Intentional social action in virtual communities

Intentional social action in virtual communities

Richard P Bagozzi; Utpal M Dholakia

Journal of Interactive Marketing: Spring 2002; 16, 2; ABI/INFORM Global pg. 2

silahkan rujuk kembali ke sumber aslinya: ini hanya sekadar catatan

Bagozzi, R. P., & Dholakia, U. M. (2002). Intentional social action in virtual communities. Journal of interactive marketing, 16(2), 2-21.

ABSTRACT

There is growing evidence of the increasing participation in, and influence of, virtual communities in digital environments. To help explain this irresistible allure, the individual and social determinants of the member's intentions to participate are investigated. Conceptualizing virtual community participation as intentional social action, we explicate the concept of "we-intentions", and use the Model of Goal-Dinscted Behavior to explain members’ we-intentions. Virtual community influences pertaining to compliance, internalization, and social identity are also elaborated on. An empirical study of regular virtual community participants (N = 157) finds that we-intentions to participate are functions of both individual determinants (positive anticipated emotions and desires), and community influences (social identity). Implications for marketing and future research opportunities are discussed.

Kayword: 

 ============

The small group tends to restore, structurally, the symbolic power. Step by step, one can see a mystical network being built, carefully yet solidly connected, leading one to speak of a cultural resurgence in social life. This is the lesson taught by these eras of the masses—eras based mainly on the concatenation of groups with splintered but exacting intentionality’s. (Maffesoli, 1996, p. 88)

Since the early characterization of digital environment1 users as "electronic sociopath? (Canter & Siegel, 1994), the conventional wisdom has changed considerably. As digital environments have become more pervasive, both with respect to the number of people using them and the different activities they are used for, there is growing realization of their social functions (e.g., Postmes, Spears, & Lea, 1998; Walther, 1996), their potent influence in bringing together far-flung, like-minded individuals (Hagel & Armstrong, 1997; Wellman & Gulia, 1999), and their role in influencing consumer opinions, knowledge, and behaviors (e.g., Williams & Cothrell, 2000). By their very nature, digital environments originate in networks—and networks thrive on social interaction, whether specialized or broad, interpersonal or group-based, social or formal (Hagel & Armstrong, 1997; Rheingold, 1991). Where digital environments are concerned, the Interactive" in "interactive marketing" pertains just as much to interactions between consumers as between marketers and consumers.

Even in the early development of digital environments, users recognized and employed the social affordances of digital networks to organize, support, and communicate, giving rise to a unique social entity, or what is now commonly known as a virtual community (Rheingold, 1998). We view virtual communities to be mediated social spaces in the digital environment that allow groups to form and be sustained primarily through ongoing communication processes. Here, we conceive of groups as two or more individuals, each aware of his or her membership in the group, each aware of the others who belong to the group, and each aware of the positive interdependence as they strive to achieve mutual goals (Johnson & Johnson, 1987). The mutual goals of the virtual community may be functional, such as the symbiotic exchange of useful information regarding products, or hedonic—the creation and consumption of a positive, confluent experience through interaction. In either case, the community acts as an important reference group for its individual participants. For some participants, the virtual community may supplement existing primary and secondary reference groups, but for many others, it may actually replace other influential reference groups (e.g., Constant, Sproull, & Kiesler, 1996).

'By digital environment we refer mainly to the internet and the World Wide Web, and include such socially oriented virtual forums as electronic mail, bulletin board 'menu, multiuser dungeons, internet relay chat (IRO, and other Weis-hosed chat rooms.

Membership, frequency, and extent of participation in virtual communities are driven by volitional choice, and may be terminated by the member relatively effortlessly. In spite of this seemingly tenuous hold, researchers find that virtual communities are playing a bigger and bigger role in many aspects of member life—from forming and maintaining friendships and romantic relationships (Park & Floyd, 1996; Walther, 1996), to learning (Constant et al., 1996), to forming opinions, purchasing, and consuming products and services (Kozinets, 1999; Hagel & Armstrong, 1997). Researchers active in trying to understand the sustained, even increasing, allure of virtual communities have primarily adopted (1) the social network analysis paradigm (e.g., Wellman, 1999), (2) an ethnographic methodology (e.g., Kozinets, 1999), or have focused on (3) the unique characteristics of the digital environment, in understanding how mediated communication differs from face-to-face communication (e.g., Danet, Ruedenberg, & Rosenbaum-Tamari, 1998; Walther, 1996). Interestingly, in spite of practitioner interest, little research in marketing has focused on the consumer psychology of virtual community participation or its antecedents (however, see Kozinets, 1999).

To complement existing conceptual approaches and to focus more explicitly on understanding the role of individual and social influences in attracting members to the virtual community, we adopt a social psychological lens and conceptualize member participation as intentional social action. Building on this conceptualization, we study the role of attitudes, perceived behavioral control, desires, and anticipated emotions, all individual characteristics, and compliance, identification, and social identity, all social influences exerted by the community—on the individual member's intention to participate in the virtual community.

Our objectives in this research are threefold. First, we wish to elaborate on the "we-intentions" (i.e., group intentions) concept, argue for its distinctiveness, and establish its value for understanding the individual's participation in a virtual community, as well as its importance for interactive environment marketers. We present a brief overview of some pertinent philosophical ideas to fully explain the concept. Second, we present the Model of Goal-Directed Behavior (MGB) (Perugini & Bagozzi, 2001) and extend it to apply to group behavior in order to explicate the individual and social variables that shape the member's we-intention to participate in virtual community interaction. Integrating ideas from attitude theory, motivational research, and social identity theory, this model enables a rich understanding of the allure of virtual communities for individual participants. Finally, our goal is to test the hypotheses implied by the MGB in the context of virtual community participation. Our contribution in this regard is also to the substantive social psychological discipline, to enhance knowledge regarding the determinants of social intentional action. Taken together, we hope to provide digital environment marketers with both a rich conceptual understanding and practical insights regarding the allure of virtual communities and their implications.

We begin with a brief review of the extant virtual community research, followed by an elaboration of the "we-intentions" concept. This is followed by a detailed discussion of the individual and social bases of member participation in the virtual community, along with a presentation of the research hypotheses. The results of an empirical study, employing a structural equation modeling methodology are then presented. The article concludes with a discussion of the marketing significance of issues considered in, and arising from, the analysis.

AN OVERVIEW OF VIRTUAL COMMUNITY RESEARCH

Virtual communities may be of many different types. Some are tightly bound, densely knit groups of individuals who know one another well, and use the digital environment primarily as a way of augmenting their existing social relationships (Wellman & Gulia, 1999). In contrast, others are far-flung, sparsely connected networks of individuals who come together only in the mediated digital environment and have little chance of ever meeting physically. Some exist for social reasons, such as to enable like-minded individuals to meet; others exist primarily for commercial reasons (Hagel & Armstrong, 1997 call these "communities of transaction"). Irrespective of type, one characteristic that all virtual communities share is that text-based communication in the digital environment is the primal formative and shaping force for their evolution, growth, and sustenance.

Rheingold (1991) describes the essence of virtual communities: “People in virtual communities use words as screens to exchange pleasantries and argue, engage in intellectual discourse, conduct commerce, exchange knowledge, share emotional support, make plans, brainstorm, gossip, feud, fall in love, find friends and lose them, play games, flirt, create a little high art, and a lot of idle talk." Pictures, animations, or voice may nowadays augment the written word, but text still remains the primary medium of virtual community communication. Given our objectives, our interest in this research is in text-based virtual communities formed for primarily social purposes such as online chat rooms on the World Wide Web, rather than in communities focused on commercial or information exchanges. It is important to note that it is such social communities that exert the greatest influence on participants' knowledge and opinions regarding products and services through normative or informational mechanisms (or both), and ultimately influence consumer behaviors. Because of this, social scientists have referred to such communities as "information neighborhoods" (Burnett, 2000).

