management focus learning in online forums

European Management Journal Vol. 21, No. 5, pp. 565–577, 2003 2003 Elsevier Ltd. All rights reserved.Pergamon

Printed in Great Britain0263-2373/$30.00doi:10.1016/S0263-2373(03)00106-3

Management Focus

Learning in OnlineForumsGERARDINE DESANCTIS, Duke UniversityANNE-LAURE FAYARD, INSEADMICHAEL ROACH, Duke UniversityLU JIANG, Duke University

Information and communication technologiesafford different levels and types of support forlearning networks. We draw on our studies ofvideo-conferenced classrooms, group discussionspaces, and online communities to suggest a frame-work for understanding how learning networks canbenefit from various e-learning venues. We showhow the design of computer-mediated environ-ments influence the kinds of learning processes thatare likely to unfold as business professionals inter-act with one another across time and space barriers.The extent to which participants experience thesetypes of learning depends upon how the electronicenvironments are structured and, more importantly,on how participants manage their interaction pro-cesses. Though all venues provide access to distrib-uted social resources, some settings are more effec-tive than others in addressing the specific learningneeds of knowledge workers. 2003 Elsevier Ltd. All rights reserved.

Keywords: Learning, Virtual teams, Knowledgemanagement, Online communities

In ancient times, learning networks occurred withinone’s immediate family, community, and tribe.Today our learning networks are more tentative anddiffuse. We join schools, places of work, andextended families that are diverse, distributed, andwithout the strong, longstanding ties characteristic oftraditional learning networks. Often the groups wejoin interact via technology; participants may be mereacquaintances, or even strangers. In matters as vitalas learning one’s profession, upgrading new skills, ormaking sense of information in everyday life, we findourselves a part of many networks that are, inessence, new organizational forms — not the triedand true settings of yesteryear.

Learning networks provide opportunities for seekingand providing information, for forming relationships

European Management Journal Vol. 21, No. 5, pp. 565–577, October 2003 565

among participants, and for creating shared mind-sets. As learning networks span the globe, becomeculturally diverse, and operate across traditional timeand space boundaries, many questions arise regard-ing their feasibility and effectiveness. Can peopleshare information, form strong working relation-ships, and develop insight and understanding whentheir interaction occurs primarily via electronicmedia? Can e-based learning networks be built tolast, or are they destined to be short-lived and basedon loose ties? Do different e-based venues spawn dif-ferent forms of learning, or is the propensity forgroup learning similar across various electronicforums?1

This paper has two goals: first, to suggest a frame-work for understanding how e-based venues influ-ence the learning process for participants; and,second, to consider how participants, in turn, suc-cessfully interact in these venues to form effectivelearning networks. Each technology venue affordssomewhat different possibilities for learning; and dif-ferent kinds of learning networks emerge dependingon how participants interact with one another withinthe venue.

Technology-mediated Venues forLearning

Relative to traditional venues for group learning,technology-mediated forums offer potential advan-tages and disadvantages. For example, on the upside,we know that e-based venues afford the participationof a larger and more diverse set of people than tra-ditional non-technology venues. People from mul-tiple time zones and organizations can join distrib-uted classrooms, group discussion spaces, and onlinecommunities. Increases in the size and diversity of

LEARNING IN ONLINE FORUMS

the learning network afford a larger and morediverse set of information resources, and enhancedopportunity for information sharing and idea gener-ation. These, in turn, should improve learning effec-tiveness and promote a more robust network (Guzzoand Dickson, 1996; Hiltz, 1994). On the downside,however, computer-mediated communication canreduce overall communication within a social net-work (Hollingshead and McGrath, 1995; Sarbaugh-Thompson and Feldman, 1998; Straus, 1997), whichmay hamper network survival and growth. If parti-cipants lack dissimilar backgrounds, common attri-butes, and a history of interaction, knowledge trans-fer may become more difficult (Marshall and Novick,1995; Ko et al., 2003). The interruptions and lags asso-ciated with asynchronous communication can dis-rupt conversational routines and potentially hindertacit learning, which relies on deep and subtle infor-mation exchange. Finally, lack of face-to-face contactamong participants can reduce group identity andmake relational ties more fragile (Shapiro et al., 2002).

Given both positive and negative impacts, how canwe anticipate the effects of e-mediated venues ongroup learning? As a start, we can compare venuesin light of their relative affordances for learning(Gaver, 1993; Gibson, 1977, 1979).

Technology Affordances: Three e-Based Venues

Figure 1 identifies important dimensions that can beused to compare e-based learning venues, along withtheir corresponding implications for group com-munication and learning. We consider only threevenues, but our general analytic approach can be

Figure 1 Technology Affordances and their Implications for Learning in e-Venues

European Management Journal Vol. 21, No. 5, pp. 565–577, October 2003566

applied to other learning venues, or to differentimplementations of these three venues. Our approachis built on media richness theory (Daft and Lengel,1986), which stresses the importance of highinformation-carrying capacity in media for tasks withhigh equivocality (i.e., uncertainty in the meaning ofthe message); social presence theory (Short et al.,1976), which emphasizes the need for high salienceof others’ presence in tasks that are highly interper-sonally involving; and language game theory(Wittgenstein, 1974), which emphasizes the cognitiveprocess of building shared mental models throughdynamic, patterned discourse with others. E-venuescan be compared based on their likelihood to pro-mote open (public) versus closed (private) partici-pation, their likelihood of creating a sense of co-presence (immediate versus diffused), and their like-lihood of promoting conversation with meaningfulstructural routine such that members can share a cog-nitive mindset with others when they join and inter-act from afar.

