics of the survey participants.
4.1
Research Question
The aim of our study is to answer the research question: “How do the Big Five
personality traits influence the persuasiveness of Cialdini’s persuasive strategies among the
Canadian population?” Our hypotheses (see Fig. 1) are as follows:
H1: Agreeableness and Conscientiousness positively influence Authority, while
Openness and Neuroticism negatively influence Authority.
H2: Agreeableness, Conscientiousness and Openness positively influence
Commitment.
H3: Agreeableness and Neuroticism positively influence Consensus, while
Conscientiousness and Openness negatively influence Consensus.
H4: Agreeableness and Extraversion positively influence Liking, while
Conscientiousness and Openness negatively influence Liking.
H5: Agreeableness, Conscientiousness, Extraversion and Neuroticism positively
influence Reciprocation.
H6: Conscientiousness, Extraversion and Neuroticism positively influence
Scarcity.
The above hypotheses are based mostly on the prior findings by Alkış and Temizel [
4
].
Those based on the findings of these authors are shown as regular lines in Fig. 1, while
the new ones we formulated are shown as bold lines. The solid and dashed arrows
represent positive and negative relationships respectively. The new relationships (bold
arrows) are briefly discussed. In H3, based on the findings by Lane and Manner [
17
] and
Moore and McElroy [
18
], we hypothesized that people high in Neuroticism and low in
Conscientiousness will be more responsive to Consensus. Regarding Neuroticism, Lane
and Manner [
17
] found that neurotic people are more likely to follow the crowd, e.g.,
getting addicted to texting and using the mobile phone, both of which have become the
norm in the age of mobile communication. Regarding Conscientiousness, Moore and
McElroy [
18
] found that highly conscientious people are less likely to post pictures on
their Facebook wall: an activity which has become the norm in the age of social
networks. Further, we hypothesized that people low in Neuroticism will be more
susceptible to Authority (see H1) because Karim et al. [
19
] found that students, who are high in
Neuroticism, are more likely to break university’s laws against unethical behaviors,
such as plagiarism (a defiance of authority). Finally, we hypothesized that people high
in Conscientiousness will be more responsive to Scarcity (see H6) because Sofia et al.
[
15
] found that Conscientiousness positively influences Scarcity.
4.2
Measurement Instruments
Prior validated scales were used to measure Cialdini’s persuasion principles and the
Big Five personality traits. The 27 validated items in Kaptein et al.’s [
7
] 32-item
Susceptibility to Persuasive Strategies (STPS) scale were used; the other five items, which
were not validated during the scale’s development process, were dropped. They include
Authority (4 items), Commitment (6 items), Consensus (4 items), Liking (3 items),
Reciprocity (6 items) and Scarcity (4 items). The STPS is a 7-point Likert scale, ranging
from Completely Disagree (1) to Completely Agree (7). It measures how susceptible
people are to Cialdini’s principles. Regarding personality, Gosling et al.’s Ten-Item
Personality Inventory (TIPI) scale was used to measure the Big Five, with two items
measuring each trait [
6
]. The TIPI uses a Likert scale, ranging from Disagree Strongly
(1) to Agree Strongly (7). All of the 27 items in the STPS scale were randomly presented
to participants at the beginning of the survey. Similarly, the TIPI scale items were
randomly presented to participants after responding to the STPS questions.
4.3
Participants
The survey was approved by the University of Saskatchewan (UofS) Research Ethics
Board. Participants were recruited by email, on the UofS website, Facebook and AMT.
Those on AMT were paid $0.8 each, while those on other platforms were given a chance
to win a C$50 gift card. About 310 subjects participated in the study. After cleaning,
we were left with 216 participants, whose country of origin is Canada for analysis. We
did this to have a monocultural population, which could be compared with the Turkish
population. Table 1 shows the participants’ demographics: 31.5% males and 65.3%
females. 32.9% were between the age of 18 and 24, while the other 77.1% were above.
About 35.2% were students at the time of the survey, while the other 64.8% were not.
