Virtual Teams: Effects of Technological Mediation on Team ... · proliferation of teams that do not...

Virtual Teams: Effects of Technological Mediation onTeam Performance

James E. DriskellFlorida Maxima Corporation

Paul H. RadtkeNaval Air Warfare Center Training

Systems Division

Eduardo SalasUniversity of Central Florida

Recent advances in networking environments and telecommunications have led to theproliferation of teams that do not work face-to-face but interact over a computer-mediated communications network. Although some have asserted that virtual teamstranscend boundaries of time or distance, others have claimed that working remotely ina mediated team environment differs in significant ways from working face-to-face. Inthis article, the authors examine the effects of technological mediation on teamprocesses such as cohesiveness, status and authority relations, counternormative be-havior, and communication. They discuss conditions under which distance matters invirtual team interaction.

Recent advances in networking environmentsand telecommunications have led to the prolif-eration of teams that do not work face-to-facebut interact over a computer-mediated commu-nications network. We use the term virtual teamto refer to a team or group whose members aremediated by time, distance, or technology.Other closely related terms that have been usedto describe this type of environment includecomputer-mediated communications and com-puter-supported cooperative work. Althoughthere are differences in the type of technologyused and the types of communication enabled invirtual team environments, for our purposes, thecore feature of a virtual team is that it is one inwhich interdependent group members work to-

gether on a common task while they are spa-tially separated.

Cairncross (1997) has coined the phrase “thedeath of distance,” suggesting that distance mayno longer be a limiting factor in our ability tocommunicate and is quickly becoming irrele-vant to the way people interact. Proponents ofthis view presume a future (or present) in whichall time and space restrictions have been re-moved from the communication process andwhere “face-to-face communication can bedone across oceans if video conferencing facil-ities are available” (Burgelman, 2000, p. 3).However, most researchers take issue with theview that the technology that mediates humaninteraction is seamless or transparent. For ex-ample, Olson and Olson (2000) argued that“distance matters” and that group members whoare remotely located or distributed from oneanother are likely to face obstacles in coordi-nating group efforts. The general consensus isthat the nature of interaction in virtual teamsmay differ in a number of important ways from“normal” face-to-face team interaction. How-ever, if distance matters to how group membersinteract, then how does it matter? How does thisphenomenon—that virtual team members mustwork interdependently but remotely in a com-

James E. Driskell, Florida Maxima Corporation, WinterPark, Florida; Paul H. Radtke, Naval Air Warfare CenterTraining Systems Division, Orlando, Florida; EduardoSalas, Department of Psychology, University of CentralFlorida.

The views expressed herein are those of the authors anddo not reflect the opinion, policy, or views of the Depart-ment of Defense.

Correspondence concerning this article should be ad-dressed to James E. Driskell, Florida Maxima Corporation,507 North New York Avenue, R-1, Winter Park, Florida32789. E-mail: [email protected]

Group Dynamics: Theory, Research, and Practice Copyright 2003 by the Educational Publishing Foundation2003, Vol. 7, No. 4, 297–323 1089-2699/03/$12.00 DOI: 10.1037/1089-2699.7.4.297

297

puter-mediated environment—impact the na-ture of team interaction and performance?

The purpose of this article is to examine theeffects of technological mediation on team pro-cesses such as cohesiveness, status and author-ity relations, counternormative behavior, andcommunication. We first present a model of theeffects of technological mediation on team per-formance. We then examine how performing ina virtual team environment may impact teamprocesses. Finally, we discuss factors, such asthe nature of the task, that may moderate theeffects of technological mediation on teaminteraction.

Team Interaction in Computer-MediatedEnvironments

In face-to-face interaction, group membersshare the same physical location, can see andhear one another, receive messages in “realtime” as they are produced, and send and re-ceive information simultaneously and in se-quence. During face-to-face interaction, groupmembers can see another’s nods and gestures;they can observe eye contact, facial expressions,and posture; they can hear the other’s tone ofspeech and dialect; they are aware of the timingof speech and who responds to whom; and theyexperience the immediacy of interacting andbeing involved with a physically present teammember. These types of contextual cues provideimportant information about the individual withwhom one is interacting (i.e., Does the speakergive evidence of being competent or experi-enced? Is the speaker a high-status or low-statusteam member?), how the message is being con-veyed (i.e., Is the speaker angry, tense, or con-fident?), and whether the message is being con-veyed successfully (i.e., Are others paying at-tention? Do they look puzzled?). Groups whosemembers are distributed apart from one anothermay lose some of these communicativecapabilities.

A substantial body of research on computer-mediated communication has been conducted inthe past several decades. Some studies haveindicated that the distribution of team membersover remote networks tends to impair team in-teraction in comparison with face-to-face inter-action. McLeod (1992) found that computer-mediated interaction led to an increase in thetime required to make a decision and a decrease

in team member satisfaction. Straus (1997) andWarkentin, Sayeed, and Hightower (1997) re-ported that virtual teams developed lower cohe-siveness than face-to-face teams. However,other studies have produced conflicting resultsthat are somewhat more difficult to interpret.For example, whereas Dubrovsky, Kiesler, andSethna (1991) found that status distinctionsamong team members were reduced in comput-er-mediated groups, Weisband, Schneider, andConnolly (1995) found little evidence of statusequalization. Although some researchers havereported a greater incidence of counternorma-tive or uninhibited behavior in computer-medi-ated groups (Siegel, Dubrovsky, Kiesler, &McGuire, 1986), others have failed to find thesedifferences (Walther & Burgoon, 1992). A rea-sonable and cautious interpretation of the evi-dence at this point is that, indeed, distance mat-ters—that working remotely in a mediated teamenvironment is different from working face-to-face—but that the manner in which media-tion affects team interaction warrants closerexamination.

There are several overarching points that arerelevant to the following discussion. First, whenwe use the term computer-mediated communi-cations, we must remain cognizant of the factthat there are a number of different types ofmediated environments. Clark and Brennan(1991) described several characteristics of face-to-face and distributed settings that determinethe nature of communication:

Copresence: Group members occupy the same physi-cal location.

Visibility: Group members can see one another.

Audibility: Group members can hear one another.

Cotemporality: Communication is received at the ap-proximate time it is sent.

Simultaneity: Group members can send and receivemessages simultaneously.

Sequentiality: Group members’ speaking turns stay insequence.

As shown in Table 1, computer-mediated en-vironments differ according to the communica-tion capabilities that are enabled. In face-to-faceinteraction, group members share the samephysical location, can see and hear one another,receive messages in real-time as they are pro-duced, and send and receive information simul-taneously and in sequence. Teams that are dis-tributed over various types of computer-medi-

298 DRISKELL, RADTKE, AND SALAS

ated environments lose certain capabilities. In avideoconference setting, distributed groups mayinteract over networked computer systems andexchange live video as well as audio and text.On the other hand, what we have termed com-puter chat refers to the computer-mediated elec-tronic dialogue between two or more groupmembers, in which the users exchange mes-sages via text in real-time. Group communica-tion over this type of distributed environment iscotemporal, simultaneous, and sequential, butgroup members lack the capability to see oneanother and to hear the timing or intonation oftheir speech.