1. Video-conferenced Classrooms extend the place-dependent face-to-face classroom so that two ormore otherwise unconnected social networks canbe joined together. The technology provides abridge for linking face-to-face groups, and there isthe hope that the disparate locales can think andact as one. (We assume here that participants arein physical classrooms, such as at a college or uni-versity, and not in a virtual setting, such as whereparticipants attend from home.) The technologyincludes audio connections, video connections,and computerized connections (e.g., shared white-boards and projection systems). If these operateduring times of informal interaction, not just dur-


ing designated meetings, the effect is to create acommon ‘media space’ for group interaction (Blyet al., 1993). Relative to the other venues, video-conferenced classrooms offer privacy (one has tobe invited into the room and cannot ‘drop in’ with-out an invitation) and a high sense of co-presence(the gathering has defined boundaries, and parti-cipants can see and hear one another in ‘live’mode). An instructor or other leader encourages astructured dialogue among parties across the mul-tiple venues so that there is turn-taking, attentionto speakers, and a discussion chain that is suf-ficiently organized and meaningful such thatpeople can contribute and learn.

2. Group Discussion Spaces offer a venue for a work-ing group, such as a project team, consulting team,or student learning team. The discussion spacemay include places to store group documents andtrack team progress; often the group can cus-tomize the design of the space to meet their parti-cular needs or preferences. A key feature of groupdiscussion spaces is their use by pre-establishedgroups, so access is private (not public). In ourexperience, these venues are used to supplementother learning venues for the group, such as face-to-face meetings and e-mail, and they are bettersuited for reinforcing existing social networks thanfor establishing new ones. The venues can be usedfor synchronous chats but are more frequentlyused for asynchronous interaction, especially ifgroup members are highly mobile or living in dis-tant locations. The diffused learning environmentresults in what Goffman (1963) calls a ‘multi-focused gathering’ (as opposed to the fully-focused gathering found in traditional meetings orclassrooms). Group members are likely to spendmuch of their communication signaling one other,seeking feedback, and checking for the reactionsof others. Issues of social identity and trust cometo the fore, and learning is heavily relational, asopposed to cognitive.

3. Online Communities are open, Internet-basedforums that anyone can join to discuss topics ofmutual interest. Not all online communities arededicated to learning; for example, some exist forentertainment and others to produce software pro-ducts (e.g., Linux). Our concern is with learning-oriented communities, such as those that discussbusiness topics, political interests, or medical mat-ters. These online communities are akin to volun-tary associations that provide a common venue forinteraction among professionals from a wide setof organizations (Aldrich, 1999). Successful com-munities are longstanding; people visit repeatedly,and participants form a sense of mutual identity.Whereas video-conferencing bridges learning net-works, and group discussion spaces supplementpre-existing networks, online communities areused to create entirely new social networks. Com-munication is via lean, text-based media. The


environment is virtual in the sense that communi-cation boundaries are diffused over time andspace, participants are strangers who rarely meetface-to-face (if at all), and the forum can be readilycreated and destroyed (deleted). The group struc-ture in online communities is fluid and fragile.

Implications for Learning

The three e-based venues progressively decrease intheir media richness and sense of co-presence, as theybecome more public, diffused, and less structured.So, we would expect learning to become more chal-lenging as one moves from venue (1) to (3). Thevenues differ in their affordances for three generaltypes of group learning: declarative and proceduralinformation exchange, transactive learning, andsense-making.2 These are learning processes, embed-ded in communication, as opposed to specificaccomplishments or goals. They reflect increasinglysophisticated forms of interaction as a function of thekinds of knowledge the group is sharing.

Declarative and Procedural Information Exchange occurswhen people seek and provide relatively objective orfactual knowledge with one another. Declarativeknowledge is a discussion of fact (what); and pro-cedural knowledge is a discussion of method (how).The two often occur together as a process of knowl-edge swapping. The exchange is equity governedsuch that people provide knowledge to others inreturn for receiving knowledge. For example, a par-ticipant might ask, ‘Where can I find a good referencefor statistical modeling?’ or ‘What companies offerdata mining tools suitable for use by multi-languageoperators?’ or ‘I’d like to do a Balanced Scorecardassessment for my company. Any suggestions onhow to get started?’ Respondents might offer refer-ences, lists, guidelines, opinions, and so on. The spe-cifics of the information being discussed may be com-plex, but the discussion is not because it lackschallenge, interpretation, debate, or other forms ofcomplex conversation. Declarative and proceduralinformation exchange is a relatively rudimentaryprocess of knowledge sharing that is well suited toe-based venues. The exchange can be between twoparties in the group or hundreds. Large volumes ofknowledge exchange are possible, and transactionscan occur rapidly. In fact, the larger and more openor public the venue, the greater the possibilities forparticipants to find the information they need in theresource pool of participants, and the wider the rangeor frequency of questions people are likely to ask.The focus tends to be more on the knowledge itselfthan on the relationship between the parties who areseeking or providing the knowledge.

Transactive Learning is the process of sharing infor-mation about the capabilities and boundaries ofknowledge that exist among members of a group(Wegner, 1986). Transactive learning is an investment


in the ‘groupness’ of the collective as members ident-ify their expertise (e.g., ‘Shawn has worked for asoftware firm and can probably help you out withthat problem’), recount their successes and failures(‘I worked all night to develop an implementationplan, but I am really stuck at this point’), acknowl-edge each other’s abilities (e.g., ‘Jacques can siftthrough 100 balance sheets and quickly find the oneswith errors’), etc. Through queries and replies about‘who knows’ or ‘where to find,’ group members dis-cover the frontiers and limits of their learning net-work (Moreland et al., 1998). Some transactive knowl-edge is shared in the process of declarative andprocedural information exchange. But, whereaslearning in the former case tends to be equity gov-erned, transactive knowledge is decidedly relationaland incorporates information about the persons whoare interacting not just the information itself. In thisway, transactive learning is more challenging. Asense of co-presence, or immediacy in the socialoccasion, should facilitate transactive learning. If thegathering has defined boundaries, such as a definablestart and finish and/or a fixed set of group members,exchange of transactive knowledge is easier than ifthe social gathering is diffused.

