Visibility and anonymity effects on attraction and group cohesiveness

This study investigated attraction and group cohesiveness under different visibility and anonymity conditions for social categories that differed in their capacity to be visually cued. Using computer-mediated communication in 36 mixed gender (visually cued category) and nationality (non-visually cued category) groups, we manipulated social category salience (via discussion topic), and anonymity vs. visibility (via live video links). Under high salience, the effects of anonymity versus visibility were moderated by availability of visible category cues. Visibility increased attraction and cohesiveness for visually-cued groups, whereas anonymity increased attraction and cohesiveness for non-visually cued groups. Path analysis showed that, under high salience, effects of visibility and anonymity were mediated by self-categorization processes, triggered by prototypicality of self in the case of non-visually cued groups under anonymity. In low salience conditions, visibility directly cued attraction independently from self-categorization, in line with relational attraction processes.

Introduction

Attraction and group cohesiveness are fundamental aspects of groups because they are key indicators of the sense in which the group and its members are valued and form a distinct and united entity. They can be used to predict many other group-based outcomes, such as social influence, conformity, group productivity, and performance (Forsyth, 1990; Hogg, 1993). Social identity theory differentiates group- based social attraction from personal attraction (Hogg & Hains, 1996; Hogg & Hardie, 1991, 1992; Hogg, Hardie, & Reynolds, 1995; Hogg & Turner, 1985; Turner, 1982). Whereas personal attraction is the product of specific bonds between people and is unaffected by group considerations in principle, group-based social attraction arises from categorization of self and others in terms of the group. In this paper we examine how visibility and anonymity deferentially affect attraction and cohesiveness in different ‘kinds’ of group, through their impact on self-categorization processes. For certain groups that are visually denoted, self-categorization processes of depersonalization proceed through visual fit. However, because this route is unavailable for groups that are not visually designated, depersonalized attraction proceeds through a different route that benefits from lack of visibility: anonymity.

Both the interpersonal-intergroup continuum proposed in social identity theory (Tajfel, 1978), and the  conceptualization of personal versus  group level self-categorizations proposed in self- categorization theory (SCT: Turner, 1982, 1985), predict that group-based processes will be more pronounced in contexts where group salience is high. In practice however, cross-cutting social identifications are present in many group interactions (Crisp, Ensari, Hewstone, & Miller, 2002; Mullen, Migdal, & Hewstone, 2001) and therefore sources of ingroup social attraction include not only the immediate group, but also wider social categories, such as gender and nationality, to which group members belong. However, the flexibility of the self emphasized in self-categorization theory (Turner, Oakes, Haslam, & McGarty, 1994) means that the relative contribution of these different identity-based sources of attraction in real world groups will be highly dependent on contextual shifts in their relative salience during group interaction. High salience of a given category, such as gender or nationality, cued by the subject of discussion in an interacting group, is likely to give rise to self-categorization with that category, and attraction to other interacting group members who share social category membership. Members of a group sharing the same wider social category are also likely to perceive greater cohesiveness between them when that category is salient during interaction.

Our particular concern here is with the different contextual conditions affecting social category ‘fit’ (the differences perceived and/or expected between one group or category and another; Oakes, 1987), and the resulting salience, that give rise to these different sources of attraction and cohesiveness in discussion groups. We focus primarily on understanding how the visibility and anonymity of group members interacts with the capability of different salient social categories to be visually cued, and the consequential effects on attraction and cohesiveness in newly formed groups. We also consider how the capability of different social categories to be visually cued can affect the process by which the fit between self and the group is established, and delineate two specific processes by which group-member visibility or anonymity, respectively,  promote  group  attraction  and  cohesiveness.  We  use  evidence  of  the  presence  of depersonalization processes to distinguish social attraction from interpersonal attraction. Finally, we explore an additional relational process by which visibility can enhance group attraction and cohesiveness.