Results
Measurement Model
We began our analyses by assessing the indicator reliability, internal consistency
reliability, convergent validity and discriminant validity of the constructs in our
measurement models. A Partial Least Square Structural Equation Model (PLS-SEM) was built
using SmartPLS [
20
] for each persuasive strategy, with the personality traits being the
exogenous constructs. SmartPLS is a software tool for building path models. Indicator
Reliability: Indicators with an outer loading greater than 0.5 were retained; otherwise,
they were dropped. For Authority model, one item each was dropped from Openness,
Neuroticism and Authority; for Commitment model, one item each was dropped from
Openness and Extraversion; for Liking model, one item each was dropped from
Extraversion, Neuroticism and Liking; for Reciprocity model, one item each was dropped
from Openness, Agreeableness and Extraversion; and for Scarcity model, one item each
was dropped from Openness, Conscientiousness, Agreeableness and Scarcity. Internal
Consistency Reliability, for each construct, was evaluated using the composite
reliability criterion, which was greater than 0.7. Convergent Validity was evaluated using the
Average Variance Extracted, which was greater than 0.5 for each construct.
Discriminant Validity was evaluated using the crossloading criterion. No indicator loaded higher
on other constructs than the one it was meant to measure [
20
].
5.2
Data-driven Partial Least Square Structural Equation Model
For easy visualization of supported hypotheses and comparison of the significant
relationships across the persuasive strategies (see Fig. 2), we tabulated the path coefficients
from the PLS-SEM models (see Table 2). Overall 45% of our hypotheses (see the
grayed cells) are supported, while 55% of them are not supported (as indicated by
“NS”). The amount of variance of each strategy explained by its model (symbolized by
R2) ranges from 3% (Scarcity) to 15% (Liking). We found Conscientiousness,
Agreeableness and Openness are the most consistent predictors of Cialdini’s persuasive
strategies. For example, Conscientiousness is a predictor of three strategies: Commitment
(β = 0.17, p < 0.05), Liking (β = -0.36, p < 0.001) and Reciprocity (β = 0.25, p < 0.001).
These path coefficients (ranging from 0.17 to 0.36) are relatively higher than the others.
This indicates Conscientiousness is the strongest predictor of Cialdini’s persuasive
strategies among the Canadians, followed by Agreeableness (a predictor of Authority,
Commitment and Liking) and Openness (a predictor of Authority, Consensus and
Liking). Lastly, Neuroticism turns out to be the least predictor of the persuasive strategies,
as it only predicts Consensus. It is noteworthy that none of the personality traits predicts
Scarcity among the Canadians. Though Conscientiousness and Neuroticism have
relatively high path coefficients (0.11 and 0.13 respectively), they are not significant.
We have shown that personality traits influence Cialdini’s persuasive strategies.
Overall, our results reveal that Conscientiousness, Agreeableness and Openness are the most
consistent predictors of Cialdini’s persuasive strategies, with Conscientiousness being
the strongest. As shown in Table 2, 45% of our hypotheses (10 out of 22 relationships)
were validated. Our first hypothesis (H1) is partially validated. We found that people
high in Agreeableness and low in Openness are more responsive to Authority.
However, our hypothesis that Conscientiousness and Neuroticism positively and negatively,
respectively, influence Authority are not supported. Our second hypothesis (H2) was
also partially supported. We found that people high in Agreeableness and
Conscientiousness are more responsive to Commitment. However, we could not find any
relationship between Openness and Commitment among the Canadian population.
Regarding our third hypothesis (H3), two of the four relationships were supported but one was
not, indicating H3 is partially supported. Specifically, we found that Canadians low in
Openness and high in Neuroticism are more susceptible to Consensus. However, we
did not find any relationship between Conscientiousness and Agreeableness, on one
hand, and Consensus, on the other hand. Further, three of the relationships in our fourth
hypothesis were supported, but one was not supported. We found that Canadians low
in Openness and Conscientiousness are more susceptible to Liking but we could not
validate that Extraversion influences Liking. Regarding our fifth hypothesis, only one
of the four relationships are validated: Canadians high in Conscientiousness are more
responsive to Reciprocity. Finally, our sixth hypothesis was completely invalidated.
We found no relationship between the personality traits Conscientiousness,
Extraversion and Neuroticism and the persuasive strategy Scarcity. This indicates it is more
difficult to predict Scarcity compared to other persuasive strategies among the
Canadian population. This is similar to the findings by Alkış and Temizel [
4
], where Scarcity
also turns out to be one of the least predictable in terms of the number of influencers
(two in number) and its amount of variance explained by its predictors. We discuss in
details the validated relationships in the next subsections.
