You are on page 1of 16

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/282409294

Social influence or personal preference? Examining the determinants of usage


intention across social media with different sociability

Article  in  Information Development · August 2015


DOI: 10.1177/0266666915603224

CITATIONS READS

31 2,815

2 authors, including:

Yongqiang Sun
Wuhan University
94 PUBLICATIONS   3,916 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Electronic Commerce View project

E-Learning View project

All content following this page was uploaded by Yongqiang Sun on 17 February 2016.

The user has requested enhancement of the downloaded file.


Article
Information Development
1–15
Social influence or personal preference? ª The Author(s) 2015
Reprints and permission:
sagepub.co.uk/journalsPermissions.nav
Examining the determinants of usage DOI: 10.1177/0266666915603224
idv.sagepub.com
intention across social media with
different sociability

Nan Wang
Wuhan University

Yongqiang Sun
Wuhan University

Abstract
Previous studies have identified that social influence and personal preference are two key mechanisms to
understand users’ social media adoption behavior. However, whether or not these two mechanisms play equal
roles under different contexts has rarely been empirically investigated. To fill this gap, in this study, we propose
sociability as a criterion to classify social media and examine the determinants of usage intention across social
media with different sociability. An empirical study of 118 low-sociability and 123 high-sociability social media
users found that social influence factors play a more important role for high-sociability media users while atti-
tude has stronger impacts on intention for low-sociability media users. Implications for research and practice
are also discussed.

Keywords
social media, social influence, theory of reasoned action, sociability, we-intention, Weibo, Wechat, China

Received June 4, 2015. Accepted for publication August 7, 2015.

Introduction following the emergence of a new social media ser-


Social media services, which refer to a group of vice, namely Wechat (a social media service based
applications based on the Web 2.0 techniques and on the famous instant messaging tool QQ), the num-
user generated content (UGC), including a lot of ber of users of Weibo (a Twitter-like service in
well-known tools such as Facebook and Twitter China) decreased from 281 million by the end of
(Kaplan and Haenlein, 2010), are rapidly changing 2012 to 253 million by the end of 2013, with a
the way we are sharing information and keeping decrease rate of nearly 10%. The report further
social relationships (Kane et al., 2014; Kaplan and showed that about 37.4% of the users who chose to
Haenlein, 2010; Hanna et al., 2011). Following the leave Weibo switched to Wechat. Therefore, it is
proliferation of social media services, the competi- interesting to understand why two different types
tion between different social media service providers of social media have different destinies and the
has become more and more stiff. As Kaplan and underlying mechanisms behind this change.
Haenlein have stated, social media is ‘‘a very active
and fast-moving domain’’ so that ‘‘what may be up-
Corresponding author:
to-date today could have disappeared from the vir-
Yongqiang Sun, School of Information Management, Wuhan
tual landscape tomorrow’’ (Kaplan and Haenlein, University, Wuchang District, Wuhan, Hubei, China. Tel. 86-
2010: 64-65). According to a report by China Inter- 18672389832.
net Network Information Center (CNNIC) in 2014, Email: syq@mail.ustc.edu.cn

Downloaded from idv.sagepub.com at WUHAN UNIV LIBRARY on September 5, 2015


2 Information Development

Previous studies on social media adoption have Theoretical background


investigated two key mechanisms underlying the Social media adoption
social media adoption behavior. The first mechanism,
based on the technology acceptance theories (Venka- Social media adoption is one of the most frequently
tesh, 2000; Venkatesh et al., 2003; Venkatesh et al., discussed issues in social media research (Ngai
2012), focuses on a variety of individual perceptions et al., 2015). Previous studies view social media adop-
about social media, such as perceived usefulness, per- tion behavior from three major perspectives, includ-
ceived enjoyment or the overall perception attitude ing the personal preference perspective, the social
(e.g., Lin and Lu, 2011; Sun et al., 2015). In contrast, influence perspective and the personality perspective
the second mechanism, based on the social influence (as summarized in Table 1). The personal preference
model (Kelman, 1958; Dholakia et al., 2004), views perspective stresses the key values that users can gain
social media adoption behavior as a collective action from the social media usage behavior and proposes
and proposes that user behaviors are affected by others that when users’ personal needs are satisfied they will
through three social influence processes, namely, com- be more likely to use social media. From this perspec-
pliance, internalization and identification. Although tive, prior literature has investigated the impacts of
previous studies have provided certain empirical sup- perceived usefulness and perceived enjoyment based
port for the impacts of both social influence and per- on the Technology Acceptance Model (Lin and Lu,
sonal preference factors, whether or not these two 2011), a variety of value perceptions including purpo-
mechanisms have equal contributions to user behaviors sive value, self-discovery, maintaining interpersonal
has been rarely empirically tested. interconnectivity, social enhancement, and entertain-
Further, prior studies on social media adoption tend ment value based on use and gratification theory
to treat social media as a broad term to apply to a vari- (Cheung et al., 2011), and the overall perceptions
ety of technologies, including microblogs and social about consequences of using social media (e.g., atti-
networking sites (Kaplan and Haenlein, 2010; Kane tude) based on the Theory of Reasoned Action or the
et al., 2014), while the distinctions between and the Theory of Planned Behavior (Al-Debei et al., 2013).
user behaviors under different types of social media In contrast, the social influence perspective argues
have been rarely examined. Kaplan and Haenlein that ‘‘the use of social technologies can make sense
(Kaplan and Haenlein, 2010) have proposed that only when a group of individuals are willing to use
social media can be classified into different categories and continue to use the technology together’’ (Cheung
according to two dimensions, namely: social presence / and Lee, 2010: 24), such that individual behaviors in
media richness and self-presentation / self-disclosure. social media can be conceptualized as an intentional
Thus, it is interesting to know whether or not users will social action (Cheung and Lee, 2010). Thus, we-
behave differently in different types of social media intention, which is defined as ‘‘commitment of an
such as Weibo and Wechat. individual to participate in joint action, and involves
Considering the two different explanatory in an implicit and explicit agreement between the par-
mechanisms and the distinctions between different ticipants to engage in that joint action’’ (Tuomela,
types of social media, the research objective of this 1995: 2), is used to study online social activities.
study is: Unlike individual intention, which describes an indi-
vidual act in isolation, we-intention ‘‘highlights the
to explore the different roles of social influence and per-
individual commitment in collectivity and the social
sonal preference factors in different types of social nature of group action’’ (Cheung and Lee, 2010:
media (e.g., Weibo and Wechat). 25), so it should be more appropriate to capture the
individual behavior in the social media context.
The rest of the paper is organized as follows. In the From this perspective, previous studies have iden-
next section, previous studies on social media (e.g., tified three mechanisms to explain the social influ-
theories, factors and typologies) are reviewed and the ence process, including compliance, internalization
research model and hypotheses are proposed. Then, and identification (Kelman, 1958). Specifically, com-
the survey study to test our research model is pliance process reflects the social influence based on
described and the data analysis results are reported. reward approaching or punishment aversion; interna-
Finally, the theoretical and practical implications of lization process captures the social influence induced
the study are discussed. by individuals’ acceptance of others’ beliefs; and

Downloaded from idv.sagepub.com at WUHAN UNIV LIBRARY on September 5, 2015


Wang and Sun: Social influence or personal preference 3

Table 1. Literature review on social media adoption.