Attributes of Virtual Communities

Regardless of geographic dispersion or organizational emphasis, virtual communities share several characteristics. First, most virtual communities are organized around some distinct interest, which to a lesser or greater extent provides its raison d’être. This shared interest may pertain to a particular product or topic (such as Ford Probe cars, or gardening, for instance) or an affliction (such as cancer or Parkinson's disease), or a demographic attribute (single people over 50). From a marketing standpoint, such virtual communities may be thought of as market fragments—whit one specific product, or many different products and services of relevance to it, depending on the scope of the shared interest.

Second, as in real communities, virtual community members feel a "consciousness of kin"—an intrinsic connection toward other members, and a collective sense of separation from nonmembers (Wellman & Gulia, 1999). Such group affiliation not only colors the individual's opinions, ideas, and positions on specific issues, but also provides the impetus to return to the community in the future. Not only that, the interpersonal ties shared by virtual community members have also been shown to increase the willingness to share information and resources with other members to provide support and to commit to goals identified by the group (Walther, 1996; Wellman, 1999).

Third, most virtual communities create and use shared conventions and language (such as jargon, emoticons, or acronyms), maintain social roles, establish boundaries, enact rituals, show commitment to communal goals, and follow norms of interaction (such as "netiquette"). Through these functions, virtual communities are able to provide many of the same benefits to members as traditional communities, in spite of their physical dispersion and mediated environment Shirley (1995) cites Eric Hochman, a member of an early virtual community called ECHO (East Coast Hang Out), as offering the following comments:

I think ECHO is, in some ways, its own separate world, one with its own mythology, jargon, and social order, in other words, it has its own culture. An interesting one, because rather than it being an external thing that we adapt to, or have imposed on us, we're collectively creating it, here and now, as we post."

Fourth, unlike many traditional media where individuals consume content passively, content is created by community members through active participation. This content creation acts as an important shaping force of the community's character, and determines not only its influence on participants, but also the status and influence of individual members (Werry, 1999). Moreover, since digital environments facilitate the archiving of past content inexpensively, virtual communities come to represent an aggregation of collective expertise on individual topics, difficult to match elsewhere, and create a capital of knowledge, increasing its value for all members. Such member-generated content also provides the opportunity for integration into digital media advertising programs to raise their credibility and effectiveness (Werry, 1999).

Finally, because most virtual communities, whether Web-based chat rooms, or Usenet newsgroups, or even email lists or old-fangled bulletin-boards, are still based predominantly on text, most cues used in the traditional face-to-face community settings such as nonverbal expressions and social characteristics are filtered out (Kiesler & Sproull, 1992; Walther, 1996). This "filtering out" elevates the importance of communication content as the community's shaping force, allowing individual members the strategic freedom to express themselves (Postmes et al., 1998; Spears & Lea, 1994). Rheingold (1993) describes this facet of virtual communities well: "It's like a neighborhood pub or coffee shop. It's a little like a salon, where I can participate in a hundred ongoing conversations with people who don't care what I look like or sound like, but who do care about how I think and communicate. There are seminars and word fights in different corners."

Marketing practitioners initially adopted a commercial, albeit simplistic, lens when considering the value of virtual communities from a tactical standpoint. This is exemplified by Rayport and Sviokla (1995) in an early discussion on the marketing value of virtual communities: "The successful marketspace will invite consumers into a communal experience ... (making) shopping a transaction involving not just goods and services, but also (a positive) experience." Werry (1999) summarizes this early marketing approach to virtual communities well: “... (In marketing) 'Community' became a polite way of talking about audience, consumer demographics, and market segmentation, while seeming sensitive to Internet users, their culture and community" (p. 4). Recognizing the importance of member-generated content and social interaction, more recent thinking has adopted the "organic ecosystem" metaphor, focusing on managerial practices such as "seeding" conversations, "planting" provocative ideas, and carefully considering such organic attributes of the virtual community as size, intimacy, continuity, and growth as decision variables in their management (Hagel & Armstrong, 1997). But all of these approaches emphasize commercial aspects inordinately, ignoring for the most part the social psychological processes that make virtual communities so popular and influential in the first place.

Difference From Traditional Communities

In spite of many similarities to traditional communities, there is one essential difference. For the individual member, membership, involvement, and communication in virtual communities is driven by volitional choice—unlike traditional "bounded" communities where membership may be imposed involuntarily by chance of birth, proximity of residence, or the happenstance of geographic relocation. An individual member can terminate his or her membership in the virtual community conveniently and effortlessly—often simply by ending the navigation session and never returning to the virtual community's domain. In spite of this seemingly tenuous hold on individual members, there is growing evidence that participation in such virtual communities is immersive and protracted. With the passage of time, the virtual community becomes a central venue for many members, where they seek and appear to find companionship, social support, and a sense of belongingness (Wellman & Gulia, 1999). As adoption of and participation in such communities increases, these forums also appear to be gaining in importance as sources of information, and shapers of the participants' world view, not just regarding the community's organizing topic, but much broader attitudes and opinions, and importantly from our perspective, their consumption decisions and practices (Ilagel & Armstrong, 1997).

There has been considerable interest among researchers in the communication domain in trying to understand the sustained allure of virtual communities. The paradigm of social network analysis has been frequently used to examine this member attraction (e.g., Wellman & Gulia, 1999). This paradigm has uncovered valuable insights: from the breadth of communication topics found in these communities (Wellman & Gulia, 1999), to the strength of weak ties (Constant et al., 1996), to the impact of filtered-out cues on communication (e.g., Kiesler & Sproull, 1992), and group dynamics (Postmes, Spears, & Lea, 2000). A second research approach has focused on obtaining a better understanding of the unique characteristics of the digital environment, and how they are used by members to construct community. For instance, Danet and her colleagues (1998) find incidence of play and performance in virtual chat rooms facilitated by the creative use of text, and by the emphemerality, speed, and interactivity of the medium. Similarly, Postmes and his colleagues show that the anonymity afforded by computer-mediated groups allows deindividuation effects to occur—enhancing susceptibility of group members to situational group norms (Postmes et al., 1998; 2000). Finally, Walther (1996) reviews the literature on computer-mediated communication and use, and finds that in addition to impersonal and interpersonal uses, digital environments sometimes facilitate communication surpassing normal interpersonal levels—he labels such communication as "hyper personal," when "users ... exchange information, ... build impressions and compare values" (p. 33). On the whole, this line of research has shown that configurations of the digital environment implicate social processes, both by accentuating the influence of existing social influences, and creating new ones (discussed in detail later). In contrast, as mentioned before, relatively little academic research in marketing has examined virtual communities, though marketing practitioners have evinced considerable interest (e.g., Williams & Cothretl, 2000; Hagel & Armstrong, 1997).

Our interest is in contributing to this growing literature on virtual communities by seeking a better understanding of the individual and social determinants of action that drive member participation in the virtual community. Our emphasis is squarely on the individual community member. Adopting a social psychological lens, our interest is in studying the individual psychological mechanisms and the processes of community influence that attract participants to these mediated communication forums and shape their social action. We posit that virtual community participation constitutes intentional social action in that the community member acts intentionally (i.e., engages in purposive and goal directed action that remains under the individual's volitional control), and that these actions have a collective basis in that both what is done and why it is done in the virtual community are determined by the community's social characteristics (Bagoni, 2000).