Sense-making is the process of developing sharedmental models that enable a group to coordinate itsefforts, respond to novel events, absorb information,and detect and reduce errors (Weick, 1995). Sense-making occurs through dense dialogue that includesinformation interpretation, providing of opinion, try-ing out new ideas, and reflecting on results of indi-vidual and group actions. It also may include storytelling or narratives that convey nuances in the mean-ing of knowledge. Whereas the other types of learn-ing focus more on explicit knowledge, sense-makingemphasizes tacit knowledge. When a group engagesin sense-making, a common view of the world isinteractively produced, challenged, and reproducedover time. The group is able to absorb new knowl-edge, to change, and avoid becoming insular(Goodwin and Heritage, 1990). Sense-making con-tributes to the formation of a community of practice(Wenger, 1998). Sense-making is difficult to experi-ence in any venue, but it is especially difficult whendialogue is scattered or disjointed. E-based venuesthat emulate face-to-face meetings, such as video-conferenced classrooms, are more likely to fostersense-making than asynchronous or text-basedvenues, since, in the former, dialogue can be rich andrapid, and non-verbal cues are available. Group dis-cussion leaders can facilitate sense-making via con-versational structuring or use of technologicaldevices, such as topic organizers, voting tools, orbrainstorming modules that organize the discourse.As venues become more open, diffused, and free-flowing, conversation management becomes moredifficult.

Figure 2 presents a comparison of the three e-basedvenues for their relative potential to promote collab-


Figure 2 Anticipated Levels of Learning in Three e-Mediated Venues

orative learning. Venues that offer open participationare most likely to foster information exchange. Thosethat offer the greatest co-presence are most likely tosupport transactive learning, and those that offereffective conversational structuring are most likely topromote sense-making. The three kinds of learningare cumulative, so we would expect sense-making tobe most readily achieved in video-conferencedclassrooms venues and most difficult to achieve inonline communities. Video-conferenced classrooms,group discussion spaces and online communities areprogressively more challenging venues for promot-ing group learning.

Research Findings

As noted earlier, technology affordances do not fullydetermine communication patterns. Learning isemergent, and groups can produce complex com-munication even in lean media environments (Abel,1990; Lea and Spears, 1991; Markus, 1994; Carlsonand Zmud, 1999). In our research, we have attemptedto document group struggles and successes in realiz-ing the learning benefits of various e-based venues.Our case studies illustrate how groups can success-fully leverage the technological affordances of e-based venues to yield rich learning experiences. Theresults provide lessons in how the downsides of e-based venues can be overcome by groups to yieldlearning that spans the three levels shown in Figures1 and 2.

Video-conferenced Classrooms

We studied video-conferenced classrooms linkingstudents enrolled in courses at INSEAD campuseslocated in Fontainebleau, France and Singapore. Thetwo facilities were designed to operate as a commonmedia space (Bly et al., 1993) such that when students


stepped into the room in their local facility there wasthe sense of being in a shared facility with the othercampus. Participants could see, hear, and inter-actively read and write with people in the othervenue. The shared venue was relatively small (15–25students across the two sites), team-based, and heav-ily interactive. Students in the courses were full-timeMBA students with 5–6 years of prior work experi-ence and differing cultural and language back-grounds. Prior to participating in the video-conferenced courses, some of the students had meteach other face-to-face, but most had not. Further,many of the participants had never been to the othergeographic location, so they had no prior sense of acommon social network other than the fact that theywere all a part of the INSEAD organization. Thelearning network challenge was to build relation-ships within and between the two locations. Inaddition, there was the hope that the video-basedlearning experience would contribute to the largerinstitutional goal of developing a common com-munity across two campuses.

Figure 3 summarizes key observations from thisresearch. The observations are based on analyses ofvideo-recordings of class sessions (including informaltime before and after class sessions and duringbreaks), interviews with participants, and live partici-pant observation. (More detailed reporting of theresults is provided in Fayard, 2002).

Case Example Participation in the video-conferencedcourse was voluntary; the course was elective (ratherthan required). The course was heavily project-ori-ented and aimed to achieve experiential learning.Students worked in teams with members drawn fromthe two campuses. For the case described here, therewas one faculty at each site, 10 students in Singaporeand 14 in Fountainebleau. There were eight class ses-sions of three hours, each with a 15-minute break inthe middle, over a period of two months. In addition

Figure 3 Learning Experiences in Video-conferenced Classrooms


to classes, group work sessions of 1–1.5 hours alsowere held in the video-conferencing facility andrecorded. Each video-conferencing facility includedan electronic white board, visible at both sites, videocameras, and a TV monitor. Tables at each site wereconnected at their ends via the video, resulting in onelong virtual table.

A key challenge in video-conferenced meetings, as inany e-venue, is establishing a sufficient physical andlinguistic co-presence so that participants can beginto experience a shared social context — a sense of‘us’ rather than ‘you’ versus ‘me,’ and a sense of allof us being ‘here’ rather than some ‘here’ and others‘there.’ A common social context helps groups tomake inferences about one another’s knowledge andengage in learning (Cook and Brown, 1999). Further,there is the matter of using the course experience todevelop a common context for the institution as awhole. Several approaches proved important to pro-moting a common social context in this video-confer-enced course.

First, one student at each site was appointed as ‘host’to help involve participants in the discussion; thehost was asked to organize the technology for thesession, guarantee that nobody talked off camera,monitor the picture, etc. The host was considered acrucial role because instructors were not able to man-age the technical details while leading the class dis-cussion. However, we found that the appointed hostssoon gave way to emergent facilitators who weremuch more effective in mediating the group inter-action than the appointed hosts. Our emergent stud-ent facilitators understood the technology better thanthe others, were more aware of the constraints andlimits introduced by it, and endeavored to enhanceinteraction by modifying the setting and supportingothers’ behaviors. They also felt at ease among theparticular set of people in the class and wererespected by them. These facilitators, along with the


faculty leadership, provided governance of the dis-tributed classroom.

Second, various social practices emerged that provedcrucial to conversation management. For example,students designed a practice to encourage fair turntaking. It was hard to interrupt when persons at theother site were talking, and waving hands for atten-tion was not effective for gaining airtime. Thus, whensomeone in Singapore wanted to comment on whatsomeone in Fountainebleau was saying, the personin charge of the remote control — who was able tonotice the subtle non-verbal behavior because theywere co-located — focused the camera on the personwho desired to speak. Over time, this became a con-vention for assuring that both ‘sides’ were involvedin the conversation. These sorts of emergent socialpractices were critical to maintaining the viability ofthe e-based venue over time.