Perspective Theory Factors Key References

Personal preference Technology Acceptance Model Perceived usefulness (Cheung et al., 2011; Lin and Lu, 2011;
Use and gratification theory Perceived enjoyment Al-Debei et al., 2013)
Theory of Reasoned Action Value perceptions
Attitude

Social influence Social influence model Subjective norm (Cheung and Lee, 2010; Cheung et al., 2011;
Theory of Reasoned Action Group norm Qin et al., 2011; Choi and Chung, 2013)
Social identity
Personality Big-Five model Neuroticism (Hughes et al., 2012; Correa et al., 2010)
Extraversion
Openness
Agreeableness
Conscientiousness

identification process describes the social influence sociable and talkative. Openness-to-experience refers
based on liking or respect of another person. The com- to the extent to which one has broad interests and seeks
pliance process is consistent with the mechanism novelty. Agreeableness measures how friendly people
through which subjective norm influences individual are, and conscientiousness refers to a person’s work
behaviors in the Theory of Reasoned Action (TRA), ethic, orderliness and thoroughness. Previous studies
given that both of them capture the extent to which have confirmed that these five personality factors are
individuals would like to comply with the expecta- related to users’ social media adoption behavior
tions of significant others (Fishbein and Ajzen, (Hughes et al., 2012; Correa et al., 2010).
1975). Subjective norm is thus usually operationa- Because personality is difficult to be manipulated by
lized as a construct to reflect the compliance process. practitioners, in this study we pay more attention to the
Considering that group norm refers to the shared val- factors relevant to personal preference and social influ-
ues or goals among group members, it is appropriate ence. Specifically, consistent with Tsai and Bagozzi
to be used to reflect the internalization process, which (2014), we try to integrate the personal preference per-
is based on the similarity of values held by group spective and the social influence perspective by consid-
members (Cheung and Lee, 2010; Tsai and Bagozzi, ering social influence factors and attitude, which is the
2014). Finally, Bagozzi and Lee (2002) conceive overall perception of individual values as the antece-
social identity as the reflection of the identification dents of users’ we-intentions to use social media.
process because social identity describes cognitive
self-awareness of group membership, affective com-
mitment to the group and group-based self-esteem, Media sociability
consistent with the internalization process. This oper- Although social influence factors and attitude may
ationalization is also widely used in many empirical influence social media adoption intention, whether or
studies such as Cheung and Lee (2010) and Tsai and not their influential strengths vary across different types
Bagozzi (2014). of social media is unknown. To solve this problem,
The third theoretical perspective focuses on the rela- we need to classify social media in terms of certain cri-
tionships between personality factors and social media teria. According to Kaplan and Haenlein (2010), social
adoption behavior by drawing upon the Big-Five media can be categorized into a variety of types based
model (Goldberg, 1990). Specifically, the Big-Five on two dimensions, as shown in Table 2.
model points out five broad personality traits, including The first dimension is termed as the social presence
neuroticism, extraversion, openness, agreeableness and / media richness dimension, which is based on social
conscientiousness. Neuroticism as a measure affect and presence theory (Short et al., 1976) and media rich-
emotional control captures the extent to which an indi- ness theory (Daft and Lengel, 1986). Specifically,
vidual has good control over emotions. Extraversion social presence theory proposes the intimacy (inter-
describes the extent to which one is adventurous, personal vs. mediated) and immediacy (asynchronous

Downloaded from idv.sagepub.com at WUHAN UNIV LIBRARY on September 5, 2015


4 Information Development

Table 2. Classification of social media (adapted from Kaplan and Haenlein (2010)).

Social presence / Media richness

Low Medium High

Self-presentation / High Blogs Social networking sites (e.g., Virtual social worlds (e.g., Second
Self-disclosure Facebook) Life)
Low Collaborative projects Content communities (e.g., Virtual game worlds (e.g., World of
(e.g., Wikipedia) YouTube) Warcraft)

vs. synchronous) of the medium as two key features In this study we try to classify social media into dif-
relevant to users’ perceptions of social presence, such ferent types according to the social presence percep-
that individuals will feel higher social presence for a tion generated when users interact with other users
interpersonal and synchronous medium than for a via social media. Consistent with prior literature, we
mediated and asynchronous medium (Short et al., use the term sociability (Phang et al., 2009; Preece,
1976). Media richness theory assumes that media pos- 2001) to capture this media characteristic. Sociability
sess different abilities to solve ambiguity and reduce refers to an online community feature that supports ‘‘a
uncertainty, such that rich media (e.g., video) may state of being sociable’’, where community members
deliver more information in a given time interval than can feel pleasant when interacting with each other
poor media (e.g., text) (Daft and Lengel, 1986). through the supporting technology (e.g., social media)
Kaplan and Haenlein (2010) consider that media (Phang et al., 2009; Preece, 2001). Although there are
richness and social presence are closely related, so many social media with different levels of sociability,
they take them together as one dimension to classify we try to take two social media tools namely Wechat
social media. However, media richness and social and Weibo in China as the representatives of high-
presence may capture different aspects of the medium. and low-sociability social media respectively. Specif-
For example, a video to introduce a digital product can ically, Weibo, which is also known as Sina Weibo, is
enhance consumers’ cognitive understanding of the a microblogging service in China that enables users to
product, but cannot improve social interactions. In this follow other users for pushed information and post or
situation, high media richness may not necessarily lead forward information to others, just like Twitter. In
to high social presence. Similarly, social interactions contrast, Wechat, which is initiated by Tencent.com,
may be improved through some social support tools, integrates mobile instant messaging services with
which are based on poor media (e.g., text). In this social network services (like Facebook), enabling
regard, we would like to treat social presence and users to share information with their friends. Their
media richness as two different dimensions and we distinctions can be summarized as shown in Table 3.
focus on the social presence dimension in this study. These two social media are different in tie symme-
The second dimension is called as self-disclosure / try, tie strength, and network nature. Specifically,
self-representation, which captures the extent to Wechat, like Facebook, requires symmetrical rela-
which the medium requires users to disclose their per- tionships confirmed by both parties for a tie to exist,
sonal information or allows them to represent them- while Weibo permits a tie to exist even if only one
selves (Kaplan and Haenlein, 2010). For example, person initiates it in terms of asymmetrical relation-
users can participate in collaborative projects in Wiki- ships (Kane et al., 2014). In this way, Wechat builds
pedia without disclosing their personal information, a closed space for friends to reciprocally share their
while users need to disclose much personal informa- information, while Weibo develops an open space,
tion in social networking sites (e.g., Facebook). Here, such that anyone who is interested in others’ informa-
Wikipedia and Facebook reflect the media with tion can check their information directly without the
low self-presentation / self-disclosure and high self- others’ permissions. Second, the relationships
presentation / self-disclosure respectively. In our between most Wechat users are based on strong ties
study, because we focus on the self-disclosure tech- such as friendship (Granovetter, 1973) and their rela-
nologies, which require users to disclose their per- tionships in the social media are derived from their
sonal information (Lowry et al., 2011), this offline relationships (Kane et al., 2014). In contrast,
dimension is not applicable in our classification. the relationships between Weibo users are based on