The important point that this classificationillustrates is that all of the settings in Table 1(other than the face-to-face setting) may be de-fined as a computer-mediated environment.They are all alike in that group members workon a common task but do not share the samespatial location. However, they differ quite con-siderably regarding the communication capabil-ities that are enabled. Therefore, when we dis-cuss virtual teams, we must be aware that vir-tual teams may operate in different types ofcommunication environments and that the typeof communication environment implementedwill have a significant impact on teaminteraction.

A second broad point of consideration is thatjust as there are different types of technologi-cally mediated environments, there are differenttypes of teams. Teams may be ad hoc (a tem-porary team assembled solely for a specifictask) or intact (an existing team in which teammembership is stable). Teams may work to-

gether on a task over time, or interaction may beshort and time limited. Teams may meet ini-tially prior to interaction, or team members maybe completely anonymous. Team members mayexpect future interaction with one another afteran initial task is performed, or the task may bea one-shot transaction. Teams can be composedof many members or few. All of these factorsplay a role in how teams develop and interact ina mediated environment.

Finally, a third point is that the distribution ofteam members over computer-mediated sys-tems can disrupt team interaction under someconditions and facilitate interaction under otherconditions. The focus of most research on com-puter-mediated team performance has been onthe disadvantages incurred when team membersmust perform apart from one another. Clearly,coordination of team activities can becomemore difficult when team members are not inthe same physical location. However, teammembership can be a double-edged sword. Forexample, research indicates that under certainconditions, being in a team can lead to negativeconsequences such as increased pressure forconformity and impaired decision making(Mullen, Anthony, Salas, & Driskell, 1994).Thus, it follows that there are some circum-stances in which separating team members apartfrom one another may serve to overcome thenegative consequences of team membership.Therefore, the question is not simply “How isperformance impaired in virtual teams?” In-stead, we expect that performing in a virtualteam environment will affect teams in ways thatmay be advantageous as well as disadvanta-

Table 1Characteristics of Face-to-Face and Mediated Environments

Type of environment

Media characteristics

Copresence Visibility Audibility Cotemporality Simultaneity Sequentiality

Face-to-face X X X X X XReal-time audio/video

(videoconference) X X X X XAudio-only (telephones,

conference calls) X X X XReal-time electronic dialogue,

text-only (computer chat) X X XE-mail

Note. From “Grounding in Communication,” by H. H. Clark & S. E. Brennan, in L. B. Resnick, J. M. Levine, & S. D.Teasley (Eds.), Perspectives on Socially Shared Cognition (p. 142). Washington, DC: American Psychological Association.Copyright 1991 by the American Psychological Association.

299VIRTUAL TEAMS

geous. As a practical strategy, we may wish tomitigate the costs of computer-mediated com-munication in circumstances in which it inter-feres with effective team functioning; at thesame time, we want to realize the benefits incircumstances in which computer-mediatedcommunications may facilitate team interaction.

A Research Model

Figure 1 presents a model that illustrates theeffects of computer-mediated communicationson team performance. This model adopts a gen-eral input–process–output framework. The in-

put factor, shown in column 1, is the computer-mediated environment. Note that, as we dis-cussed above, the term computer-mediatedcommunications is a general term and that spe-cific types of computer-mediated environmentsoffer different communicative capabilities. Infact, the type of communications environmentover which team members interact is one of theprimary moderators that we intend to examinein the following sections.

The second column in Figure 1 describesseveral team processes of interest. Research hasexamined the effect of computer-mediated en-vironments on cohesiveness (Straus, 1997;

Figure 1. Input–process–output model of the effects of mediation on team interaction.CMC � computer-mediated communications.


Warkentin et al., 1997), status and authorityrelations (Dubrovsky et al., 1991; Hiltz, Turoff,& Johnson, 1989; Silver, Cohen, & Crutchfield,1994), counternormative behavior (Dubrovskyet al., 1991; Kiesler, Zubrow, Moses, & Geller,1985; Siegel et al., 1986, Sproull & Kiesler,1986), and communication (Sellen, 1995;Thompson & Coovert, 2003). Although this re-view must by necessity be selective, we chosethese specific team processes to examine be-cause they have been examined in prior reviewsof team performance in general and in reviewsof team performance in computer-mediated en-vironments (cf. Hollingshead & McGrath,1995; Straus, 1997) and because evidence sug-gests these team processes are affected by tech-nological mediation.

In Baron and Kenny’s (1986) terms, we viewprocess variables as mediators. That is, teamprocess variables are patterns of group interac-tion through which input factors influence oraffect team outcomes (see Hackman & Morris,1975). In reality, this input–process–output re-lationship is often not as linear or static as thismodel implies. If we consider team perfor-mance over time, team outcomes such as theperformance of the group can in turn impactteam processes such as cohesiveness. More-over, given the specific interests of the re-searcher, variables such as cohesiveness can beconsidered as input variables and outcome vari-ables as well.

The third column describes performance out-comes. Note that we restrict our focus to theeffects of technological mediation on perfor-mance in task groups. Furthermore, althoughperformance can be defined by more specificoutcome measures such as accuracy, speed, andvariability of performance, it is appropriate forour purposes to simply address team perfor-mance, broadly defined.

Finally, moderators are factors that affect thestrength of the relation between two variables(Baron & Kenny, 1986) or that can account forobserved variation in a relationship. Typically,each specific research literature, such as theresearch on group status or on communication,has its own idiosyncratic cluster of moderatorsthat are dictated by research concerns and find-ings within that domain. However, there aresome moderators that emerge at least to someextent across most domains. We examine threesuch variables that seem most relevant to virtual

teams: (a) type of computer-mediated environ-ment, (b) type of task, and (c) temporal context.

We have previously described differences intypes of computer-mediated environments (seeTable 1). Types of computer-mediated environ-ments (e.g., audio–video vs. audio only vs. textonly) differ considerably in the richness of com-munications afforded (Daft & Lengel, 1986).Thus, we would expect the effect of technolog-ical mediation on team processes to vary de-pending on the type of computer-mediated com-munications environment.

Perhaps no factor has a greater effect on teaminteraction than the type of task that the team isengaged in. There have been a number of at-tempts to classify group tasks along a variety ofdimensions, such as the cognitive versus phys-ical requirements of the task or the requirementfor cooperation or interdependence (cf.McGrath, 1984; Shaw, 1973; Steiner, 1972).One of the most recent and comprehensive clas-sification systems relevant to teams has beenoffered by Devine (2002), who identified 14team types on the basis of the type of work thatthey are engaged in, and further classified themalong seven contextual dimensions. Driskell,Hogan, and Salas (1987) also classified grouptasks on the basis of the primary activities orbehaviors required of team members, resultingin six task categories: (a) mechanical/technicaltasks, requiring the construction or operation ofthings; (b) intellectual/analytic tasks, requiringgeneration of ideas, reasoning, or problem solv-ing; (c) imaginative/aesthetic tasks, requiringcreativity or artistic endeavor; (d) social tasks,requiring training, supporting, or assisting oth-ers; (e) manipulative/persuasive tasks, requiringmotivation or persuasion of others; and (f) log-ical/precision tasks, requiring performance ofroutine, detailed, or standardized tasks. Weadopt this typology, focusing on the behaviorsrequired for task completion, to examine theeffects of type of task. We would expect theeffect of technological mediation on team pro-cesses to vary depending on the type of task.