Third, a common social context was developedthrough informal conversation, joking, playing withthe camera, and visual display of the environmentacross the two sites. A friendly and informal atmos-phere developed with teasing exchanged across thetwo locales. The weather was a frequent topic of con-versation, and students occasionally pointed the cam-era out of the window so that participants on theother continent could view their weather. Experimen-tation was also important. For example, during thesecond class session, one student in Singapore tooka paper board into the classroom and placed it thescreen. He drew the tables in Fountainebleau andwrote down the names of students he knew. Asecond student entered the room and did the same,followed by another who did the same. Students alsotried out different locations for the camera for vari-ous meeting topics and purposes. Group memberstook on mutual responsibility for developing aneffective classroom environment over time.

Declarative learning exchange largely gave way toprocedural learning in this course; there was exten-sive discussion — especially early on — about pro-cesses for managing the equipment, conversation,and teamwork. Declarative learning exchangeoccurred largely during lectures, but procedurallearning occurred on a more routine basis within andacross the two sites. Even more dominant was trans-active learning, which proved difficult and extremelytime consuming for students as they invested inlearning about each other’s capabilities and ongoingprogress in their projects. Investment in transactivelearning was critical and difficult. Each team had tocontinually assess who knew what and determinetheir mutual capabilities. Most teams invested con-siderable time in developing and maintaining trust,deciding upon roles, monitoring progress, and mak-ing adjustments as needed. Teams that invested timein getting to know one another tended to developmore productive working relationships and weremore satisfied with the course experience, in contrast


to teams that spent little time on transactive learningactivities. As the course progressed, and especially inthe project work, there were many occurrences ofsense-making. Students extensively discussed thedefinition of the project — how they interpreted theproblem, their goals and understanding about thefinal output, etc. Discussions were lively, withdebates within and across the sites. Planned brain-storming sessions in which group members gener-ated ideas without criticism seemed to facilitatemovement toward sense-making. Idea generationwas followed with idea interpretation, challenge, andgradual building of shared ideas for executing theproject.

Group Discussion Spaces

We studied small group discussion spaces used byexecutive MBA student teams at Duke University. Aspart of a global MBA program, executives meet face-to-face at the start of each term, after which theyreturn to their jobs — in Europe, Asia, and the Amer-icas — and use Internet-based tools for five monthsof distance learning. The executives work in groupsof four-to-six people and use the group discussionspaces for analysis of cases, drafting of reports, work-ing on technical problems, and general coordinationof team work activities. The MBA program emphas-izes team-based learning, and executives are givenmany assignments to complete as a team. Further,team members are encouraged to rely on one anotherfor individual learning, that is, to seek and providehelp to one another as they progress through thecoursework.

We studied 18 teams across two academic terms, fora total of 10 months of activity. During these 10months the teams met face-to-face (for traditionalclassroom learning) for three weeks at the start of theMBA program and, again, for two weeks at the startof their second academic term (approximately half-way through the 10-month period). The teams hadaccess to asynchronous group discussion spaces,synchronous chat, and audioconferencing tools, aswell as email and file sharing. We report here on ouranalysis of learning-oriented discourse in the groupdiscussion venue, which supports threaded conver-sations in a bulletin-board format. Due to time zonedifferences and travel schedules, most of the teamsrelied on this asynchronous group discussion spaceas their primary venue for interacting at a distance.

The learning network challenge for these teams asthey used the group discussion space was to deepenrelationships they initiated while face-to-face so thatthey could move past ‘getting to know you’ to themore challenging work of sense-making. Their worktasks were complex and required deep levels oflearning, not just surface-level exchange of infor-mation. The teams had to be able to work effectivelyas a group, to maintain mutual identity and trust,


to engage in dense dialogue, to discuss competinginterpretations of information, to generate creativeideas, and to identify and resolve conflicts. Further,their work schedules and family lives put immensepressure on participants to be able to work efficientlyand to meet deadlines with a minimal number of set-backs.

We sampled six weeks of discourse (two weeks ineach of the five-month terms) and coded all passages(i.e., meaningful statements) within the messagesposted to the group discussion space. In all, 8076passages were coded.3 As shown in Figure 4, declara-tive and procedural exchange were the dominantform of learning in the teams, followed by transactivelearning and sense-making. This concurs with ourexpectations (Figure 2), in that declarative and pro-cedural knowledge are the simplest forms ofexchange that groups can undertake, followed bytransactive learning. It is notable that there wasnearly as much transactive learning as informationexchange and, in turn, nearly as much sense-makingas transactive learning. Perhaps more noteworthy arethe differences across teams and the fact that vari-ation across teams was lowest for transactive learn-ing. (See the standard deviation values in Figure 4.)When we split the sample into the relatively-highversus relatively-low performing teams, we observedan interesting trend. (See Figure 5.) Compared to thelower-performing teams, the higher-performingteams engaged in slightly less declarative and pro-cedural knowledge exchange, slightly more trans-active knowledge exchange, and significantly moresense-making.4

Case Example Here we describe Team 62, one of thehigher-performing executive MBA teams. Theamount of declarative and procedural knowledgeexchange in this team was slightly below average(15%); transactive knowledge exchange was average(16%), and sense-making was above average (18%).Four men and one woman composed the team; twolived in the United States, and the others lived in Bra-zil, Japan, and Indonesia. Their backgroundsincluded finance, accounting, and engineering; and

Figure 4 Summary of Learning Discourse in a Sample of Executive Student Teams


Figure 5 Three Types of Knowledge Exchange in aGroup Discussion Space for Higher-versus-lower Per-forming Executive Student Teams

they were employed in the auto industry, telecom-munications, engineering, consulting, and education.This degree of team diversity is typical in the globalMBA program. Likewise typical, the members didnot know one another prior to their enrollment in theMBA program.