6.1
Influence of Openness on Authority, Consensus and Liking
Our results reveal individuals low in Openness (LO) are more likely to be susceptible
to Authority and Consensus. In contrast, individuals high in Openness (HO) are less
likely to be susceptible to Authority because they are more independent-minded and
daring [
21
]; thus, they are more likely to challenge or disobey authority. Also, they are
less susceptible to Consensus because they are more creative, imaginative and more of
initiators than imitators [
22
]. They tend to do novel things rather than copy others or
follow the status quo. Palmer [
23
] classified the LO and HO individuals as adaptors
and innovators respectively. According to Palmar [
23
], adaptors are concerned about
“doing things better,” while innovators are concerned about “doing things differently.”
For example, with regard to information seeking, as cited in Heinström [
14
], given their
openness to new experiences, innovators tend to seek information more widely and
enthusiastically by using multiple sources. On the other hand, adaptors, given their
closeness, doubt their abilities and thus are more prone to conformity and vulnerable to
social pressure and authority. Regarding Liking, LO individuals are more likely to be
susceptible. As explained by Alkış and Temizel [
4
], closed individuals feel more
comfortable with familiar situations and experiences. As a result, in unfamiliar situations,
they are likely to seek the opinion of and trust those they like. In a nutshell, all three
findings (Openness negatively influencing Authority, Consensus and Liking) replicate
those of Alkış and Temizel [
4
], discussed in subsection 6.5.
Influence of Conscientiousness on Commitment, Reciprocity and Liking
Our results show that individuals who are high in Conscientiousness (HC) are more
likely to be susceptible to Commitment and Reciprocity, but are less susceptible to
Liking. Regarding Commitment and Reciprocity, HC individuals are more self-disciplined,
dependable and responsible [
24
]. As such, they have higher inclination to keep their
commitment and return favor [
4
]. For example, in organizations, Conscientiousness has
been shown to be one of the most consistent predictors of job performance (which is
highly correlated with organizational commitment) [
24
]. One possible explanation for
Conscientiousness negatively influencing Liking is that, given that individuals high in
Conscientiousness (HC) are more thoughtful, they are less likely to agree with or say
“yes” to a person just because they like the person. Thus, they may have more
“comprehensive” or “deeper” reasons for agreeing with or saying “yes” to a person than by
the “superficial” appearance or the “mere” liking of a person.
6.3
Influence of Agreeableness on Authority, Commitment and Liking
Agreeable people are known to be more compliant, straightforward and altruistic (i.e.,
sympathetic and willing to help others) [
21
]. Thus, as we found, they are more willing
to listen to superiors (Authority), do as they promise (Commitment) and agree with or
say “yes” to those they like (Liking). These were also found by Alkış and Temizel [
4
].
6.4
Influence of Neuroticism on Consensus
Individuals high in Neuroticism (HN) are more likely to be more susceptible to
Consensus given their anxiety, as they may not want to do things wrong. Just like LO
individuals, they may neither like to initiate new ventures nor embark on undertakings in
unexplored territories characterized by uncertainties, so they seek social proof. This
finding is similar to that of Lane and Manner [
17
], who investigated individuals’ most
used apps and found and a link between the use of travel apps and neurotic people.
They explained that neurotic people may use travel apps mostly “to overcome their fear
of being lost or of not being prepared for changing weather conditions” (p. 259) [
17
].
6.5
Comparison of Relationships between Canadian and Turkish Populations
We compare our findings based on Canadian participants (Can) with those based on the
Turkish participants (Tur) by Alkış and Temizel’s [
4
] (see Table 3). The grayed out
cells indicate the Turkish findings (relationships) we are able to replicate. As one can
see, our results replicate to a large extent the prior findings by Alkış and Temizel [
4
].
In their study, just as in our study, they found that Agreeableness, Conscientiousness
and Openness were the most consistent predictors of the persuasive strategies. In their
study, at least, two of these traits are co-predictors of five persuasive strategies,
excluding Scarcity. In our study, they are co-predictors of three persuasive strategies
(Authority, Commitment and Liking). Further, just as in our study, Neuroticism was the least
predictor of the persuasive strategies.