Visibility can have powerful effects on social categorization (Biernat & Vescio, 1993). Some social identity researchers have argued that visibility of ingroup members enhances attraction and cohesiveness by increasing the meaning and salience of group membership (Abrams, Wetherell, Cochrane, Hogg, & Turner, 1990; Bhattacharya, Rao, & Glynn, 1995; Elsbach & Glynn, 1996; Mael & Ashforth, 1992; Waldzus & Schubert, 2000). However, research that has specifically examined social identity processes in groups under conditions of visual anonymity has concluded that anonymity, rather than visibility, can strengthen the common identity basis for group perceptions and behavior by reducing visually cued perceptions of interpersonal differences within the group (the SIDE model; Lea & Spears, 1991; Postmes, Spears, Lea, & Reicher, 2000; Reicher, Spears, & Postmes, 1995; Spears & Lea, 1992; Spears, Postmes, Lea, & Watt, 2001). Indeed, anonymity has been observed to increase group attraction when group salience is high, and furthermore the anonymity effect is mediated by self-categorization, signifying group-level social attraction (Lea & Spears, 1992; Lea, Spears, & de Groot, 2001).

Taken together, these two arguments suggest that either visibility or anonymity can increase group- based attraction by increasing group-level cues, or decreasing interpersonal or other extraneous cues, respectively. Apparently neither visibility nor anonymity can be regarded as having universal group-level social attraction effects and their effects likely depend on moderating contextual factors. One important factor may be the congruence between social categories that provide a basis for visual fit and their underlying fit on the group dimension (e.g., real or expected intergroup differences on a discussion topic). Gender is one example of a category where identity is visually cued because the appearance of men and women is recognizably different (age and ethnicity are other examples). Thus on a gendered discussion topic, where men and women are expected to disagree (but agree with each other), visual fit should enhance depersonalization and group attraction by reinforcing the underlying fit on the discussion topic.

Other social categories, like nationality, may provide no visible cues that correspond with the underlying intergroup dimension. Thus in a group discussion between British and Dutch based around differences in European nationality there may be only weak visual fit. In this case visibility may actually detract from depersonalization by making salient individual differences and alternative category cues that cross-cut the underlying differences along national lines. However, as the SIDE model proposes, conditions of anonymity mean that these individuating or cross-cutting cues are reduced and so the salience of national identity can be enhanced.

The absence of factors that could undermine or detract from identity does not mean that national identity will automatically be salient. As early work on the SIDE model showed, if individual identity is currently salient, we would not expect anonymity to effect or enhance the salience of group identity by itself (Spears, Lea, & Lee, 1990). However, if the group discussion is one that provides comparative fit in terms of the real or expected differences along national lines, we would expect nationality to be salient to some degree.

Anonymity may also more actively enhance depersonalization, but via a different route from the case where visibility capitalizes on visual fit. In this case perceivers cannot use the visual cues associated with the category as a whole (visual fit), but must infer interchangeability within the category and self-categorization (depersonalization) from other cues. With regard to the effects of anonymity in the crowd, Reicher (1987) proposed that crowd members infer social category attributes from prototypical members of the crowd. However, when group members are anonymous because they literally cannot see other group members, as here, we propose that they use the self as the most available prototype on which to base depersonalization and self-categorization processes.

There is considerable evidence that people use the self as an anchor to infer category properties when group attributes are unclear (Cadinu & Rothbart, 1996; Otten, 2002). We think this is especially likely when the self is the only exemplar available, or at least the most prominent. In these conditions the self is likely to be seen as prototypical of the group, and used as a basis for inferring group properties. However, the content of the group discussion, and the underling fit it provides should also confirm and reinforce this prototypicality process, by providing increasing evidence that the self and group match up in their agreement on the topic, thereby enhancing self-categorization in group terms. In short we think that for categories characterized by low visual fit, depersonalization may proceed by defining the self as prototypical of the salient group (in the absence of contradictory or cross-cutting evidence under anonymity), and by inferring a match between self and category (self-categorization), as the interchangeability with group members on the underlying discussion topic emerges  (as anticipated consensus is confirmed). Thus, anonymity should enhance prototypically of the self, which then enhances self-categorization in terms of nationality in this case.