The team started to use the group discussion spaceduring the initial three-week period of their programin which they were co-located for classroom learning.They used the discussion space to post notes aboutteam responsibilities, project progress, deadlines, andpointers to where specific files or other team infor-mation could be located. They also used the board toplan dinners (while co-located) and other team meet-ings. From the start, the message postings by thisteam tended to be brief (200 words or less) andpresented in the form of itemized lists. Even in thecase of detailed analysis of technical problems (suchas discussion of how to do a sensitivity analysis for a


specific case), the messages were presented in cogent,step-by-step fashion. Communication was frequent,and members notified one another of their where-abouts, schedules, and progress on tasks. Theirapproach to using the group discussion space createda sense of co-presence; even members who were notactively working on a specific team task ‘checked in’and made themselves visible and available to others.Two members who were coordinating on a sub-taskused the bulletin board to post messages to oneanother so that others could monitor their progressand offer thoughts if interested. The tone of messageswas uniformly positive and often closed with an invi-tation for others to disagree or postalternative/opposing ideas. As an example:

One Team suggestion: my understanding from the SupportGroup in Durham is we can really clog up our Board withattachments. I don’t have a feeling when this is, but asLeader of the last project, I want to make sure we can stillhave an effective Board dialog….I would like to suggestany file over 50K be put on the FTP. Any comments orother ideas? Great work Team, as usual.

Although messages were polite and positive, mem-bers also were willing to challenge one another,admit failure, and actively promote continualimprovement. For example:

Geez Kai! You get a virtual smack on the head for this one!:) :) I can’t believe you threw out two weeks worth of workand re-did the whole thing! It is very nice though, BRAVO.I was working with Georgio yesterday (he was in NYC)and we were talking about time series in general and Iwent back and ran my data with the time series approachand then realized I had to break out the data like you did.However, I could never have gotten as far as you did. I dohave a much better understanding now, but still need tomake the leap into prediction. I got the playing with thedata thing down….

Conversations were dense and often became verytechnical as the team struggled with analyzing prob-lems and interpreting results:

When you deseasonalize or reseasonalize by dividing ormultiplying by the seasonal factor, this means you’re treat-ing the seasonal relationship as multiplicative. For instance,reseasonalizing with a seasonal factor of 1.5 changes 100 to150 and 200 to 300. The change is 50 in the first case and100 in the second case. If you use the seasonal factor as anindependent variable, this treats the relationship as addi-tive in the sense that it will add or subtract a fixed amountfrom each forecast for a given season.

At the same time, the group did not get bogged downin a task. They consistently pushed to complete workand move on. They developed both routines of con-versation (e.g., posting messages and replies at leastfive times per week…letting team members know ifthey would be unavailable and for how long, etc) androutines for managing their work tasks (e.g., settingdeadlines, posting progress, conducting audio-conferenced meetings with planned agendas set inthe group discussion space). They gave directions to


one another and accepted direction from one another.A high degree of mutual respect emerged on thisteam.

Overall, Team 62’s success can be attributed to theirregular use of the group discussion space, theirdevelopment of organized and efficient routines forinteraction, a tone of mutual respect and challenge intheir discourse, and a willingness to modify routinesover time. A structure emerged in their interaction.They used the group discussion space to manage thepresent, monitor progress, take time for sense-making, and plan for the future. Three members weremore active than others in posting messages to thegroup discussion space, but everyone on the teamwas involved in the conversation on a regular basis.This group used the technology to build an effectivelearning network.

Online Communities

The online communities we studied differed from theother venues in that these were created by individ-uals or organizations as open spaces for discussion.In many cases the forums were hosted by Internetservice providers, such as Yahoo.com. The parti-cipants generally were not members of the sameschool or other organization; they did not share acommon work task or goal. They came together asstrangers to discuss topics of mutual interest. Thelearning network challenge in such a venue is to forma meaningful learning network out of the scatteredset of individuals who join the discussion. Ourresearch interest was in identifying the attributes ofthe more enduring learning communities, to observetheir dynamics, and to understand how they grewand learned.

We report here on our research on open online com-munities devoted to the topic of knowledge manage-ment. Contributors to these e-based venues includetechnical professionals, general managers, consult-ants, students, and researchers. The forums are read-ily suited to declarative and procedural informationexchange, where people can seek and provide infor-mation without forming ongoing relationships. Weexpected these venues to include relatively simplequestion-and-answer discussions about software,systems design, and use of knowledge managementsystems. But we were curious as to whether, andhow, some of these communities emerged to formmore enduring learning networks that engaged intransactive learning and sense-making.

Our research included two phases. First, we com-pared 40 online communities using readily observ-able indicators. We discovered three dominant typesof communities, which we summarize in Figure 6.Next, we examined discussions within these com-munities and tracked their development. Our fieldnotes from one of the communities serves as the basis


Figure 6 Three Types of Online Communities

for our case example. We do not have direct meas-ures of learning for these communities, but we canmake inferences based on the general indicators anddiscussion field notes.

Type 1 communities have been in existence an aver-age of 11.5 months. They have notably fewer uniquecontributors per month than other communities, alower retention rate, and a small degree of overlapwith other communities. A relatively small coregroup of people stays with the community through-out its life. Discussions in these communities gener-ally are not moderated or organized by a leader.Fewer new discussion topics and replies are intro-duced by contributors each month in these communi-ties, in comparison to the other two types of com-munities. However, as in the other communities,contributors are more likely to reply than start dis-cussion, suggesting a willingness to provide answersor comments to message postings. Willingness toreply to queries suggests at least some miminal inter-active learning-type of activity in the forum. Messageproduction and word count are relatively low, andthe discussion density tends to be low; people are nothaving extended conversations with one another somuch as they are seeking and providing information.We call these Type 1 online communities ‘infor-mation kiosks’ because they appear to providequestion-and-answer forums. The learning networkin these kiosks is smaller, more isolated, and less pro-ductive than in the other types of communities.Nonetheless, the network holds together over timeand has a small core group of people who regularlyreturn to contribute to the discussion. Informationkiosks would seem ideal for people who want to


drop in and out of a learning network in search ofrelatively efficient declarative and procedural knowl-edge exchange.