THE DISTINCTION BETWEEN THE INDIVIDUAL'S PERSONAL AND GROUP INTENTIONS

The intention concept commonly studied by social psychologists is that of a personal intention. Eagly and Chaiken (1993, p. 168) define such a personal intention as a "person's motivation in the sense of his or her conscious plan to exert effort to carry out a behavior." Researchers to date have primarily scrutinized personal intentions, where the target is a singular subject. However, we believe that the allure of a virtual community for an individual member derives from the collectivity, the positive experience of congregating and communicating in the mediated environment, together, as a group (see Postmes et al., 2000, for a similar view). Consequently, virtual community members are likely to perceive themselves as members of the group, and form participation intentions in relation to this plural target Gilbert (1989) points out that action in relation to such plural subjects requires different conceptual schemes than the more common theme of singular action (see Bagoni, 2000, for a detailed explication).

Philosophers have recently given a great deal of attention to this construct of group intentions, using such labels as 'collective intentions" (Searle, 1990), "we-intentions" (Tuomela, 1995), and "shared-intentions" (Bratman, 1997). For instance, Bratman (1997) expresses a "shared intention" in the form "I intend that we act." In a similar vein, Tuomela (1995, p. 2) defines a "we-intention" as a "commitment of an individual to participate in joint action, and involves an implicit or explicit agreement between the participants to engage in that joint action." In this article, we conceptualize a virtual community member's participation intention as a group intention, based on the premise that members regard themselves as part of the social fabric of the virtual community in certain specific ways (Bagoni & Lee, 2002a). Then, based on the extant philosophical literature cited above, a set of criteria may be specified to stipulate the constitution of this group intention, which we henceforth refer to as the virtual community member's "we-intention."

First, Tuomela (1995, pp. 154-146) maintains that a we-intention occurs under the following conditions: (1) two or more members of a collectivity agree to jointly perform an action on behalf of the collectivity, (2) each member intends to perform his or her own part contributory to the group action, (3) members individually and mutually believe that the opportunity for joint action is likely to occur, and the members will perform their parts, and finally, (4) the intention to perform one's part is determined, in part, by the aforementioned individual and mutual beliefs (plus presumably the assurances and obligations entailed by the prior agreement). Second, Tuomela maintains that the beliefs required for we-intentions are purely subjective and need not be true, and a member could, in principle, be the only agent in a group with the we-intention (assuming that the four conditions listed above are satisfied). Finally, Tuomela (1995, p. 129) asserts that we-intentions require individual commitment, and commitment to provide a mutual support in the sense that a member is "not only committed to performing the pre-assigned part ... but he is also committed to furthering" the joint action (such as helping others in performing their parts, when needed). Bratman (1997) proposes a somewhat similar conceptualization of we-intentions, which he terms "shared intentions." Tuomela's (1995) and Bratman's (1997) specifications of group intentions are conceptual frameworks based on logical criteria and pre-suppose what philosophers term intentionalism (i.e., "... the view that ...individual human beings must see themselves in a particular way in order to constitute a collectivity ... intentions ... are logically prior to collectivities" [Gilbert, 1989, p. 12]).

THE INDIVIDUAL BASES OF INTENTIONAL ACTION

Classic formulations of attitude theory maintain that intentions to act are functions of individual and normative influences (e.g., Eagly & Chaiken, 1993). The Theory of Planned Behavior (TPB) suggests that one's personal intention to enact behavior is a function of the individual's attitude toward the behavior (i.e., the behavior's evaluation), his/her subjective norms (i.e., the perceived social pressure to perform or not perform the behavior), and perceived behavioral control (i.e., the perception of how easy or difficult it is to perform the behavior). The TPB has been used extensively because of its parsimony (Ajzen, 1991). Modifying the TPB to apply to group behavior, we might hypothesize that the virtual community member's we-intention to participate will be determined by his or her attitudes toward participation, his or her perceived pressure from online group members and other central people to participate, and his or her perceived control over the act of participation.

The Model of Goal-Directed Behavior (MGB) subsumes the TPB and has been recently shown to improve the predictive and explanatory power of the TPB (Perugini & Bagozzi, 2001). In addition to the constructs in the TPB, the MGB introduces three classes of individual bases to better explain purposive behavior and its affective implications. One addition to the TPB under the MGB is the incorporation of anticipated emotions—pre-factual appraisals where the individual imagines the affective consequences of goal attainment and goal failure before deciding to act (Gleicher et al., 1995)—as predictors.

Recent research shows that negative anticipated emotions influence the individual's intentions over and above the variables in the TPB. For instance, Parker, Manstead, and Stradling (1995) found that anticipated regret lowered expectations that one would commit certain car driving violations. Similarly, Bagozzi, Baumgartner, and Pieters (1998) found that both positive and negative anticipated emotions influence volitions to exercise and to diet. The rationale for the effects of anticipated emotions is based on the argument that people, when deliberating to act or not in goal-directed situations such as virtual community participation, take into account the emotional consequences of both enacting and not enacting that behavior (Bagoni et al, 1998). This is done through a particular form of counterfactual thinking that Gleicher et al. (1995) term "prefactuals." Gleicher et al. (1995, p. 284) suggest that "...individuals may think about imaginary alternatives to events in terms of the implications of these events for the future ... People's ... behavior ... may well be determined by what the counterfactuals imply for the future..." For example, anticipated negative outcomes may affect decisions through counterfactual processes in two ways:

First, when a person generates a counterfactual that reverses a negative outcome, he or she is likely to make the attribution that there is an effective action that can be taken in the future ... (Second) when an individual thinks about a counterfactual in advance, the motivation to avoid this negative affect influences behavioral choices (Gleicher et al., 1995, pp. 294-295).

Parallel processes can be hypothesized for positive outcomes, where it is assumed that people are motivated to make choices promoting positive affect. Thus, we expect that positive emotions result when the individual imagines the pleasant aspects of the experience when she succeeds in chatting with the other virtual community members beforehand, and negative emotions when she imagines what will happen when she is not able to do so. We posit that these anticipated emotions in turn positively influence the individual's intentions to participate in the virtual community (through their effect on desires, as developed below).

A second addition to the TPB under the MGII is the role of past behavior. A recent meta-analysis examined 64 studies and found robust evidence for the impact of frequency of past behavior on both intentions and future behaviors (Oullette & Wood, 1998). Other authors have proposed partitioning the effects of past behavior into frequency and recency effects (e.g., Bagozzi & Warshaw, 1990). Although seemingly related, frequency and recency effects are conceptually distinct and therefore might carry independent information. For instance, one may have a long history of performing a given behavior without having performed it recently (e.g., a person who curtailed her browsing on the Web for pleasure, because of a time-consuming job), or one may have recently taken up an activity with no prior experience (as when purchasing a computer and using the internet for the first time). Recency of behavior should influence future behavior to the degree that availability and anchoring/adjustment biases occur in information processing (e.g., Tversky & Kahneman, 1974), and to the degree that the activity, whether established or not, has been recently initiated. Based on this discussion, we expect that the more frequent and recent the past participation of the member, the more positive the we-intentions to participate in the virtual community in the future.

Finally, the third augmentation introduced by the MGB is desires as a mediator between attitudes, anticipated emotions, and subjective norms on the one hand, and intentions on the other hand.² Desires are also hypothesized to partially mediate the effects of perceived behavioral control. A number of researchers over the years have pointed out that attitude theory and/or the TPB fail to consider how decisions become energized (e.g., Bagozzi, 1992, pp. 184-186; Fazio, 1995, pp. 271-272). The criticism is that attitudes, subjective norms, and other commonly specified direct determinants of intentions provide reasons for acting but do not incorporate the motivational content needed to induce an intention to act. Drawing on arguments on philosophy, Bagozzi (1992) proposed that desires can function as the proximal determinants of intentions and in so doing channel the effects, if any, of the classical `sreasoned" antecedents (e.g., attitudes) on decisions and intentions.

