The formation of social identity and self-identity based on knowledge contribution in virtual communities: An inductive route model

Computers in Human Behavior 43 (2015) 229–241

Contents lists available at ScienceDirect

Computers in Human Behavior journal homepage: www.elsevier.com/locate/comphumbeh

The formation of social identity and self-identity based on knowledge contribution in virtual communities: An inductive route model Zhi-chao Cheng, Tian-chao Guo ⇑ School of Economics & Management, Beihang University, No. 37, Xueyuan Road, Haidian District, Beijing City 100191, China

a r t i c l e

i n f o

Article history:

Keywords: Knowledge contribution Social interaction tie Social identity Self-identity Self-esteem Virtual communities

a b s t r a c t Based on the social network perspective and work group perspective, this study brings social interaction tie and membership esteem together as the mediating variables between knowledge contribution and social identity to construct an inductive route model, aiming to understand how social identity and self-identity form based on knowledge contribution behaviors in virtual communities. To assess the theoretical model, an online survey was conducted in an interest-based discussion community, Baidu Post Bar (China), and yielded 348 useable responses. Both social interaction tie and membership esteem were found to have mediating effects between knowledge contribution and social identity. In addition, knowledge contribution was found to have a direct influence on social identity. The results also showed that self-identity can form through an inductive route. Our findings have implications for both practice and theory. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction In recent years, virtual communities have attracted much attention from researchers and practitioners. Following Chou, Min, Chang, and Lin (2010), this study focuses on an interest-based virtual community (VC) in which a group of people share their opinions, insights, perspectives and experiences with each other, develop relationships, and collectively seek to attain goals through computer-mediated communication as a means of information exchange (Lee, Vogel, & Limayem, 2002). This type of VC is also called an online forum, bulletin board, or (electronic) discussion group. User-generated content (i.e., knowledge) has long been recognized as a vital factor for VCs’ survival and success (Shiue, Chiu, & Chang, 2010); therefore, many studies have focused on the motivation of knowledge contribution in virtual communities. The literature on knowledge contribution shows that a variety of factors affect this behavior, including personal factors (personality traits, performance expectancy, sense of self-worth, reputation, altruism, self-efficacy, professional experience; Bock, Zmud, Kim, & Lee, 2005; Mooradian, Renzl, & Matzler, 2006; Wang & Lai, 2006) and social factors (social capital, social presence, sense of belonging, social identity, online relationship commitment; Ma & Yuen, 2011; Shen, Yu, & Mohamed, 2010; Wasko & Faraj, 2005; Zhao, Lu, Wang, Chau, & Zhang, 2012). Studies have also indicated that ⇑ Corresponding author. Tel.: +86 15010710697. E-mail address: [email protected] (T.-c. Guo). http://dx.doi.org/10.1016/j.chb.2014.10.056 0747-5632/Ó 2014 Elsevier Ltd. All rights reserved.

recognition from the site (Jeppesen & Frederiksen, 2006) and outcome expectancy (Chiu, Hsu, & Wang, 2006) play key roles in an individual’s willingness to contribute knowledge. However, little research has revealed the mechanism that underlies the influence of knowledge contribution on the effective running of VCs. There is a self-running mechanism in VCs based on knowledge contribution. On the one hand, knowledge contribution can lead to interaction between members, which contributes to the formation of social identity and self-identity (Postmes, Spears, Lee, & Novak, 2005; Stryker & Vryan, 2006). On the other hand, social identity and self-identity play key roles in VCs’ development (Shen et al., 2010). When social identity is significant, individuals are assimilated to a group-specific prototype, and group-specific thoughts and behaviors become the individuals’ own thoughts and behaviors; thus, the individuals will work hard to help achieve group goals (Fielding & Hogg, 2000). Therefore, social identity contributes to members’ participation in and loyalty to VCs (Dholakiaa, Bagozzia, & Pearob, 2004; Lin, 2008). Meanwhile, self-identity is associated with a relevant social role or in-group role, forming a set of identity standards that guide identity-relevant behaviors (Stets & Burke, 2000). Therefore, there is a strong relationship between self-identity and role-relevant behavior intention (Jostein, Paschal, & Silje, 2010), and thus, self-identity as a contributor may predict members’ contribution behaviors in VCs. The self-running mechanism of VCs involves knowledge contribution leading to the formation of social identity and self-identity, which can reversely facilitate members’ sustained participation

230

Z.-c. Cheng, T.-c. Guo / Computers in Human Behavior 43 (2015) 229–241

and knowledge contribution in VCs. The process through which social identity and self-identity form based on knowledge contribution is the core of the self-running mechanism. However, there has been little research that has examined the process of formation of social identity and self-identity based on knowledge contribution behaviors in VCs. A few studies have argued that based on contribution behaviors, individuals can form social identities through an inductive path. However, these studies have not fully revealed the mechanism that underlies the inductive path or addressed how and what compositions constitute the inductive formation path. This study constructs an inductive route model to identify the mechanism of the inductive social identity formation path. Meanwhile, the inductive route model may also predict the formation of self-identity.

2. Theoretical background

tional connection with a group (Meyer & Allen, 1991). The small differences between these two conceptions may be that social identification accentuates both task involvement and an emotional connection with the group, whereas affective commitment more accentuates an emotional connection with the group (Ashforth et al., 2008; Dávila & Jiménez, 2012). For example, in VCs, both lurkers and contributors may feel an emotional connection with the group, but lurkers cannot feel task involvement in the group (Blanchard & Markus, 2004). Because of involving emotional attachment to and involvement in a group, social identification is more enduring and long term than self-categorization (Ashforth et al., 2008; Postmes, Haslam, et al., 2005). The items that measure social identification accentuate a sense of belonging rather than similarity: ‘‘I see myself as a member of this group’’, ‘‘I have a strong sense of belonging to this group’’, ‘‘I feel connected to this group’’ (Chiu et al., 2006; Postmes, Spears, & Lea, 1999; Swaab, Postmes, van Beest, & Spears, 2007), and so forth.

2.1. The two forms of social identity

2.2. The two formation paths of social identity

Tajfel (1972) first introduced the concept of social identity as ‘‘the individual’s knowledge that he belongs to certain social groups together with some emotional and value significance to him of this group membership’’. On the basis of this definition, Ellemers, Kortekaas, and Ouwerkerk (1999) and Bergami and Bagozzi (2000) proposed that social identity consists of three dimensions: a cognitive dimension (the cognitive assimilation of the self to the group prototype—self-categorization), an evaluative dimension (a positive or negative evaluation attached to the group membership—group self-esteem), and an emotional dimension (a sense of affective connection with the group—affective commitment). Although the above dimensions are widely recognized, social identity formation studies often adopt unidimensional social identity, and the connotation of social identity is inconsistent across these studies (Ashforth, Harrison, & Corley, 2008); thus, two research trends have emerged. In one research trend, social identity refers to self-categorization, which represents a response to the immediate perceptual environment, whereby individuals define themselves based on the degree to which they are similar to or different from others in their surroundings (Deaux & Martin, 2003). Social categorization of the self cognitively assimilates the self to the in-group prototype and, thus, depersonalizes self-conception and highlights the similarity between members (Brown, 2000; Tajfel, 1981; Turner, Hogg, Oakes, Reicher, & Wetherall, 1987). Self-categorization arises based on the situational significance of the group features (i.e., race, gender, visual similarities, and so forth) and thus tends to be more situationally and contextually determined (Postmes, Haslam, & Swaab, 2005). The items that measure self-categorization are ‘‘the extent of overlap between my image and the group image’’, ‘‘I am similar with others in the group’’ (Foels, 2006; Kim & Park, 2011; Lee, 2004), and so forth. In the other research trend, social identity refers to social identification, which may emerge based on individuals’ contribution behaviors, social interactions and social relationships in social networks or organizations (Postmes, Haslam, et al., 2005; Rink & Ellemers, 2007; Van Dick, 2001). Social identification accentuates the sense of belonging that forms on the basis of the individuals’ perception and acceptance of the shared task and goal (Wegge & Haslam, 2003); therefore, social identification can be conceptualized as ‘‘the experience of personal involvement in a group so that persons feel themselves to be an integral part of that group along with the emotional significance of this identity’’ (Ellemers, De Gilder, & Haslam, 2004; Hagerty, Lynch-Sauer, Patusky, Bouwsema, & Collier, 1992; Rink & Ellemers, 2007). This conception is quite similar to affective commitment that describes the emo-

