SIDE-VIEW: A social identity account of computer-supported collaborative learning

Much is written about the benefits of computer-supported collaborative learning (CSCL) but many accounts report problematic interactions and less than desirable outcomes for attempts at CSCL. A theoretical approach is presented that is grounded in a social identity perspective of groups and seeks to promote and support successful collaborations. Using this approach, along with examples from laboratory and field studies, key concepts such as cohesion, participation, accountability and group norms are addressed and contrasted with more traditional approaches. It is suggested that design features that ‘collectivise’ rather than ‘personalise’ CSCL can promote psychological issues that are crucial for successful collaborations (i.e. group cohesion, accountability to the group, and increased participation), and norms for success can be developed through intergroup comparisons. The theoretical approach is also presented as a platform from which further investigations and predictions can be made.

Introduction

Computer-supported collaborative learning (CSCL) aims to combine the benefits of small group learning with those of computer-mediated communication (CMC). Proponents of collaborative learning claim that the active exchange of ideas within small groups increases interest among the participants and also promotes critical thinking (e.g. Kumar, 1996). In addition, according to Johnson and Johnson (1994), collaborative teams can achieve at ‘higher levels of thought’ and retain information longer than students who work individually. Amongst other advantages, the use of CMC enables students to be brought together irrespective of geographical distance, to make use of the asynchronous aspect of CSCL to overcome temporal difficulties, to allow time for reflection, and also have control over archived resources (Kim, 2008). Moreover CSCL can be implemented in both distance and blended learning. However, the CSCL literature is littered with reports of high dropout rates and anecdotal evidence of poorly performing collaborations (e.g. Chambers, 2000; Cole, M., 2009; Palloff & Pratt, 1999), and it is acknowledged that there are limitations in the knowledge of how best to support learning communities in CSCL (Stahl, Rohde & Wulf, 2007).

A review, by the authors, of the substantial CSCL literature suggests a high proportion of descriptive case studies that simply outline the technological tools for CSCL and do not address the conditions under which use of these tools is likely to result in successful collaborations, while factors influencing success or failure are often attributed post-hoc. This, we suggest, significantly impedes progress in the successful use of CSCL. Similarly, Suthers (2006), in a review of the CSCL research community, observed that studies generally belong to one of three categories: descriptive, experimental, and iterative design approaches, and therefore lack the predictive features that could be used to inform CSCL design. Where more theoretical approaches have been taken, these tend to be based upon underlying assumptions about the nature of groups and communication technologies and their basic incompatibility, leading to attempts to overcome ‘inadequacies’ in the technology to support groups online (e.g., Yamaguchi, Bos & Olson, 2002). The questioning of this underlying assumption of groups and CMC is central to the theoretical approach outlined in this paper which aims to establish the conditions under which successful collaborative learning can flourish and, to this end, utilise, rather than attenuate, the characteristics of CMC technology. We begin below by outlining the theoretical background which is grounded in the well established social identity approach (e.g. Tajfel, 1978; Turner, Hogg, Oakes, Riecher & Wetherall, 1987). We then go on to apply this approach to CSCL giving consideration to four fundamental dimensions of successful group behaviour (cohesion, accountability, participation and norms) setting out the theoretical predictions, as well as empirical evidence, and suggestions for future developments.

SIDE-VIEW: a social identity approach to CSCL 

Our point of departure from other CSCL perspectives derives from an analysis of various types of social cues in CMC and their relative availability under different interaction conditions (Lea & Giordano, 1997), combined with a theoretical analysis of the self and of group behaviour grounded in social identity and self-categorization theories (Tajfel, 1978; Tajfel & Turner, 1986; Turner et al. 1987). The resultant SIDE analysis of CMC (Social Identity model of Deindividuation Effects) has been developed and elaborated over more than a decade (Lea, Spears & Rogers, 2003; Postmes, Spears & Lea, 1998; Spears & Lea, 1992, 1994; Spears, Lea & Postmes, 2007). A substantial body of empirical work testing predictions of group behaviour with email, computer conferencing, bulleting boards, and video conferencing technologies under different conditions has been carried out both in our laboratory (e.g., Lea & Spears, 1991, 1992; Lea, Spears & de Groot, 2001; Lea, Spears & Watt, 2007; Postmes, Spears, Sakhel, & De Groot, 2001; Rogers & Lea, 2005) and in independent labs (e.g., Douglas & McGarty, 2002; Flanagin, Tiyaamornwong, O'Connor & Seibold, 2002; Sassenberg & Kreutz, 2002; Nowak, 2003; Scott, 1999).