Downloaded from idv.sagepub.com at WUHAN UNIV LIBRARY on September 5, 2015


Wang and Sun: Social influence or personal preference 5

Table 3. Distinctions between low- and high-sociability media.

Wechat Weibo
Dimensions (e.g., high-sociability media) (e.g., low-sociability media)

Tie symmetry Bilateral / Symmetric Unilateral / Asymmetric


Tie strength Strong tie Weak Tie
Network nature Relationship network Information network

weak ties, such that a relationship between users in internalization process (Kelman, 1958). If an individ-
Weibo may not exist in real life (Kane et al., 2014). ual perceives that the values of other individuals are
Third, the network among Wechat users is a relation- congruent with her/his own value system, s/he will
ship network which focuses on reciprocal social inter- internalize others’ opinions, take others’ evaluations
actions, while the network among Weibo users is an as her/his own evaluations, and conduct similar beha-
information network that emphasizes on the sharing viors. Bagozzi and Dholakia (2002) argue that if a
of opinion or information (Hughes et al., 2012). person finds that s/he and other members share com-
Therefore, when interacting with other members, mon values or goals, that person will be more likely to
users will feel a stronger sense of being sociable in commit to a joint action. The relationship between
Wechat than in Weibo, suggesting that it is appropri- group norm and we-intention has also been empiri-
ate to use Wechat and Weibo to represent high- and cally confirmed in the social media context (Cheung
low-sociability social media respectively. et al., 2011). Therefore, we propose that
Considering the distinctions of social media in
sociability and the two perspectives to explain social H1b: Group norm is positively associated with we-
media adoption behavior (e.g., personal preference vs. intention to use social media.
social influence), we develop our research model to
examine the relative influential strengths of social Previous studies have proposed to use social iden-
influence factors and attitude under social media with tity to reflect the identification process of social influ-
different sociability (see Fig. 1). ence (Bagozzi and Lee, 2002). Specifically, social
identity is regarded as a second-order construct with
three components, cognitive, affective and evaluative
Hypotheses social identity (Ellemers et al., 1999). Cognitive
Social influence factors social identity refers to ‘‘a cognitive sense of the self
The impact of subjective norm on we-intention to use as a representative of a social category’’ (Tsai and
social media can be supported by both the social influ- Bagozzi, 2014: 147) through the self-categorization
ence model (Dholakia et al., 2004; Cheung and Lee, process, which is based on judgments on similarities
2010) and the TRA (Fishbein and Ajzen, 1975). The with other group members and dissimilarities with
mechanism underlying the relationship between sub- non-members. Affective social identity refers to an
jective norm and we-intention relies on the compli- emotional involvement with the group or a feeling
ance process, which suggests that individuals would of attachment and belongingness to the group (Tsai
like to comply with the expectations of persons whose and Bagozzi, 2014; Cheung and Lee, 2010). Evalua-
opinions are important to them. Within the social tive social identity refers to the evaluation of self-worth
media context, because users are more exposed to (Ellemers et al., 1999) or group-based self-esteem
other people’s influences through social interactions, (Bagozzi and Lee, 2002) derived from membership.
it is more possible for users to behave under social When an individual has a strong social identity
pressures from others (Cheung and Lee, 2010; toward the social media, s/he will be more likely
Cheung et al., 2011). Therefore, we propose that to commit to a joint action with other members
through identification processes (Tsai and Bagozzi,
H1a: Subjective norm is positively associated with 2014; Cheung and Lee, 2010; Cheung et al., 2011).
we-intention to use social media. Therefore, we propose that

According to the social influence model, group H1c: Social identity is positively associated with we-
norm can exert its impact on we-intention through the intention to use social media.

Downloaded from idv.sagepub.com at WUHAN UNIV LIBRARY on September 5, 2015


6 Information Development

Social Influence

Social
Subjective
Media Type
Norm

H3a-c H4
Personal Preference

Group We -
Attitude
Norm H1a-c Intention H2

Social
Identity

Figure 1. Research model.

Attitude Consistent with Kaplan and Haenlein (2010), we have


The TRA suggests that individual behaviors or beha- proposed sociability as a criterion to classify social
vioral intentions are determined by their evaluations media. For social media with low and high sociability,
of the consequences of behaviors (Fishbein and Ajzen, social influence factors and attitude may play differ-
1975). Although the TRA was originally used to explain ent roles. In general, for high-sociability social media
I-intention, which captures individuals’ personal (e.g., Wechat), in which members keep frequent
actions (Cheung and Lee, 2010), in this study, we pro- social interactions and close social relationships,
pose that we-intention, which describes an individual’s social influence factors may play a more important
intention to commit to a joint action, should be formed role. In contrast, for low-sociability social media
based on his/her own desires to conduct the behavior (e.g., Weibo), where the social dependency among
(Tsai and Bagozzi, 2014). Specifically, drawing upon users is relatively weak, a user’s behavior may be
the use and gratification theory, Cheung et al. (2011) largely determined by his/her personal preference or
have proposed that the fulfillment of a variety of per- attitude. This is consistent with Kaplan and Haenlein
sonal needs can influence individuals’ we-intention to (2010), who argue that the higher the social presence,
use social media. Attitude, which reflects an overall eva- the larger the social influence that the group members
luation of their value expectations, can be taken as a have on each other’s behavior.
proxy of the need fulfillment. Further, in their study Specifically, the influential strength of subjective
on knowledge contribution in virtual communities, Tsai norm, which exerts its influence on we-intention
and Bagozzi (2014) explicitly include attitude as a through the compliance process, depends on the
determinant of we-intention. Within the social media extent to which individuals can be aware of others’
context, we expect that attitude can influence we- expectations and would like to comply with those
intention to use social media too. Thus, we propose that expectations (Kelman, 1958). High-sociability social
media, compared to low-sociability social media,
H2: Attitude is positively associated with we-intention enable users to keep frequent social interactions with
to use social media. other users, facilitating them to learn each other’s
expectations (Nahapiet and Ghoshal, 1998). Further,
because users in high-sociability social media have
Moderating role of social media type developed strong social ties, they are more likely
Although social influence factors and personal prefer- to comply with others’ expectations. Thus, high-
ence factors (e.g., attitude) can both influence users’ sociability social media make users be more exposed
we-intention to use social media, their influential to social influences. In an empirical study, Logan
strengths may vary across different social media. (2014) has found that normative beliefs are significant