For desires to function as motivators, Davis (1984) maintains that decision makers must be aware of their desires and accept them as motivating reasons for acting. Frankfurt (1988) also sees a special role for desires. In his view, desires function as determinants of decisions when decision makers give self-reflective consideration to their desires and come to endorse them as motivators to act. Empirical support for the motivational role of desires can be found in Perugini & Bagozzi (2001). Recent research has elaborated on the distinction between desires and intentions demonstrating several structural differences with regard to level of abstraction, temporal construal, and perceived feasibility (Perugini & Bagozzi, '2002). We hypothesize that desires will be strong determinants of the individual's we-intentions, and the remaining antecedents in the MGB will influence we-intentions through desires (except for PBC), which is expected to have both indirect and direct effects on we-intentions. We now develop some additions to the MGB by considering varieties of social influences exerted by the virtual community on the individual member.

--------------

² Belk, Cer, and Askcgaard (2000) consider important aspects of desire but construe it in a narrower sense than we do. Our conceptualization of and role for desires in consumer research are similar to the treatment of desires in the theory of action and theory of mind by philosophers (see below). See also Bagoni (1992) and Perugini and Bagoni (2001, 2002).

VIRTUAL COMMUNITY INFLUENCES ON THE INDIVIDUAL MEMBER

After examining the individual bases of the individual's participation, we turn now to the influences of the group, represented by the virtual community, on the individual's we-intention. The MGB is primarily a model of individual goal-directed behavior and only considers one aspect of social influence: namely, felt subjective norms, which reflect social pressure from significant, others to perform a focal act. However, there is reason to question whether subjective norms fully capture group effects (see Eagly & Chaiken, 1993, for a detailed discussion of this issue).

Compliance Processes

One problem with subjective norms is that although diffuse in scope, they do not cover many important aspects of group behavior. Subjective norms reflect the influence of expectations of others and constitute what Kelman (1974) terms "compliance." The person holding the subjective norms are believed to be motivated by the need for approval from significant others. Subjective norms are typically operationalized in terms of this felt influence in a rather general sense by specifying "other people whose opinions are important to me" as the source of expectations (Ajzen, 1991). For the individual member, such "other people" could either be virtual community members or members of other primary reference groups such as family or friends, or both. A particular group influence, if any, is not singled out, and the comparative function of norms (Kelley, 1952) from any of these social groups is not incorporated directly into the subjective norms concept and its measures. Because of volitional choice, inconspicuous participation, and low barriers of exit associated with participation in virtual communities, we expect that compliance processes will play a less influential role in determining the individual's we-intentions compared with other community influences. Nevertheless, we expect a positive mediated effect of subjective norms on the individual's participation we-intentions through desires in the MGB (see Figure 1). Two other forms of interpersonal influence specified by Kelman (1974) are more pertinent to the study of community influence: internalization and identification.

FIGURE I

Model of Goal-Directed Behavior with Proposed Social Determinants Introduced

Internalization Processes

Internalization is the adoption of a decision based on the congruence of one's values with the values of another. Eagly and Chaiken (1993, p. 639) suggest that such "values" can be construed broadly to encompass beliefs and attitudes, as well as more abstract moral tenets. For a virtual community member, internalization occurs when the individual finds his or her values match those of other group members. A cigar connoisseur may find others who love cigars just as much, or a gardener may meet others equally knowledgeable about azaleas and daffodils, in virtual communities devoted to these topics. In each case, there is a great over-lap of values, facilitating internalization. Further, we expect internalization to play an especially important role for the virtual community member because of volitional choice: individuals is likely to seek out and maintain member-ship in virtual groups with overlapping values.

Internalization is thought to arise primarily from the information communicated between group members, and to reside in the personal meaning of that information for the participant (e.g., Deutsch & Gerard, 1955; French & Raven, 1959). Internalization processes are represented in the current research through the effects of group norms. Here, social influence is captured by shared values or goals perceived by the individual between oneself and other members of the virtual community. To the extent that the member's values and goals are congruent with those of other members of the community, we expect that his/her we-intentions will increase. This is also posited to result, in part, from a reorganization of means-ends frame-works (Kelman, 1974). It is not sufficient for a person to merely perceive reference group influence in order to form an intention to participate in the community. Rather, the individual has to realize that he/she and other community members share one or more common goals in this regard—whether to swap notes about the best brand of cigars, or new varieties of hybrid seeds (see Method). Consistent with this view, recent research has shown that group norms have a strong influence in computer-mediated groups (Postmes et al, 1998). Group norms emerge in such groups through interaction, and are often inferred by members from the text-based communications used here (Postmes et al., 2000). In the present research, we posit that group norms will positively influence both desires and participation we-intentions (see Figure 1).

Identification Processes

Identification refers to one's conception of self in terms of the defining features of a self-inclusive social category (in this case, the virtual community) that renders the self stereotypically "interchangeable" with other group members, and stereotypically distinct from outsiders (Hogg, 1992). Because of identification, the individual develops we-intentions to maintain a positive self-defining relationship with virtual community members. Further, one's self-esteem is boosted to the extent that one's ego-ideal overlaps with that of the others, and acting as the other acts or wants one to act reinforces one's self-esteem. Identification resembles aspects of normative and informational influence (Deutsch & Gerard, 1955), as well as referent power (French & Raven, 1959), and is characterized by the community member's social identity.

We hypothesize that social identity has main effects on we-intentions, where group action is the common referent. Our rationale draws from social identity and social categorization theories (e.g., Hogg & Abrams, 1988; Tajfel, 1981; Terry, Ho, & Duck, 1999) and is supported by the view of communication researchers who show that digital environments may be perceived as socially rich environments allowing social identity to develop for, and subsequently influence, individual members (Postmes et al., 1998). Tajfel (1978) suggested that a person achieves a social identity through self-awareness of one's membership in a group, and the emotional and evaluative significance of this membership. Building on these insights, Ellemers, Kortekaas, and Ouwerkerk (1999, p. 372) recently proposed that three components comprise one's social identity: "a cognitive component (a cognitive awareness of one's membership in a social group—self-categorization), an emotional component (a sense of emotional involvement with the group—affective commitment), and an evaluative component (a positive and negative value connotation attached to this group membership—group-based self-esteem)."

The self-categorization process that constitutes the cognitive aspect of social identity postulates a cognitive categorization process where similarities between self and group members are accentuated as are comparisons of dissimilarities with nonmembers, and the self is perceptually and behaviorally depersonalized in terms of the relevant group prototype (Hogg, 1992). The cognitive self-categorization process has been shown to be operant for communicators in mediated digital environments (Spears & Lea, 1994). Second, emotional meaning of group membership is also thought to be important to the social identification process of the individual (Tajfel, 1978). This emotional component of social identity may be labeled affective commitment, and may be characterized in a manner similar to Allen and Meyer (1996, p. 253), who define affective commitment as "identification with, involvement in, and emotional attachment to," the focal group. Ellemers et al. (1999) studied experimentally formed groups and found that the affective component of social identity influenced in-group favoritism (in terms of evaluative ratings and outcome allocations). Similarly, in the organizational context, Bergami and Bagozzi (2000) found that the cognitive component of social identity had an indirect effect, and the affective component had a direct effect, on organizational citizenship behaviors.