Another potential source of group attraction and cohesiveness arises not from identification with a collective entity, but from the network of interpersonal relations within a group. Although the social identity approach in general argues that interpersonal attraction (e.g., Byrne, 1971) does not have any direct implications for group cohesiveness, groups that are primarily based on attachments among group members (such as friendship groups), have been distinguished from identity-based groups (Prentice, Miller, & Lightdale, 1994). However, rather than pointing to the existence of fundamentally different group types, this distinction refers more correctly to different attachment processes within groups. With respect to group development, groups which form on the basis of common identity may also develop networks of interpersonal bonds, just as interpersonal networks may also develop a shared group identity. In their analysis of the social self, Brewer and Gardner (1996) accordingly make a distinction between the collective self (corresponding to perceiving the self in terms of group membership) and the relational self, where the latter corresponds to identities grounded in dyadic personal relationships and small face- to-face groups that are essentially networks of interpersonal relationships (Brewer & Gardner, 1996, p. 83). Where the relational self is concerned, group attraction is isomorphic with attraction to individual group members, and perceptions of strong ties within a unified group, or group cohesiveness, follows from a pattern of consistently positive interpersonal evaluations.

Analytically, one way to distinguish this relational basis for attraction and cohesiveness from the social identity-based attraction processes described above is from the degree to which it occurs independently from conditions of high group salience (Tajfel, 1978; Turner, 1985) and from group level processes such as self-categorization (Hogg & Hardie, 1991). Combining this perspective with our analysis of visibility and anonymity, we argue that visibility should also provide a source of attraction, independent from the depersonalization processes described above, that is based on the availability of interpersonal cues and attraction commensurate with the relational self.

In the following study, discussion groups interacted under visibility or anonymity conditions. The group was composed of males and females (visually cued categories) and different British and Dutch nationalities (non-visually-cued categories). We predicted an asymmetry of effects under high category salience such that visibility would enhance group-based attraction and cohesiveness for gender groups (‘visually cued depersonalized attraction’) and anonymity would enhance group-based attraction and cohesiveness for nationality groups (‘anonymous depersonalized attraction’). In addition, we predicted that under low category salience, visibility would increase attraction and cohesiveness for both groups (‘visually cued relational attraction’). As well as assessing differences between groups on the main outcome measures of attraction and group cohesiveness, as a function of visibility and anonymity, we measure prototypicalty of the self and self-categorization as indicators of group-based processes. Under high salience conditions we predicted that visibility would directly increase self-categorization in gender groups, but that anonymity effects on self-categorization in nationality groups would be mediated through perceptions of the self as prototypical of nationality.

To continue reading, download the full text pdf.

Background

This paper was part of the output from a 2-year research project “Visual anonymity effects in group computer-mediated communications” funded by the Economic and Social Research Council (ESRC) under the Virtual Society? Research Programme. Russell Spears and Susan Watts were members of the project.

Non-technical summary of the project

The proposed research will focus on  various aspects of social behaviour that are of importance as the Internet emerges as the ubiquitous global communication medium of the virtual society. We identify visibility versus anonymity as a key variable influencing communication and behaviour in the computer medium. Two laboratory experiments will be conducted on group interactions using a video conferencing system to test the predicted social psychological processes responsible for the effects of anonymity. The following research questions will be addressed.

Does the technical capacity of the medium to overcome national barriers to communication equate with a social capacity for Internet communicators to disregard national stereotypes, or can the medium accentuate apparent differences between groups leading to increased stereotypical perceptions with consequent deleterious effects on intergroup relations?

Secondly, does a lack of individuating, personal information in the medium promote impersonal interactions, a loss of social cohesion within groups, and reduced adherence to group norms, or does it change the basis interaction between individuals by promoting group identification, cohesion, attraction and normative behaviour?

Thirdly, how will the advent of desktop video conferencing to Internet communications affect intergroup interactions?

Will it “repair” Internet communications by offering a greater degree of equivalence with face-to-face interaction, or will it merely undermine the advantages of the medium that accrue precisely because of the reduced interpersonal dimension and degree of personal anonymity that it offers?

How does video-mediated communication differ from face-to-face interaction and with what effects on the individual and the group?

The answers to these and similar questions fundamentally turn on the relative balance between the interpersonal versus group bases for interactions in the computer medium relative to face-to-face interactions and how visibility and anonymity can affect the balance.

The results of the proposed research will influence attitudes towards the computer medium and its effects, and offer guiding principles for the use of video conferencing systems.