The reason for the emergence of the two social identity forms may be that there are two distinct formation paths of social identity: a deductive path and an inductive path. Through the deductive path, individuals can assimilate themselves to a social category, whereas through the inductive path, individuals may integrate themselves into a social structure (Postmes, Spears, et al., 2005); that is, self-categorization emerges through the deductive path, whereas social identification forms through the inductive path. The deductive path is a top-down process through which superordinate categories can shape a social identity. Group members may form a social identity based on the shared properties that differentiate their in-group from other groups (Postmes, Spears, et al., 2005). This property may be a feature (e.g., skin color, religion), a common interest, or other related factors (such as some form of entitativity or essence; see also Lickel, Hamilton, & Sherman, 2001). It should be noted that it is not the case that the group members need to like each other or identify their similarities as individuals. Rather, they identify and share a certain common feature that is given meaning at a super-individual level and in the intergroup environment (Postmes, Spears, et al., 2005). Based on these common features, individuals can deduce group properties to construct an internalized social identity composed of stereotypes and norms. This is the deductive social identity known as self-categorization in this study. The inductive path is a bottom-up process through which social identity can be shaped based on individual contribution behaviors (Jans, Postmes, & Van der Zee, 2012). Contribution behaviors may lead to interaction and communication between individuals (Postmes, Spears, et al., 2005). It is through interaction and community that individuals can perceive the shared task and goal between group members, which can characterize the shared identity (Wegge & Haslam, 2003). There have been a few studies on the inductive formation path of social identity, and inconsistencies can be found across these studies. On the one hand, Jans et al. (2012) argued that individual contributions of group members may contribute to the formation of a social identity. They experimentally examined the influence of diversity on the formation of social identity and indicated that the distinctiveness may be integrated as the essential property and thus as the shared cognitive representation of the group. They also suggested that the inductive path is a process through which individuals make active contributions to the emergence of a shared identity simply because they have an opportunity (or ‘‘voice’’). On the other hand, in another study, Postmes, Spears, et al. (2005) argued that the formation of an inductive identity does not

Z.-c. Cheng, T.-c. Guo / Computers in Human Behavior 43 (2015) 229–241

necessarily depend on the existence of differences within the group (Postmes, Spears, et al., 2005). An inductive identity may be derived more from a sense of shared tasks, social interactions and social relationships, all of which may emerge on the basis of individuals’ contribution behaviors (Postmes, Haslam, et al., 2005; Postmes, Spears, et al., 2005). These inconsistencies may be due to different research purposes and contexts. In an experimental setting, social interactions and social influences may not exist or have an effect, but in a real environment such as VCs, which can be seen as a social network, social interaction and social influence are significant and may contribute to the formation of social identity (Postmes, Haslam, et al., 2005). No research has empirically examined and revealed the internal mechanism of the inductive formation path of social identity in real environments, including virtual communities. The present study intends to fill these gaps. Because this study focuses on an inductive route model on the basis of knowledge contribution behaviors, in this study, social identity refers to social identification rather than self-categorization. Because people in VCs are anonymous and physically separated from each other, an individual’s social identification within VCs plays an important role in his/her integration into and contribution to the VC (Chiu et al., 2006; Huemer, Becerra, & Lunnan, 2004).

3. Research model and hypotheses 3.1. Inductive route model This study intends to construct an inductive route model that describes the mechanism that underlies the inductive formation path, in other words, to answer how and what compositions constitute the inductive formation path. The most important characteristic of the inductive formation path is the bottom-up process through which individuals can perceive their shared cognition, but a shared cognition may forms not necessarily depending on different task contributions by group members (Postmes, Spears, et al., 2005). Therefore, there may also be moderating variables between knowledge contribution and social identity. On the one hand, a virtual community, as a social network, consists of community members and the ties among them (Shiue et al., 2010). Wasserman and Faust (1994) suggested that a virtual community can be seen as a social network and be characterized by social interaction ties. Wellman and Wortley (1990) argued knowledge sharing is positively associated with social interaction ties. VCs build social relationships among individuals and then facilitate trust and deeper intimacy among them (Ren et al., 2012). Jones and Volpe (2011) indicated there appears to be a clear connection between social interaction ties and identification with groups. On the other hand, the virtual community that runs based on usergenerated content can be seen as a work group (Pollack, 1996). Knowledge contribution is also a task behavior through which individuals can gain a sense of achievement; this is also known as membership esteem (Luhtanen & Crocker, 1992). Based on these two perspectives, we think that both the social interaction tie and membership esteem may be the moderating variables between knowledge contribution and social identity. Meanwhile, knowledge contribution may also have a direct effect on social identity. Therefore, there may be three paths that lead to social identity and constitute the inductive route model. In addition, a few studies argue that self-identity can be perceived through interaction between group members (Ashforth, 2001; Stryker & Serpe, 1982); this is consistent with the inductive formation path of social identity. Therefore, this study also examines whether self-identity can form through the inductive route.

231

Based on the above analysis, we developed our research model—an inductive route model—as is depicted in Fig. 1. The justification for the relationships proposed and corresponding hypotheses will be discussed as follows.

3.2. Social interaction tie and the social network perspective The concept of ‘‘social interaction tie’’ comes from the theory of social capital. Social capital has been defined as ‘‘the sum of the actual and potential resources embedded within, available through, and derived from the network of relationships possessed by an individual or social unit’’ (Nahapiet & Ghoshal, 1998). Nahapiet and Ghoshal (1998) proposed that social capital consists of three distinct dimensions: structural, relational, and cognitive. The structural dimension refers to the impersonal configuration of linkages between members in a social network and the extent to which the members are connected with each other (Chiu et al., 2006). This dimension is manifested as a social interaction tie. The relational dimension describes the nature of the connections between members. This dimension is manifested as interpersonal relations, such as trust and reciprocity. The cognitive dimension expresses the extent to which members share a common representation, interpretations and understanding. The critical resources of this dimension are shared visions, goals, and languages. Studies on social capital focus on the influence of knowledge sharing and innovation in either the traditional or the virtual environment (Tsai & Ghoshal 1998; Yli-Renko, Autio, & Sapienza, 2001). However, in VCs, such as in an interest-based online discussion forum, participants interact with each other, aiming to build interpersonal relations and a sense of belonging. Therefore, social capital can be regarded as an important resource that reflects the features of social relations in a social system. Ainhoa (2007) suggested that social identity can form based on social relations. Social capital theory primarily aims to identify how the social structure of an entity serves as a resource that elicits rich output (Coleman, 1988); therefore, the structural dimension—social interaction tie—is the core of social capital (Granovetter, 1985). In addition, in VCs, members have contact with each other only through online interactions, and the relational and cognitive dimensions of social capital are developed mainly based on social interaction tie (Chiu et al., 2006; Hsiao & Chiou, 2012). For these reasons, we use social identity ties rather than social capital as the mediating variable between knowledge contribution and social identity. This is consistent with Hsiao and Chiou (2012), who also used social interaction tie rather than social capital to study the influence of the former on VC loyalty. Knowledge contribution, social interaction tie and social identity together comprise the social network perspective of the inductive path. Interest-based VCs have defined topics and attract participants to develop an interest in them (Stanoevska-Slabeva, 2002). For example, Baidu Post Bar (www.tieba.baidu.com) is the most popular Chinese-speaking community wherein various types of interest groups discuss defined topics such as history, music, film stars, and so forth. In interest-based communities, discussions emerge based on a new viewpoint or thought that was initially posted by an individual. Hence, knowledge contribution is the important determinant of the interaction between group members (Carmel, Roitman, & Yom-Tov, 2012; Schwämmlein & Wodzicki, 2012). Baumeister and Leary (1995) suggested that people innately have a need for belonging. He defined the belonging need as ‘‘a need to form and maintain social relations with others’’. People who frequently post within virtual communities can often receive replies from others. To satisfy their need for belonging, they may positively correspond to and communicate with the repliers, and thus,

232

Z.-c. Cheng, T.-c. Guo / Computers in Human Behavior 43 (2015) 229–241

Social interaction tie H7 H2

H1 H4 Knowledge contribution

H9

Social identity

H6

H3

Selfidentity

H8

H5

Membership esteem Fig. 1. Research model.

social interaction ties may be shaped (Ma & Yuen, 2011). Therefore, we hypothesize the following: H1. A member’s knowledge contribution positively affects his/her social interaction tie in the VC. On the one hand, the social interaction tie can contribute to the group members’ perception of the shared task and goal and result in a shared social identity (McLeish & Oxoby, 2011). Social interaction ties are an efficient channel for the flowing and exchanging of information between group members (Ashforth et al., 2008). Through the exchange of viewpoints on topics in which group members are commonly interested, task interdependencies can form between the group members and contribute to their perception of the shared task and goal, which may serve as the common cognition of the group (Ashforth et al., 2008). An interdependent task and common purpose focus motivations in the community as a whole and induce feelings of common identity in online communities (Ren, Kraut, & Kiesler, 2007). On the other hand, social interaction ties cultivate the social attachment to a group and interpersonal attachment. Attachment refers to group members’ affective connection to and caring for a virtual community in which they become involved (Ren et al., 2012). First, a facet of social interaction ties is members’ familiarity with the group, which increases the liking of the group. Zajonc (1968) and Milgram (1977) demonstrated a ‘‘mere exposure effect’’ whereby the more familiar one is with objects, symbols, or people, the more one likes them. In virtual communities, social ties make a group and its activities repeatedly visible to a member, which may increase the member’s liking for and thus attachment to the group (Ren et al., 2012). Second, social interaction ties are a major determinant of the extent to which group members build relationships with each other (McKenna, Green, & Gleason, 2002). More exchanges among group members, for example, through social expression, provide opportunities for group members to build interpersonal attachment and create both liking and trust (Chua & Balkunje, 2013). Meanwhile, based on a common interest, interactions between members can cultivate members’ cognition of the shared task and goal, which increases the perception of similarity between each other. Similarity can contribute to interpersonal attraction (Ren et al., 2007). Therefore, social interaction tie contribute to the intimate relationship between individuals. Moreover, because all individuals are anonymous and visual information about the individuals is not available, members’ impressions of their intimate co-workers can be extended to the group as a whole, and thus, the attachment to intimate co-workers can be extended