A third factor that we would expect to have asignificant effect on the relationship betweentechnological mediation and team interaction isthe temporal context of the team. McGrath(1997) has noted that research is often con-ducted with ad hoc laboratory groups with nopast or anticipated future, in contrast to dynamicgroups that are intact and perform over a longer

301VIRTUAL TEAMS

period of time (see also Hollingshead &McGrath, 1995; McGrath, 1990). Teams thatperform over time gain experience with the taskand the communications technology, whichmay reduce performance decrements (Hollings-head, McGrath, & O’Conner, 1993). Althoughthere are a number of temporal distinctions thatcan be drawn related to the synchronicity, pac-ing, and sequencing of performance, we focuson the simple distinction between ad hoc teamsthat interact for a single session and dynamicteams that interact over time. We would expectthe effect of technological mediation on teamprocesses to vary depending on the temporalcontext of the team.

These moderators may have an impact on therelationships depicted in Figure 1 at two sepa-rate points. A moderator such as the type of taskmay impact the extent to which a variable suchas cohesiveness affects team performance (re-flected in the rightmost arrow from the “Mod-erator” box shown in Figure 1). However, therelationship between cohesiveness and perfor-mance is not unique to the topic of virtualteams, and the extent to which this generalrelationship is moderated by the type of task iscaptured within that more general research lit-erature. Our more specific interest is in exam-ining factors that moderate the extent to whichthe computer-mediated communications envi-ronment affects team processes (the leftmostarrow from the “Moderator” box in Figure 1).Thus, our focus, for example, is on whether theeffects of technological mediation on cohesive-ness may vary according to the type of task theteam is engaged in.

In summary, this model is intended to orga-nize or structure examination of the effects oftechnological mediation on team interaction.We propose that technological mediation mayimpact team performance because of changes incohesiveness, status, counternormative behav-ior, and communication and that these changesmay be moderated by factors such as the type ofcommunications environment, the type of task,and the temporal context of the team. In thefollowing sections, we use the framework ofthis model to examine each of the team pro-cesses identified. In each case, we briefly de-scribe the team process, discuss how that pro-cess may be affected within a virtual teamenvironment, discuss possible performance ef-fects, and examine potential moderators.

Cohesiveness

Cohesiveness is considered to be one of themost fundamental aspects of groups. Golem-biewski (1962) described it as “the essentialsmall group characteristic” (p. 149). Moreover,research has shown that cohesive groups tend tointeract more (Back, 1951), agree more readily(Lott & Lott, 1961), report greater satisfactionwith the group (Curtis & Miller, 1986), and atleast under some circumstances, outperformless cohesive groups (Mullen & Copper, 1994).However, there is some concern that virtualteams may experience greater difficulty in de-veloping strong relational bonds that underliegroup identity, cohesiveness, and trust (Jarven-paa & Leidner, 1999; Rocco, 1998; Warkentinet al., 1997).

Within the research literature, cohesivenesshas been viewed as group pride, loyalty, sharedunderstanding, bonding, interpersonal attrac-tion, trust, task commitment, and mutual aid,leading some researchers to question the clarityand uniformity of existing conceptualizations(Levine & Moreland, 1990). However, it is im-portant to note that even the earliest conceptu-alizations of cohesiveness treated it as a multi-dimensional construct. Festinger (1950) pro-vided the seminal definition of cohesiveness asthe forces that act on team members to remainin the team, including interpersonal attraction toother team members, group prestige, and attrac-tion to the activities in which the group engages.Most subsequent research on group cohesive-ness has elaborated one or more of these threemajor dimensions of cohesiveness.

Mullen and Copper (1994), in a meta-analy-sis of the effects of group cohesiveness on per-formance, examined separately the three com-ponents of cohesiveness first identified by Fest-inger. Interpersonal attraction reflects affectiverelations or attraction to other team members.Group pride reflects group prestige, loyalty, andnormative bonds. Task commitment reflectscommitment to the task or goals of the team.Mullen and Copper found that group cohesive-ness had a positive effect on performance andfurther found that this relationship was primar-ily a function of the task commitment compo-nent of cohesiveness.

Furthermore, some research has shown thatthe components of cohesiveness may have dif-ferential effects depending on the type of task.


For example, on an additive task, group mem-bers pool individual products to produce a teamoutcome but are not required to coordinate ac-tions among themselves. Zaccaro and Lowe(1986) found that higher task-based cohesive-ness led to greater performance on an additivetask (folding paper tents), whereas interpersonalcohesiveness had no effect. However, on a “sur-vival exercise” disjunctive task that requiredteam members to communicate and coordinateindividual efforts, Zaccaro and McCoy (1988)found that team performance was affected byboth task cohesiveness and interpersonal cohe-siveness. Craig and Kelly (1999) also found thatboth task cohesiveness and interpersonal cohe-siveness enhanced performance on an interac-tive group creativity task.

Cohesiveness in Virtual Teams

Generally speaking, researchers have sug-gested that the process of team developmentmay be more complex in a virtual environment.In contrast to face-to-face interaction in whichindividuating information on team members isabundant, some have argued that members ofvirtual teams are more anonymous and deindi-viduated. Thus, interaction that is mediated bytechnology may lead to less intimacy and diffi-culty in establishing relationships among teammembers. Furthermore, some research hasshown that weaker relational ties in virtualteams can lead to lower cohesion (Straus, 1997;Warkentin et al., 1997).

However, we have noted that cohesiveness iscomposed of three primary components—inter-personal attraction, group pride, and task com-mitment—and we propose that technologicalmediation may have a differential effect onthese three components of cohesiveness. First, itis likely that a weakening of social cues or areduction in personalizing information amongvirtual team members may lead to weaker af-fective bonds and a decrease in intimacy(Straus, 1997; Weisband & Atwater, 1999).Thus, we would expect technological mediationto have a negative impact on cohesiveness,when defined as interpersonal attraction.

Second, technological mediation may affectthe group pride component of cohesiveness. Tothe extent that normative bonds may be weak-ened in computer-mediated teams, the distribu-tion of team members over a computer-medi-

ated network may lead to lower commitment tothe team. However, the impact of mediation ongroup pride may be minimized if group mem-bers possess a strong preexisting sense of prideand loyalty to the group or organization towhich they belong. In other words, ad hoc teamsthat have never met prior to the virtual teaminteraction may have difficulty in developing astrong team commitment, whereas intact teamsmay have developed strong normative bondsprior to interaction in a virtual team setting.Thus, we would expect technological mediationto have a negative impact on cohesiveness,when defined as group pride, in some situationsbut not in others.

Third, technological mediation may also im-pact the task commitment component of cohe-sion. Task commitment is an instrumental bond,rather than an interpersonal or normative bond,reflecting attractiveness of and satisfaction withthe task or the group’s activities (Mullen &Copper, 1994). Some studies suggest teammembers mediated by technology experienceless satisfaction with the task, although this maybe a function of the requirements of the task orthe newness of the task experience (Straus,1996; Straus & McGrath, 1994). In general, wewould expect technological mediation to have anegative impact on cohesiveness, when definedas task commitment.