Type 2 communities, which we call ‘associations,’ arenotably younger than the others and nearly half areaffiliated with a (non-virtual) professional society,such as the International KM Institute or the GlobalKnowledge Economics Council. We speculate thatthe societies create online discussion forums as exten-sions of their real-world counterparts in order to cre-ate awareness and attract new members. In somecases contributors appear to know one another, or atleast to know of one another, often via their presencein other KM communities. Contributor retention ishigher than in the information kiosks, and the dis-cussions are deeper and more productive (in termsof word count, density, and the number of repliesrelatively to discussions). Also, contributors tend tojoin other online communities (especiallyassociations), in search of additional contacts andinformation sources. Type 2 learning networks —while vibrant — are more overlapping with othernetworks as participants seek out additional forumsin which to gain knowledge. We speculate that parti-cipants in these types of online communities areespecially interested in exchanging transactiveknowledge as part of the process of building a pro-fessional social network.

Type 3 communities are the closest of the three typesto what Wenger (1998) and others have termed, ‘com-munities of practice.’ Such communities are charac-terized by frequent, patterned social interaction. Theyproduce high volumes of knowledge generated by


large numbers of participants in the network. Denseinteractions help them build coherence and hom-ogeneity and promote shared understandings andpractices associated with sense-making. A core groupof participants sustains the network, but the com-munity is also able to absorb newcomers and growover time. Communities of practice are not insular orisolated from other networks; there is some degreeof overlap with other networks. Together, these attri-butes and behaviors result in learning networks thatare able to sustain production of knowledge overtime, share it efficiently, and facilitate innovation(Brown and Duguid, 1991; Lave and Wenger, 1991).Our case example is of one of these communities ofpractice.

Case Example We studied the KMF5 communitybetween the period December 1999 (its foundingdate) and April 2001. The community was startedwith a clear and detailed welcome message in whichthe founder articulated the purpose, goals, andintended environment for this online forum:

The [KMF] is a learning community dedicated to buildingknowledge about public sector knowledge management.We aim to provide an environment where members cancreate and share knowledge about public sector knowledgemanagement issues… Over the next twelve months theChapter will pursue three themes: understanding how toimplement knowledge solutions in a public sector environ-ment; gaining a better understanding of the people aspectsof knowledge management; raising the profile of knowl-edge management among senior public sector managersthrough education…Our ultimate aim is to be the pre-eminent source of public sector knowledge managementknowledge.

The discussion then started with a series of postingsof article titles and contents on the KM topic. Earlyon, a newcomer, Tim, proceeded to post excerpts ofarticles from books and other sources related to KM.At first, few people replied; then one participantengaged Tim in conversation about the content ofone of his postings. Soon thereafter, Tim encouragedthose visiting the site to introduce themselves andjoin the conversation: ‘If we want to make this siteas successful as it can be, then I am of the view thatit would be reasonable if we shared some personalinformation: who am I, what do I do, what do Iknow, what can I be helpful with?’ From thereon, thecommunity grew rapidly, as people began to intro-duce themselves, provide comment on Tim’s post-ings of article content, and pose queries and topicsfor discussion. The founder was not a high contribu-tor to the forum, though he appeared on a regularbasis to welcome newcomers, announce events, pro-vide reference information, and to encourage thegroup in their discussion. The founder seemed toserve as a moderator/facilitator of the forum.

Early on, the conversations in this forum were notdeep, but there was extensive knowledge seeking


and providing as people posted information aboutthemselves, articles, events, and so on. Peoplerequested and provided opinions, but the threads ofconversation were relatively short and lacked pen-etrating exploration or debate. This general patternof declarative and procedural information exchangebecame a standard routine. Over time, however, theconversation grew beyond this form of learning toinclude deeper dialogue. By October of 2000, con-tributors were providing personalized responses thatincluded detailed reflections and ideas. Most of thediscussion concerned KM definitions, nuances, gen-eral issues and debates. At times the discussions werevery general, almost philosophical, as contributorsraised questions about the meaning of KM terms orhow KM should be managed. At other times the dis-cussions were highly focused, centering on documentmanagement, for example, or protection of intellec-tual property. Individuals described their ownorganizational experiences with KM, including tri-umphs and challenges. The conversations were notparticularly technical (i.e., about computers orsoftware). They were jargon-rich, but jargon was notthe center of the discussion. In all, 130 people contrib-uted to the discussion over the period we studied. Ofthese, 20 constituted a core group in the sense thatthey repeatedly contributed to the discussion andtheir tenure in the community was average or above.

Reviewing the contents of the KMF forum as a whole,we see that knowledge exchange occurred at all threelevels: declarative and procedural, transactive, andsense-making. The online community venue lendsitself to declarative and procedural knowledgeexchange, so it is no surprise that the learning startedat this level. But the KMF community was able tomove into sense-making over time. Our review of theforum contents suggests that this success was facili-tated by their investment in transactive learningabout one another (as encouraged by Tim andreinforced by their founder over time). The groupwas able to develop a meaningful organization out ofwhat was otherwise a formless e-venue. KMF startedwith a clear purpose; it was effective in absorbingnewcomers; there were roles (e.g., moderator) androutines (e.g., conversational habits). Perhaps mostimportant, the KMF community developed a senseof identity and organizational culture. As one con-tributor noted, ‘we are a diverse group but with avery definite culture of knowledge exchange in a sup-portive environment…’ Throughout the forum, thetone was friendly, professional, and personalized.Contributors signed their messages, almost alwayswith their first name and often with their completename, position, organization, and location infor-mation. The forum developed a helping and welcom-ing atmosphere that nurtured debate, challenge, andidea generation — hallmarks of a community of prac-tice, i.e., a sustainable learning network.