to the group as a whole. Individuals in VCs gain greater satisfaction from their social attachments (Whitton & Kuryluk, 2012) and frequent social interactions (Kline & Stafford, 2004). This indicates that social attachment and social interaction tie may be the similarly important antecedents of feelings of belonging in the context of network groups (Easterbrook & Vignoles, 2013; Jones & Volpe, 2011). Based on the arguments above, we hypothesize the following: H2. A member’s social interaction tie positively affects his/her social identity with the VC. 3.3. Membership esteem and the work group perspective According to social identity theory, it is assumed that people strive to enhance their positive feelings about their own self-conception and self-esteem through belonging to a group (Ellemers et al., 1999; Tajfel & Turner, 1986). Self-esteem can be classified as personal and collective (Foels, 2006). Personal self-esteem refers to an individual’s feelings of self-worth based on his own traits and abilities, whereas collective self-esteem describes a group member’s feelings of self-worth based on membership (Foels, 2006; Luhtanen & Crocker, 1992). Collective self-esteem can be assessed by four dimensions—private collective self-esteem, public collective self-esteem, membership esteem, and importance to identity—that reflect different facets of collective self-esteem (Crocker, Luhtanen, Blaine, & Brodnax, 1994; Luhtanen & Crocker, 1992). Membership esteem refers to the degree to which a group member believes him/herself to be worthy of and important to a social group (Luhtanen & Crocker, 1992). Only this dimension of collective self-esteem was obtained from members’ contribution to the group and thus is more suitable for our study. Membership esteem is also known as organization-based self-esteem in a work group or organization context (Pierce, Gardner, Cummings, & Dunham, 1989). Scholars have suggested that individuals form a self-concept around work and that their work experiences play an important role in fostering their self-esteem (Pierce & Gardner, 2004). On the one hand, in task- and goal-oriented communities, an individual tends to attain extrinsic benefits such as task achievement based on his/her contribution behaviors (Chou et al., 2010). Knowledge contribution has been ranked as one of the top factors for a VC’s success (Leimeister, Sidiras, & Kremar, 2004), and thus, a member’s knowledge contribution behaviors can give him a sense of self-worth. On the other hand, members’ contribution behaviors in VCs are autonomous. Studies have shown positive relationships between autonomy and membership esteem (McAllister & Bigley,

Z.-c. Cheng, T.-c. Guo / Computers in Human Behavior 43 (2015) 229–241

2002; Vecchio, 2000). Based on these reviews, we hypothesize the following: H3. A member’s knowledge contribution positively affects his/her membership esteem. Posts with high quality can attract more replies, such as comments and compliments from others; thus, the quantity of the replies may be regarded as a sign of personal achievement. Social interaction ties represent more replies. Based on many replies, people can feel good and improve their self-esteem (Mallan & Giardina, 2009). Pierce and Gardner (2004) suggested that task/ work achievement can enhance the individual’s membership esteem. Therefore, we hypothesize the following: H4. A member’s social interaction tie is positively related to his/ her membership esteem. Ainhoa (2007) indicated that there are two dimensions that underlie the process of social identification: a cognitive dimension and an evaluative dimension. They argued that individuals seek to obtain a positive self-evaluation, which may results from belonging to a social group, and the evaluations associated with the social group (Ainhoa, 2007). De Cremer and Oosterwegel (2000) suggested that people with high collective self-esteem are more willing to accept their esteemed social identity, and thus, there is a positive relationship between collective self-esteem and in-group identity. Membership esteem depends on members’ contribution and participation, so membership esteem may contribute to members’ sense of belonging (Jans et al., 2012). Based on membership esteem, members can gain high satisfaction, which may lead to members’ sense of evolvement and social identification (Easterbrook & Vignoles, 2013; Pierce, Gardner, Dunham, & Cummings, 1993). Rousseau (1998) indicated that members who are important for their group may be characterized by deep social identity. Based on these reviews, we hypothesize the following: H5. A person’s membership esteem positively affects his/her social identity with the VC. 3.4. Knowledge contribution and social identity In addition to the above two perspectives of the inductive formation path, people’s individual task contributions may also shape their social identity (Jans et al., 2012). A few studies have found that in VCs, people may express various opinions rather than conform to others in the VC (Kim, 2006; Lee, 2004). People share their lives, comments, opinions, and deeply felt emotions in VCs (Nardi, Schiano, Gumbrecht, & Swartz, 2004); these contents tend to be personal. To attract readers, people often embed novelty in their posts; hence, the contents that people present in VCs are more or less diverse. Jans et al. (2012) experimentally examined the influence of sharing differences on the formation of social identity and indicated that distinctiveness may be integrated as the essential property of groups; thus, social identity may form on the basis of individual contributions. Based on these conclusions, we hypothesize the following: H6. A member’s knowledge contribution positively affects his/her social identity with the VC. 3.5. Self-identity The conception of self-identity, which is also called role identity, comes from identity theory (Stryker, 1980, 1982, 1987), which pro-

233

poses that an individual’s self-concept is defined in terms of the special roles that a person occupies in the social structure or identifies with. Central to identity theory is the link between self-identity and role-related behavioral intentions (Terry, Hogg, & White, 1999). According to identity theory, people categorize themselves in specific social roles, and these roles then guide people’s intentions and behaviors so that people spontaneously act in accordance with their self-identities (De Bruijn, Verkooijen, de Vries, & van den Putte, 2012; Hagger & Chatzisarantis, 2006). The link between selfidentity and behavioral intention has been confirmed by many studies. Hogg, Terry, and White (1995) suggested there is an important difference between self-identity and social identity. Self-identity focuses on role identities, such as social roles (e.g., mother, father), social types (e.g., motorcyclist, exerciser, green consumer), group roles (e.g., manager, employer, employee), whereas social identity focuses on membership inherent in collectives, such as groups and organizations. Social identity accentuates belonging to a group. In virtual communities, roles can be classified, such as moderator, contributor, lurker (Stanoevska-Slabeva, 2002). Although the role of contributor is very important for virtual communities, it is not imposed. A few studies have argued that self-identity can be perceived through interaction and communication between group members; this is inconsistent with the inductive formation path of social identity. We consider that self-identity can also form through inductive path; i.e., the inductive formation path of social identity can also predict the formation of self-identity. Identity theory is originated in sociological tradition that focuses on properties of social structures in which people build their ties to others (Stryker, 1982). Research has suggested that identity forms in the course of social interactions that take place within a social structure (Ashforth, 2001; Stryker, 1982). Meanwhile, according to the symbolic interactionism, the meaning of roles is socially formed based on social processes such as interaction, observation, feedback, and so forth (Deaux & Martin, 2003; Sluss & Ashforth, 2007). Therefore, through frequent social interaction, members may perceive the roles they play in groups. Social interactions play an essential role in the survival and success of a VC and thus can be seen as contribution behaviors, which can serve as the indicator of a role and contribute to members’ perception of their self-identity as a contributor (Ashforth, 2001). Therefore, we hypothesize the following: H7. A member’s social interaction tie positively affects his/her self-identity as a contributor in the VC. Weick (1995) argued that people perceive their identities by projecting themselves into a social context and observing the consequences. Ashforth (2001) suggested task behaviors and task achievement, such as the quantity and quality of output, may serve as the observable indicators of identity, thus, can contribute to people’s perception of their self-identity as a specific role. Membership esteem come from task achievement and thus can serve as a strong indicator of self-identity as a specific role. Therefore, we hypothesize the following: H8. A person’s membership esteem positively affects his/her selfidentity as a contributor in the VC. On the one hand, in VCs, social identity refers to members’ behavior involvement and the acceptance of group norms, which encourage members to contribute more (Shen et al., 2010). Therefore, members with high social identity with the VC tend to make more contributions to the community and are keenly aware of their roles as a contributor. Social identity can strengthen members’ affective commitment to the group (Hagerty et al., 1992);

234

Z.-c. Cheng, T.-c. Guo / Computers in Human Behavior 43 (2015) 229–241

hence, members with high social identity may be more willing to make contribution and accept their roles as a contributor. Therefore, we hypothesize the following: H9. A member’s social identity in the VC is positively related to her/his self-identity as a contributor in the VC.

