Purpose – This paper has two purposes. First, to provide insight into the formation of completely online small groups, paying special attention to how their work practices develop, and how they form identity. Second, to pursue conceptual development of a more multi-level view of completely online group experience, which can be made visible through analysis of the unique interaction logging system used in this study. p. 1

Findings – Completely online group formation is explicated as a socio-technical system. The paper identifies themes of tool uptake and use, and patterns of interaction that accompany group formation and development of completely online group practices. These patterns show little respect for the boundaries of space and time. It then shows how groups who are paired together for two non-sequential activities develop a common internal structural arrangement in the second activity, and are viewable as groups in the larger course context in four of six cases. p. 1

Definition of small groups? Is there an absolute size requirement, or is it more that it's a subgroup within a larger group/community? p. 2

Small groups are powerful engines for getting work done. When small work groups are formed, members must get to know each other, make sense of their tasks and coordinate task progress and completion. Completely online small groups may never meet face to face, And this means they face at least two unique challenges when they are starting out. First, no technology available today recreates the experience of a shared, physical work space (Heath and Luff, 1992; Teasley et al., 2002; Birnholtz et al., 105 2007). Second, when completely online groups form, members must make sense of available technology, each other and the work at hand simultaneously (Laffey et al., 2006). In this study we describe how one small group meets this challenge in the context of a completely online graduate level software design course. We then compare the evolution of interaction patterns for that group with other groups assigned the same tasks and using the same tools. Our comparisons make group development in a completely online context visible, and the deep case study informs our analysis of the common patterns that emerge from the data. p. 2

By describing patterns of completely online group formation and development we illustrate the smallest and arguably most productive social unit in a new form of organization; those built through distributed and networked connections instead of physical co-location. Such virtual group behavior is hard to anticipate or identify in many prominent settings like Facebook, but occurs with some regularity in online graduate level courses, where the diversity of member experience and background is similar to what we find in virtual work groups. Prior studies of virtual work group formation describe patterns of high early expectations, followed by the cyclical movement between individual and group identity oriented behaviors, similar to studies of physically co-located groups (Smith, 2005). p. 2

While studies of online learning groups offer a window into patterns of group formation, sustained evolution of group work practices in virtual organizations are studied in a range of traditional organizational contexts and newer, more technologically mediated organizational forms. Watson-Manheim et al. (2002) explore the nature of virtual work as an array of continuities and discontinuities across time, space, organization and culture, referencing online learning groups as one example. p. 2

In spite of this extensive study, organizations of all kinds maintain a strong preference for collocation over virtualization (Kiesler et al., 2005). Kling and Courtright (2003, 2004) note the rarity of community formation in completely online settings. Powell et al. (2004) describe 47 studies examining virtual work teams, which individually and collectively do not build a clear picture of how virtual organizations p. 2

Group formation online has been predominantly studied in the aforementioned contexts like Wikipedia or FOSS, where there is a task orientation, or in online social networks, where there is a social-emotional orientation (Adamic and Adar, 2005; Backstrom et al., 2008; Lazer et al., 2009). p. 3

Differentiating between task orientation and social-emotional orientation? p. 3

We work to close this gap by describing completely online group formation where such phenomena presently exist, while drawing on literature from communities of practice (Wenger, 1998) and small group development (McGrath, 1984; Mcgrath, 1991; McGrath and Hollingshead, 1993; Arrow et al., 2000; McGrath et al., 2000), which describe how organizational and group practices emerge and are reified. We present a single, revelatory case study analysis of one group, showing how they simultaneously make sense of their work, each other and collaborative technology (Yin, 2009). We then analyze the unique, bi-directional interaction logs produced by the collaborative technology the groups work in (explained in section 3) using time-sliced social network analysis. The changes in network structure within and between groups in this course reinforce our findings from the case study, and reveal that patterns of completely online group formation and development can be made visible. p. 3

The purpose of this paper is to provide a detailed explication of how groups cooperate and develop characteristics of community as described by Wenger in a tool facilitated, virtual environment. The first focus of this case study is one small group assigned to work together in the context of an online graduate school course. The second focus is the online graduate student course community. This setting is chosen because the duration of engagement, tasks and expectations are stable, which allows us to focus our examination on the effect of tools on the undertaking and completion of tasks, and the corresponding formation of group identity and community. We describe how member cooperation evolves into a shared repertoire, which includes tools as a salient dimension of the community itself. p. 3

RQ1. How do members of a small group experience cooperation in a completely online environment? RQ2. How do members of a small group participate in completely online communities? RQ3. How and to what extent are the experiences of a small cooperative group, and other groups in the same context, revealed through analysis of the social structure derived from bi-directional (read and post) trace data? p. 3