Downloaded from idv.sagepub.com at WUHAN UNIV LIBRARY on September 5, 2015


Wang and Sun: Social influence or personal preference 7

for Facebook users, but insignificant for Twitter users, is proposed to have stronger impact for low-sociability
where Facebook and Twitter can be classified as high- social media. For users of low-sociability social media,
and low-sociability social media respectively in terms because the social interactions with other members are
of Kaplan and Haenlein (2010). Thus, we propose that not so frequent and the social ties between members are
relatively weak, they tend to adopt individualistic
H3a: Subjective norm has stronger impact on we- decision making which is ‘‘regulated largely by indi-
intention to use social media for the social media with vidual likes and dislikes and cost-benefit analyses’’
high sociability than those with low sociability. (Triandis et al., 1990: 1007), suggesting that attitude
plays a pivotal role in their decision making. In con-
According to the social influence model, group norm trast, for users of high-sociability social media, their
affects we-intention through the internalization process decision making may heavily rely on other members’
(Kelman, 1958). A key to the internalization process opinions or social pressures, which engender a ‘‘have-
relies on the extent to which an individual shares com- to-do’’ versus ‘‘want-to-do’’ mentality (Tsai and
mon values or goals with other members and the extent Bagozzi, 2014). Therefore, we propose that
to which one would like to accept the shared values and
goals. High-sociability social media, compared to low- H4: Attitude has stronger impact on we-intention to use
sociability social media, enable users to interact and social media for the social media with low sociability
exchange opinions with each other, facilitating the for- than for those with high sociability.
mation of shared values or visions (Nahapiet and
Ghoshal, 1998). Through social interactions, users in
high-sociability social media may develop a more sim- Research method
ilar decision-making pattern or cognitive style (Naha-
piet and Ghoshal, 1998), leading users to easily accept
Research setting and respondents
the shared values. Thus, media sociability can facilitate Two mobile social media applications, Wechat and
the internalization process and strengthen the impact of Weibo, were used in this study to examine the pro-
group norm on we-intention. So we propose that posed research model and hypotheses. Both of these
social media are widely used by Chinese users.
H3b: Group norm has stronger impact on we-intention Wechat, which is provided by Tencent, integrates
to use social media for the social media with high socia- instant messaging services with social network site
bility than for those with low sociability. services, enabling users to keep frequent social inter-
actions with each other. In contrast, Weibo, which is
Social identity can influence we-intention through provided by Sina, is a microblogging service like
the identification process (Kelman, 1958), which Twitter, enabling users to freely share information
relies on individuals’ respect to commitment and through a broadcast mechanism. These two social
attachment to the group. For high-sociability social media can be used to reflect the high- and low-
media, which require more social interactions during sociability social media in terms of their differences
usage, users like to pay more attention to the ongoing in tie symmetry, tie strength and network.
relationships with other members, suggesting stronger University students were treated as the target popula-
impact of social identity (Tsai and Bagozzi, 2014). tion for this study because previous studies had shown
Further, compared to low-sociability social media, that university students are the major users of social
high-sociability social media can enable users to form media (Cheung et al., 2011). According to the Research
a collectivistic rather than individualistic behavioral Report on Chinese Social Media Users (2014) provided
orientation (Nahapiet and Ghoshal, 1998), leading by the China Internet Network Information Center
them to care more about their social identities (Tsai (CNNIC)1, 68.2% of social media users were aged
and Bagozzi, 2014). Therefore, we propose that below 30 and 49.9% of them had the education level
of bachelor or above, suggesting that taking university
H3c: Social identity has stronger impact on we-intention students as the targeted population was appropriate.
to use social media for the social media with high socia-
bility than for those with low sociability.
Data collection
Unlike social influence factors, which play more A convenience sample of Wechat and Weibo users was
important roles for high-sociability social media, attitude created by inviting student volunteers to participate in

Downloaded from idv.sagepub.com at WUHAN UNIV LIBRARY on September 5, 2015


8 Information Development

Table 4. Demographics.

Weibo WeChat

Frequency Percentage Frequency Percentage

Gender Male 56 47.5 69 56.1


Female 62 52.5 54 43.9
Age 18 – 25 117 99.2 119 96.7
Above 25 1 0.8 4 3.3
Education Below bachelor 7 5.9 15 12.2
Bachelor 108 91.5 105 85.4
Master or above 3 2.5 3 2.4
Experience Below 6 months 8 6.8 22 17.9
6 months to 1 year 22 18.6 32 26.0
1 – 2 years 36 30.5 51 41.5
2 – 4 years 51 43.2 18 14.6
Above 4 years 1 0.8 0 0

our study. First, we informed 40 university students simultaneously. Compared with covariance-based
about the survey in class and asked them to participate SEM, PLS is more suitable for exploratory study and
in the survey voluntarily. An invitation message with relatively small samples (Hair et al., 2011; Gefen
the URL of the online questionnaire was posted on et al., 2011). Thus, PLS was appropriate for the cur-
these initial volunteers’ Wechat and Weibo informa- rent study. Specifically, SmartPLS was used as the
tion pages and could be forwarded to other users analytic tool. According to the two-step analytical pro-
through the snow-balling approach. Consistent with cedures (Hair et al., 1998), the measurement model and
prior research (Cheung and Lee, 2010), subjects were the structural model were evaluated respectively.
told to follow an instruction to identify their friends
in the Wechat and Weibo and they needed to refer to
these friends when answering the questionnaire. Measurement model
Finally, we obtained 123 and 118 valid responses for Reliability and validity of the constructs were assessed
Wechat and Weibo respectively. Their demographical in the measurement model. Reliability can be evaluated
profiles were shown in Table 4. by checking the composite reliability (CR) and average
variance extracted (AVE). As shown in Table 5 and
Measures Table 6, the CR values for all the constructs were
above 0.9 and the AVE values for all the constructs
All constructs were measured using multi-item scales
were above 0.8, beyond the suggested threshold values
adapted from previous studies (see Appendix 1). Specif-
0.7 and 0.5 (Fornell and Larcker, 1981), indicating that
ically, the measures for we-intention, subjective norm,
these constructs had good reliabilities.
social norm, and social identity were adapted from
The validity assessment includes convergent and
Cheung and Lee (2010) and Tsai and Bagozzi (2014).
discriminant validity assessments. Specifically, con-
The measures for attitude were adapted from Ajzen
vergent validity and discriminant validity can be eval-
(1991) and Tsai and Bagozzi (2014). Seven-point scales
uated by checking the loadings and cross-loadings. As
were used for all the measures. Since the survey
shown in Table 7 and Table 8, the loadings on the
was conducted in China, all the instruments were
expected constructs were significantly high (above
translated into Chinese, adopting a translation com-
0.8) and the loadings on the expected constructs were
mittee approach (Van de Vijver and Leung, 1997).
higher than the loadings on other constructs, suggest-
ing that the convergent and discriminant validities for
Data analysis all the constructs were good (Fornell and Larcker,
Partial least squares (PLS) approach was used to test 1981). The discriminant validity also can be evaluated
the research model. As a second-generation struc- by comparing the square roots of AVEs and correla-
tural equation modeling (SEM) technique, it can tions. As shown in Table 5 and Table 6, the square
estimate the measurement model and the structural roots of AVEs for all the constructs were greater than