Table 1 Demographic information. Variables

Items

%

Gender

Male Female

76.7 23.3

Identity

Middle school student High school student Undergraduate Postgraduate Worker

6.6 17.5 46.0 3.2 26.7

Age

<15 16–20 21–25 26–30 31–40 >40

5.2 37.4 39.9 8.9 6.9 1.7

Membership history

Less than six months About six months About one year About two years About four years More than four years

19.0 17.2 20.1 26.7 5.7 11.2

4. Method 4.1. Sample and collection To test these hypotheses, we conducted an online survey at Baidu Post Bar (www.tieba.baidu.com), which was launched on the Web in 2005 and is now the most popular online community in China. Baidu Post Bar can be categorized as an interest VC in which members spontaneously get together and discuss a common interest. There are over two hundred thousand1 active groups and subgroups in Baidu Post Bar, such as the Chinese history group, the pop music group, the plant group, the football group, the NBA group, the Michael Jordan group, and so forth. One of these groups may be the largest domain forum in China, as the NBA group is possibly the largest online basketball forum in China. According to a news report, the traffic spikes of Baidu Post Bar have exceeded one billion1. In addition, according to the website statistics2 from December 2013, there are over one million members in the NBA group, and per day, more than one hundred thousand members click the online ‘‘sign in’’ button in the group. Baidu Post Bar is the most comprehensive and popular online interest community in China; thus, the data collected from it can ensure the generalizability of the findings. We obtained the permission of the moderators of Baidu Post Bar to post a link to our online questionnaire. The collection of data lasted one month, and a total of 348 valid questionnaires were collected and used in the data analysis. Table 1 lists the demographic statistics of the respondents. The members’ personal data are available in their Baidu Space, so we collected a new random demographic sample. To examine the possibility of non-response bias, we compared our study sample with this new sample in terms of gender, age, identity and tenure of membership. There was no significant difference between these two samples, which suggests that our study sample is representative. It should be noted that the gender ratio reflects a difference in community participation rather than community enrollment between the genders. 4.2. Measures To assure the validation of the instrument, survey items were mostly adapted from scales developed and validated by previous studies. Three items measured knowledge contribution. Two items were adapted from Igbaria, Parasuraman, and Baroudi (1996) and Kim, Zheng, and Gupta (2011): ‘‘I contribute my knowledge often to others in the group which I joined (in Baidu Post Bar)’’ and ‘‘I post my knowledge often in this group’’. Another item was adapted from Chiu et al. (2006): ‘‘How many knowledge posts do you create per month in this group?’’ The interaction tie refers to the strength of online social relations based on social interaction and was measured with three items adapted from Chiu et al. (2006): ‘‘I maintain close relationships with some members in this group’’, ‘‘I have frequent communication with some members in this group’’ and ‘‘I know some members in this group on a personal level’’. Member1 The data are from the Phoenix, a well-known news website in Hong Kong (see http://finance.ifeng.com/roll/20100928/2660578.shtml). 2 The data are available at the Baidu Post Bar website (see http://tieba.baidu.com/ sign/index?kw=nba&type=3&pn=#current_forum).

ship esteem was operationalized with two items adapted from Bagozzi and Dholakia (2002) and Luhtanen and Crocker (1992): ‘‘I am a valuable member of this group’’ and ‘‘I am important to this group’’. Social identity was measured with items adapted from Chiu et al. (2006) and Swaab et al. (2007), which are the same as those in Section 2.2. Finally, self-identity was measured with three items adapted from Yun and Silk (2011) and Smith et al. (2007): ‘‘I think of myself as a contributor in this group’’, ‘‘I hardly make any contributions to this group’’, and ‘‘I think of myself as a member who is concerned with making contributions to this group’’. Except for the question ‘‘How many knowledge posts do you create per month in this group?’’ all items were measured using five-point Likert scales ranging from ‘strongly disagree’ to ‘strongly agree’. The item ‘‘How many knowledge posts do you create per month in this group?’’ is answered with the average number of knowledge contributions per month. To normalize the data, however, we transformed the average volume of knowledge contribution per month to a five-point scale where 1 = less than once per month, 2 = approximately 5 times per month, 3 = approximately 10 times per month, 4 = approximately 20 times per month, and 5 = more than 40 times per month (Chiu et al., 2006). This ordered categorical scale was tested and adjusted in advance to make it correspond to and consistent with the measurements of the two other items of knowledge contribution. All items were then translated into Chinese by the authors. To ensure the accuracy of translation, the questionnaire was verified and refined by one behavior science professor and three senior doctoral students, who were quite familiar with and had conducted in-depth research on online behavior. The questionnaires were then pretested by students who often participated in Baidu Post Bar to ensure that the questionnaire items’ wordings and the translation had logical consistency and contextual relevance and were comprehensible (Bock et al., 2005). 4.3. Data analysis Structural equation modeling was used to test the models shown in Fig. 1. The AMOS 7.0 software was employed for this purpose. We first tested the reliability and validity of the measurement model with a confirmatory factor analysis (CFA). After the reliability and validity were established, we examined the research model. Finally, we also tested the mediation effect using the Mplus 7 software.

235

Z.-c. Cheng, T.-c. Guo / Computers in Human Behavior 43 (2015) 229–241 Table 3 Measurement loadings.

5. Results 5.1. Measurement model Because the three items that measured knowledge contribution used different measuring methods, we conducted a principal component factor analysis to test whether one factor could be extracted from these three items. Bartlett’s test of sphericity showed that the Kaiser–Meyer–Olkin (KMO) statistic of 0.704 was significant at a level of 0.001, indicating the appropriateness of using a principle component factor analysis on the data. Only one factor, which explained 72.8% of the variance, was extracted; item loadings were 0.871, 0.890 and 0.796, all above the required threshold of 0.5. The construct reliability and validity were examined by CFA. First, the construct reliability was assessed with Cronbach’s alphas and composite reliability (Nunnally, 1978). As shown in Table 2, the Cronbach’s alpha values ranged from 0.812 to 0.904, and composite reliability ranged from 0.820 to 0.903, indicating an adequate reliability (Hatcher, 1994). Convergent validity was assessed by average variance extracted (AVE) from the constructs and by checking the loadings, as shown in Table 2. The average variance extracted (AVE) from every construct was greater than 0.5, which suggests good convergent validities of the constructs (Baggozi & Yi, 1988). In addition, Table 3 reports the loadings of the items in our research model. As expected, all item loadings are significantly higher than 0.5 (Teo & King, 1996). Finally, the discriminant validity of our constructs was examined by comparing the square roots of AVE for the individual constructs to the shared variances between constructs (Fornell & Larcker, 1981). As shown in Table 4, the square roots of the AVE for the individual constructs, the diagonal elements, are all greater than their corresponding correlation coefficients with other constructs, confirming discriminate validities. Data collected via a single self-report are susceptible to common method variance (CMV), i.e., ‘‘the variance that is attributable to the measurement method rather than the constructs that the measures represent’’. To avoid CMV, we first conducted a principal component factor analysis, and no general factor was found to exceed the acceptable range (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Then, we used a one-factor model approach to test CMV (Harris & Mossholder, 1996; Liang, Saraf, Hu, & Xue, 2007). We connected all items to one latent variable and constructed a one-factor model; then, we tested this model with confirmatory factor analysis (CFA). The overall fit indices of this one-factor model performed very poorly: GFI = 0.578, AGFI = 0.424, CFI = 0.541, TLI = 0.457, v2 = 1371.067, df = 77, v2/df = 17.806. All these indices were far beyond the acceptable range and far worse than the original model (as shown in Table 5). Taken together, these results suggest that CMV did not pose a significant threat to interpreting our present findings (Harris & Mossholder, 1996). 5.2. Structural model The research model was tested using AMOS 7.0. The overall fit indices of the research model are presented in Table 5. As shown, all the overall fit indices of the research model perform well; the Table 2 Cronbach’s alpha, composite reliability and AVE. Constructs

Cronbach’s alpha

Composite reliability

AVE

Knowledge contribution Social interaction tie Membership esteem Social identity Self-identity

0.812 0.904 0.882 0.865 0.884

0.820 0.903 0.885 0.883 0.884

0.606 0.769 0.794 0.693 0.701

Constructs

Standard loading

Knowledge contribution KC1 KC2 KC3

0.835 0.836 0.646

Social interaction tie SIT1 SIT2 SIT3

0.891 0.896 0.827

Membership esteem ME1 ME2

0.910 0.869

Social identity SI1 SI2 SI3

0.762 0.909 0.818

Self-identity SEI1 SEI2 SEI3

0.889 0.815 0.839

Table 4 Correlations between latent constructs. Constructs

KC

SIT

ME

SI

SEI

Knowledge contribution (KC) Social interaction tie (SIT) Membership esteem (ME) Social identity (SI) Self-identity (SEI)

0.778 0.692 0.495 0.559 0.557

0.877 0.574 0.476 0.561

0.891 0.557 0.542

0.832 0.652

0.837

Notes. The diagonal numbers are the square root of AVE.

Table 5 Goodness of fit indices for the structural model. Goodness of fit indices 2

v

df 2

v /df GFI AGFI CFI RMSEA

Results

Desired levels

123.473 68 1.815 0.953 0.927 0.982 0.048

Smaller – <3 >0.90 >0.80 >0.90 <0.05

goodness-of-fit index (GFI), adjusted goodness-of-fit index (AGFI) and comparative fit index (CFI) all perform above the threshold values, and the root mean square error of approximation (RMSEA) was less than 0.05. The hypotheses were all tested, and the findings are presented in Fig 2. We elaborate the findings in the following sequence: the social network perspective for the inductive formation of social identity (H1 and H2), the work group perspective for the inductive formation of social identity (H3–H5), the direct influence of knowledge contribution on social identity formation (H6), and finally the influence of the inductive route on self-identity formation (H7–H9). First, H1 and H2 are supported as expected. Regarding the relationship between knowledge contribution and social interaction tie (i.e., H1), the path coefficient is very high, up to 0.65 (p < 0.001). As to the relationship between social interaction tie and social identity (i.e., H2), the path coefficient is 0.28 (p < 0.001). Both H1 and H2 were confirmed, indicating the significance of the social network perspective. Second, H3–H5 are also supported. Both knowledge contribution and social interaction tie are positively associated with membership esteem (H3: path coefficient = 0.31, p < 0.001; H4: path coefficient = 0.40, p < 0.001). Membership esteem also exerts a

236

Z.-c. Cheng, T.-c. Guo / Computers in Human Behavior 43 (2015) 229–241

Social interaction tie (R2 =0.423) -0.03 0.28**

0.65** 0.40** 0.27**

Knowledge contribution

0.32**

Social identity (R2 =0.389)

0.31**

0.18*

Self -identity (R2 =0.557)

0.56**

Membership esteem (R2 =0.414) Fig. 2. Model testing results. Note.