In total, we replicated nine of the eighteen significant relationships (i.e., 50%) found by
Alkış and Temizel [
4
]. For Authority, irrespective of nationality, the results show that
people who are low in Openness (β = -0.14 for Can and -0.2 for Tur) and high in
Agreeableness (β = 0.25 for both groups) are more suspecible to Authority. However, there
is a difference between the two populations: Conscientiousness is only a predictor of
Authority for the Turkish population. This indicates that, while those high in
Conscientiousness in the Turkish population are more responsive to Authority, this may not
be the case with the Canadian population. One possible explanation may be the inherent
cultural differences. Turkey, for example, is a collectivist culture [
16
], in which those
in position of authority are highly respected because of the wide power distance
between the leaders and followers. So, the more conscientious a person, the more
ressponsible s/he may be to authority [
19
]. Another possible explanation is that the Turkish
group is more homogeneous than the Canadian group, as it comprises only students
from the university, where conscientiousness and authority (e.g., professors, scholars,
etc.) play a vital role in learning and academic success [
25
]. Thus, there is a higher
chance of predictability of Authority by Conscientiousness among the Turkish student
population. For Commitment, irrespective of nationality, the results show that people
who are high in Conscientiousness (β = -0.17 for Can and -0.31 for Tur) and high in
Agreeableness (β = -0.18 for Can and -0.15 for Tur) are more suspecible to
Commitment. However, there is a difference between the two populations: Openness is only a
predictor of Commitment for the Turkish population. This indicates while those high
in Openness in the Turkish population are more responsive to Commitment, this may
not apply to the Canadian population. For Consensus, irrespective of nationality, the
results show that people who are low in Openness (β = -0.18 for Can and -0.22 for Tur)
are more suspecible to Consensus. However, there is a difference between the two
populations: while Neuroticism is only a predictor of Consensus among the Canadian
population, Agreeableness is only a predictor of Consensus among the Turkish population.
Putting it all together, individuals high in Neuroticism and low in Openess are more
responsive to Consensus among the Canadian group, while individuals high in
Agreeableness and low in Openess are more responsive to Consensus among the Turkish
group.
Further, for Liking, irrespective of nationality, the results show that people who are
low in Openness (β = -0.17 for Can and -0.12 for Tur), low in Conscientiousness (β =
-0.36 for Can and -0.12 for Tur) and high in Agreeableness (β = 0.19 for Can and 0.29
for Tur) are more susceptible to Liking. While the influence of Conscientiousness on
Liking is stronger among the Canadian group, the influence of Agreeableness on Liking
is stronger among the Turkish group. However, there is a difference between the two
populations: Extraversion is only a predictor of Liking for the Turkish population (β =
0.12), with those high in this trait being more responsive to Liking. For Reciprocity,
irrespective of nationality, the results show that people who are high in
Conscientiousness (β = 0.25 for Can and 0.14 for Tur) are more susceptible to Reciprocity. However,
Extraversion and Agreeableness only influence Reciprocity among the Turkish group,
with people high in these traits being more susceptible to Reciprocity. Finally, for
Scarcity, Extraversion and Neuroticism only influence Reciprocity among the Turkish
group, with people high in these traits being more susceptible to Scarcity. To wrap up,
one possible explanation why more personality traits are involved in predicting the
persuasive strategies (e.g., Reciprocity) in the Turkish study than the Canadian study is
that the demographic of the former (all students) is more homogeneous than the mixed
demographic of the latter (see Table 1).
6.6
Implications and Contribution
Though prior research has shown that Commitment is the strongest Cialdini’s
persuasive strategy, followed by Reciprocity and Liking [
9
], it reveals that personality has a
role to play as well in terms of which of the persuasive strategies may be more effective
for certain individuals. For example, in the e-commerce domain, McElroy et al. [
26
]
submit that “firms that can understand their customers’ personality and buying
behavior will have a competitive advantage in the marketplace” (p. 818) [
26
]. Thus, it
becomes important for research findings on this subject to be presented to provide
designers of persuasive apps a broader body of knowledge to draw from when making
decisions about persuasive strategies to apply to their users, who are characteristically
different. In this regard, our contributions are as follows: By using a different culture, we
have shown and replicated the findings of Alkış and Temizel [
4
] that
Conscientiousness, Agreeableness and Openness are the strongest and most consistent predictors of
Cialdini’s principles of persuasion, which designers can leverage in tailoring persuasive
strategies for users in persuasive apps. We have also shown that some relationships
between persuasive strategies and the Big Five may not generalize to all cultures. They
may vary depending on the studied population.
6.7
Limitation
The first limitation of our study is the use of self-report and the shorter version of the
Big Five scale to measure personality. However, we believe, to a large extent, our result
reflects that of the longer version, as research has shown short scales can be as reliable
as long scales [
27
]. A second limitation is the comparison of our findings to those of a
prior study [
4
], e.g., the usage of different measurement instruments in both studies and
the time difference between the conduction of both studies, which may have impacted
our results. However, our study, with respect to the similarities in both studies, provides
insight into how personality traits influence persuasive strategies irrespective of culture.