Downloaded from idv.sagepub.com at WUHAN UNIV LIBRARY on September 5, 2015


Wang and Sun: Social influence or personal preference 9

Table 5. Reliability and correlations (Wechat).

CR AVE ASI CSI ESI GN SN UA WI

ASI 0.952 0.908 0.953


CSI 0.910 0.836 0.530 0.914
ESI 0.959 0.921 0.483 0.541 0.960
GN 0.928 0.865 0.459 0.590 0.491 0.930
SN 0.956 0.916 0.477 0.457 0.494 0.673 0.957
UA 0.933 0.823 0.447 0.254 0.184 0.244 0.286 0.907
WI 0.954 0.911 0.504 0.412 0.459 0.445 0.561 0.389 0.954
Note: ASI ¼ Affective social identity, CSI ¼ Cognitive social identity, ESI ¼ Evaluative social identity, GN ¼ Group norm,
SN ¼ Subjective norm, UA ¼ Attitude, WI ¼ We-intention.

Table 6. Reliability and correlations (Weibo).

CR AVE ASI CSI ESI GN SN UA WI

ASI 0.952 0.908 0.953


CSI 0.933 0.875 0.649 0.935
ESI 0.964 0.931 0.657 0.589 0.965
GN 0.933 0.875 0.646 0.720 0.576 0.826
SN 0.945 0.897 0.538 0.630 0.435 0.588 0.947
UA 0.950 0.863 0.420 0.444 0.382 0.527 0.538 0.929
WI 0.975 0.950 0.370 0.398 0.410 0.463 0.506 0.630 0.975
Note: ASI ¼ Affective social identity, CSI ¼ Cognitive social identity, ESI ¼ Evaluative social identity, GN ¼ Group norm,
SN ¼ Subjective norm, UA ¼ Attitude, WI ¼ We-intention.

the correlations, confirming that these constructs had Table 7. Cross loadings (Wechat).
good discriminant validities.
Consistent with prior studies (Cheung and Lee, ASI CSI ESI GN SN UA WI
2010; Tsai and Bagozzi, 2014), social identity was ASI1 0.952 0.504 0.442 0.403 0.439 0.433 0.462
conceptualized as a reflective second-order construct. ASI2 0.954 0.505 0.477 0.472 0.470 0.418 0.498
The loadings for the three first-order constructs were CSI1 0.450 0.908 0.467 0.519 0.430 0.194 0.348
above 0.8 for both Wechat and Weibo samples (see CSI2 0.516 0.921 0.520 0.558 0.407 0.267 0.404
Table 9). ESI1 0.446 0.535 0.959 0.457 0.463 0.157 0.431
ESI2 0.480 0.503 0.960 0.484 0.485 0.195 0.450
GN1 0.426 0.583 0.511 0.939 0.676 0.212 0.439
GN2 0.430 0.510 0.395 0.921 0.570 0.244 0.385
Structural model SN1 0.448 0.423 0.454 0.633 0.963 0.314 0.569
The PLS results for Wechat and Weibo samples are SN2 0.467 0.455 0.493 0.658 0.952 0.229 0.501
shown in Figure 2. UA1 0.413 0.235 0.197 0.221 0.278 0.906 0.361
The results showed that subjective norm had a sig- UA2 0.421 0.173 0.144 0.149 0.185 0.918 0.355
nificant effect on we-intention for Wechat (b ¼ 0.347, UA3 0.381 0.284 0.158 0.295 0.318 0.897 0.344
t ¼ 3.350) but had an insignificant effect on we- WI1 0.431 0.427 0.476 0.453 0.593 0.344 0.959
intention for Weibo (b ¼ 0.171, t ¼ 1.372), lending WI2 0.536 0.357 0.398 0.394 0.473 0.402 0.951
partial support to H1a.
Social identity was found to have a significant effect
on we-intention for Wechat (b ¼ 0.306, t ¼ 2.600) but t ¼ 0.180) and Weibo (b ¼ 0.052, t ¼ 0.315), so H1b
insignificant effect for Weibo (b ¼ 0.084, t ¼ 0.595), was not supported.
suggesting that H1c was partially supported. Attitude was found to have significant effects on
Group norm was found to have insignificant we-intention for both Wechat (b ¼ 0.186, t ¼ 2.277)
effects on we-intention for both Wechat (b ¼ 0.024, and Weibo (b ¼ 0.470, t ¼ 5.746), supporting H2.