Conclusion: Guidelines for LearningNetworks

Strong learning networks are difficult to form in anyvenue, and the challenges are especially great wheninteraction occurs primarily through the conduit ofe-media. Given a set of e-based options from whichto choose, managers can evaluate the options basedon the ways in which participation, co-presence, andconversational structuring are likely to be experi-enced by group members. Evaluation can proceed ina systematic fashion, first assessing the technologicalaffordances of available venues and, from there,anticipating communication and collaborative learn-ing processes. Once an e-venue is selected, groupscan work to produce declarative and proceduralinformation exchange, transactive learning, andsense-making, realizing that these types of learningare progressively more difficult to experience inonline settings.

We offer the following general guidelines for groupsas they endeavor to experience collaborative learninginside e-based venues:

❖ Aim for frequent interaction. Frequent communi-cation facilitates formation of the network andhelps to sustain it over time. E-venues that arevibrant with ongoing communication are morelikely to experience declarative and proceduralinformation exchange, transactive learning, andsense-making. Regular contribution to the conver-sation, even if just ‘checking in,’ signals a groupmember’s presence in the group and will improvethe retention rate in the network.

❖ Foster a mindset of viewing the technology not somuch as a conduit that links distributed people orsites but as a platform for group discourse — amedia space for interaction. In this way, the venueexists not so much to link nodes into a networkbut to provide a shared context in which grouplearning evolves.

❖ Over time, aim for deep discussion, that is, for dis-cussion that includes not only information seekingand providing but also discussion of group mem-bers’ capabilities and changing needs, and dis-cussion that includes challenge, reflection, anddebate. These latter types of discussion are denseand will move the group toward transactive learn-ing and sense-making. Note that deep discussiontakes time to develop. Even in rich, video-linkedmedia spaces, dense discussion rarely happensearly on but rather after the network is established.

❖ Emphasize the importance of speech that is positiveand respectful in tone. Learning is more likely tobe nourished if the conversational atmosphere istolerant and people feel a comfortable degree ofco-presence. Even when conversations includecriticism and debate, communication can be posi-tive. Mutual respect is critical to development of ahealthy learning network, and more sense-making


will occur in groups that exhibit a consistentlyrespectful tone in their online conversation.

❖ Recognize the importance of facilitators, especiallyemergent facilitators. Some facilitators serve astechnology ‘translators’ who are willing to helpwith adaptation of new technology to collabor-ative work (Mackay, 1990; Nardi and O’Day, 1999;Orlikowski et al., 1995). They buffer group mem-bers from difficulties in operation of the tech-nology and encourage learning-oriented com-munication within the group. Other facilitators actas leaders or moderators of the online conver-sation. Both types of facilitators contribute to thesuccess of the learning network.

❖ Work to develop a relatively large core group ofparticipants who provide stability in the networkand foster growth. Avoid insularity or a coregroup that is too small in size to foster learning.

❖ Recognize the importance of routines for inter-action (e.g., regular online contribution, turn-tak-ing, productive debate or challenging oneanother). Conversational routines provide struc-ture that facilitates participation and aidsinterpretation of knowledge as it is shared insidethe group. Specific routines can be advocated (e.g.,asking people to say something about their workbackgrounds in advance of commenting on anissue), but most routines cannot be fully definedin advance. Soft (pliable) routines are more con-ducive to learning than rigid conversational rou-tines. Routines should evolve over time as a func-tion of group needs and preferences.

❖ Encourage groups to experiment with the tech-nology and the conversations they conduct withinit. Through experimentation groups can discoversocial practices that meet their unique needs andinterests.

To conclude, this paper has described some ways tothink about online venues of today and their use forgroup learning. Future research will examine suc-cesses and failures not only within e-venues butacross venues, as group work increasingly spansmultiple e-venue settings, and the concept of mediaspace expands to include entire sets of mediatedlearning environments. Learning networks willevolve to intertwine many groups, communities, andinstitutions, and the challenges for managers willmagnify. No doubt, the study of learning in onlineforums will remain a vibrant area of research formany years to come.

Acknowledgements

The research was supported by National Science FoundationAward SES-0135602 and by a grant from the Hartman Center.The views expressed here are solely those of the authors andnot of the research sponsors.

Notes

1. In this paper we use the term ‘learning network’ to referto the group of people (the social network) who interact for


learning purposes. We use the term e-based learning ‘ven-ue’ or ‘forum’ to refer to the technology-mediated setting inwhich the people interact with one another.

2. Although a more refined analysis of social learning cer-tainly is possible, these three categories capture the majorlearning activities that occur as people interact with oneanother in a group setting. For more detailed typologies ofgroup knowledge exchange, readers are referred to Alaviand Leidner (2001), Choo (1998), and Edmondson et al.(2001).

3. We thank Ms Karen Rivers for assistance in coding the datafor this analysis.

4. Note that we are reporting general trends and not tests ofstatistical significance.

5. The community name and its contributors are disguised inthis paper in order to protect their privacy.

References

Abel, M.J. (1990) Experiences in an exploratory distributedorganization. In Intellectual Teamwork, eds J. Galagher, R.Kraut and C. Egido, pp. 489–510. Erlbaum, Norwood, NJ.

Alavi, M. and Leidner, D.E. (2001) Review: knowledge man-agement and knowledge systems: conceptual foun-dations and research issues. MIS Quarterly 25(1), 107–136.

Aldrich, H. (1999) Organizations Evolving. Sage, ThousandOaks, CA.

Bly, S., Harrison, S. and Irwin, S. (1993) Media spaces: bring-ing people together in a video audio and computingenvironment. Communications of the ACM Special Issue36(1), 29–47.

Brown, J.S. and Duguid, P. (1991) Organizational learning andcommunities-of-practice: toward a unified view of work-ing, learning, and innovation. Organization Science 2,40–57.

Carlson, J.R. and Zmud, R.W. (1999) Channel expansiontheory and the experiential nature of media richness per-ceptions. Academy of Management Journal 42(2), 153–170.

Choo, C.W. (1998) The Knowing Organization. Oxford Univer-sity Press, New York.

