The rush to judgment about the social effects of the new communications media has branded them as positive and negative in equal measure. Alienation from “real world’ relationships coupled with a lack of social regulation within the medium is balanced by liberation from the influences, inequalities, and identities to which people are subjected in face-to-face interaction. The authors argue that such general conclusions may in fact be turned upside down and propose that these media may actually strengthen social bonds but also reinforce power inequalities. Reviewing evidence of their research with university students, employing the social identity model of deindividuation effects, the authors show how these technologies can often be more “social,’ and socially regulated, than face-to-face interaction.
Researchers of computer-mediated communication, the Internet, and Information Communication Technologies (ICTs) in general, have pointed to a range of social costs and benefits of these new media. Identified costs include alienation from social relations in the “real world.” Social benefits include the liberation of users from social influences and inequalities to which people are exposed in face to face interaction (group pressure, power relations), while freeing us from the restrictions of our everyday identities. In other words, these communication technologies can be seen as setting up sometimes unwanted barriers with everyday life, while also eliminating unwanted boundaries from everyday life. We provide a theoretical and empirical critique of both of these assumptions, arguing that in important respects these characterizations may often be the wrong way round.
The Internet and ICTs have been accused of having a range of harmful social effects. For example, the influential Homenet study suggests that the Internet leads us to neglect the “strong ties” of our everyday lives (e.g., Kraut et al., 1998). The Internet can become addictive, and social contact often falls below the standards set by face-to-face interaction, producing frustration, flaming, and diverting us from our immediate social responsibilities. In short, concerns of social cohesion and social conduct have put these new media in the dock. We propose that this verdict is premature. Already doubts are being cast about whether the Internet is damaging to existing social relationships (see McKenna & Bargh, 2000).
In the following we also question whether being ensconced in the Internet, really is as damaging to social conduct and character as some theorists and popularisers have supposed.
The second main theme moves us from the social to the political realm. Much has been made of the power of computer communication and Internet as technologies of democratization, and even liberation. Researchers have pointed to the equalizing nature of communication (e.g., Kiesler & Sproull, 1992) as well as the possibilities of freeing ourselves from restrictive or even stigmatized identities (e.g., Turkle, 1995). While not denying these possibilities, when we examine how ICTs are actually used, and by whom, the picture is somewhat less rosy. Where power and hierarchies have become flatter within the organization, there is little evidence that power is less centralized nor that this has been driven by these technologies. Moreover, these media continue to be characterized by the relative social exclusion of disadvantaged groups both from the technology and within it (e.g., Thomas & Wyatt, in press). We certainly do not have all the answers to these apparent paradoxes but we can point to processes by which power relations are reinforced as well as transcended.
Our theoretical critique examines the assumptions, embedded in social psychological and communications theories that have fostered the two dominant themes identified above. Many current theoretical approaches have their roots in the engineering concept of communication bandwidth (Spears, Lea & Postmes, 2001), proposing that limiting information exchange can have deleterious effects on social communication. The implications for the social effects of technology follow directly (“technological determinism”). They are positive where social influences are obstacles for individual performance (as in the realm of Group Decision Support Systems, where determinist perspectives have tended to be utopian). The social effects are negative where social influence regulates behavior and keeps it in check (as is the case in flaming and anti-normative behaviors, where dystopian views prevail). Underlying both of these visions is the assumption that face-to-face interaction is a more truly, richly “social” than interaction in cyberspace. Consequently the ICTs are held to lack qualities of truly social interaction.
As we will argue below the assumptions are more clear-cut than is indicated by the empirical support they receive. There is little compelling evidence that the “social nature” of CMC can be determined in such a direct manner from technical efficiency, whether this is cast in terms of social presence, information richness, or reduced social cues. For example, research suggests that in some contexts, the use of ICTs stimulates anti-normative behavior, whereas in other contexts behavior is more normative, and social influence stronger (Spears, Lea, Postmes, in press). Likewise, contradictory results have emerged for status equalization (Schofield, 1999; Spears & Lea, 1994; Straus, 1997; Weisband, Schneider & Connolly, 1995), social cohesion and attraction (Lea, Spears, & De Groot, 1999; Walther, 1996), and decision-quality (Chun & Park, 1998; Postmes & Lea, in press).
It is important not to underestimate the importance of information-based perspectives, however. Their assumptions are widely shared, and as a consequence we find these conceptions are applied to a broad range of social effects. In response some theorists have dismissed all perspectives that claim to identify the ways in which technology can produce social consequences, concluding that technological determinism is too limited to explain the rich interaction between technological and social influences. This theoretical response has fuelled an equally one-sided counter-movement, which assumes that technologies are principally constructed by their social uses (“social determinism”; see Spears et al., 2000, for a review).
Taking a more interactionist approach we try to plot a course between both forms of determinism, arguing that many of the same social psychological processes and principles apply in virtual communication as in face-to-face interaction, and that the social dimension can actually be enhanced by these media. The empirical part of the paper elaborates this point by presenting experimental evidence from our own research program using the social identity model of deindividuation effects (SIDE; Lea & Spears, 1991; Spears & Lea, 1992; 1994; Postmes, Spears & Lea, 1999).