What makes communities function well in the physical world is some combination of mutual engagement, joint enterprise and shared repertoire (Brown and Duguid, 1991; Wenger, 1998). The challenge of community formation in a virtual world includes these same dimensions, as well as new challenges. These challenges are the same as those faced by online cooperative work in general, most notably the context poor nature of the tools used. Social awareness in an online environment makes mutual engagement in general more likely, and the sort of rebelliousness that can demonstrate commitment to community possible. Similarly, Wenger’s notion of joint enterprise can be managed or assigned in most online cooperative environments with great efficiency, and is therefore more readily manufactured in the virtual world than it is in the physical world. Online group formation and development 107 p. 4

Recent theoretical analysis by Stahl (2006) suggests that small groups of 3 to 5 are “where the action is” for understanding online group cooperation. p. 4

Stahl’s focus on small group cognition in online cooperation is compelling, as it extends existing notions of organic systems of knowing, as suggested by Hutchins (1995). p. 4

A socially grounded theory of group behavior in general is posited by Wenger (Lave and Wegner, 1991; Wenger, 1998), who observes that practice is the construction of meaning in everyday life, and that our understanding of that meaning construction in any context – online or in person – must reconcile the duality of established protocols (what Wenger references as reification) with the participation of actors who apply and transform those protocols in the context of community. These systems are, in Wenger’s theory, dynamic and social. p. 4

Shared repertoire is the socially negotiated component of a community of practice, and in a completely online setting, is therefore highly dependent on the types of tools available for interaction. In a shared physical setting, this includes the ability to work through many different problems at once, taking advantage of the richness of verbal and non-verbal communication people use when situated around the same table. p. 5

Wenger describes the effect of communities of practice on individual identity. There 108 is a powerful set of dynamics at play between the individuals who belong to communities, how those individuals participate (or “non-participate”) in those communities, the negotiation of identity within the context of a particular community and the juxtaposition of membership in different communities with an individual’s identity. p. 5

First, the definition and continuum between group and community is neither clear nor consistently applied. Second, there is wide variation in the meaning of “online”. Regarding the first challenge, we agree with Kling and Courtright (2003, 2004), who argue there is extensive misuse of the term community by researchers examining online behavior. Frequently, community is used as an aspirational construct or without consideration of the criteria for identifying community, like those outlined by Brint (2001). At the same time, completely online small groups display characteristics described in Wenger’s (1998) explication of communities of practice, such as the reification of practices, shared repertoire and mutual engagement, which makes CoP an important and relevant framework for the study of group formation and development. As a matter of definition, McGrath’s (Mcgrath, 1991; McGrath and Hollingshead, 1993) enumeration of the purposes of small groups: meeting member needs, sustaining the group and completing group work, surely apply. From this core definition of the small group hundreds of studies using different methodological and theoretical foundations emerge. p. 5

McGrath et al. (2000), speaking of group research more generally, observed that after over 100 years of investigation of small groups in social psychology, we still know very little about the antecedents to highly effective and cohesive groups; p. 5

More recently, small groups have begun to be characterized as small, complex systems (Arrow et al., 2000). This framing enables researchers to recognize and address the challenges of connecting laboratory experiments to real world phenomena, but it also further blurs the boundary between small groups and the communities p. 5

within which they are inevitably situated. Our incorporation of theories of communities with theories of small groups reflect the diminished parsimony of these two theories in a completely online environment. p. 6

Groups represent a more canonical, task oriented and identifiably bounded combination of people than communities, which are characterized by Wenger as having the distinguishing features of mutual engagement, joint enterprise and shared repertoire among members. Communities represent a less explicitly defined set of norms than groups. p. 7

We believe study of completely online groups situated within a larger virtual community provides two advantages for gathering insights about small groups as socio-technical systems. First, at the small group level of analysis there is no established shared repertoire unique to the group. These members are new to the socio-technical context, and new to each other. In contrast, studies of online groups who are part of a larger community with established practices may produce as much insight about a particular group and its member’s personalities p. 7

or the virtual community’s established practices as it does about the interplay of tools, cooperative work and the development of community. Second, at the community level of analysis, we describe the variance of structure and interaction in six completely online groups who share the same tools, tasks and socio-technical context. (Our case study group is one of the six.) This knowledge of the socio-technical context is methodologically novel, but contributes to understanding of the relationship between small groups and wider communities in completely online environments. p. 8

called Sakai and another open source tool, the Context Aware Notification System (CANS). CANS provides awareness of each member’s activity within the course management system for all members, and logs all activity with reference to the course context, activity, and members. CANS records both read and post behaviors. Members use Sakai wikis, discussion boards, chat rooms and file storage areas during the course, and receive daily e-mail digests of fellow member activity from CANS. There are 16 students in the online course in this study. p. 8

The course activities centered on software design for performance support systems. p. 8

Highly guided (scripted?) activities - individual, peer to peer (pair?) and group work. p. 8

Three of the students are in the core group we build our case study around. Each of the three group members is interviewed three times. The first author interviews the three group members before they become a group, and then once again following each of two, three week cooperative tasks. p. 9