Performance Effects

The empirical evidence for the effect of co-hesiveness on performance is mixed. Studiesreport that cohesiveness may have a positiveeffect on performance (Tziner & Vardi, 1982),no effect on performance (Terborg, Castore, &DeNinno, 1976), or either positive or negativeeffects depending on the performance standardsof the group (Schachter, Ellertson, McBride, &Gregory, 1951). However, meta-analyses of thecohesiveness–performance literature reportoverall positive mean effects of cohesiveness onperformance (Evans & Dion, 1991; Mullen &Copper, 1994; Oliver, Harman, Hoover, Hayes,& Pandhi, 1999). Furthermore, the Mullen andCopper (1994) meta-analysis reported that thestrongest effect of cohesiveness on performanceoccurred when cohesiveness was operational-ized as task commitment and that the interper-sonal attraction and group pride components ofcohesiveness had weaker effects on perfor-

303VIRTUAL TEAMS

mance. This evidence suggests that cohesive-ness enhances performance and that it does soprimarily because cohesive team members aremore committed to successful task perfor-mance. This suggests that team task perfor-mance would be most degraded through thenegative effect of technological mediation ontask commitment and that there is less of apotential effect on performance of reduced so-cioemotional bonds or normative bonds. Inother words, technological mediation is morelikely to impact team performance through itseffect on task commitment rather than throughits effect on socioemotional bonds or normativebonds.

In brief, we would expect technological me-diation to affect the different components ofcohesiveness, with the potential to reduce inter-personal attraction, group pride, and task com-mitment. In the following, we consider factorsthat may moderate the effect of technologicalmediation on cohesiveness.

Moderators

Type of task. In keeping with our focus onthe three primary components of cohesiveness,we propose that the effect of technological me-diation on cohesiveness will differ according tothe type of task that the team is performing andthe components of cohesiveness that are mostrelevant to that task. To the extent that techno-logical mediation impairs interpersonal attrac-tion, this should have a greater impact on cohe-siveness for tasks in which interpersonal rela-tions are most salient. Thus, we would expectthe impact of technological mediation on theinterpersonal attraction component of cohesive-ness to be greater for social tasks or manipula-tive/persuasion tasks that emphasize affectiverelations. To the extent that technological me-diation impairs group pride, this would have agreater impact on tasks that require strongshared beliefs and normative consensus. Wewould expect the impact of technological me-diation on the group pride component of cohe-siveness to be greater for teams organized forideological or special-interest purposes. Finally,to the extent that technological mediation im-pairs task commitment, this would have agreater impact on tasks that are primarily per-formance or productivity oriented and that em-phasize instrumental relations. Thus, we would

expect the impact of technological mediation onthe task commitment component of cohesive-ness to be greater for mechanical/technical orintellectual/analytic tasks. Unfortunately, thereis little empirical evidence to support thesepredictions.

Type of computer-mediated environment.Two major approaches to understanding the ef-fects of different types of communications me-dia, theories of social presence (Short, Wil-liams, & Christie, 1976) and media richness(Daft & Lengel, 1984), hold that the capacity totransmit communicative information (visual,verbal, and contextual cues) is progressivelyrestricted as one moves from face-to-face toaudio–video to audio-only to textual modes ofcommunication. The type of information that isrestricted includes verbal and nonverbal expres-sive cues, body language, gaze, gesture, appear-ance, and voice tone. This loss of contextualinformation across various communicationmodes may differentially impact the three pri-mary components of cohesiveness.

It is likely that the loss of expressive contex-tual information, especially in audio-only andtext communication modes, will lead to weakerinterpersonal bonds. Straus (1997) found thatcomputer-mediated teams (engaged in com-puter conferencing via text messaging) reportedless cohesiveness, measured as interpersonal at-traction, than did face-to-face teams. Weisbandand Atwater (1999) reported that teams com-municating via a text-messaging network likedother team members less than did face-to-faceteams. Examining trust across different types ofmedia, Bos, Olson, Gergle, Olson, and Wright(2002) found that teams using text had thegreatest difficulty establishing trust, and thatteams using video conferencing and audio con-ferencing took longer to establish trust thanface-to-face teams. Thus, although it is reason-able to expect that strong interpersonal bondscan develop in less rich modes of communica-tion such as chat or e-mail, it is likely that thesebonds will develop more slowly and withgreater difficulty.

The richness of the mode of communica-tion may also moderate the effect of techno-logical mediation on task commitment.Doherty-Sneddon et al. (1997) found that taskefficiency and understanding were lessened invideoconferencing and audio-only communi-cations versus face-to-face communications.


Straus and McGrath (1994) similarly foundthat productivity, or the time it took to com-plete a given task, was less for computer-mediated communications (computer confer-encing via text messaging) than with face-to-face communications. Straus and McGrathalso reported that computer-mediated teammembers had a harder time understanding oneanother than did face-to-face teams, and Straus(1996) found that computer-mediated teammembers were less satisfied with the task pro-cess than face-to-face members. Thus, wewould expect that richer modes of communica-tion may allow more satisfactory task participa-tion and accomplishment (especially for moredemanding tasks), which would be reflected ingreater task commitment.

Finally, the richness of the mode of commu-nication may moderate the effect of technolog-ical mediation on group pride. No studies thatwe are aware of have tested this relationshipdirectly. However, group pride is related to nor-mative consensus, shared beliefs, and satisfac-tion with the team, and some research hasshown that less rich modes of communicationcan result in longer time to reach consensus andagreement (Connolly, Jessup, & Valacich,1990; Hiltz, Johnson, & Turoff, 1986) andlower identification with the group (Cappel &Windsor, 2000). Sellen (1995) examined face-to-face teams, teams who interacted via audio–video communications, and teams who inter-acted via audio-only communications. Shefound that compared with face-to-face groups,technology-mediated teams (with or withoutvideo) exhibited clear symptoms of depersonal-ization and psychological distance. Moreover,she found that overall, the use of audio versusvideo mattered less than whether group mem-bers were mediated or interacted face-to-face.Sellen concluded that some forms of interactionmay be fundamentally altered when team mem-bers do not share the same space. Thus, for thedevelopment and maintenance of strong norma-tive bonds, the important distinction may bebetween mediated and face-to-face teams, withthe type of mediated communication environ-ment less relevant.

Temporal context. Some research has sug-gested that team processes develop more slowlyin virtual teams and that the weaker relationalties observed in virtual teams may be the simpleresult of these teams needing longer to develop

(Griffith & Neale, 1999). Walther and Burgoon(1992) found that members of computer-medi-ated groups initially rated each other unfavor-ably. Yet over the course of several weeks,group members’ ratings of composure/relax-ation, informality, receptivity, trust, and socialorientation became more positive. Thus, deficitsin the development of team processes in virtualteams may be temporary, and some researchsuggests that virtual teams can function as ef-fectively as face-to-face teams provided theyhave sufficient time to develop.

Furthermore, most of the research that hassuggested a weakening of cohesiveness in vir-tual teams examined teams that had never met,had no history, and were not likely to meetagain after the interaction ceased. Research sug-gests that these ad hoc teams may experiencedifficulty in development of team processessuch as trust (Jarvenpaa & Leidner, 1999).However, some research has suggested thattrust can develop in virtual teams but that itdevelops in a different manner from the gradualdevelopment of trust in face-to-face groups.Meyerson, Weick, and Kramer (1996) arguedthat such groups may develop “swift trust,”based on existing bonds of organizational mem-bership rather than on interpersonal feedbackfrom the task.