Third, the evaluative component of social identity—group-based self-esteem—has been de-fined as the positive or negative value connotation attached to group membership (Ellemers et al., 1999, p. 372), and arises from evaluations of self-worth derived from membership. Group-based self-esteem has been found to promote actions that produce in-group welfare (e.g., Long & Spears, 1997). The construct validity and the measures of these components of social identity have been demonstrated by Bergami and Bagozzi (2000) and Ellemers et al. (1999). We hypothesize that each of these three components of social identity should influence desires and we-intentions (see Figure I).

METHOD

Subjects and Procedure

A total of 157 active virtual community members (39% female) participated in this research. We wed the screening condition that respondents had to engage in participation in a "virtual chat room" at least 5 hours a week. The self-reported average time spent by participants in the virtual community was 9.58 hours per week (SD = 20.36). Participants were recruited using a snow-ball procedure by six research assistants, and varied in age from 17 to 67 years (mean age = 28.4 years). The study was introduced as an "opinion survey," and participants first listed their favorite online chat room. The most-frequently mentioned chat room was Yahoo Chat, followed by AOL Chat, Blackplanet.com, and MSN. Topics ranged from sports-related (e.g., hockey), to specific city (New York City), age group (teenchat), and general interests (tattoos, stock investing, fitness, etc.). Participants were then asked to "imagine that you are logging on to the online chat room to talk with the group of friends that you regularly talk to." Though the virtual communities mentioned typically have thousands of active members, the self-reported average size of this virtual community group was 4.69 persons (SD = 1.34). The survey was administered to all participants using a pencil-and-paper method. Details of the measures are provided next.

Measures

Attitudes. Four items were used to measure the attitudes construct. Subjects were asked to respond to the following: "On the following scales, please express your attitude toward chatting in the virtual chat mom with the group of friends you identified above sometime during the next 2 weeks." Four 7-point semantic differential items were then presented anchored by "foolish-wise," "harmful-beneficial," "bad-good," and "punishing-rewarding."

Subjective norms. Two 7-point items were used to measure subjective norms and were introduced with the statement, "Please express how strongly most people who are important to you feel you should or should not chat in the virtual chat room with the group of friends you normally chat with". Then the two items were presented as follows:

Most people who are important in my life think I (circle appropriate number): should 1: 2: 3 4:5:6:7: should not chat in the virtual chat-room with friends somethne during the next 2 weeks.

Most people who are important to me would (circle appropriate number): approve 1: 2: 3:4:5:6:7 disapprove of me chatting in the virtual chat room with friends sometime during the next 2 weeks.

Perceived Behavioral Control. Two items were used to measure perceived behavioral control. The first was a 7-point item that asked respondents to react to the query, "How much control do you have over chatting in the virtual chat room with the group of friends you identified above during the next 2 weeks?" The end points of the scale were anchored with "no control" and "total control" and had "moderate control" as the midpoint. The second item asked participants to respond to the statement, "For me to chat in the virtual chat room with the group of friends I mentioned above during the next 2 weeks is:", and used a 7-point difficult-easy scale.

Past Behavior. This was measured with two items. The first item asked, "How many times in the past 2 weeks did you chat together in the virtual chat-room with the group of friends you identified above?" The respondent then entered a number in a blank space: " time(s) with my friends." The second item asked, "How many times did you chat together in the virtual chat room with the group of friends you identified above in a typical 2-week period over the past 6 months?" The respondent again entered a number in a blank space: “time(s) with my friends." The first item captures the "recency" aspects of past behavior, while the second item obtains a robust measure of the past behavior's frequency.

We-Intentions. Two items were used to measure the we-intentions construct The first was a 5-point scale strongly disagree-strongly agree item in response to the statement, "I intend that our group (i.e., the group of online friends that I identified before) chat in the virtual chat room together sometime during the next 2 weeks." The second item was also a 5-point strongly disagree-strongly agree item in response to the statement "We (i.e., the group of online friends that I identified above) intend to chat in the virtual chat room together sometime during the next 2 weeks."

Anticipated Emotions. A 17-item scale developed by Bagozzi et al. (1998) was used to measure anticipated emotions. After verifying that respondents had a goal to chat in the virtual community with their online friends, seven positive anticipated emotions (excited, delighted, happy, glad, satisfied, proud, self-assured) were introduced with the statement, "If I am able to chat in the virtual chat room with the group of friends I identified above during the next 2 weeks, I will feel:" and were measured on 7-point scales with "not at all" and "very much" as anchors, and "moderately" as a mid point. Similarly, 10 negative anticipated emotions (angrily, frustrated, guilty, ashamed, sad, disappointed, depressed, worried, uncomfortable, anxious) were introduced with the statement, "If am unable to chat in the virtual chat room with the group of friends I identified above during the next 2 weeks, I will feel:" and were measured with the same response alternatives as with positive anticipated emotions.

Desires. Three items were used to measure desires. The first was a 7-point disagree-agree item that asked participants to respond to the statement, "I desire to chat in the virtual chat room with the group of friends I mentioned above during the next 2 weeks." The second item asked participants to react to the statement, "My desire for chatting in the virtual chat room with the group of friends I mentioned above during the next 2 weeks can be described as:", and provided seven response alternatives: "no desire at all," "very weak desire," "weak desire," "moderate desire," "strong desire," "very strong desire," "very, very strong desire." The last item presented the statement, "I want to chat in the virtual chat room with the group of friends I mentioned above during the next 2 weeks." A 7-point "does not describe me at all" to "describes me very well" scale was used to record the subject's response.

Croup Norms. Group norms were indicated with 5-point items measuring the degree of shared goals between the self and each of the group members. The items were introduced with the following statements: "Chatting together in the virtual chat room sometime during the next 2 weeks with the group of friends you often chat with can be considered a goal. For each of the people listed below, please estimate the strength with which each holds the goal." The responses were recorded using "very weak," "weak," "moderate," "strong," and "very strong" alternatives. To operationalize group norms, the item measuring strength of shared goals by the self was used as one indicator, and the average of the items measuring strength of shared goals for the other group members was used as a second indicator.

Social Identity. The cognitive component of social identity (i.e., self-awareness as a member of the virtual community) was measured by two items. The first was developed by Bergami and Bagozzi (2000) to measure self-awareness of group membership and consists of an 8-point visual and verbal representation of one's perceived overlap between one's self-identity and identity of the group. The second item instructed participants to "Please indicate the degree to which your self-image overlaps with the identity of the group of online friends you mentioned above as you perceive it" and used a 7-point scale anchored with "not at all" and "very much" and with 'moderately" in the middle.

Affective social identity was measured by two items based on research pertaining to commitment (e.g., Allen & Meyer, 1996). The first was a 7-point item asking, 'How attached are you to the group of your online friends that you chat regularly with?", and had "not at all attached; I have no positive feelings toward the group" and "attached very much; I have very strong positive feelings toward the group" as end points and "moderately attached" as a mid point. The second items asked, "How strong would you say your feelings of belongingness are toward the group you mentioned above?" and used a 7-point scale anchored with "not at all" and "very much" and with "moderately" in the middle.

Evaluative social identity was measured using two 7-point items based on a scale developed by Luhtanen and Crocker (1992). The first item stated: "I am a valuable member of the group mentioned above," and responses were anchored by "does not describe me at all" and "describes me very well" with "describes me moderately well" as a mid point. The second item stated, "I am an important member of the group mentioned above," and used the same response scale as the first item.