Within this approach, an important distinction is made between interpersonal cues, which identify and individuate the communicators, and cues to group identity and social category membership. Interpersonal cues rely more on physical appearance and non-verbal cues such as gesture and eye gaze for their communication in real-time and as such are best communicated in face-to-face interaction. While video conferencing degrades the quality of these cues (Parkinson & Lea, 2009), text-based CMC, which relies solely on the verbal channel, provides the slowest and least efficient communication of interpersonal cues. On the other hand, group and social category membership can be communicated easily and relatively independently from bandwidth considerations (e.g., through text, or static image). Even an email address can communicate many category cues such as gender, work or college affiliation, country of residence, and so on, whereas cues that uniquely identify an individual and express the individual's socio-emotional relationship with you are far more subtle and difficult to communicate (Lea & Spears, 1995).

The importance of this distinction is in its implications for identity and action. As a consequence of their more efficient communication in CMC, group level cues have relatively more opportunity to define the situation, reinforce the communicator's self-definition and shape the interaction. The degree to which this influence is achieved depends upon the social context established for the interaction and how it frames the communicator's self- definition. According to the social identity approach, the self comprises both unique individual aspects and shared aspects, or social identities, derived from membership of different social groups and categories. When the social context makes one or other social identity salient, individual behaviour tends to be more in line with the normative values, attitudes and behaviours associated with the respective social group or category. In other words the communicators act more as group members when a group identity is salient than when their unique individual identities are salient. When a salient group identity is combined with features of the communication environment that promote the communication and perception of group level cues over interpersonal cues then perceptions, attitudes, behaviour and interactions become yet more group influenced.

This fundamental prediction has chiefly been demonstrated with respect to anonymity in CMC where various forms of social influence and social attraction have repeatedly been shown to be positively enhanced by combining group salience with anonymous communication. Under these conditions, individual attitudes and attitude change are more influenced by the group, group norms are more efficiently communicated throughout the group, knowledge-sharing is increased, and decision-making is more in line with prevailing group norms (Cress, 2005; Lea & Spears, 1991; Postmes & Spears, 2002; Postmes, Spears & Lea, 2002; Postmes, Spears, Sakhel, & De Groot, 2001). At the same time, attraction to the group, appreciation of the group's communicative style, group cohesion, trust, and social support all increase (Lea & Spears, 1992; Lea, Spears & de Groot, 2001, Lea, Spears & Watt, 2007; Lee, 2004; Postmes & Spears, 2002; Postmes, Spears & Lea, 2000; Spears, Lea, Corneliussen, Postmes & Ter Haar, 2002; Tanis & Postmes, 2005).

The overall effect of combining group salience with anonymity is therefore to reinforce group normative behaviour. Moreover, some of these studies have shown that different forms of anonymity (e.g., visual, nominal) and different ways of achieving group salience, as well as the content of the communication itself interact to produce subtly different effects in line with SIDE predictions (see Spears, Lea & Postmes, 2007 for a review of these issues). Together they provide a powerful toolkit for influencing group collaboration, and we have begun to apply these findings to develop an alternative social identity-based perspective on CSCL system design and practice, which we term SIDE-VIEW (Lea, Rogers & Postmes, 2002).