Status and Authority Relations

Several decades of research conducted withmock juries, military teams, and problem-solv-ing groups have revealed how status differencesamong group members structure the nature ofgroup interaction. In a typical work group, thosewith higher status assume a leadership positionand command more of the group’s resources—they talk more, they direct group activities, andthey are more likely to exert their opinion dur-ing decision making. In brief, individuals whoare perceived as having high status within thegroup command more of the group’s resourc-es—they dominate conversation, their ideas areaccepted more often, and they are seen as morecompetent and leaderlike (see Driskell &Mullen, 1990; Driskell, Olmstead, & Salas,1993).

The theory of status characteristics and ex-pectation states provides one of the most com-prehensive and well documented explanationsof how status differentials emerge and are main-

305VIRTUAL TEAMS

tained in small groups (see Berger, 1992; Wag-ner & Berger, 1993; Webster & Driskell, 1983).According to this theory, status and influence intask groups are determined by the performanceexpectations formed for group members relativeto one another. The higher the performanceexpectations formed for a group member are,the more likely that person is to be given op-portunities to perform in the group, initiate in-teraction, receive positive evaluations andagreement from others, and exert influence.

To illustrate this process, consider a decision-making group that has initially gathered to per-form a task. The group members are task ori-ented, they want to achieve a successful out-come of the task, and they search forinformation regarding each others’ capabilitiesto help them achieve this goal. It is in eachindividual’s self-interest, in order to accomplishthe task, to defer to others on the basis of theothers’ expected task contributions. In otherwords, it is in Person A’s best interest to deferto Person B (i.e., to allow B to command moretime speaking, to accept B’s influence attempts,etc.) if Person B seems to possess superior taskcapability. Therefore, deference is exchangedby some members of the group for the perceivedsuperior task contributions of others.

There are several types of characteristics ofindividuals that are typically salient in face-to-face groups and that provide a basis for theformation of performance expectations. We willdiscuss two types, status characteristics and sta-tus cues. Status characteristics are externalcharacteristics that differentiate individuals,such as race, gender, occupation, and evenphysical attractiveness (Webster & Driskell,1983). In most cases, the status hierarchy thatemerges within the small group reflects thesecultural stereotypes: Males, Whites, and thoseof higher occupational status enact a more pro-active role and command more of the group’sresources, whereas females, ethnic minorities,and those of lower occupational status are lessactive and generally more compliant duringgroup deliberations. There are two types of sta-tus characteristics: (a) diffuse status character-istics such as race, gender, and age, whichevoke broad, general expectations for perfor-mance, and (b) specific status characteristics,which include specific skills or abilities andhave more specific, delimited expectations forperformance. In brief, status characteristics are

visible characteristics of individuals that struc-ture such group behaviors as who speaks towhom, who speaks more often, and whose ideasare more likely to be solicited and accepted.

Status cues are expressive behaviors that peo-ple exhibit, such as body position, intonation,gaze, gestures, and other expressive cues. Ingeneral, individuals who speak rapidly, withfew verbal disfluencies or hesitations, who se-lect the head of the table, who speak more often,and who maintain eye contact, especially whilespeaking, are perceived as more competent andoccupy a higher position in the group statushierarchy (see Berger, Webster, Ridgeway, &Rosenholtz, 1986; Erickson, Lind, Johnson, &O’Barr, 1978; Mullen, Salas, & Driskell, 1989;Ridgeway, 1987). These types of expressivestatus cues may have as strong an informationalimpact as one’s “formal” status. For example,when expressive cues and formal status providecontradictory information (e.g., when someoneis a task group leader by virtue of his or herorganizational position but sounds hesitant andlooks confused), we attend to the informationprovided by both types of characteristics. In thisexample, the group leader may be afforded lessinfluence in the group than when his or herexpressive behavior suggests competence.

Status in Virtual Teams

A number of studies have examined the ef-fects of computer-mediated environments onstatus and authority relations in teams.Dubrovsky et al. (1991) examined the effects ofstatus on groups who communicated eitherface-to-face or through e-mail. Their results in-dicated that in groups communicating via e-mail, the high-status team member’s relativedominance in participation over the low-statusteam member was reduced compared with face-to-face groups, resulting in greater equality ofinteraction. Note that status distinctions werediminished but not eliminated; high-status teammembers participated more than low-statusteam members in both the face-to-face and e-mail groups. Sproull and Kiesler (1986) re-ported that the use of e-mail breaks down statusbarriers, resulting in a less hierarchical patternof communication that is less likely to reflectdifferences in rank or status among group mem-bers. Sproull and Kiesler (1991) concluded,“The high status group member participated


less when the group communicated electroni-cally . . . the low status members spoke more”(p. 62). However, other studies have shown noeffect of technological mediation on reducingstatus differentials. For example, Linville, Lieb-haber, Obermayer, and Fallesen (1991) exam-ined the effect of computer-mediated interac-tion on a simulated military task and found littleevidence that status or leadership was impaired.Weisband et al. (1995) and Silver et al. (1994)found status differences to persist in both face-to-face and computer-mediated groups. Saun-ders, Robey, and Vaverek (1994) reported thatstatus differentials were maintained among hos-pital personnel (physicians and nurses) whencommunicating via computer conferencing.

Although the available research suggests thattechnological mediation may affect the statusstructure of the group, there is a good bit that isnot known at this point. For example, the statuscharacteristics theory describes a processwhereby certain evaluated characteristics thatdifferentiate group members lead to the forma-tion of performance expectations, which thendetermine behavioral inequalities in the group,such as the rate of speaking. Figure 2 illustratesthis status3 expectations3 behavior process.One question that is unclear from the existingliterature is, at what point in this process doestechnological mediation affect status? We pro-pose that there are three primary mechanismsthrough which mediation may impact statusprocesses, differing in the stage at which thisimpact occurs. First, it is clear that some com-puter-mediated communications systems may

block the transmission of status information(status characteristics and status cues) that dif-ferentiates group members. Thus, in somecases, because information that differentiatesother group members is not available or acces-sible, initial status differentials are not estab-lished as they would be in normal face-to-faceinteraction. Thus, computer-mediated systemsmay block the transmission of differentiatinginformation on group members that initiates theoperation of status processes.

A second possibility is that the effects ofstatus cues are dampened in computer-mediatedenvironments. Status cues such as strong eyecontact while speaking, fluid gestures, and awell-moderated voice tone have been shown tostructure interaction in face-to-face settings(Driskell et al., 1993). However, these expres-sive behaviors may not have the same impactwhen expressed by a remote group memberover a computer-mediated network. Indeed,some have argued that gestures and expressionsmay lose their interactional significance whenabstracted from the environment in which theyare produced and mediated via a video image(Doherty-Sneddon et al., 1997; Heath & Luff,1992). Thus, it is possible that status cues maybe dampened in a computer-mediated environ-ment and performance expectations that areformed on the basis of these weakened cuesmay be attenuated.

A third possibility is that the manner in whichexpectations are translated into behavior maydiffer in a computer-mediated environment. Thestatus structure in a group is a normatively

Figure 2. Stages at which technological mediation may impact status processes.