Tests of Hypotheses

Structural equation modeling was used to test the models shown in Figure 1. The LISREL 8.3 program was employed for this purpose (Joreskog & &whom, 1999). The goodness-of-fit of the overall models was assessed with chi-square tests, the root mean square error of approximation (RMSEA), the standardized root mean square residual (SRMR), the nonnormed fit index (NNFI), and the comparative fit index (CFI). Discussions of these indices may be found in Bender (1990), Brown and Cudeck (1993), and Marsh, Balla, and Hau (1996). Satisfactory model fits are indicated by no significant chi-squared tests, RMSEA and SRMR values less than 0.08, and NNFI and CFI values greater than or equal to 0.90. Two indicators were used to operationalize most latent variables, except for desires where three indicators were used. This meant using individual measures for all constructs, except for attitudes where the four measures were combined into two indicators by avenging pairs of measures (Bagozzi & Heatherton, 1994). Because the measures of affective and evaluative social identity were highly correlated, they were modeled as indicators of one affective-evaluative factor (see Table 1 below). Social identity was thus treated as a second-order factor, with self-categorization and affective-evaluative reactions as first-order factors loading on it. All tests of models were performed on covariance matrices (Cudeck, 1989).

RESULTS

Table 1 presents the means, standard deviations, and reliabilities of all the measures used. The reliabilities are generally high except for the measures of perceived behavioral control, where the reliability was 0.65. To test for convergent and discriminant validity, we ran a confirmatory factor analysis for the 12 latent variables underlying the extended MGB shown in Figure 1 (i.e., attitudes, positive anticipated emotions, negative anticipated emotions, subjective norms, group norms, past behavior, desires, we-intentions, perceived behavioral control, self-categorization, affective commitment, and group-based self-esteem). This model fit well, confirming convergent validate (258) = 405.58, r » .00, RMSEA = .056, SRMR = .045, NNFI = .94, and CFI = .95. Discriminant validity can be scrutinized by inspection of the correlations among latent variables and their standard errors. The findings in this regard are shown in Table 2, where it can be seen that all correlations are less than 1.00, when we examine the confidence intervals for each. Only the correlation between affective commitment and group-based self-esteem suggests a problem of discriminant validity; thus our decision here to combine the four measures as indicators of a single affective commitment/group-based self-esteem factor. It should be emphasized that the correlations displayed in Table 2 have been corrected for attenuation and that the raw Pearson product-moment correlations of measures across the latent variables are actually lower than these correlations.

TABLE1

Means. Standard Deviations. and Reliabilities of Construct Measures

Scale	Mean	Standard Deviation		Number of Measures	Range
Attitudes	18.97	4.02	.88	4	0-28
Subjective norms	9.18	2.99	.89	2	0-14
Perceived behavioral control	9.04	2.66	.65	2	0-14
Past behavior	13.06	18.53	.95	2	NA
Positive anticipated emotions	34.59	12.47	.94	9	0-63
Negative anticipated emotions	23.49	11.90	.93	12	0-84
Desires	13.72	15.25	.88	3	0-21
Group norms	6.25	1.97	.90	2	0-14
Cognitive social identity	8.35	3.02	.82	2	0-15
Affective/evaluative social identity	18.97	5.90	.95	4	0-28
We-intentions	7.16	1.83	.91	2	0-10

Figure 2 presents the results for the test of the model in Figure I. This model fits well overall: x2 (282) = 438.36, r » .00, RMSEA = .063, SRMR = .058, NNFI = .93, and GFI = .94. As hypothesized under the MGB, desires mediate the effects of antecedents on we-intentions ( β= .63, SE = .10), and positive anticipated emotions influence desires (β = .04, SE = .01). The addition of social identity to the MGB results in a strong effect for this variable on desires: y = .42, SE = .11. Very high levels of explained variance result for desires (R2 = .77) and we-intentions (R2= .69). Inspection of the standardized coefficients in Figure 1 suggests that positive anticipated emotions and social identity have roughly equal effects on desires.

The findings up to this point show that the model of Figure 1 is consistent with the data. To see if desires fully mediate the effects of the antecedents on we-intentions, we added direct paths from attitudes, positive anticipated emotions, negative anticipated emotions, and subjective norms to we-intentions. The results show that none of the direct paths is significant; i.e., g ₁ = .15 (t = 1.30), g ₂= -.03 (t = -1.67), g ₃ = -.02 (t = -1.29),         g ₄ = -.04 (t= -.079)respectively. Thus, desires fully mediate the effects of the antecedents on we-intentions, as hypothesized under the extended MGB.

Table 2

Correlations Among Latent Variables for Examining Discriminant Validity (Standard Errors in Parentheses)

FIGURE 2

Path coefficients and R2 values for the extended Model of Goal-Directed Behavior (factor loadings, error variances, and correlations among exogenous variables omitted for ease of interpretation

GENERAL DISCUSSION

In contrast to the psychology literature which addresses personal intentions, we construed intentions differently to apply to the group setting characterized by the virtual community (see also Bagozzi & Lee, 2002a). Unlike the traditional personal intention that refers to an action that one will do alone (e.g., "I intend to check my email this evening"), we scrutinized we-intentions to capture the social behavior. We-intentions are intentions expressed either in the form, "I intend that we act jointly", or the form, 'I intend that our group performs group activity X." Such we-intentions have only recently been considered by philosophers (e.g., Bratman, 1997; Tuomela, 1995), who have understandably focused on conceptual and logical aspects of the concept, and not on measurement and hypothesis testing concerns. In a sense, a we-intention reflects the intention of a person who construes him- or herself as a social category (e.g., "member of virtual community n and acts as an agent of the group in concert with other group members. We posit that we-intentions are drivers of participation in virtual communities and the expression of its allure for the individual member.

Marketers have evinced considerable interest in measuring personal intentions in traditional settings where consumption has the individual as the referent. In virtual communities in contrast, social interaction is the objective and the draw for the individual participant. The resulting joint communication and the positive experience are the direct products that are consumed by members. We suggest that in this case, we-intentions, encapsulating joint behaviors by the collectivity, are more appropriate and should be measured instead by marketers for predictive or inferential purposes.

The findings in our study show that we-intention decisions to participate in virtual communities are functions of three antecedents. Positive emotions in anticipation of achieving one's goal of participation in virtual communities function as strong determinants of we-intentions and reflect an individual-level criterion. Social identity also drives decisions to participate in virtual communities and reflects a group-level criterion. Positive anticipated emotions and social identity were found to produce their effects on we-intentions through the mediating role of desires. Desires perform a transformative function to motivate decisions to participate with fellow group members in the virtual community.

We investigated a particular kind of group action in the present study and scrutinized the individual's participation decision from the vantage point of a member of the group constituted by the virtual community. Such friendship groups have already been formed, and the shared intention to participate in community interaction together is in a sense an ongoing agreement to act jointly in the future, whenever predetermined conditions arise, such as at a particular time or occasion. Our focus on ongoing groups in the virtual community setting restricts the scope of inquiry, but was deemed necessary for purposes of manageability and to control boundary conditions. We thus avoid issues concerning certain understandings and commitments that respondents made prior to formation of their groups. Our emphasis is on the sustenance of we-intentions, a form of "continuing" we-intentions, but we recognize that the creation of the virtual community and die initiation of first-time we-intentions are worthy of future study. But we wish to note that our emphasis complements nicely the work of social science researchers studying online groups without any preexistence and prior knowledge of group members (e.g., Postmes et al., 1998; Spears & Lea, 1994).

Our focus on online chat rooms constitutes what social psychologists have referred to as fully cooperative group action (Bagozzi & Lee, 2002b). Here, members perform individual actions contributory to the group action—such as sharing information regarding a particular product or object Such groups typically involve consumer-to-consumer communications, and when compared with other groups considered next, are likely to be the most influential in shaping the consumer's enduring opinions and behaviors.