⁄⁄

p < 0.01; ⁄p < 0.05.

Social interaction tie a1

d1 b1 a3

Knowledge contribution

d2

Social identity

c1’

a2

Selfidentity

d3

b2

Membership esteem Fig. 3. Mediation effect analysis model.

positive effect on social identity (H5: path coefficient = 0.18, p < 0.05). All these positive relationships indicate the significance of the work group perspective. Third, support is provided for the direct relationship between knowledge contribution and social identity (H6: path coefficient = 0.27, p < 0.001). Fourth, H8 and H9 are supported, but H7 is not. Both social identity and membership esteem are positively associated with self-identity (H8: path coefficient = 0.32, p < 0.001; H9: path coefficient = 0.56, p < 0.001). The effect of social interaction tie on self-identity (i.e., H7) is not significant. To further examine the intermediate effect, we separately tested the direct relationship between social interaction tie and self-identity, and the result shows the path coefficient being as high as 0.48 (p < 0.001). As seen in Fig. 2, the intervening constructs fully mediated the relationship between social interaction tie and social identity because the direct path coefficient was completely eliminated. Finally, the explained variances of social interaction tie, membership esteem, social identity and self-identity are 42.3%, 41.4%, 38.9% and 55.7%, respectively, showing a good predictive validity of the model (Straub, Boudreau, & Gefen, 2004). 5.3. Mediation effect As shown in Fig. 3, the mediation effect analysis model is a multiple mediation model. Bootstrapping is the most powerful and reasonable method for obtaining confidence limits for specific indirect effects under most conditions (Preacher & Hayes, 2008), so this study used bootstrapping to test the multiple mediation model via

MPLUS7.0 software. Because the mediation effects in this study is complex, to clearly present mediating effects, this study primarily investigated the indirect effects of knowledge contribution on social identity and self-identity through social interaction tie and membership esteem, and then made a contrast analysis between these indirect effects (Preacher & Hayes, 2008). In Fig. 3, a1, a2, b1, b2, c10 , d1, d2, and d3 are the effects of the previous variables on the subsequent variables; for example, a1 expresses the effect of knowledge contribution on social interaction tie. The indirect effect of knowledge contribution on social identity through social interaction tie is a1b1 + a1a3b2; the indirect effect of knowledge contribution on social identity through membership esteem is a2b2 + a1a3b2; the indirect effect of knowledge contribution on self-identity through social interaction tie is a1d1 + a1a3b2d2 + a1b1d2 + a1a3d3; and the indirect effect of knowledge contribution on self-identity through membership esteem is a2b2d2 + a2d3 + a1a3b2d2 + a1a3d3. The products of effects represent the specific indirect effects through the corresponding paths. For example, a1a3b2 expresses the specific indirect effect through the path of ‘‘knowledge contribution—social interaction tie—membership esteem—social identity’’. Testing results are shown in Tables 6–8. First, Table 6 shows all the specific indirect effects through social interaction tie and membership esteem. The specific effect through the path of ‘‘knowledge contribution – social interaction tie – self-identity’’ (that is, a1 ⁄ d1) is not significant because zero lies between the lower and upper limits (0.046, 0.075). Because zero is not included in the 95% confidence interval, all other specific indirect effects are significant.

237

Z.-c. Cheng, T.-c. Guo / Computers in Human Behavior 43 (2015) 229–241 Table 6 All specific indirect effects through social interaction tie and membership esteem. Point Estimate

Product of coefficients SE

a1 ⁄ b1 a1 ⁄ a3 ⁄ b2 a2 ⁄ b2

0.135 0.033 0.038

a1 ⁄ d1 a1 ⁄ b1 ⁄ d2 a1 ⁄ a3 ⁄ b2 ⁄ d2 a1 ⁄ a3 ⁄ d3 a2 ⁄ d3 a2 ⁄ b2 ⁄ d2

0.015 0.045 0.011 0.103 0.120 0.013

Bootstrapping BC 95% CI Z

Lower limit

From knowledge contribution to social identity 0.031 4.332 0.012 2.884 0.017 2.276 From knowledge contribution to self-identity 0.031 0.477 0.014 3.253 0.004 2.537 0.019 5.300 0.032 3.699 0.006 2.116

Upper limit

0.074 0.010 0.005

0.196 0.056 0.071

0.046 0.018 0.002 0.065 0.056 0.001

0.075 0.072 0.019 0.142 0.183 0.024

SE = standard error, BC = bias corrected confidence intervals, 5000 bootstrap samples.

Table 7 Indirect effects of knowledge contribution on social identity through social interaction tie and membership esteem. Point estimate

Social interaction tie Membership esteem C1

0.168 0.071 0.097

Product of coefficients

Bootstrapping BC 95% CI

SE

Z

Lower limit

Upper limit

0.031 0.026 0.040

5.512 2.784 2.445

0.112 0.024 0.021

0.232 0.131 0.173

C1 = contrast of the two indirect effects, SE = standard error, BC = bias corrected confidence intervals, 5000 bootstrap samples.

Table 8 Indirect effects of knowledge contribution on self-identity through social interaction tie and membership esteem. Point estimate

Social interaction tie Membership esteem C2

0.174 0.246 0.073

Product of coefficients

Bootstrapping BC 95% CI

SE

Z

Lower limit

Upper limit

0.034 0.038 0.052

5.105 6.431 1.403

0.113 0.169 0.167

0.248 0.320 0.043

C2 = contrast of the two indirect effects, SE = standard error, BC = bias corrected confidence intervals, 5000 bootstrap samples.

Second, Table 7 reveals that the indirect effect of knowledge contribution on social identity through social interaction tie is 95% likely to range from 0.113 to 0.248, and the estimated effect is 0.168, which lies between these two values. Because zero does not occur between the lower and upper limits, we can conclude that the indirect effect is significant. Similarly, the indirect effect of knowledge contribution on social identity through membership esteem is also significant. It may be of interest to see whether these two indirect effects differ significantly. Examination of the contrast of these two indirect effects (C1) shows that the indirect effect through social interaction tie is larger than the indirect effect through membership esteem, with a BC 95% CI of 0.021–0.173. Third, Table 8 presents that the indirect effect of knowledge contribution on self-identity through social interaction tie is 95% likely to range from 0.054 to 0.179, and the estimated effect is 0.174. Because zero does not occur between the lower and upper limits, we can conclude that the indirect effect is significant. Likewise, the indirect effect of knowledge contribution on self-identity through membership esteem is also significant. Examination of the contrast of these two indirect effects (C2) shows that there is no significant difference between these two indirect effects because zero lies between the lower and upper limits (0.167, 0.043). Finally, it should be noted that in the mediation effect analysis model (as shown in Fig. 3), there is no direct connection between the two variables of knowledge contribution and self-identity, in order that the model can clearly interpret the view that an inductive formation path of social identity can also foster self-identity. It

is of interest to identify whether the indirect effect of knowledge contribution on self-identity changes when connecting knowledge contribution and self-identity. Hence, we connect the two variables of knowledge contribution and self-identity, and build a new mediation effect analysis model (as shown in Fig. 4). Testing results about the new model are shown in Table 9. The results show the indirect effects for social interaction tie and membership esteem still maintain at a significant level, however, the difference between these two indirect effects also becomes significant, i.e., examination of the contrast of these two indirect effects (C3) shows the indirect effect through social interaction tie is lower than the indirect effect through membership esteem, with a BC 95% CI of 0.198 to 0.009.

6. Discussion In this article, we show how social identity forms through an inductive path and show the mechanism that underlies the inductive path. As shown by our findings, there are three sub-paths that can lead to the formation of an inductive identity. First, from a social network perspective, the VC is a social network that is characterized by social interaction tie. Social interaction tie is the foundational dimension of social capital and can lead to the emergence of other dimensions: interpersonal relationships and shared cognition (Xiao, Li, Cao, & Tang, 2012). The results of our study indicate that knowledge contribution can positively affect social interaction

238

Z.-c. Cheng, T.-c. Guo / Computers in Human Behavior 43 (2015) 229–241

Social interaction tie a1

d1 b1 a3

Knowledge contribution

d2

Social identity

c1’

a2

Selfidentity

d3

b2

Membership esteem c2’ Fig. 4. Revised mediation effect analysis model.