307VIRTUAL TEAMS

supported process (Ridgeway & Diekema,1989). As we noted earlier, group members maydefer to Member A because Member A showscompetence or other evidence of helping thegroup reach its cooperative goal. Behaviors in-consistent with this status structure, as whenMember B repeatedly overrules the higher sta-tus Member A, are deemed to be a violation ofgroup norms and typically result in negativereactions toward the norm violator. In a normalgroup setting, negative reactions to a norm vi-olation may range from a disapproving glanceto a face-to-face confrontation. Given that vir-tual team members are more remote and lessimmediately accessible than members of face-to-face groups, it is possible that the pressure toconform to these norms of status allocation maybe weaker. In this case, even if status differen-tials are salient and performance expectationsare formed on this basis, there may be lesspressure to behave in a manner consistent withthese expectations. It is likely that previous re-sults showing a status equalization effect incomputer-mediated groups may be a combina-tion of (a) blocking of status information, (b)dampening of status cues, and (c) weakening ofthe normatively supported process by whichexpectations are translated into behavior.

There may be some aspects of computer-mediated communications that bolster orstrengthen status effects. For example, in nor-mal group settings, one’s position in the groupstatus structure is established on the basis ofmultiple indicators of status: A person may havethe relative disadvantage of being a lower levelemployee in a group of managers but may alsoprovide evidence of specialized skills and train-ing, professional demeanor, and competency.Research indicates that people combine multi-ple items of differentiating information to formcomposite performance expectations for others(Webster & Driskell, 1983). To the extent thatcomputer-mediated environments restrict thetransmission of differentiating information (es-pecially expressive and other contextual cues),interaction may be more likely to be patternedsimply on a single salient distinction, such asthe primary employee–manager distinction inthis example. In fact, the social identity modelof deindividuation effects suggests that whenindividuating information is scarce, relevantgroup membership cues may become more in-fluential (Postmes, Spears, & Lea, 2002). Thus,

it is possible that major status characteristicsthat are salient, such as occupational position,may have a relatively greater impact in comput-er-mediated environments in which other differ-entiating information is suppressed.

Performance Effects

It is important to note that status processes ingroups may have desirable and undesirable con-sequences. Furthermore, any potential attenua-tion of status differences in computer-mediatedgroups can likewise have desirable as well asundesirable effects. Ideally, status processesprovide structure to the team so that it can useits resources more effectively to accomplish atask. Status processes operate in groups to pro-duce a hierarchical patterning of interaction,such that those with higher status (i.e., thosewho are perceived to be more competent) tendto dominate group discussion. To the extent thatstatus differentials within the group are basedon cultural stereotypes (such as race or gender)this may result in loss of resources to the groupand undesirable barriers to equal participationfor females and ethnic minorities. In this case, aflattening of the status hierarchy may lead tomore open and equal participation and greaterresources for the group to draw upon. On theother hand, status generalization may have pos-itive effects when it operates in general accor-dance with the distribution of ability in thegroup. To the extent that status differences re-flect actual differences in ability, expertise, andcompetence, it is desirable for higher statusgroup members to be more active and influen-tial. In this case, the dampening of status effectscan lead to lower participation by those withgreater competence.

There are ways in which status can affectperformance other than by determining the ex-tent that team members participate or contributeto the task. Under conditions in which statusdistinctions are blurred, when there is uncer-tainty about status, or when team members holdconflicting expectations, status incongruencemay arise. One type of status incongruence mayoccur when Team Member A possesses highstatus (perhaps having more experience or hav-ing specific content knowledge) and is ignoredor afforded fewer opportunities to contribute tothe task. Status incongruence can lead to dissat-isfaction and lower productivity (Nixon, 1979).


Furthermore, for technology-mediated teams, itmay be more difficult to correct normative vio-lations of the status structure. In normal inter-action, when there is a discrepancy between ateam member’s status, behavior, or rewards(e.g., if the low-ability team member dominatesconversation), team members tend to act to cor-rect these discrepancies. In situations in whichstatus distinctions are unclear, these normativeviolations are less likely to be perceived oraddressed.

Moderators

Type of task. In general, a hierarchical sta-tus structure is potentially more advantageous,and more likely to be established, for tasks thatare ambiguous, uncertain, and complex versustasks that are highly structured and routine(Nixon, 1979). Thus, in general, technologicalmediation should impact status processes lessfor more routine logical/precision tasks than forintellectual/analytic tasks.

Examining the moderating effects of type oftask more precisely requires that we also con-sider the temporal context of the team and theconsequences of changes in the status structure.One point that we have touched on previously isthat status processes in teams are not passive butare normatively supported by team members,both high status and low status. Given that thestatus structure is a means to organize interac-tion within the team to achieve effective taskperformance, changes to this structure are likelyto be resisted, especially if the status structuresupports the effective accomplishment of thetask. In other words, real groups that have ahistory of interaction are likely to resist changesto an established status structure.

Straus and McGrath (1994) noted that ideageneration tasks require very little coordinationor consensus to accomplish. And, in fact, status-differentiated groups tend to do more poorly onidea generation tasks than status-undifferenti-ated groups (Silver et al., 1994). Thus, if tech-nological mediation suppresses status effects onidea generation tasks, there will probably belittle effort by team members to restore a struc-ture that does not support successful task per-formance in the first place. However, judgmenttasks are tasks for which team members mustseek consensus on a preferred alternative. Struc-turing interaction so that those who have better

ideas contribute more to team deliberationsshould result in better performance. Technolog-ical mediation may suppress status effects onjudgment tasks, but it may be temporary, asteam members seek to restore an effective statusstructure. Status differentiation may be some-what less beneficial for intellective tasks thathave a correct answer, given that as long as oneperson has the answer, consensus is not re-quired. However, to the extent that status pro-cesses operate to place those who are morelikely to have a correct answer in an advantagedposition in the group, an appropriate statusstructure should support effective performance.Again, for this type of task, technological me-diation may initially suppress status effects, butthis effect may be temporary, as team membersattempt to restore an effective status structure.

Type of computer-mediated environment.Some computer-mediated communications sys-tems restrict the transmittal of visible as well asexpressive indicators of group members’ status.For example, group members who are separatedover a computer network with no visual capa-bilities lose access to visual cues of others’status that are present in face-to-face interac-tion, such as eye contact, style of dress, select-ing the head of the table, or perhaps even therace, age, and gender of other group members.The relative anonymity of other group members(compared with face-to-face groups) may resultin less hierarchically structured and more equalparticipation across group members.

Computer-mediated environments in whichgroup members communicate primarily via text,with no visual or verbal communication, mayalso limit the transmittal of expressive statuscues such as rate of speech, verbal fluency, andloudness and tone of speech. Furthermore,many computer-mediated systems impose stan-dardized forms of communication that furtherlimit the expression of status. In fact, moreformalized means of communication may allowteam members to speak uninterrupted, whichmay in itself lead to changes in patterns of teammember interaction. By masking cues to statusor by imposing more standardized forms ofcommunications, we would expect interactionin these types of mediated environments to beless hierarchical and more equal.