The fundamental principle informing our approach is the need to establish congruity between the self-definition of group members and their collaborative learning tasks. Faced with a collaborative task, learners who perceive high congruity are more likely to engage in socially self-enhancing behaviour, such as making mutually supportive coordinated contributions to their learning group. In contrast, low congruity designs that do little to reinforce a group identity or a group collaborative environment are likely to produce less adaptive responses to collaborative learning goals such as personally self-enhancing behaviour (in which the individual is labouring for herself) or socially self-retarding behaviour (such as free-riding) (Lea, Rogers & Postmes, 2002). Elsewhere we have begun to develop and evaluate some of the design features of our approach with distance-learning and CSCL students (Lea, Rogers & Postmes, 2002; Watts, Nugroho & Lea, 2003). In this paper we further elaborate these ideas and their implications for a number of social psychological issues that are crucial for successful CSCL: 1) establishing group cohesion, 2) promoting accountability within the group, 3) ensuring effective participation, and 4) developing norms for successful collaboration. In each case we identify the unique features of a social identity approach to the issue and discuss some of the implications for CSCL system design.

Cohesion

Group cohesion is generally recognised to be an integral feature of successful collaboration and a key variable for increasing productivity in groups (Fisher & Ellis, 1990; Mullen & Cooper, 1994), and the formation of cohesive CSCL groups is therefore an important goal. We agree with this conclusion, but argue that the social psychological and technological factors that underpin cohesive CSCL groups have been misunderstood in the literature, and we propose an alternative perspective based upon our social identity approach to CMC.

In CSCL generally, the definition of group cohesion together and the techniques that supposedly encourage its development in CSCL groups have been derived from the traditional social psychological perspective of small group dynamics (e.g. Clouder, Dalley, Hargreaves, Parkes, Sellers & Toms, 2006; Duffy & Shaw, 2001). According to this perspective, the aggregate of interpersonal bonds of attraction within a group is considered to be equivalent to group cohesion, and interpersonal attraction is seen as an important determinant of both group membership and solidarity (McGrath, 1984, Hogg, 1996).

However, as noted earlier, text-based communication (e.g. email, bulletin boards, chat rooms, wikis) which are commonly utilised for CSCL convey only a fraction of the interpersonal information that is available in face-to-face interactions through cues such as appearance, gesture, facial expression, posture, and tone of voice. The implication is that much CMC technology prevents the formation of cohesive groups, and that additional measures therefore need to be taken to emulate face-to-face interactions or otherwise maximise interpersonal interactions in online learning groups in order to achieve cohesion and successful collaboration (Greenberg, 1998). This has led in turn to various attempts to overcome channel restrictions on the spontaneous communication of interpersonal information by strategically exchanging photographs of group members (e.g. Shapiro et al, 2002), or personal profiles (e.g. Brandon & Hollingshead, 1999). Alternatively, video-conferencing which supports visual communication of interpersonal cues has been advocated, or the use of face-to-face meetings supplement online activities has been argued for (e.g. Alpay, 2005). However, apart from the questionable success of these approaches to producing cohesive CSCL groups, which has yet to be demonstrated, these activities seem to undermine one or both of the key advantages that accrue from e-learning and CSCL, namely geographical and temporal flexibility.

An alternative approach to cohesiveness in CSCL lies in the application of the social identity approach to groups, which challenges the traditional formulation of cohesiveness based on interpersonal attraction. According to the social identity approach interpersonal attraction is not a necessary condition for group formation or cohesion.

Instead, group cohesion is a function of the perceived similarity between the self and others in terms of the defining characteristics of the ingroup (Hogg & Turner, 1985). In other words, cohesion is dependant on relevant shared social identities rather than interpersonal bonds. When group members identify with a salient group, this increases their attraction to the group, which in turn creates group cohesion (Hogg, 1993).

We suggest that the implication of this theoretical approach to group cohesion is that it may be much less dependent upon cue-rich interpersonal interactions in CSCL. Put simply, the amount of information required to transmit a social, or group identity, can be much smaller than is necessary to share interpersonal information and create interpersonal bonds. Whereas interpersonal cues are restricted in text-based CSCL, cues as to the membership of social categories can be very easily communicated. For example, the textual information provided in message headers can provide category information such as gender and ethnicity (often from one's name), organisation affiliation, group membership, and more. Moreover, these simple cues to social groups and categories can convey significant social and behavioural information (from our knowledge of the behavioural norms and stereotypes associated with them) that is independent from interpersonal cues.