Table 9 Indirect effects of knowledge contribution on self-identity through social interaction tie and membership esteem in the revised mediation effect analysis model. Point estimate

Social interaction tie Membership esteem C3

0.123 0.226 0.103

Product of coefficients

Bootstrapping BC 95% CI

SE

Z

Lower limit

Upper limit

0.036 0.035 0.048

3.428 6.37 2.158

0.054 0.157 0.198

0.179 0.318 0.009

C3 = contrast of the two indirect effects, SE = standard error, BC = bias corrected confidence intervals, 5000 bootstrap samples.

tie, which contributes to the formation of social identity. It is worth mentioning that the path coefficient between knowledge contribution and social interaction tie is as high as to 0.65. This may indicate that in VCs, social interaction ties form mainly on the basis of knowledge contribution; at the same time, it suggests that social interactions and social relationships are important for participants. These findings are consistent with other studies on the influence of social interaction and social relationships on social identity. Postmes, Haslam, et al. (2005) argued that social interactions and social relationships may play important roles in the emergence of a shared cognition, such as a shared task and goal, in groups that contribute to the emergence of social identity (Wegge & Haslam, 2003). In VCs, social interaction tie may represent the strength of interpersonal relationships. Zhao et al. (2012) indicated that social relationships are positively related to members’ sense of belonging to the group. Ainhoa (2007) also found that social identity can form through a social relations path. The results of these studies and our study suggest that social interaction tie provides an important mechanism based on which other variables exert influences on social identity. In VCs, these other variables could refer to interpersonal relationships or shared cognition, which may be derived from social interaction. Second, from a work group perspective, knowledge contribution can be seen as a task behavior that can cultivate self-esteem. Self-esteem can be classified as personal and collective. Collective self-esteem can also be assessed by four dimensions: private collective self-esteem, public collective selfesteem, membership esteem, and importance to identity. In VCs, there are no status differences between groups; members obtain self-esteem only based on task achievement. Therefore, only membership esteem can significantly emerge in a virtual community. This is in accordance with the study of Chiu et al. (2006), in which membership esteem was used to represent collective self-esteem. Our results confirm that knowledge contribution and social interaction tie both positively affect membership esteem, which leads to the emergence of social identity. Certainly, in VCs, knowledge

contribution is autonomous and thus can cultivate individuals’ sense of self-worth. At the same time, more interactions result in more replies, which can contribute to members’ sense of selfworth for the group. These findings are consistent with Nunnally (1978). He found members’ task experiences play an important role in determining their level of membership esteem. Finally, our findings confirm that social identity may form based directly on knowledge contribution. This finding is in accordance with Jans et al. (2012), who indicated that the diversity between contribution behaviors can be integrated into a shared cognitive representation of the group, and thus, social identity emerges on the basis of these unique contributions. By comparing these three formation paths of social identity, we can find that social interaction tie and knowledge contribution exert a more significant influence on social identity than that exerted by membership esteem. The reason for this may be that in virtual community, because of the lack of physical cues, members’ behavioral involvement and social relations play vital roles in cultivating a sense of virtual community and perceiving their own embeddedness in the community (Blanchard & Markus, 2004; Jones & Volpe, 2011). This study shows that the inductive route can also lead to the formation of self-identity as a contributor. Both social identity and membership esteem are positively associated with self-identity. Social identity plays a powerful role in members’ intention of knowledge contribution and thus can enhance members’ willingness to act as a contributor (Terry et al., 1999). Meanwhile, membership esteem can be seen as an identity label or status symbol that makes people aware of their contributor roles. There are no significant relationships between social interaction tie and self-identity, which can be explained by an intermediate effect. Social identity and membership esteem serve as the mediating variables between social interaction tie and self-identity. Self-identity directly forms based on membership esteem and social identity, both of which may be derived from social interaction. Among the formation paths of self-identity, the path between

Z.-c. Cheng, T.-c. Guo / Computers in Human Behavior 43 (2015) 229–241

membership esteem and self-identity is the most significant (as high as 0.56). We think this is because membership esteem can serve as an indicator that allows people to easily perceive their roles as contributors. The examination of mediation effects shows that social interaction tie and membership esteem both play significant mediating roles in the model. In contrast, the results indicate that social interaction tie plays a more significant mediating role between knowledge contribution and social identity than that of membership esteem, whereas membership esteem has a more significant mediating effect between knowledge contribution and self-identity than that of social interaction tie, in spit of only in the revised mediation effect analysis model. This may be because social interaction tie plays a vital role in cultivating a sense of virtual community, while membership esteem is closely associated with personal roles. Finally, among the specific indirect effects, only a1 ⁄ d1 (through the path ‘‘knowledge contribution – social interaction tie – self-identity’’) is not supported, possibly because, as shown in Fig. 2, there is no direct effect of social interaction tie on selfidentity. 7. Implications and future research 7.1. Theoretical implications This study makes several theoretical contributions to the literature. We investigated the influence of knowledge contribution on the formation of social identity in VCs, contributing to both virtual community literature and social identity formation research. Social identity has been used in many studies on VCs. However, most of them take social identity as given, without bringing to light how to enhance members’ social identification with a group (Shen et al., 2010). This study suggests that in VCs, an operational social identity can emerge only through participation. Many studies show social identity is an important antecedent of knowledge contribution intention; however, in turn, social identity must first form on the basis of contribution behaviors. Meanwhile, as a few studies have suggested, in work group and organization contexts, there are many factors such as communication, interpersonal relationship, network density, self-esteem and contribution behaviors that can play roles in the formation of social identity (De Cremer & Oosterwegel, 2000; Jans et al., 2012; Postmes, Haslam, et al., 2005); however, no study has integrated these factors to form a complete and clear formation route of social identity, especially in VCs. This study advanced the theoretical work on social identity formation by investigating how and what compositions constitute the inductive formation path of social identity. 7.2. Practical implications This study also has several practical implications. Because our study investigated an interest-based virtual community, the results provide several practical implications for the interest-based virtual community operators. First, our study suggests there is a self-running mechanism in VCs. Being motivated by personal interest, people tend to participate in discussion around a common interest and contribute knowledge to groups. Based on knowledge contribution, social identity and self-identity can emerge and then reversely affect the intention to contribute knowledge. This can be seen as a self-running mechanism. This may indicate that it is very important for communities to attract more newcomers. Based on the self-running mechanism, the more newcomers, the more people who can make contributions and integrate themselves into the communities. Virtual community operators should pay more attention to the factors such as convenience, popularity and service

239

quality to attract more newcomers. Operators should also provide more interest topics to involve more people in the communities. Second, this study suggests that social interaction tie and membership esteem are positively related to the social identity. Hence, on the one hand, VC operators could take actions to promote social interactions between the members. For example, VC operators could offer virtual credits and ranks as moral encouragements to the members who positively present knowledge and interact with others. Meanwhile, these rewards can boost members’ self-esteem. For example, the virtual ranks that VC operators provide can serve as a status symbol and encourage members’ self-esteem. Finally, because membership esteem plays a very important role in forming a self-identity, VC operators should pay more attention to the influence factor of membership esteem. 7.3. Limitations and future research There are several limitations in our study. First, the type of VC we surveyed in our study is a VC wherein members discuss common interests. The findings might not totally apply to other types of VCs, such as commercially oriented VCs, where members gather together mainly to share commercial information, and relationship VCs, where people mainly aim to maintain and strengthen their relationships with acquaintances, make friends, and so forth. Future research therefore could test the research model in different types of VCs and examine the difference among them. Second, because of the small variances explained, there could be other factors that affect social identity. A few researchers argue that online community experiences that derive from interpersonal interactions or machine interactions may influence members’ attitude towards the community and the sense of community (Keng, Ting, & Chen, 2011; Mazaheri, Richard, & Laroche, 2012). Other factors such as members’ need fulfillment and the management of the VC may also be conducive to the formation of social identity. Other research perspectives, such as the experience perspective, could be used for future research to understand the effect of online community experience on members’ psychological change. 8. Conclusions The goal of this study was to identify the mechanism that underlies the inductive formation path of social identity and selfidentity in VCs. To this end, social interaction tie and membership esteem were, respectively, adapted from the social capital literature and self-esteem research and serve as the mediating variables between knowledge contribution and social identity to theorize a model of the inductive identity formation route. Data collected from Baidu Post Bar provided empirical support for the proposed model. The results show that knowledge contribution can directly affect social identity; at the same time, social interaction tie and membership esteem play significant mediating roles in the relevance between knowledge contribution and social identity. In addition, self-identity can also form through an inductive path. Noteworthy contributions of our study include theorizing and validating the model of the inductive identity formation route. References Ashforth, B. E. (2001). Role transitions in organizational life: An identity-based perspective. Mahwah, NJ: Erlbaum. Ashforth, B. E., Harrison, S. H., & Corley, K. G. (2008). Identification in organizations: An examination of four fundamental questions. Journal of Management, 34, 325–374. Baggozi, R. P., & Yi, Y. (1988). On the evaluation of structural equation models. Journal of the Academy of Marketing Science, 16, 74–94. Bagozzi, R. P., & Dholakia, U. M. (2002). International social action in virtual communities. Journal of Interactive Marketing, 16, 2–21.