At least two other types of group actions are pertinent from a marketing standpoint to virtual communities. Under partially cooperative group action, members of a group perform coordinated individual actions, but coordination governs only a portion of the group action, something less than full cooperation. Examples of such virtual communities would be bulletin boards, mailing lists, or newsgroups devoted to specific commercial or informational objectives. Here, members may participate in response to earlier communication, or originate communication, but these actions lack the extent of mutual understanding, commitment, and coordination characteristic of fully cooperative group action. So-called partially cooperative group action in such virtual communities also lacks a sense of personal obligation or responsibility on the part of members for performing their own parts to a joint action (indeed, "jointness" is ambiguous or vaguely defined in this situation), and often lacks a feeling of social responsibility for performing extra contributions in the form of mutual support, if needed, to implement the group intention. Such groups are more likely to support transactions rather than relationships between members, and are therefore less likely to evoke the social processes of influence described herein. Indeed, virtual communities of transaction (Hagel & Armstrong, 1997) also fall into this category.

A second distinct type of pertinent group action not considered here is what may be termed as minimally cooperative group action (Baozzi & Lee, 2002b), or what Tuomela (1995) calls "co-action." According to him, co-action is "collective action in which agents—without a joint intention—have the same goal, perhaps mutually believing so and possibly interacting in various ways" (Tuomela, 1995, p. 73). An example would be sports fans standing in unison at a baseball game to hear the national anthem. Many virtual communities such as technical support groups (Constant, et al., 1996), news bulletin boards, and mailing lists, fit this bill. Many virtual communities may be organized and sustained by a commercial entity such as a firm selling a particular type of software, with the intent of collecting and managing consumer feedback. In such communities, most participants have a common goal such as "to get help in using the software," yet they cannot be said to have a joint intention. Future research may benefit from studying the bases of participation in these two types of virtual communities: partially cooperative groups and minimally cooperative groups.

This research provides some important insights for digital environment marketers. Much effort in the last 5 years or so has gone toward creating virtual communities for commercial purposes. Early simplistic thinking of "build, and they will come" has given way to a less obtrusive, hands-off "nurture and cultivate" approach—but even here, the focus of marketers has been on keeping the commercial topic (discussion regarding the product) as the underlying focus of the community. The research presented here suggests that such an emphasis may be somewhat myopic and misdirected. The group, not the product, must be the object of nurturance, for virtual community builders. In our research, even though the communities chosen by participants were vast with thousands of members, the average size of the member's "regular" group was only about five people. This suggests that the allure of these virtual communities for these participants lies in the benefits of social interaction with a small circle of friends. They develop identification with this small group, allowing group norms to form through processes of identification. A vast body of social science research suggests that such groups, once formed, are very influential in shaping and changing the member's opinions, preferences, and actions. Rather than focusing on the product or service, per se, these findings suggest that marketers should focus on providing the right conditions for the same individuals to come together and meet often enough for such groups to form, and then naturally exert their influence on participating consumers.

Further, in the interactive marketing literature, much recent research has extolled the benefits of mass customization or one-to-one marketing (e.g., Wind & Rangaswamy, 2001)— the designing of customized offerings specifically for individual consumers. Indeed, many attributes of digital environments, such as digital information modularity, the network architecture, and availability of extensive behavioral information at the individual level, all facilitate this approach. The research presented here suggests a need to temper this enthusiasm. Even in digital environments where most consumers face a glut rather than a scarcity of information sources, and the freedom to express their preferences more individually, they seem to be impelled to form small social groups to participate and create a common pool of information, often of marketing relevance. The outcome of such social interaction is likely to be a homogenization rather than a dispersion of preferences and behaviors, suggesting that group-level marketing analysis and programs may still be quite appropriate and effective in the digital world (see Kozinets, 1999, for a similar view).

Another important issue for future consideration is the characterization of the type of action, and hence referents for explanatory variables, to which we-intentions apply. Some member actions pertaining to virtual communities may have limited if any sense of jointness to them, and indeed the actions involved could be performed by one person—say retrieving some specific product-related information from a virtual community devoted to the pertinent topic. Other virtual community actions, even within fully cooperative groups, in contrast, are not separable into individual parts that could be performed separately and entirely by one community member. That is, certain virtual community actions appear to require multiple individuals' actions in concert, together, in a particular way to be meaningful. Our focus in this study on interacting with one's "regular" friends in a virtual community is just such a group action. After all, this activity can only occur when all community members are together simultaneously, and act jointly; no one individual can perform this activity by him- or herself alone. Future research on this topic should explore we-intentions for digital collectivities and circumstances where the nature or extent of "groupness" varies. In conclusion, we fully expect this area to be a buzzing hive of research activity, as virtual communities become more pervasive, accepted, and valued by more and more consumers.

REFERENCES

Ajzen, 1. (1991). The theory of planned behavior. Organization Behavior and Human Decision Processes, 50, 179-211.

Allen, MJ., & Meyer, J.P. (1996). Affective, confirmance, and normative commitment to the organization: An examination of construct validity. Journal of Vocational Behavior, 49, 252-276.

Bagoni, R.P. (1992). The self-regulation of attitudes, intentions and behavior. Social Psychology Quarterly, 55, 178-204.

Bagozzi, R.P. (2000). On the concept of intentional social action in consumer behavior. Journal of Consumer Research, 27 (3), 388-396.

Bagorzi, R.P., Baumgartner, H., & Pieters, R. (1998). Goal-directed emotions. Cognition and Emotion, 12, 1-26.

Bagozzi, R.P., & Heatherton. T.F. (1994). A general approach to representing multifaceted personality constructs: Application to state self-esteem. Structural Equation Modeling, 133, 35-67.

Bagozzi, R.P., & Lee, KM. (2002a). Multiple routes for social influence: The role of compliance, internal• intion, and social identity. Social Psychology Quarterly, forthcoming.

Bagozzi, RP., & Lee, KH. (2002b). Intentional social action and the reasons why we do things with others. Unpublished Working Paper, Rice University. Bagozzi, ltP., & Warshaw, P.R. (1990). Trying to consume. Journal of Consumer Research, 17, 127-140.

Belk, R.W., Ger, G., & Askegaard, S. (2000). The missing streetcar named Desire. In S. Ratneshwar, D.G. Mick, & C. Huflinan (Eds.), The why of consumption: Contemporary perspectives on consumer motives, goals, and desires (pp. 98-119), London, UK Routledge.

Bender, P.M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238-246.

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.

Bratman, M.E. (1997). I intend that We J. In G. Holm-strômaintikka & R. Tuomela (Eds.), Contemporary action theory, volume II (pp. 49-63). Dordrecht, the Netherlands: Kluwer.

Browne, M., & Cudeck, R. (1993). Alternative ways of assessing model fit. In KA. Bollen, & J.S. Long (Eds.), Testing structural equation models (pp. 136-162). Newbury Park, CA: Sage.

Burnett, G. (2000). Informational exchange in virtual communities: A typology. Information Research. 5, 4.

Canter, L., & Siegel, M. (1994). How to make a fortune on the information superhighway. New York, NY: Harper Collins.

Constant, D., Sproull L., & Kiesler, S. (1996). The kindness of strangers: The usefulness of electronic weak ties for technical advice. Organization Science, 7 (2), 119-135.

Cudeck, R. (1989). Analysis of correlation matrices using covariance structure models. Psychological Bulletin, 317-327.

Danet, B., Ruedenberg, L., & Rosenbaum•Tamari, Y. (1998). Hmmm...Where's that smoke coming from? Writing, play and performance on Internet Relay Chat. In F. Sudweeks, M. McLaughlin, & S. Rafaeli (Eds.), Network and Netplay: Virtual groups on the Internet (pp. 41-76). Cambridge, MA: MIT Press.

Davis, WA. (1984). A causal theory of intending. American Philosophical Quarterly, 21, 43-54.