From a social identity perspective therefore the text-based and relative anonymous nature of CMC can be exploited under salient group identity conditions to promote cohesion. Combining these observations about cohesion with the SIDE model of CMC suggests that the creation of cohesive online learning groups involves the counter-intuitive approach of providing fewer, rather than more opportunities for group members to exchange personal information and interpersonal cues and develop interpersonal acquaintanceships as individuals. The challenge instead is to focus attention on the collaborating group entity and to provide opportunities to exchange information about the collaborating group and the interactants' group memberships, to ensure that this social identity (rather than personal identities) is salient for all group members. Group cohesion then necessarily develops as a consequence of these conditions.

Previous SIDE research has demonstrated that visual anonymity increases social identification and group cohesion compared to video conferencing conditions (Lea, Spears & DeGroot, 2001). It should therefore be possible to employ anonymity to increase identification and cohesion in CSCL groups, and we have tested this idea both in the laboratory and the field. However, computer-mediated technologies vary in the degree of anonymity that they offer. Even under visually anonymity, textual information (e.g. a name) can convey significant social information. In order to promote the collaborating group as the salient social identity for all group members it is suggested that complete anonymity of individual identities could be employed in early interactions in CSCL groups. This entails not only visual anonymity, but also the removal of nominal identifiers. A laboratory study was designed to test this in which all participants undertook a group task under conditions of visual anonymity, but in some groups participants were identifiable by their name and in others only by a moniker ‘groupmember1’ for example. This study confirmed that identification at the level of the collaborating group was greater under conditions of nominative anonymity than when identified by ones name (Lea, Spears & Rogers, 2003). Moreover, in a subsequent field study of online learning groups in which mixed nationality students were participating in an initial chat meeting, increases in cohesion was found when group members were anonymous by name (Lea, Rogers & Postmes, 2002).

It may be impractical and indeed undesirable, due to accountability issues for example, to maintain nominative anonymity between members of an e-learning group for any length of time. However, there are other mechanisms which can be employed to ensure that the online learning group is the identity that remains salient during the period of collaboration. This involves focus on and developing the group identity in a number of ways (Lea, Rogers & Postmes, 2002). Groups can be encouraged to ‘collectivize’ (as opposed to personalize) their CSCL environment through the use of web page designs and icons agreed by the group. In addition, groups can be given specific tasks that require them to reflect on their group as the collaboration develops. This ensures that the social identity of the collaborating group is kept salient throughout the life of the group. These tasks contrast with traditional ‘ice- breaking’ activities used in CSCL in which personal identities are made salient during initial group meetings by personalizing communication and encouraging interpersonal acquaintanceship (e.g. Hyo-Jeong & Brush, 2008).

We have compared the two contrasting strategies of collectivising group communication versus exchanging personal information in a field study of e-learning among tutorial students enrolled in on-line distance education (Lea, Rogers & Joinson, 2003). Groups were enrolled on the same module and were assigned to complete specific collectivizing tasks or else the standard on-line tutorial activities (including personalizing tasks). It was found that ‘collectivized’ group communication had the effect of increasing the importance students attached to the online tutorial group and reducing their desire for a face-to-face meeting, particularly among students with high conferencing anxiety. It also increased accountability and consensus within the group, as well as satisfaction, particularly for students who did not take up an invitation to attend a face-to-face meeting. The results suggest that the collectivizing approach to developing CSCL groups has significant benefits, in particular for students who would be considered vulnerable to dropout, i.e. those with high anxiety about the online conferencing and those who did not attend a supplementary face-to-face session.

In sum, the social identity approach to developing cohesive CSCL groups focuses on building and making salient the shared identity of the online learning group, and encouraging communication among students as members of the same group rather than the usual communication approach that only serves to individuate students from one another and create interpersonal bonds of liking and disliking. In contrast with traditional approaches, our SIDE approach to CSCL attaches far less importance to interpersonal bonding particularly at the early stages of group formation when the establishment and development of an appropriate social identity is key. Of course, interpersonal bonds between group members may develop during collaboration and the practical application of this approach does not suppress or deny them. Neither however, does it encourage their development. Instead, social features of the collaborative environment (e.g., instructions and activities) and technical features of the communication system (e.g., anonymity) are configured so that they are congruent with the formation and development of an appropriate group identity (Lea, Rogers & Postmes, 2002). Future research intends to further explore these ideas through the development of a portfolio of tasks to support collectivization, to implement these in a variety of CSCL contexts, and to develop systems to automate such interventions.