Cook, S.D.N. and Brown, J.S. (1999) Bridging epistemologies:the generative dance between organizational knowledgeand organizational knowing. Organization Science 10(4),381–400.

Daft, R.L. and Lengel, R.H. (1986) Organizational informationrequirements, media richness, and structural determi-nants. Management Science 32, 554–571.

Edmondson, A.C., Bohmer, R.M. and Pisano, G.P. (2001) Dis-rupted routines: team learning and new technologyimplementation in hospitals. Administrative Science Quar-terly 46, 685–716.

Fayard, A.-L. (2002) The virtual stage: video-mediatedbehaviors and presentation of self in a video-mediatedenvironment. Unpublished working paper. INSEAD,Singapore.

Gaver, W.W. (1993) How do we hear in the world? Explo-rations in ecological acoustics. Ecological Psychology 5,285–313.

Gibson, J.J. (1977) The theory of affordances. In Perceiving, Act-ing and Knowing, eds R.E. Shaw and J. Bransford.Erlbaum, Hillsdale, NJ.

Gibson, J.J. (1979) The Ecological Approach to Visual Perception.Houghton Mifflin, Boston, MA.


Goffman, E. (1963) Behavior in Public Places: Notes on the SocialOrganization of Gatherings. Free Press, New York.

Goodwin, C. and Heritage, J. (1990) Conversational analysis.Annual Reviews in Anthropology 19, 283–307.

Guzzo, R.A. and Dickson, M.W. (1996) Teams in organiza-tions: recent research on performance and effectiveness.Annual Review of Psychology 47, 307–338.

Hiltz, S.R. (1994) The Virtual Classroom: Learning without Limitsvia Computer Networks. Ablex, Norwood, NJ.

Hollingshead, A.B. and McGrath, J.E. (1995) Computer-assisted groups: a critical review of the empiricalresearch. In Team Effectiveness and Decision Making inOrganizations, eds R.A. Guzzo and E. Salas, pp. 46–78.Jossey-Bass, San Francisco, CA.

Ko, D.-G., Kirsch, L.J. and King, W.R. (2003) Determinants ofinter-firm knowledge transfer in enterprise systemimplementations: Transferring knowledge from consult-ants to clients.

Lave, J. and Wenger, E. (1991) Situated Learning: Legitimate Per-ipheral Participation. Cambridge University Press, Cam-bridge.

Lea, M. and Spears, R. (1991) Computer-mediated communi-cation, de-individuation and group decision-making.International Journal of Man–Machine Studies 34, 283–301.

Mackay, W.E. 1990 Users and customizable software: a co-adaptive phenomenon. Unpublished Ph.D. dissertation,Massachusetts Institute of Technology.

Markus, M.L. (1994) Electronic mail as the medium of mana-gerial choice. Organization Science 5(4), 502–527.

Marshall, C. and Novick, D. (1995) Conversational effective-ness and multi-media communications. Information Tech-nology and People 8(1), 54–79.

Moreland, R.L., Argote, L. and Krishnan, R. (1998) Trainingpeople to work in groups. In Theory and Research on SmallGroups: Social Psychological Applications to Social Issues,eds R.S. Tindale and L. Heath, pp. 37–60. Plenum Press,New York.

Nardi, B. and O’Day, V. (1999) Information Ecologies: UsingTechnology with Heart. MIT Press, Cambridge, MA.

Orlikowski, W. J., Yates, J., Okamura, K. and Fujimoto (1995)Shaping electronic communication: the metastructuringof technology in the context of use. Organization Science6(4).

Sarbaugh-Thompson, J.S. and Feldman, M.S. (1998) Electronicmail and organizational communication: does saying ‘hi’really matter? Organization Science 9(6), 685–698.

Shapiro, D., Furst, S.A., Spreitzer, G. and Von Glinow, M.A.(2002) Transnational teams in the electronic age: are teamidentity and high-performance at risk? Journal of Organi-zational Behavior 23, 455–468.

Short, J.E., Williams, E. and Christie, B. (1976) The Social Psy-chology of Telecommunications. Wiley, London.

Straus, S.G. (1997) Technology, group process, and group out-comes: testing the connections in computer-mediatedand face-to-face groups. Human–Computer Interaction 12,227–266.

Weick, K.E. (1995) Sensemaking in Organizations. Sage, Thou-sand Oaks, CA.

Wegner, D.M. (1986) Transactive Memory: A ContemporaryAnalysis of the Group Mind. Theories of Group Behavior. InSpringer-Verlag, New York.

Wenger, E. (1998) Communities of Practice: Learning, Meaning,and Identity. Cambridge University Press, Cambridge.

Wittgenstein, L. (1974) Philosophical Investigations. MacMillan,New York.


GERARDINE ANNE-LAUREDESANCTIS, Duke Uni- FAYARD, INSEAD, 1versity, 1 Tower View, Box Ayer Rajah Avenue, Singa-90120, Durham, NC 27708- pore 138676. E-mail: anne-0120, USA. E-mail: gd@du- [email protected]

Anne-Laure Fayard isGerardine DeSanctis is Assistant Professor in Tech-Thomas F. Keller Professor nology Management atof Business at Duke Univer- INSEAD. Her researchsity, and a member of the interests include socio-faculty in Duke’s Global technical practices, collabor-

Executive MBA Program. Her extensive editorial ation, virtual teams and distributed cognition.experience includes Associate Editor for Manage-ment Science.

MICHAEL ROACH, LU JIANG, Duke Univer-Duke University, 1 Tower sity, 1 Tower View, BoxView, Box 90120, Durham, 90120, Durham, NC 27708-NC 27708-0120, USA. E- 0120, USA. E-mail: lj4@du-mail: [email protected] ke.edu

Michael Roach is a doctoral Lu Jiang is a doctoral candi-candidate in management at date in management at thethe Fuqua School of Busi- Fuqua School of Business,ness, Duke University. His Duke University. Her cur-research interests include rent research interestsorganizational design, include communication

organizational learning and communication networks. behaviour and virtual team structure and performance.


management focus learning in online forums

Documents