Downloaded from idv.sagepub.com at WUHAN UNIV LIBRARY on September 5, 2015


10 Information Development

Table 8. Cross loadings (Weibo). mediated by subjective norm (i.e., internalized nor-
mative beliefs) and/or attitude (i.e., internalized
ASI CSI ESI GN SN UA WI
value beliefs). To confirm this proposition, we con-
ASI1 0.952 0.624 0.595 0.608 0.539 0.416 0.335 ducted mediating effect analysis using Baron and
ASI2 0.954 0.613 0.657 0.623 0.486 0.385 0.370 Kenny’s (1986) method. As shown in Table 11, the
CSI1 0.578 0.935 0.571 0.673 0.562 0.399 0.378 relationship between group norm and we-intention
CSI2 0.636 0.936 0.531 0.673 0.617 0.432 0.367 was found to be fully mediated by attitude for
ESI1 0.635 0.569 0.965 0.545 0.416 0.363 0.415 Weibo, while this relationship was found to be fully
ESI2 0.634 0.567 0.965 0.567 0.422 0.373 0.376 mediated by subjective norm for Wechat. The indi-
GN1 0.665 0.689 0.601 0.925 0.539 0.512 0.398 rect effects of group norm on we-intention through
GN2 0.553 0.661 0.486 0.946 0.560 0.477 0.464
subjective norm and attitude were further analyzed
SN1 0.478 0.585 0.398 0.566 0.941 0.514 0.451
in Table 12 using Sobel’s z-test (Sobel, 1982). The
SN2 0.537 0.608 0.424 0.549 0.953 0.506 0.505
UA1 0.381 0.423 0.347 0.501 0.520 0.938 0.607 results showed that the indirect effect through sub-
UA2 0.401 0.413 0.314 0.502 0.524 0.934 0.589 jective norm was significant for Wechat (ab ¼ 0.232,
UA3 0.389 0.403 0.406 0.463 0.452 0.915 0.557 z ¼ 3.168), but insignificant for Weibo (ab ¼ 0.100,
WI1 0.359 0.370 0.418 0.447 0.511 0.605 0.975 z ¼ 1.352), with significant differences (D ¼ 0.132,
WI2 0.363 0.406 0.381 0.455 0.476 0.622 0.975 t ¼ 6.008). The indirect effect through attitude was
significant for Weibo (ab ¼ 0.249, z ¼ 4.200), but
insignificant for Wechat (ab ¼ 0.046, z ¼ 1.425), with
Table 9. Dimensions of social identity (SI). significant differences (D ¼ 0.203, t ¼ 33.384). These
results indicated that the impacts of group norm on
Wechat Weibo
we-intention were mediated by subjective norm and
Loadings t Loadings t attitude for high- and low-sociability social media
respectively.
Affective social identity .819 17.674 .887 30.850
Cognitive social identity .827 21.254 .852 23.869
Evaluative social identity .825 20.043 .866 26.974 Discussion and conclusion
The paper attempts to examine the different impacts of
Path coefficients were compared using Keil social influence factors and personal preference factors
et al.’s (2000) method (see Appendix 2) to test H3 across social media with different levels of sociability.
and H4. As shown in Table 10, except for group The data analysis results showed that these two types
norm, subjective norm (Db ¼ 0.176, t ¼ 11.902) and of factors explained over 40% of the variance of the
social identity (Db ¼ 0.222, t ¼ 13.276) were found dependent variable (e.g., we-intention) while their rela-
to have stronger impacts on we-intention for Wechat tive influential strengths varied across two social media.
than for Weibo, lending supports to H3a and H3c.
Attitude (Db ¼ 0.284, t ¼ 26.878) was found to have Implications for theory
stronger effects on we-intention for Weibo than for Several implications for theory can be derived from
Wechat, supporting H4. the study. First, this study has proposed to use socia-
bility as a criterion to classify self-disclosure social
media. Based on Kaplan and Haenlein’s (2010) two-
Post-hoc analysis dimension typology of social media, we find that the
As group norm was found to have insignificant effect self-disclosure / presentation dimension is not appli-
on we-intention for both Wechat and Weibo, the cable when focusing on self-disclosure social media,
post-hoc analysis tried to find potential explanations and that social presence and media richness should
for the insignificant effects. Specifically, previous be considered as two different dimensions rather than
studies have proposed that group norm influences one. Further, to better capture the features of social
we-intention through the internalization process media rather than users’ perceptions, we suggest to
(Cheung and Lee, 2010), which requires individuals use sociability rather than social presence as the cri-
to assimilate the group norm and transform it into terion to classify social media. More importantly,
their own normative or value beliefs. Therefore, the we have also found that media sociability can be
impact of group norm on we-intention may be reflected by tie symmetry, tie strength and network

Downloaded from idv.sagepub.com at WUHAN UNIV LIBRARY on September 5, 2015


Wang and Sun: Social influence or personal preference 11

Social Influence

Subjective
Norm
.171 [.347**]

R2=.445 [.427]

Group .052 [-.024] We -


Attitude
Norm Intention
.470** [.186*]

.084 [.306**]
Social
Identity

Figure 2. PLS results.


Note: The numbers out and in the brackets denote the path coefficients for Weibo (low-sociability) and Wechat (high-
sociability) users respectively.
*p < .05, *p < .01.

Table 10. Cross-media comparison.

Weibo (N1¼118) Wechat (N2¼123)

Path coefficient (b1) Path coefficient (b2) D|b| tspooled

GN!WI .052 .024 .028 1.453


SI!WI .084 .306** .222** 13.276
SN!WI .171 .347** .176** 11.902
UA!WI .470** .186* .284** 26.878
Note: SI ¼ Social identity, GN ¼ Group norm, SN ¼ Subjective norm, UA ¼ Attitude, WI ¼ We-intention.
**p < .01, *p < .05.

Table 11. Mediating effect analysis.

(IV, M) ! DV

Media IV M DV IV ! DV IV ! M M! DV IV ! DV Mediating?

Weibo GN SN WI .463** .588** .358** .251* Partial


GN UA WI .463** .528** .536** .178 Full
Wechat GN SN WI .445** .675** .480** .123 Full
GN UA WI .445** .250** .298** .368** Partial
Note: GN ¼ Group norm, SN ¼ Subjective norm, UA ¼ Attitude, WI ¼ We-intention.
**p < .01, *p < .05.

nature. This classification and the underlying logic for Social media users do not passively accept the social
classification can be used in future research. influences exerted by the environment, but actively
Second, we propose an integrated research model balance their personal preferences and social influ-
of social media adoption by considering both the per- ences. Therefore, this study theorizes and empirically
sonal preference and social influence perspectives. examines a research model that takes intentional

Downloaded from idv.sagepub.com at WUHAN UNIV LIBRARY on September 5, 2015


12 Information Development

Table 12. Indirect effect analysis (Sobel’s z test).

Wechat (N¼123) Weibo (N¼118)

IV!M M!DV ab IV!M M!DV ab Dab tspooled

GN!SN!WI .674** .344** .232** .588** .170 .100 .132** 6.008


GN!UA!WI .249* .185* .046 .528** .471** .249** .203** 33.384
Note: ab denotes the product of the path coefficients of IV!M and M!DV. **p < .01, *p < .05.
GN ¼ Group norm, SN ¼ Subjective norm, UA ¼ Attitude, WI ¼ We-intention.