In addition, discussion board activity and assignments completed by each group during 112 cooperative activity are observed on a daily basis, with field notes recorded during that process. Assignments and work products produced by the small group are also reviewed and coded. p. 9

We use social network analysis to build sociograms of member interactions from electronic trace data produced by CANS. p. 9

The second type of data we collect is made up of written deliverables, discussion board posts and field notes related to discussion board posts from the entire class. These are reviewed for quality and alignment with the stated objectives of the work. We analyze these first two types of data using transcript analysis and coding techniques adapted from grounded theory (Coffey and Atkinson, 1996; Charmaz, 2003) to expose themes related to small group online cooperation. Grounded theory analysis is used because, although we approached our analysis with an understanding of p. 9

existing theory on communities of practice and group development, the kinds of highly virtual, completely online configurations studied here may not be fully explained with those theories. Therefore, we thought it important to drive our analysis out from the data. Open coding of the interviews and discussion board data revealed 71 unique codes, which were refined down into 14 categories. Additional refinement by a second rater revealed four major codes that the first author and a research assistant axially coded against the full corpus of data. Through this coding and review of participation 113 data, we develop a description of this group’s cooperative experience, and chronicle how one small group becomes part of an online community in case study format. p. 10

The third type of data analyzed is log data produced by CANS. These data provide a complete record of the reading and posting behavior of each online community member over the 16 week period that makes up this study. Previous research by the authors (Goggins et al., 2010b) establishes that the bi-directional, contextualized logs produced by CANS reveal a more accurate representation of membership structure than common weblogs. p. 10

4.1 Themes of tool use and cooperation p. 10

(1) Members demonstrate explicit politeness and are cautious about risking social capital. (2) Members are frequently multi-tasking. p. 10

The absence of disagreement in an online community, particularly in a learning community, is associated with low levels of engagement and learning. p. 11

The ICTs members use do not support structured argumentation, which results in constraints on the negotiation process and diminished capacity for cooperation in the virtual environment. p. 11

our members are frequently engaged in other tasks while they participate in synchronous or asynchronous aspects of the community. “I’ve actually been driving down (the highway), talking to my husband and posting to my team” and “sometimes I get distracted when my boyfriend saws something up in Gears of War while I’m chatting with my team online” are representative of the measures of multi-tasking our informants describe. p. 11

Interacting with one set of individuals in the physical world while simultaneously interacting with people in the virtual world is a common theme across the participants in this study p. 11

This seems to contradict the info that they are assigned groups, assignments, workflow etc p. 23

ed when members must make sense of the tools, each other and their work all at once. Completely online small groups are not bounded by time, space or even membership. They gravitate toward assigned teams, and negotiate time and place within those teams and across the larger community. They do this when they are ready, and not the first time they are asked to. p. 23

Previous literature identifies high level meta patterns of group development, such as storming, forming and norming (Tuckman, 1965) and punctuated equilibrium (Gersick, 1988; Gersick, 1989) p. 24

Brint’s (2001) six dimensions of community are: p. 24

(1) (2) (3) (4) (5) (6) dense and demanding social ties; social attachments to and involvements with institutions; ritual occasions; small group size; perceptions of similarity with the physical characteristics, expressive style, way of life or historical experience of others; and common beliefs in an idea, a moral order, an institution, or a group are never all met. But many are met here. p. 25

Really? Following a group of 3 students leads us to reconsider theories of community and group development? p. 25

We think the present findings point to a need to reconsider theories of community and group development when the context is completely online; that present social theories of learning, communities of practice and group development are insufficient for this new context. p. 25

Very focused on task performance and outcome, what about learning? Is it the same? p. 25

Returning to our theoretical framing, networks of practice are more amorphous than groups in completely online settings; yet these same groups pulled together in a larger “group of groups” arrangement are not long running enough to constitute a community. Theories of small groups focused on task performance are also insufficient for explaining the role of technology in the development of group practices and structure. p. 25

Constructs from the physical world like community, network and group lack a virtual corollary. The use of these well-worn constructs in the study of technologically mediated interaction is possibly misleading. Kling and Courtright (2003) suggest that the construct of community, when applied to online settings is aspirational. Present findings suggest that “network” and “group” conjur images from the physical world that are not descriptive of the experience people have in completely technologically mediated environments. p. 25

Online community research is, we argue, analytically deterministic in the sense that tools for data mining and network p. 25

analysis that were originally developed in different contexts and for different purposes are now applied to the analysis of trace data from massive systems. Howison, Wiggins and Crowston (under review, Journal of the Association of Information Systems) provide a comprehensive critique of this practice. p. 26

Laffey et al.’s (2006) theory of social ability comes the closest to a contribution to social theories of learning in technologically mediated environments by providing a parsimonious construct that considers tools, tasks and individuals. Our study contributes an explication of how learning occurs socially through small groups. p. 26

References p. 27