Temporal context. Teams may be com-posed of members who are initially differenti-ated by status distinctions (i.e., teams of mixed

309VIRTUAL TEAMS

organizational status, gender, or ability) or theymay be composed of status equals, undifferen-tiated with regard to observable status differ-ences. In status-unequal teams, a status structuredevelops very quickly at the onset of interac-tion, on the basis of the observable status char-acteristics that differentiate team members. Instatus-equal teams, this structure develops moreslowly on the basis of the contributions thatteam members make during team interaction.There are several implications of this distinc-tion. First, in teams that interact over a period oftime, the strong initial impact of status charac-teristics declines as interaction becomes pat-terned more on the quality of the contributionsthat team members make. Thus, the effect oftechnological mediation on blocking observableindicators of status may become less relevant toteam interaction over time. Second, the impactof status characteristics in structuring interac-tion is relatively more salient when the tasksituation is ambiguous and team members donot have much information regarding the per-formance abilities of each other. This is morelikely to be the case for ad hoc, newly formedteams than for teams that are experienced andhave interacted over time.

Counternormative Behavior

Reviews of computer-mediated interactioninvariably address differences in counternorma-tive or uninhibited behavior in face-to-face andcomputer-mediated groups. In fact, this is per-haps one of the most widely cited yet unimpres-sively supported phenomena in this literature.This ambiguity leads to conflicting conclusionsdrawn by reviewers that uninhibited communi-cation is either “a common, if not universal,feature of computer-based conferences” (Selfe& Meyer, 1991, p. 170) or “a comparativelyrare occurrence in [computer-mediated commu-nications]” (Lea, O’Shea, Fung, & Spears,1992, p. 108).

It is important to note that the term counter-normative refers to behavior that deviates fromthe norm, and it can refer to behavior that ismore positive than normal or behavior that ismore negative than normal (Blanton & Christie,2003). Within the computer-mediated commu-nications literature, counternormative or unin-hibited behavior is often discussed in terms ofhostile or negative communication (i.e., flam-

ing), but this association may not be consistentand can be misleading. In fact, authors of onestudy that is often cited as providing supportingevidence for negative communications in com-puter-mediated communications (Kiesler et al.,1985) included in their measure of uninhibitedbehavior both impolite statements and swearingas well as exclamations (e.g., “Hooray!”) andexpressions of high positive regard. Thus, cau-tion is advised in that some may treat counter-normative or uninhibited behavior as both de-sirable and undesirable behaviors that deviatefrom the norm, whereas others may discusscounternormative or uninhibited behaviors asreferring specifically to negative behavior.

Counternormative Behavior in VirtualTeams

Certainly the primary concern prevalent inthe literature is with counternormative behavioras negative behavior, such as verbal aggression,expressions of hostility, or negative socioemo-tional content. Dubrovsky et al. (1991) arguedthat the lack of contextual means of communi-cating in computer-mediated environmentsforces interactants to be more forceful and lessinhibited. Further, Kiesler et al. (1985) notedthat computer-mediated systems transmit socialinformation poorly and that computer-mediatedinteraction is likely to be characterized by lessattention to others, less social feedback, anddepersonalized communication. Sproull andKiesler (1991) concluded, “People interactingon a computer are isolated from social cues. . . .This feeling of privacy makes them feel lessinhibited with others” (p. 48). However, theexisting research on the effects of computer-mediated communications on counternormativebehavior is much more equivocal. Althoughearly research studies revealed more negativesocioemotional content in computer-mediatedgroups than in face-to-face groups, more recentreports have called this relationship intoquestion.

Siegel et al. (1986) argued that the deindi-viduation stemming from the relative absenceof social cues and social feedback in computer-mediated groups would lead to greater uninhib-ited behavior. They compared three-persongroups who either met face-to-face or werephysically separated and communicated via acomputer-mediated text messaging program. In


the computer-mediated groups, they observedsignificantly greater instances of uninhibited be-havior (swearing, insults, and name calling)than in the face-to-face groups. Dubrovsky et al.(1991) found greater swearing, name calling,and threats in groups communicating via e-mailthan in face-to-face groups. Sproull and Kiesler(1986) reported more flaming in groups com-municating via e-mail, on the basis of researchsubjects’ self-reports of how much flaming theyobserved in e-mail messages versus how muchthey observed in everyday conversations.

However, other studies have found that thereis little difference in uninhibited negative be-havior between computer-mediated and face-to-face groups. Hiltz, Johnson, and Turoff (1986)compared communication in computer-medi-ated and face-to-face groups using the BalesInteraction Process Analysis categories. Al-though they found evidence of fewer socioemo-tional statements in computer-mediated groups,they also found other results that were moreequivocal. Computer-mediated groups mademore statements relating to showing tensionthan face-to-face groups on one experimentaltask; however, for the other task, this differencewas reversed. There was no difference in state-ments showing antagonism between the twogroups. Hiltz, Turoff, and Johnson (1989) foundlittle evidence of uninhibited behavior (e.g., in-sults, profanity) in either computer-mediated orface-to-face groups composed of corporatemanagers. Walther and Burgoon (1992) re-ported no difference in communications show-ing tension or composure between face-to-faceand computer-mediated groups. Walther, An-derson, and Park (1994), in a meta-analysis ofnegative communication in computer-mediatedinteraction, concluded that the overall propor-tion of communications considered negativewas quite small and that the difference betweennegative communication in computer-mediatedand face-to-face groups was minute (d � .017).

The most reasonable conclusion that can bedrawn from this research is that negative coun-ternormative behavior is not inevitable in com-puter-mediated groups but may occur undersome situations. There are several explanationsthat can account for the potential for greatercounternormative behavior in computer-medi-ated groups, including the effect of the impov-erished computer-mediated communicationsenvironment, anonymity and deindividuation,

lack of accountability, frustration, and confu-sion. We consider each of these potential expla-nations in the following.

One explanation for uninhibited behavior incomputer-mediated communication is directlyrelated to the impoverishment of the mediumitself. Normal communication involves bothverbal and nonverbal channels, and researchershave drawn a distinction between verbal (con-tent) and nonverbal (expressive) communica-tion (Ambady & Rosenthal, 1992; DePaulo,Rosenthal, Eisenstat, Rogers, & Finkelstein,1976). Moreover, when the content of a mes-sage and the expressive behavior accompanyingthe message are discrepant—for instance, whena person states “That’s horrible!” but with awarm voice tone, a smile, and open, relaxedgestures—we are likely to interpret the messageas a friendly jibe rather than an insult or a threat.Computer-mediated systems that do not conveyexpressive behavior may result in more misin-terpretations and mistakes in interpreting mes-sage intent. Computer-mediated systems that donot provide an audio channel block paralinguis-tic information, those that do not provide avideo channel block the transmission of visualexpressive behavior, and even those that pro-vide “head and shoulders” video may distortexpressive behavior when it is abstracted fromthe context in which it is produced and pre-sented via a restricted video image (Heath &Luff, 1992). Furthermore, we may find greaterevidence of overt disagreements or more stri-dent statements in computer-mediated groupssimply because more subtle forms of disagree-ment (such as shaking one’s head) are renderedless useful by the communications medium. Inthis case, one form of disagreement may serveas a substitute in computer-mediated communi-cations for another form of disagreement that issuppressed.