social action as a joint result of social influence and influences. Second, this study reveals that social
personal preference, enriching the social media adop- influence factors and attitude exert different impacts
tion literature. in different social media. Therefore, social media ser-
Third, this is the first study that, to the best of our vice providers should recognize the differences of
knowledge, compares the influential strengths of social media in sociability and execute different stra-
social influence factors and attitude across social tegies accordingly. For example, for low-sociability
media with different sociability. Although prior stud- social media (e.g., Twitter), service providers should
ies have investigated user behaviors under either pay more attention to usability improvement.
high-sociability context (e.g., Facebook) or low-
sociability context (e.g., Twitter) (e.g., Cheung and Limitations and future research
Lee, 2010; Al-Debei et al., 2013; Chen, 2013; Cheung
There are several limitations to be noted when
et al., 2011), there have been no studies empirically
applying the findings in future research. First, as this
comparing the relative influencing strengths of social
study focuses on the relative strengths of social
influence factors and attitude. Specifically, this study
influence and personal preference factors, several
indicates that social influence factors, including sub-
other factors such as perceived behavioral control
jective norm and social identity, have stronger
in the Theory of Planned Behavior (Ajzen, 1991) and
impacts on we-intention for high-sociability media,
personality factors (Correa et al., 2010; Hughes
while attitude has stronger effects on we-intention
et al., 2012) have not been considered in this study.
for low-sociability media. Group norm is found to
Thus, future research can further include other
have no significant direct effect on we-intention for
important factors to examine the robustness of our
both low- and high-sociability media. The post-hoc
study. Second, this study takes two Chinese social
analysis further suggests that the relationship between
media tools (e.g., Wechat and Weibo) as the research
group norm and we-intention is mediated by subjective
settings. Whether or not the conclusions can be
norm for high-sociability media and by attitude for
applied for other social media with different socia-
low-sociability media. These findings advance the the-
bility in other countries should be further tested in
oretical understanding on the antecedents of intentional
future research. Third, consistent with prior studies
social action in social media by identifying the bound-
(Cheung and Lee, 2010), a student sample rather
ary conditions under which each factor applies.
than a professional sample was used in data collec-
tion. This may limit the generalizability of the con-
Implications for practice clusions and require further investigations in
The findings of the study also provide some insights future. Finally, in this study, behavioral intention
to practitioners. First, this study shows that both social rather than actual behavior was taken as the depen-
influence factors and attitude can affect social media dent variable. Previous studies have shown that
users’ intentional social actions. Thus, social media behavioral intention and behavior may not be consis-
service providers should improve the functions or tent (Limayem et al., 2007), leading researchers to
usability of social media to enhance users’ attitudes, consider other important factors such as habit when
as well as set certain mechanisms to strengthen the examining the antecedents of social media adoption
social relationships between users to leverage social behaviors.

Downloaded from idv.sagepub.com at WUHAN UNIV LIBRARY on September 5, 2015


Wang and Sun: Social influence or personal preference 13

Appendix 1. Constructs and measures

Constructs Measures Source

We-Intention (WI) WI1: I intend that our group (i.e., the group that I identified above (Cheung and Lee, 2010;
interacts on [Weibo / Wechat] together sometime during the Tsai and Bagozzi, 2014)
next 2 weeks.
WI2: We (i.e., the group that I identified above) intend to interact on
[Weibo / Wechat] together sometime during the next 2 weeks.
Subjective Norm (SN) SN1: Most people who are important to my life think I should not / (Cheung and Lee, 2010;
should use [Weibo / Wechat] with friends sometime during the Tsai and Bagozzi, 2014)
next 2 weeks.
SN2: Most people who are important to me would disapprove /
approve of me using [Weibo / Wechat] with friends sometime
during the next 2 weeks.
Group Norm (GN) Interacting together sometime within the next 2 weeks with your (Cheung and Lee, 2010;
group in [Weibo / Wechat] can be considered to be a goal. For Tsai and Bagozzi, 2014)
each of the people listed below, please estimate the strength to
which each holds the goal (strong – weak)
GN1: Average of the strength of group members’ goal
GN2: Strength of self’s goal
Social Identity (SI) Cognitive Social Identity (CSI) (Cheung and Lee, 2010;
CSI1: Please indicate to what degree your self-image overlaps with Tsai and Bagozzi, 2014)
the identity of the group of friends as you perceive it.
CSI2: How would you express the degree of overlap between your
personal identity and the identity of the group you mentioned
above when you are actually part of the group and engaging in
group activities?
Affective Social Identity (ASI)
ASI1: How attracted are you to the group you mentioned above?
ASI2: How strong would you say your feelings of belongingness are
toward the group you mentioned above?
Evaluative Social Identity (ESI)
ESI1: I am a valuable member of the group that I mentioned above.
ESI2: I am an important member of the group that I mentioned above.
Attitude (UA) Please express your attitude toward using [Weibo / Wechat] to (Tsai and Bagozzi, 2014;
interact with your friends during the next two weeks: Ajzen, 1991)
UA1: Foolish – Wise
UA2: Bad – Good
UA3: Harmful - Beneficial

Appendix 2. Path coefficients comparison


method (Keil et al., 2000)
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Spooled ¼ f½ðN1  1Þ=ðN1 þ N2  2Þ  SE12 þ ½ðN2  1Þ=ðN1 þ N2  2Þ  SE22 g
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
t ¼ ðPC1  PC2 Þ=½Spooled  ð1=N1 þ 1=N2 Þ
Where
Spooled¼ pooled estimator for the variance
t¼ t-statistic with N1 þ N2  2 degrees of freedom
Ni ¼ Sample size of dataset for group i
SEi¼Standard error of path in structural model of group i
PCi¼Path coefficient in structural model of group i