Deutsch, M., & Gerard, H.B. (1955). A study of normative and informational social influence upon individual judgment. Journal of Abnormal and Social Psychology, 51, 629-636.

Eagly, A.H., & Chaiken, S. (1993). The psychology of attitudes. Fort Worth, TX: Harcourt Brace Jovanovich.

Ellemers. N.. Kortekaas, P., & Ouwerkerk, J.W. (1999). Selfcategorization, commitment to the group, and group self-esteem as related but distinct aspects of social identity. European Journal of Social Psychology, 29, 371-389.

Fazio, R.H. (1995). Attitudes as object-evaluation associations: Determinants, consequences, and correlates of attitude accessibility. In R.E. Petty & J.A. Krosnik (Eds.), Attitude strength: Antecedents and consequences (pp. 247-282). Mahwah, NJ: Erlbaum.

Frankfurt, H. (1988). The importance of what we care about. Cambridge, UK: Cambridge University Press.

French, J.R. Jr., & Raven, B. (1959). The bases of social power. In D. Cartwright (Ed ) Studies in social power (pp. 150-167). Ann Arbor, MI: University of Michigan.

Gilbert, M. (1989). On social facts. London, UK: Rou-dedge.

Gleicher, F., Boninger, D.S., Strathman, A., Armor, D., Hens, J., & Ahn, M. (1995). With an eye toward the future: The impact of counterfactual thinking on affect, attitudes, and behavior. In NJ. Roese & M.M. Olson (Eds.) What might have been: The social psychology of counterfactual thinking (pp• 283-304). Mahwah, NJ: Lawrence Erlbaum Associates.

Hagel, J., & Armstrong, A.G. (1997). Net gain: Expanding markets through virtual communities. Boston, MA: Harvard Business School Pres.

Hogg, M.A. (1992). The social psychology of group cohesiveness: From attraction to social identity. New York, NY: NYU Press.

Hogg, MA., & Abrams, D. (1988). Social identifications: A social psychology of intergroup relations and group processes. London, UK: Routledge.

Johnson, D.W., & Johnson, F.P. (1987). Joining together: Group theory and group skills, 3rd edition. Englewood Cliffs, NJ: Prentice Hall.

Joreskog, K., & Sorbom, D. (1999). Lisrel 8: User's reference guide, 2nd edition. Chicago, IL Scientific Software International.

Kelley, H.H. (1952). Two functions of reference groups. In G.E. Swanson, T.M. Newcomb, EL Hartley (Eds.), Readings in social psychology, 2nd edition (pp. 410-414). New York, NY: Holt.

Kelman, H.C. (1974). Further thoughts on the pro-cesses of compliance, identification, and internalization. In J.T. Tedeschi (Ed.), Perspectives on social power (pp. 125-171). Chicago, IL: L. Aldine.

Kiesler, S., & Sproull, L. (1992). Group decision making and communication technology. Organization Behavior and Human Decision Processes, 52, 96- 123.

Kozinets. R.V. (1999). Etribalized marketing? The strategic implications of virtual communities of consumption. European Management Journal, 17 (3), 252-264.

Long, K., & Spears, R. (1997). The self-esteem hypothesis revisited: Differentiation and the disaffected. In R. Spears, PJ. Oakes, N.E. Ellemers, & SA. Haslam (Eds.), The social psychology of stereotyping and group life (pp. 299-317). Oxford, UK: Blackwell.

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.

Maffesoli, M. (1996). The time of the tribes: The decline of individualism in mass society. Thousand Oaks, CA: Sage.

Marsh, H., Balla, J., & Hau, K.T. (1996). An evaluation of incremental fit indices: A clarification of mathematical and empirical properties. In G.A. Marcoulides & RE. Schmacker (Eds.), Advanced structural equation modeling: Issues and techniques (pp. 315- 353). Mahwah, NJ: Erlbaum.

Oulette, JA, & Wood, W. (1998). Habit and intention in everyday life: The multiple processes by which past behavior predicts future behavior. Psychological Bulletin, 124, 54-74.

Park, M., & Floyd, K. (1996). Making friends in cyber-space. Journal of Computer-Mediated Communicalion [Online], 1 (4). Available: http://www.ascus-c.org/jcmc/

Parker, D., Manstead, A.S.R., & Stradling, S.G. (1995). Extending the theory of planned behavior: The role of personal norm. British Journal of Social Psychology, 34, 127-137.

Perugini, M., & Bagozzi, R.P. (2001). The role of de-sires and anticipated emotions in goal-directed behaviors: Broadening and deepening the Theory of Planned Behavior. British Journal of Social Psychology, 40, 79-98.

Perugini, M., & Bagozzi, RP. (2002). The distinction between desires and intentions. Working Paper, University of Essex.

Posunes, T., Spears, R., & Lea, M. (1998). Breaching or building social boundaries? SIDE-effects of computer-mediated communication. Communication Re-search, 25, 6, 689-715.

Posunes, T., Spears, R., & Lea, M. (2000). The forma-tion of group norms in computer-mediated communication. Human Communication Research, 26 (3), 341-371.

Rayon, J.F., & Sviolda, _J.J. (1995). Exploiting the virtual value chain. Harvard Business Review, (November-December)

Rheingold, H. (1991). A slice of life in my virtual community. In L.M. Harasim (1¹4.), Global networks (pp. 57-80). Cambridge, MA: MIT Press. Rheingold, H. (1993) The virtual community: Homesteading on the electronic frontier. New York, NY: Harper Collins.

Searle, J.R. (1990). Collective intentions and actions. In P. Cohen, J. Morgan, & M. Pollack (Eds.), Intentions in communication (pp. 401-415). Cambridge, MA: MIT Press.

Shirley, C. (1995) Voices from the net. Emeryville, CA: ZD Press.

Spears, R., & Lea, M. (1994). Panacea or panapticon? The hidden power in computer-mediated communication. Communication Research, 21 (4), 427-459.

Tajfel, H. (1978). Social categorization, social identity and social comparison. In H. Tajfel (Ed.), Differentiation between social groups: Studies in the psychology of intergroup relations (pp. 61-76). London, UK: Academic Press.

Tajfel, H. (1981). Human groups and social categories: Studies in social psychology. Cambridge, UK: Cam-bridge University Press.

feny, DJ., Hogg, MA, & Duck, J.M. (1999). Group membership, social identity, and attitudes. In D. Abrams & MA Hogg (Eds.), Social identity and social cognition (pp. 280-314).0xfond, UK: Blackwell.

Tuomela, R. (1995). The importance of us: A philosophy study of basic social notions. Stanford, CA: Stan-ford University Press.

Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185, 1124-1131.

Walther, J.B. (1996). Computer-mediated communication: Impersonal, interpersonal and hyperpersonal interaction. Communication Research, 23, 1, 3-43.

Wellman, B. (1999). The network community: An introduction to networks in the global village. In B. Wellman (Ed.), Networks in the global village (pp. 1-48). Boulder, CO: Westview Press.

Wellman, B., & Gulia, M. (1999). Net-surfers don't ride alone: Virtual communities as communities. In B. Wellman (Ed.), Networks in the global village (pp. 331-366). Boulder, CO: Westview Press.

Werry, C. (1999). Imagined electronic community: Representations of virtual community in contemporary business discourse [Online]• First Monday, 4 (9). Available: http://www.firstmonday.dk/.

Williams, R.L., & Cothrell, J. (2000). Four smart ways to run online communities. Sloan Management Re-view, Summer, 81-91.

Wind, J., & Rangaswamy, A. (2001). Customerization: The next revolution in mass customization. Journal of Interactive Marketing, 15 (1), 13-32.