Accountability

In a review of collaborative learning methods, Slavin (1993) reports that a key aspect of all approaches concerns the development of individual accountability within the group. In other words, the success of the group must depend on the individual contributions of every group member which ensures that group members are held responsible for their share of the workload. Slavin (1993) reports that when individual accountability is in place alongside group goals, the achievement effects of collaborative learning are consistently positive. Low individual accountability has been argued to be a particularly important concern in conditions where group members cannot see one another (e.g. Chiu & Hsu, 2004), and in order to increase levels of individual accountability in CSCL it has been suggested that individual efforts should be monitored, unique tasks given to group members, and evaluations of individual efforts should be fed back to students (Paloff and Pratt, 1999; Chiu & Hsu, 2004).

However, emphasizing individual accountability can undermine the cohesiveness of the group entity (Graham & Misanchuk, 2004), and social identity theory suggests this is particularly troublesome if the salience of personal identity is promoted over a common shared group identity. Moreover, in theoretical terms, just as attraction can be viewed as interpersonal or group-based, we believe that so too can a useful distinction be made between personal accountability of one individual to another and social accountability between a group member and the collaborating group as a whole. Arguably, the latter is likely to be more closely associated with individual participation and performance levels in group situations. While the mechanisms described in the section above to promote cohesion and group salience in e-learning should reduce the salience of interpersonal ties and hence personal accountability, nevertheless individuals should feel accountable as group members to the extent that they identify with the group. The distinction here is subtle, but can have important psychological consequences. The crucial point is whether one is accountable as an individual because of external evaluation, or accountable as a group member because of feelings of common fate.

Indeed, we investigated this in a laboratory study of online groups in which group members who were collaborating on a decision-making task via computer-conferencing were either completely (i.e., visually and nominatively) anonymous or else identifiable by name to other group members (Lea, Spears, Watt & Rogers, 2000). Anonymity was found to influence accountability through two independent and opposing processes. First, anonymity reduced group members’ feelings of personal accountability to others (individual accountability). However, at the same time it increased members' identification with the group which had the effect of increasing accountability within the group (social accountability). Social accountability, founded on shared group identity can therefore work under the relatively anonymous conditions of on-line collaboration to offset any deleterious effects anonymity may produce by reducing personal accountability. We suggest that if accountability is located within the collaborating group itself then increasing group identity salience, identification and cohesion can reduce the necessity to introduce external conduct regulators, since that function is performed by the group itself. Future research is ongoing to further disentangle these two elements of accountability, their functions and the mechanisms by which we can support social accountability.

Participation

It has been discussed above, that both accountability and group cohesion can be underpinned by a shared group identity. However, it is not clear how or even if these factors, manifest themselves in terms of increased motivation to participate in CSCL. It is the motivation to participate that is the driving force behind individuals working hard for their group, and without this motivation participants are unlikely to achieve satisfactory levels of performance. This is particularly pertinent given the noted problems of CSCL discussed earlier. Indeed, Hrastinski (2009, p.78) states that ‘If we want to enhance online learning, we need to enhance online learner participation’. Moreover, he suggests that it is important to consider online participation as more than simply ‘doing’, but also ‘belonging’, i.e. feeling attached to a group with a sense of community (Hrastinski, 2009). This is central to the social identity approach to e-learning outlined here and elsewhere (e.g. Rogers & Lea, 2005). The challenge for us now therefore, is to ensure that individual group members are motivated to work for and increase the performance of their group throughout the collaborative period.