A SIDE look at anonymity in computer-mediated communication
The two themes outlined above, that CMC is less social but can also be liberating, formed for us a focus to question some of the assumptions and conclusions of established approaches to communication in virtual environments. The current vision seemed to us overstated if not wrong. First, the assumption that computer communication was less social or socially regulated than face-to-face communication is questionable, and depends to a considerable degree on how “social” is defined. Second, we were also wary of the more positive views of the Internet as a political liberator where social differences might be left behind at will. We address these social and political themes in turn.
The idea that interaction via computers is less social or socially regulated follows from restricted bandwidth principles in communication theory, and from restricted definitions of self and social interaction in social psychology. From the perspective of social identity theory and its sister-theory self-categorization theory (Tajfel & Turner, 1986; Turner, 1985), there are important group dimensions to self and social interaction. In these terms, the group is not just an external entity or collection of individuals with which we interact, but it is also internal and identity defining. Similarly “social” interaction is not just a question of interpersonal interaction between individuals: in (inter)group situations, social interaction implicates individuals acting as members of groups or social categories.
These broader definitions of self and situation open the way to a redefinition of social interaction and communication that does not necessarily depend on the quantity or even the quality of information transfer. The social properties of communication do not have to rely exclusively (if at all) on what is transmitted down the wire. The social dimensions of self may already partly determined by how the communicators define themselves beforehand. We have argued that information restrictions of the medium may actually privilege more “social” levels of self-definition (as groups and categories; Spears & Lea, 1992). This is because cues to category membership may be both discrete (simple cues), discreet (subtly communicated, sometimes in language style) and easily discerned (because they reflect shared and sometimes chronically salient features). The individuating cues associated with personal identity are by contrast potentially infinite, complex, and much more abundant in the broader bandwidth of face-to-face communication.
This is one cornerstone the SIDE model. We have argued that the visual anonymity characteristic of computer based communication can have important social psychological effects (albeit determined by the interaction of identity with context rather than being generic). Similar phenomena have been studied by deindividuation theorists investigating the consequences of anonymity in the crowd (e.g., Diener, 1980; Zimbardo, 1969; see Postmes & Spears, 1998). Deindividuation theory proposes that behavior becomes socially deregulated under conditions of anonymity and group immersion, as a result of reduced self-awareness (e.g., Diener, 1980; Prentice-Dunn & Rogers, 1989; Zimbardo, 1969). This notion has been imported by researchers of the ICTs and the Internet to explain antisocial behavior, much as it has been applied to crowd behavior.
However, our conclusions are quite different from those of deindividuation theory.
Although we do not dispute many of the effects obtained in this literature, we do contest the explanation of them as resulting from social deregulation. On the contrary we argue, in line with social identity accounts of collective behavior (Reicher, 1987), that deindividuation effects are highly socially regulated because anonymity can increase the salience and impact of contextually relevant group identities at the expense of personal identity (“depersonalization” rather than deindividuation). To this extent we argue that self-definition and social behavior in anonymous interaction is not only as social as face to face interaction, but in certain important senses it may be more so. Although this claim is controversial it is supported by a meta-analysis of the deindividuation literature (Postmes & Spears, 1998). There is actually no strong evidence for counter-normative behavior under conditions of anonymity, and much stronger evidence of conformity to local norms that could be inferred form the context or group identity. The implication is that if ICTs create deindividuating conditions, behavior can be highly socially regulated in these media.
We have referred to the effect of anonymity on identity salience as the cognitive dimension of SIDE because it influences the accessibility of contextually relevant identities. This dimension of the SIDE model is also defined as cognitive in order to distinguish it from the strategic dimension of the SIDE model (Reicher, Spears & Postmes, 1995; Spears & Lea, 1994). The strategic dimension refers to the actual expression of behavior associated with contextually salient identities, but that takes into account social constraints affecting behavior. For example, accountability to a powerful outgroup may lead members of the ingroup to suppress behaviors that are consistent with their identity, but which transgress outgroup norms and could be punished (Reicher and Levine, 1994a). Conversely the presence of ingroup members may give the groups the collective strength to express behaviors that might otherwise be punished (Reicher & Levine, 1994b), as well as strengthening accountability to ingroup norms (Barreto & Ellemers, 2000). Self-presentation here is not just a question of individual impression management but crucially depends on the identity that is salient, and the relation to the (powerful) audience. As we shall see a number of important features of ICTs come into play when considering this strategic dimension (e.g., the ability to dissimulate and play with identity).
To summarize, the SIDE model consists of two dimensions, relating to self-categorization (cognitive) and self-expression (strategic), and analyzes the conditions that facilitate or impede these. This framework has helped us to account for the richness and variety of effects in computer-mediated communication whilst not losing sight of the underlying social psychological process. We now proceed to present empirical evidence, mostly from our own research program, to back up our claims that these are media which do not escape either the power of the social or social power.