Downloaded from idv.sagepub.com at WUHAN UNIV LIBRARY on September 5, 2015


14 Information Development

Note Fornell C and Larcker DF (1981) Evaluating structural


1. http://www.cnnic.net.cn/hlwfzyj/hlwxzbg/sqbg/201408/ equation models with unobservable variables and mea-
P020150401351309648557.pdf surement error. Journal of Marketing Research 18:
39–50.
Gefen D, Rigdon EE and Straub D (2011) An update and
References extension to SEM Guidelines for Administrative and
Ajzen I (1991) The theory of planned behavior. Organiza- Social Science Research. MIS Quarterly 35: iii–xiv.
tional Behavior and Human Decision Processes 50: Goldberg LR (1990) An alternative ‘‘Description of Per-
179–211. sonality’’: The Big-Five factor structure. Journal of Per-
Al-Debei MM, Al-Lozi E and Papazafeiropoulou A (2013) sonality and Social Psychology 59: 1216–1229.
Why people keep coming back to Facebook: Explaining Granovetter MS (1973) The strength of weak ties. Ameri-
and predicting continuance participation from an can Journal of Sociology 78: 1360–1380.
extended theory of planned behaviour perspective. Hair JF, Anderson RE, Tatham RL, et al. (1998) Multivariate
Decision Support Systems 55: 43–54. Data Analysis. Upper Saddle River, NJ: Prentice-Hall.
Bagozzi RP and Dholakia UM (2002) Intentional social Hair JF, Ringle CM and Sarstedt M (2011) PLS-SEM:
action in virtual communities. Journal of Interactive Indeed a silver bullet. Journal of Marketing Theory and
Marketing 16: 241–263. Practice 19: 139–152.
Bagozzi RP and Lee K-H (2002) Multiple routes for Hanna R, Rohm A and Crittenden VL (2011) We’re all
social influence: The role of compliance, internaliza- connected: The power of the social media ecosystem.
tion, and social identity. Social Psychology Quarterly Business Horizons 54: 265–273.
65: 226–247. Hughes DJ, Rowe M, Batey M, et al. (2012) A tale of two
Baron RM and Kenny DA (1986) The moderator-mediator sites: Twitter vs. Facebook and the personality predic-
variable distinction in social psychological research: tors of social media usage. Computers in Human Beha-
Conceptual, strategic, and statistical considerations. vior 28: 561–569.
Journal of Personality and Social Psychology 51: Kane GC, Alavi M, Labianca G, et al. (2014) What’s differ-
1173–1182. ent about social media networks? A framework and
Chen R (2013) Member use of social networking sites–an research agenda. MIS Quarterly 38: 275–304.
empirical examination. Decision Support Systems 54: Kaplan AM and Haenlein M (2010) Users of the world,
1219–1227. unite! The challenges and opportunities of social media.
Cheung CM, Chiu P-Y and Lee MK (2011) Online social Business Horizons 53: 59–68.
networks: Why do students use Facebook? Computers Keil M, Tan BCY, Wei K-K, et al. (2000) A cross-cultural
in Human Behavior 27: 1337–1343. study on escalation of commitment behavior in software
Cheung CM and Lee MK (2010) A theoretical model of projects. MIS Quarterly 24: 299–325.
intentional social action in online social networks. Deci- Kelman HC (1958) Compliance, identification, and inter-
sion Support Systems 49: 24–30. nalization: Three processes of attitude change. Journal
Correa T, Hinsley AW and De Zuniga HG (2010) Who of Conflict Resolution 2: 51–60.
interacts on the Web? The intersection of users’ person- Limayem M, Hirt SG and Cheung CMK (2007) How habit
ality and social media use. Computers in Human Beha- limits the predictive power of intention: The case of
vior 26: 247–253. information systems continuance. MIS Quarterly 31:
Daft RL and Lengel RH (1986) Organizational information 705–737.
requirements, media richness, and structural design. Lin K-Y and Lu H-P (2011) Why people use social net-
Management Science 32: 554–571. working sites: An empirical study integrating network
Dholakia UM, Bagozzi RP and Pearo LK (2004) A externalities and motivation theory. Computers in
social influence model of consumer participation in Human Behavior 27: 1152–1161.
network-and small-group-based virtual communities. Logan K (2014) Why isn’t everyone doing it? A comparison
International Journal of Research in Marketing 21: of antecedents to following brands on Twitter and Face-
241–263. book. Journal of Interactive Advertising. Forthcoming.
Ellemers N, Kortekaas P and Ouwerkerk JW (1999) Self- Lowry PB, Cao J and Everard A (2011) Privacy concerns
care categorization, commitment to the group, and versus desire for interpersonal awareness in driving the
group self-esteem as related but distinct aspects of social use of self-disclosure technologies: The case of instant
identity. European Journal of Social Psychology 29: messaging in two cultures. Journal of Management
371–389. Information Systems 27: 163–200.
Fishbein M and Ajzen I (1975) Belief, Attitude, Intention Nahapiet J and Ghoshal S (1998) Social capital, intellectual
and Behavior: An Introduction to Theory and Research. capital, and the organizational advantage. Academy of
Reading, MA: Addison-Wesley. Management Review 23: 242–266.

Downloaded from idv.sagepub.com at WUHAN UNIV LIBRARY on September 5, 2015


Wang and Sun: Social influence or personal preference 15

Ngai EW, Tao SS and Moon KK (2015) Social media into the technology acceptance model. Information
research: Theories, constructs, and conceptual frame- Systems Research 11: 342–365.
works. International Journal of Information Manage- Venkatesh V, Morris M, Davis GB, et al. (2003) User
ment 35: 33–44. acceptance of information technology: Toward a unified
Phang CW, Kankanhalli A and Sabherwal R (2009) Usabil- view. MIS Quarterly 27: 425–478.
ity and sociability in online communities: A compara- Venkatesh V, Thong J and Xu X (2012) Consumer accep-
tive study of knowledge seeking and contribution. tance and use of information technology: Extending the
Journal of the Association for Information Systems 10: Unified Theory of Acceptance and Use of Technology.
721–747. MIS Quarterly 36: 157–178.
Preece J (2001) Sociability and usability in online commu-
nities: Determining and measuring success. Behavior &
Information Technology 20: 347–356.
About the authors
Short J, Williams E and Chistie B (1976) The Social Psy-
chology of Telecommunications. Hoboken, NJ: John Nan Wang is an associate professor of the School of Infor-
Wiley & Sons, Ltd. mation Management, Wuhan University. She obtained her
Sobel ME (1982) Asymptotic confidence intervals for indi- PhD degree from the City University of Hong Kong and
rect effects in structural equation models. In: Leinhardt the University of Science and Technology of China. Her
S (ed.) Sociological Methodology. Washington, DC: research interests include e-commerce and social media. Her
Jossey-Bass Publishers, 290–312.
work has appeared in several international journals including
Sun Y, Wang N, Shen X-L, et al. (2015) Location infor-
Decision Support Systems and International Journal of
mation disclosure in location-based social network
Information Management. Contact: School of Information
services: Privacy calculus, benefit structure, and gen-
der differences. Computers in Human Behavior 52: Management, Wuhan University, Wuchang District, Wuhan,
278–292. Hubei, China. Email: kewang@mail.ustc.edu.cn
Triandis HC, McCusker C and Hui CH (1990) Multimethod
probes of individualism and collectivism. Journal of Yongqiang Sun is an associate professor of the School of
Personality and Social Psychology 59: 1006–1020. Information Management, Wuhan University. He obtained
Tsai H-T and Bagozzi RP (2014) Contribution behavior in his PhD degree from the City University of Hong Kong and
virtual communities: cognitive, emotional and social the University of Science and Technology of China. His
influences. MIS Quarterly 38: 143–163. research interests include e-commerce, knowledge man-
Tuomela R (1995) The Importance of Us: A Philosophical agement, virtual community, and human-computer interac-
Study of Basic Social Notions. Stanford, CA: Stanford tions. His work has appeared in several international journals
University Press,
including Information Systems Research, Journal of AIS,
Van de Vijver FJR and Leung K (1997) Methods and Data
Decision Support Systems, and Information & Management.
Analysis for Cross-Cultural Research. Thousand Oaks,
Contact: School of Information Management, Wuhan Uni-
CA: Sage Publications.
Venkatesh V (2000) Determinants of perceived ease of use: versity, Wuchang District, Wuhan, Hubei, China. Email:
Integrating control, intrinsic motivation, and emotion syq@mail.ustc.edu.cn. Tel.: 86-18672389832.

Downloaded from idv.sagepub.com at WUHAN UNIV LIBRARY on September 5, 2015


View publication stats

You might also like