240

Z.-c. Cheng, T.-c. Guo / Computers in Human Behavior 43 (2015) 229–241

Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497–529. Bergami, M., & Bagozzi, R. P. (2000). Self-categorization, affective commitment and group self-esteem as distinct aspects of social identity in the organization. British Journal of Social Psychology, 39, 555–577. Blanchard, A. L., & Markus, M. L. (2004). The experienced ‘‘sense’’ of a virtual community: Characteristics and processes. The Data Base for Advances in Information Systems, 35, 65–79. Bock, G. W., Zmud, R. W., Kim, Y. G., & Lee, J. N. (2005). Behavioral intention formation in knowledge sharing: Examining the roles of extrinsic motivators, social-psychological forces, and organizational climate. MIS Quarterly, 29, 87–111. Brown, R. (2000). Social identity theory: Past achievements, current problems and future challenges. European Journal of Social Psychology, 30, 745–778. Carmel, D., Roitman, H., & Yom-Tov, E. (2012). On the relationship between novelty and popularity of user-generated content. ACM Transactions on Intelligent Systems and Technology, 3, 69. Chiu, C. M., Hsu, M. H., & Wang, E. T. G. (2006). Understanding knowledge sharing in virtual communities: An integration of social capital and social cognitive theories. Decision Support Systems, 42, 1872–1888. Chou, S. W., Min, H. T., Chang, Y. C., & Lin, C. T. (2010). Understanding continuance intention of knowledge creation using extended expectation-confirmation theory: An empirical study of Taiwan and China online communities. Behavior and Information Technology, 29, 557–570. Chua, A. Y. K., & Balkunje, R. S. (2013). Beyond knowledge sharing: Interactions in online discussion communities. International Journal of Web Based Communities, 9, 67–82. Coleman, J. (1988). Social capital in the creation of human capital. The American Journal of Sociology, 94, 95–120. Crocker, J., Luhtanen, R., Blaine, B., & Brodnax, S. (1994). Collective self-esteem and psychological well-being among white, black and Asian college students. Personality and Social Psychology Bulletin, 20, 503–513. Dávila, M. C., & Jiménez, G. G. (2012). Organizational identification and commitment: Correlates of sense of belonging and affective commitment. The Spanish Journal of Psychology, 15, 244–255. De Bruijn, G. J., Verkooijen, K., de Vries, N. K., & van den Putte, B. (2012). Antecedents of self identity and consequences for action control: An application of the theory of planned behaviour in the exercise domain. Psychology of Sport and Exercise, 13, 771–778. De Cremer, D., & Oosterwegel, A. (2000). Collective self-esteem, personal selfesteem, and collective efficacy in in-group and outgroup evaluations. Current Psychology, 18, 326–339. Deaux, K., & Martin, D. (2003). Interpersonal networks and social categories: Specifying levels of context in identity processes. Social Psychology Quarterly, 66, 101–117. Dholakiaa, U. M., Bagozzia, R. P., & Pearob, L. K. (2004). A social influence model of consumer participation in network- and small-group-based virtual communities. International Journal of Research in Marketing, 21, 241–263. Easterbrook, M., & Vignoles, V. L. (2013). What does it mean to belong? Interpersonal bonds and intragroup similarities as predictors of felt belonging in different types of groups. European Journal of Social Psychology, 43, 455–462. Ellemers, N., De Gilder, D., & Haslam, S. A. (2004). Motivating individuals and groups at work: A social identity perspective on leadership and group performance. Academy of Management Review, 28, 459–478. Ellemers, N., Kortekaas, P., & Ouwerkerk, J. W. (1999). Self-categorisation, commitment to the group and group self-esteem as related but distinct aspects of social identity. European Journal of Social Psychology, 29, 371–389. Fielding, K. S., & Hogg, M. A. (2000). Working hard to achieve self-defining group goals: A social identity analysis. Zeitschrift für Sozialpsychologie, 31, 191–203. Foels, R. (2006). Ingroup favoritism and social self-esteem in minimal groups: Changing a social categorization into a social identity. Current Research in Social Psychology, 94, 38–53. Fornell, C., & Larcker, D. (1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18, 39–50. Granovetter, M. S. (1985). Economic action and social structure: The problem of embeddedness. The American Journal of Sociology, 91, 481–510. Hagerty, B. M., Lynch-Sauer, J., Patusky, K. L., Bouwsema, M., & Collier, P. (1992). Sense of belonging: A vital mental health concept. Archives of Psychiatric Nursing, 6, 172–177. Hagger, M. S., & Chatzisarantis, N. L. D. (2006). Self-identity and the theory of planned behaviour: Between- and within-participants analyses. British Journal of Social Psychology, 45, 731–757. Harris, S. G., & Mossholder, K. W. (1996). The affective implications of perceived congruence with culture dimensions during organizational transformation. Journal of Management, 22, 527–547. Hatcher, L. (1994). A step-by-step approach to using the SAS system for factor analysis and structural equation modeling. Cary, NC: SAS Institute Inc. Hogg, M. A., Terry, D. J., & White, K. M. (1995). A tale of two theories: Critical comparison of identity theory and social identity theory. Social Psychology Quarterly, 58, 255–269. Hsiao, Cheng-Chieh, & Chiou, Jyh-Shen (2012). The effect of social capital on community loyalty in a virtual community: Test of a tripartite-process model. Decision Support Systems, 54, 750–757.

Huemer, L., Becerra, M., & Lunnan, R. (2004). Organizational identity and network identification: Relating within and beyond imaginary boundaries. Scandinavian Journal of Management, 20, 53–73. Igbaria, M., Parasuraman, S., & Baroudi, J. J. (1996). A motivational model of microcomputer usage. Journal of Management Information Systems, 13, 127–143. Jans, L., Postmes, T., & Van der Zee, K. I. (2012). Sharing differences: The inductive route to social identity formation. Journal of Experimental Social Psychology, 48, 1145–1149. Jeppesen, L. B., & Frederiksen, L. (2006). Why do users contribute to firm-hosted user communities: The case of computer-controlled music instruments. Organization Science, 17, 45–63. Jones, C., & Volpe, E. H. (2011). Organizational identification: Extending our understanding of social identities through social networks. Journal of Organizational Behavior, 32, 413–434. Jostein, R., Paschal, S., & Silje, H. (2010). The role of self-identity in the theory of planned behavior: A meta-analysis. Journal of Applied Social Psychology, 40, 1085–1105. Keng, C. J., Ting, H. Y., & Chen, Y. T. (2011). Effects of virtual-experience combinations on consumer-related ‘‘sense of virtual community’’. Internet Research, 21, 408–434. Kim, J. (2006). Conformity behavior in computer-mediated group decision-making: The interaction between virtual self-representation and personality traits. Unpublished doctoral dissertation. Michigan State University. Kim, J., & Park, H. S. (2011). The effect of uniform virtual appearance on conformity intention: Social identity model of deindividuation effects and optimal distinctiveness theory. Computers in Human Behavior, 27, 1223–1230. Kim, H. W., Zheng, J. R., & Gupta, S. (2011). Examining knowledge contribution from the perspective of an online identity in blogging communities. Computers in Human Behavior, 27, 1760–1770. Kline, S., & Stafford, L. (2004). A comparison of interaction rules and interaction frequency in relationship to marital quality. Communication Reports, 17, 37–41. Lee, E. (2004). Effects of visual representation on social influence in computer mediated communication: Experimental tests of the social identity model of deindividuation effects. Human Communication Research, 30, 234–259. Lee, F., Vogel, D., & Limayem, M. (2002). Virtual community informatics: What we know and what we need to know. In: Proceedings of the 35th Hawaii international conference system science (pp. 2863–2872), 7–10 January 2002, Big Island, HI, USA. IEEE Computer Society. Leimeister, J.M., Sidiras, P., & Kremar, H.C. (2004). Success factors of virtual communities from the perspective of members and operators: An empirical study. In: The 37th Hawaii international conference on system sciences, Hawaii. Liang, H. G., Saraf, N., Hu, Q., & Xue, Y. J. (2007). Assimilation of enterprise systems: The effect of institutional pressures and the mediating role of top management. MIS Quarterly, 31, 59–87. Lickel, B., Hamilton, D. L., & Sherman, S. J. (2001). Elements of a lay theory of groups: Types of groups, relational styles, and the perception of group entitativity. Personality and Social Psychology Review, 5, 129–140. Lin, H. F. (2008). Determinants of successful virtual communities: Contributions from system characteristics and social factors. Information & Management, 45, 522–527. Luhtanen, R., & Crocker, J. (1992). A collective self-esteem scale: Self-evaluation of one’s social identity. Personality and Social Psychology Bulletin, 18, 302–318. Ma, W. W. K., & Yuen, A. H. K. (2011). Understanding online knowledge sharing: An interpersonal relationship perspective. Computers & Education, 56, 210–219. Mallan, K., & Giardina, N. (2009). Wikidentities: Young people collaborating on virtual identities in social network sites, 14. Mazaheri, E., Richard, M. O., & Laroche, M. (2012). The role of emotions in online consumer behavior: A comparison of search, experience, and credence services. Journal of Services Marketing, 26, 535–550. McAllister, D. J., & Bigley, G. A. (2002). Work context and the (re)definition of self: How organizational care influences organization-based self-esteem. Academy of Management Journal, 45, 894–904. McKenna, K. Y. A., Green, A. S., & Gleason, M. E. J. (2002). Relationship formation on the internet: What’s the big attraction? Journal of Social Issues, 58, 9–31. McLeish, K. N., & Oxoby, R. J. (2011). Social interactions and the salience of social identity. Journal of Economic Psychology, 32, 172–178. Meyer, J. P., & Allen, N. J. (1991). A three-component conceptualization of organizational commitment. Human Resource Management Review, 1, 61–89. Milgram, S. (1977). The familiar stranger: An aspect of urban anonymity, in the individual in a social world: Essays and experiments, S. Milgram (Ed.), Reading (pp. 51–53). MA: Addison-Wesley. Mooradian, T., Renzl, B., & Matzler, K. (2006). Who trusts? Personality, trust and knowledge sharing. Management Learning, 37, 523–540. Nahapiet, J., & Ghoshal, S. (1998). Social capital, intellectual capital, and the organizational advantage. The Academy of Management Review, 23, 242–266. Nardi, B. A., Schiano, D. J., Gumbrecht, M., & Swartz, L. (2004). Why we blog. Communications of the Association for Computing Machinery, 47, 41–46. Nunnally, J. C. (1978). Psychometric theory. New York: McGraw-Hill. Ainhoa, F. R. (2007). Networks and identifications: A relational approach to social identities. International Sociology, 22, 683–699. Pierce, J. L., & Gardner, D. G. (2004). Self-esteem within the work and organizational context: A review of the organization-based self-esteem literature. Journal of Management, 30, 591–622. Pierce, J. L., Gardner, D. G., Cummings, L. L., & Dunham, R. B. (1989). Organizationbased self-esteem: Construct definition measurement and validation. Academy of Management Journal, 32, 622–648.