A robust finding in many empirical studies is that individuals working in groups are less productive than when working alone, and this has been termed ‘social loafing’ (Latane, Williams & Harkins, 1979). A similar concept is that of ‘free- riding’ whereby individuals make a strategic decision to reduce their contribution to a group task (Haslam, 2004)1. Despite the many factors pertaining to mediate social loafing and free-riding, one of the most commonly alluded to explanation, especially in the e-learning literature, is personal identifiability (e.g. Palloff & Pratt, 1999). This is also referred to as evaluation potential or individual accountability. Indeed, research by Harkins and Szymanski (1989) suggests that even the prospect of evaluation potential can eliminate social loafing. Again, as with the previous section, the implication of this approach for CSCL is to increase the individual over the group in order to motivate the group members and ensure they do not engage in loafing behaviour. While some aspects of accountability have been discussed previously, greater analysis of social loafing is warranted.

The social identity approach offers an alternative perspective to learner motivation and social loafing. Haslam (2004, p.252) suggests that “one significant determinant of task productivity is the congruity between a person’s self- definition and features of the task environment” (hereafter referred to as the congruity hypothesis). If an individual defines him/herself in terms of their unique personal identity rather than a shared group identity and the task requires group collaboration then the incongruous nature of these factors is likely to result in free-riding behaviour (cf. Worchel, Rothgerber, Day, Hart, & Butemeyer, 1998). In other words, those who define themselves in terms of unique personal identities should be best equipped to perform tasks that appear to demand and reward personalized and independent input, but those who define themselves in terms of a shared sense of social identity should do best in tasks that demand collaborative participation. As the benefits of CSCL accrue from the collaboration of group members our approach suggests that a salient social identity will serve to establish congruity and therefore increased participation.

We tested this congruity approach to participation in CSCL was tested in a field study of 132 distance-learning students enrolled on a level one module. Ten tutorial groups (approx. 6 students per group) in a control condition were compared with eleven groups in which ‘collectivising’ tasks described above were implemented (Lea, Rogers & Joinson, 2003). It was found that students who reported low motivation perceived more difficulty in free riding when interactions were within the ‘collectivised’ group communication structure. In addition, the use of these structures supplanted the need for face-to-face tutorials, highlighting the useful of this approach to CSCL. Current research is investigating the specific processes whereby participation rates can be enhanced using this congruity approach across a number of CSCL contexts.

Norms for success

Our approach uses the social identity of the group in CSCL to promote cohesion, accountability to the group members and participation through congruity with the collaborative nature of the activity. However, the effect of identification on motivation is not a direct one, but is mediated by a number of factors (Van Kippenberg, 2000). One fundamental aspect of the social identity approach is that the sense of psychological oneness that identification with the group elicits induces individuals to take on the group's norms and to experience the group's goals and interests as their own (Haslam, 2004). However, whether identification results in higher motivation to participate fully and perform well in CSCL is dependent importantly on the specific goals and norms of the group. It may be the case for example, that the group does not agree with an imposed learning goal, or that the norm of the group is to do as little work as possible, rather than to perform well. In these situations high identification with the group may increase motivation to assume group norms and goals but is unlikely to lead to improved participation and performance. Norms are socially constructed and in CMC have been shown to be limited by the boundaries of the ingroup (Postmes, Spears & Lea, 2000). It is therefore crucial that appropriate norms are developed within the ingroup and encouraged throughout collaborations.

One way in which group norms can be manipulated is through the use of intergroup contact. It is usual in CSCL that students are split into small groups and so different groups working on similar problems or tasks can be utilised to develop appropriate norms. Indeed, the online educational literature, claims that intergroup contact helps groups to benefit from shared experiences and learning strategies (both positive and negative), and can provide groups with motivational support and new insights (De Simone, Lou, & Schmid, 2001; Lou & MacGregor, 2004). However, little discussion is given to the motivational processes that lead to these findings. In contrast, intergroup relations are central to the social identity approach, which can therefore provide some theoretical underpinning for these observations as well as suggesting additional implications. An intergroup framework is explicit in the social identity approach in which a group is defined in terms of its relationships with other groups (Tajfel, 1978; Hogg & Abrams, 1988). Tajfel (1982) notes that in order for groups to protect, enhance or achieve a positive social identity for members of the group, it is necessary for the group to be positively distinct from other groups. Moreover, predictions from the SIDE model suggest that in relatively anonymous computer-mediated communication, intergroup differentiation should be exacerbated through depersonalising perceptions and focusing on the shared group identity (Riecher, Spears & Postmes, 1995; Spears & Lea, 1994). In addition, reactions to perceived status differences between groups are predicted, and can vary, particularly in response to unfavourable comparisons which can both threaten the ingroup social identity and motivate group members to improve its standing (Ouwerkerk & Ellemers, 2002).