Z.-c. Cheng, T.-c. Guo / Computers in Human Behavior 43 (2015) 229–241 Pierce, J. L., Gardner, D. G., Dunham, R. B., & Cummings, L. L. (1993). Moderation by organization-based self-esteem of role condition-employee response relationships. The Academy of Management Journal, 36, 271–288. Podsakoff, P., MacKenzie, S., Lee, J., & Podsakoff, N. (2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88, 879–903. Pollack, B. N. (1996). Interpersonal communication and feelings of belonging in work groups: The impact of the sociophysical environment. Dissertation Abstracts International. Section B: The Sciences and Engineering, 57, 4072. Postmes, T., Haslam, S. A., & Swaab, R. (2005). Social influence in small groups: An interactive model of social identity formation. European Review of Social Psychology, 16, 1–42. Postmes, T., Spears, R., & Lea, M. (1999). Social identity, normative content, and ‘‘deindividuation’’ in computer-mediated groups. In N. Ellemers & R. Spears (Eds.), Social identity: Context, commitment, content (pp. 164–183). Oxford, UK: Blackwell. Postmes, T., Spears, R., Lee, A. T., & Novak, R. J. (2005). Individuality and social influence in groups: Inductive and deductive routes to group identity. Journal of Personality and Social Psychology, 89, 747–763. Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40, 879–891. Ren, Y. Q., Harper, F. M., Drenner, S., Terveen, L., Kiesler, S., Riedl, J., et al. (2012). Building member attachment in online communities: Applying theories of group identity and interpersonal bonds. Management Information Systems Quarterly, 36, 841–864. Ren, Y. Q., Kraut, R., & Kiesler, S. (2007). Applying common identity and bond theory to design of online communities. Organization Studies, 28, 377–408. Rink, F., & Ellemers, N. (2007). Diversity as a basis for shared organizational identity: The norm congruity principle. British Journal of Management, 18, 17–27. Rousseau, D. M. (1998). Why workers still identify with organizations. Journal of Organizational Behavior, 19, 217. Schwämmlein, E., & Wodzicki, K. (2012). What to tell about me? Self-presentation in online communities. Journal of Computer-Mediated Communication, 17, 387–407. Shen, K. N., Yu, A. Y., & Mohamed, K. (2010). Knowledge contribution in virtual communities: Accounting for multiple dimensions of social presence through social identity. Behaviour and Information Technology, 29, 337–348. Shiue, Y. C., Chiu, C. M., & Chang, C. C. (2010). Exploring and mitigating social loafing in online communities. Computers in Human Behavior, 26, 768–777. Sluss, D. M., & Ashforth, B. E. (2007). Relational identity and identification: Defining ourselves through work. Academy of Management Review, 32, 9–32. Smith, J. R., Terry, D. J., Manstead, A. S. R., Louis, W. R., Kotterman, D., & Wolfs, J. (2007). Interaction effects in the theory of planned behavior: The interplay of self-identity and past behavior. Journal of Applied Social Psychology, 37, 2726–2750. Stanoevska-Slabeva, K. (2002). Toward a community-oriented design of internet platforms. International Journal of Electronic Commerce, 16, 71–95. Stets, J. E., & Burke, P. J. (2000). Identity theory and social identity theory. Social Psychology Quarterly, 63, 224–237. Straub, D., Boudreau, M. C., & Gefen, D. (2004). Validation guidelines for IS positivist research. Communications of the Association for Information Systems, 13, 380–427. Stryker, S. (1980). Symbolic interactionism: A social structural version. Palo Alto, CA: Benjamin/Cummings. Stryker, S. & Serpe, R. (1982). Commitment, identity salience, and role behavior: Theory and research example. In W. Ickers & E. Knowles (Eds.), Personality, roles and social behavior (pp. 199–219). New York, NY: SpringerVerlag. Stryker, S., & Vryan, K. D. (2006). The symbolic interactionist frame. Handbooks of Sociology and Social Research, 3–28.

241

Stryker, S. (1987). Identity theory: Developments and extensions. In K. Yardley & T. Honess (Eds.), Self and identity (pp. 89–104). New York: Wiley. Swaab, R., Postmes, T., van Beest, I., & Spears, R. (2007). Shared cognition as a product of, and precursor to, shared identity in negotiations. Personality and Social Psychology Bulletin, 33, 187–199. Tajfel, H. (1981). Human groups and social categories: Studies in social psychology. Cambridge, UK: Cambridge University Press. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In Stephen Worchel & William Austin (Eds.), Psychology of intergroup relations (pp. 7–24). Chicago: Nelson Hall. Tajfel, H. (1972). Social categorization (English translation of ‘‘La categorisation sociale’’). In S. Moscovici (Ed.), Intro-duction d la psychologie sociale, 1 (pp. 272–302). Paris: Larousse. Teo, T. S. H., & King, W. R. (1996). Key dimensions of facilitators and inhibitors for the strategic use of information technology. Journal of Management Information Systems, 12, 35–53. Terry, D. J., Hogg, M. A., & White, K. M. (1999). The theory of planned behaviour: Self-identity, social identity and group norms. British Journal of Social Psychology, 38, 225–244. Tsai, W., & Ghoshal, S. (1998). Social capital and value creation: An empirical study of intrafirm networks. Academy of Management Journal, 41, 464–476. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherall, M. S. (1987). Rediscovering the social group: A self-categorization theory. Oxford: Blackwell. Van Dick, R. (2001). Identification in organizational contexts: Linking theory and research from social and organizational psychology. International Journal of Management Reviews, 3, 265–283. Vecchio, R. P. (2000). Negative emotion in the workplace: Employee jealousy and envy. International Journal of Stress Management, 7, 161–179. Wang, C. C., & Lai, C. Y. (2006). Knowledge contribution in the online virtual community: Capability and motivation. Lecture Notes in Computer Science, 4092, 442–453. Wasko, M., & Faraj, S. (2005). Why should I share? Examining social capital and knowledge contribution in electronic networks of practice. MIS Quarterly, 29, 35–57. Wasserman, S., & Faust, K. (1994). Social network analysis: Methods and applications. New York: Cambridge University Press. Wegge, J., & Haslam, S. A. (2003). Group goal setting, social identity and selfcategorization: Engaging the collective self to enhance group performance and organizational outcomes. In S. A. Haslam, D. van Knippenberg, M. J. Platow, & N. Ellemers (Eds.), Social identity at work: Developing theory for organizational practice (pp. 74–99). London: Taylor & Francis. Weick, K. E. (1995). Sensemaking in organizations. Thousand Oaks, CA: Sage. Wellman, B., & Wortley, S. (1990). Different strokes for different folks: Community ties and social support. American Journal of Sociology, 96, 558–588. Whitton, S. W., & Kuryluk, A. D. (2012). Relationship satisfaction and depressive symptoms in emerging adults: Cross-sectional associations and moderating effects of relationship characteristics. Journal of Family Psychology, 26, 226–235. Xiao, H. L., Li, W. F., Cao, X. B., & Tang, Z. M. (2012). The online social networks on knowledge exchange: Online social identity, social tie and culture orientation. Journal of Global Information Technology Management, 15, 4–24. Yli-Renko, H., Autio, E., & Sapienza, H. J. (2001). Social capital, knowledge acquisition, and knowledge exploitation in young technology based firms. Strategic Management Journal, 22, 587–613. Yun, D., & Silk, K. J. (2011). Social norms, self-identity, and attention to social comparison information in the context of exercise and healthy diet behavior. Health Communication, 26, 275–285. Zajonc, R. B. (1968). Attitudinal effects of mere exposure. Journal of Personality and Social Psychology, 9, 1–27. Zhao, L., Lu, Y. b., Wang, B., Chau, P. Y. K., & Zhang, L. (2012). Cultivating the sense of belonging and motivating user participation in virtual communities: A social capital perspective. International Journal of Information Management, 32, 574–588.