Applying this theoretical approach to CSCL, we suggest that the presence of a relevant outgroup can serve to strengthen the boundary of the ingroup and thus increase group identification (NB. the focus of attention is, again, at the group level, rather than an interpersonal level). Indeed, we found that in a field study of online learning groups, participants rated their own group as more distinct under conditions of a specific intergroup comparison (Lea, Spears & Rogers, 2003). Having established group cohesion through a shared group identity, one task required students to compare the summaries of their own group work, with those of another group working on a similar project. They were then required to offer feedback to the other group on issues such as structure and content. Although there was no explicit competition between groups (i.e. there were not limited resources), only comparison, this still had significant effects on the collaborating groups. Most notably, during the intergroup comparison phase of the collaborations, levels of task conflict (as distinct from interpersonal conflict) in the group increased, and resulted in positive outcomes for the group through enhanced critical analysis. One explanation is that the ingroup boundaries were reinforced by the intergroup comparison and so enabled productive disagreement and debate that was not a threat to the shared group identity. This strengthening of group boundaries and therefore identity has also been shown to play a positive part in reducing social loafing (Worchel et al., 1998).

While the introduction of an intergroup context can provide practical effects of sharing content and group working experience, it can also have significant effects on the definition of the ingroup in terms of group norms. In addition to reinforcing group boundaries, intergroup comparison can provide what is termed referent informational influence, i.e. information from relevant others about what appropriate norms are (Hogg & Abrams, 1988). Comparison groups are peer groups and so they share a superordinate identity as part of the class. Indirectly this means that ingroup norms are more likely to be in keeping with externally imposed standards and goals required for the class set by the instructor. We suggest that one implication of this is that the intergroup comparison can provide a means of ‘reality testing’ whereby optimism and complacency that may be buoyed by a strong shared group identity (cf. groupthink, e.g. Janis, 1982) could be checked and addressed.

In sum, for CSCL to be successful, appropriate norms need to be in place and these can be encouraged by employing intergroup mechanisms. Indeed, e-learning offers ideal conditions for intergroup comparison and should therefore be utilised to this end. Moreover, there are many dimensions to intergroup behaviour such as whether it involves a comparison or competition, if feedback is given to groups or not, and to what extent comparisons are favourable or unfavourable, with a wealth of social identity literature addressing some of these issues. Future research will exploit this literature and apply it to the CSCL domain as we believe that between-group aspects of online collaborative learning have the potential to significantly enhance the group experience and learning potential of students.

Discussion

The SIDE-VIEW of CSCL as outlined above is a theoretically grounded approach that has significant empirical support and implications for both design and practice. Rather than a focus on collaborating groups in terms of differing interpersonal aspects, it is suggested that the group be seen as a unit that shares a unique social identity, and that this can be harnessed and supported using the inherent characteristics of CMC technology. Design features that ‘collectivize’ rather than ‘personalize’ CSCL can promote psychological issues that are crucial for successful collaborations such as group cohesion, accountability to the group (rather than as an individual), and increased participation due to congruity between the group goals and ones self-definition. In addition, norms for success can be developed through intergroup comparisons that enhance group boundaries and provide opportunities for norm testing. Ongoing research is further investigating these concepts and seeking to enhance and refine our approach. Moreover, the theoretical background from which SIDE-VIEW is developed (i.e. the social identity approach) is a contemporary meta-theory that can be used to address many more aspects of CSCL than are covered here, and so can be used as an ideal framework to further support a range of CSCL issues. One specific example that is currently being investigated is that of leadership in CSCL groups. For example we can draw on social identity research into leadership (e.g. Haslam, 2004) to determine how best to choose leaders and what behaviours they should engage in. There are also many other dimensions of group behaviour that could be explored and we welcome any suggestions as to possible future directions.