A number of researchers have argued that therelative anonymity and loss of social cues incomputer-mediated interaction lead to greaterdepersonalization or deindividuation. The tradi-tional perspective holds that anonymity (such asinduced by wearing a mask or being in a largecrowd) leads to a loss of self-awareness, and inthis state of lowered self-attentiveness, peoplefail to regulate their behavior with the prevail-ing standards of performance (Carver &Scheier, 1981; Mullen & Baumeister, 1987).Thus, those “lost in a crowd” are more likely to

311VIRTUAL TEAMS

become deindividuated and commit atrocities(Mullen, 1987), and those depersonalized in avirtual team environment are less likely to beconcerned with prevailing team standards andmore likely to exhibit counternormative behav-ior. The social identity model of deindividua-tion (SIDE) provides an alternative and compet-ing explanation (Postmes & Spears, 1998;Spears & Lea, 1992). According to the SIDEperspective, deindividuation leads to greater,not less, conformity under certain conditions.More specifically, if group standards are salient(i.e., group membership is emphasized), thendeindividuation will lead to greater adherenceto group norms. Further, if the reference groupsupports the expression of uninhibited behavior(as SIDE proponents imply that many onlinegroups do), then this increased responsivity togroup norms will lead to greater uninhibitedbehavior.

A related argument is that deindividuated vir-tual team members, isolated from the physicalpresence of other team members, may be lessaccountable for their actions. Normative bondsare consensual—we adhere to norms because itis the right thing to do. But we also conform tonorms because of the coercive power of others.For example, I may choose not to insult some-one sitting across the table from me because heor she may take umbrage and respond. Virtualteam members are under less coercive control,stemming from the lack of immediate presenceof other team members, and are less account-able for their actions. Furthermore, in ad hocone-shot or single-meeting virtual teams, notonly do we not have someone who can figura-tively reach over and thump us on the ears, butthere is no potential for future interaction inwhich transgressions can be righted. Thus, theloss of the coercive power of other physicallypresent team members may lead to greater un-inhibited behavior in virtual teams.

It is reasonable to expect that there is greater“mechanical friction” (i.e., frustration with newor awkward procedures and systems) involvedin using computer-mediated systems, whichmay lead to greater annoyance and emotionality(Poole, Holmes, Watson, & DeSanctis, 1993). Itis further likely that these procedural difficultiesand resulting frustrations would subside overtime. However, because of the short-term, tem-porary nature of many research studies, there is

little evidence available regarding whether orhow this type of adaptation occurs.

Finally, there may be greater counternorma-tive behavior in computer-mediated teams be-cause interaction is more confusing than in face-to-face communication (Thompson & Coovert,2003). Thompson and Coovert claim that virtualteams have greater difficulty establishing mu-tual or shared knowledge because they lack thedirect knowledge gained from firsthand, imme-diate experience with other team members.They found that in comparison with teams thatworked face-to-face, teams that interacted viatext-based computer conferencing reportedgreater confusion and less understanding ofteam discussions. These results suggest thattechnological mediation may lead to greatercounternormative behavior, especially uninten-tional norm violations, because virtual teammembers may find it more difficult to under-stand just what the rules are.

Performance Effects

There is little direct research on the effects ofcounternormative behavior on performance incomputer-mediated groups. However, assumingthat team norms support performance, adher-ence to team norms should generally result ingreater productivity (Postmes, Spears, & Cihan-gir, 2001). Moreover, negative or hostile com-munications may lead to greater task-irrelevantcommunications and may disrupt team interac-tion and interpersonal relations among teammembers.

Moderators

Type of task. Many of the studies that haveexamined counternormative behavior in com-puter-mediated groups have used judgment orchoice dilemma tasks, including studies thathave reported greater counternormative behav-ior in computer-mediated groups versus face-to-face groups (Dubrovsky et al., 1991; Siegel etal., 1986) as well as those reporting no differ-ence (Hiltz et al., 1989). It is likely that therelative anonymity, lack of accountability, frus-tration, and confusion that may stem from in-teraction in computer-mediated environmentswould have similar effects on a broad range oftasks. However, the tendency for these condi-tions to result in negative socioemotional out-


bursts may be greater for tasks in which socio-emotional relations are more salient, such associal or persuasive tasks, than in more routin-ized mechanical/technical or logical/precisiontasks.

One further characteristic of the task is rele-vant to the effect of deindividuation on coun-ternormative behavior. The SIDE model arguesthat deindividuation leads to greater identifica-tion with the group and stronger normativecommitment when social identity is salient(Postmes, Spears, & Lea, 1998). That is, fortasks in which group members perceive strongmembership to the group, deindividuation mayreinforce conformity to group norms (i.e., teammembers are less self-aware and more group-aware). When group members do not identifystrongly with the group, deindividuation mayweaken social influence and increase counter-normative behaviors.

Type of computer-mediated environment.Research on counternormative behavior has al-most exclusively compared face-to-face groupswith groups communicating via e-mail or text-based computer conferencing systems (cf.Dubrovsky et al., 1991; Hiltz et al., 1989;Kiesler et al., 1985; Siegel et al., 1986; Sproull& Kiesler, 1986; Walther & Burgoon, 1992).We would expect deindividuation or loss ofself-awareness to be more prevalent the morethat the communications medium restricts thetransmission of individual cues. Thus, wewould expect deindividuation to be greatest fortext-based communications systems, relativelyless likely to occur for audio-based systems, andeven less likely to occur for audio–video com-munications. Frustration with the communica-tions medium and confusion with team deliber-ations are also likely to be greater for morerestrictive communications environments. Onthe other hand, the loss of accountability stemsprimarily from the physical separation of teammembers, and this is likely to be salient in anycommunications environment in which teammembers are separated from one another.

Temporal context. Sproull and Kiesler(1986) and Hiltz et al. (1989) examined coun-ternormative behavior in intact work teams.Sproull and Kiesler reported greater counternor-mative behavior in e-mail communications, andHiltz et al. reported no differences betweenteams communicating via text conferencing andthose interacting face-to-face. The vast majority

of other studies of counternormative behaviorexamined newly formed groups whose time to-gether lasted approximately 90 min or less.Walther (1997) has argued that short-term com-puter-mediated groups are more task oriented,impersonal, and hostile than long-term groups.He states that groups that anticipate future in-teraction increase social information seeking,show more positive affect, and enact more re-lationally positive communication than ad hocone-shot groups. Indeed, Walther et al. (1994)reported that differences in the extent of nega-tive communications between face-to-face andcomputer-mediated groups were reduced wheninteraction was longer term versus timerestricted.

Communication

Communication is a multimodal process thatincludes both verbal and nonverbal compo-nents. As people speak, they also gesture toelaborate speech, they change body position andposture, and they vary facial expression andgaze. In fact, one primary characteristic of com-munication is that the literal meaning of speechunderspecifies the speaker’s intended meaning.To infer the speaker’s intended meaning re-quires that the listener supplement speech con-tent with contextual information that is not rep-resented in speech. This contextual informationmay include a speaker’s gaze or eye contact,facial expressions, gestures, posture and bodymovements, as well as paralinguistic cues suchas intonation and loudness of speech. For ex-ample, expressive behaviors such as gesturescan serve multiple functions: they may supple-ment and elaborate speech content, accent orpunctuate speech, substitute for speech, clarifyambiguity, regulate the timing and sequence ofcommunication, and convey emotion (Driskell& Radtke, in press). All of these functions canserve to more fully convey information to thelistener. The promise of computer-mediatedcommunications systems is to produce an envi-ronment that captures the richness of face-to-face communication.

Communication in Virtual Teams

Communications systems that do not supportthe transmission of contextual information areviewed as impoverished and less informative

313VIRTUAL TEAMS

(Fussell & Benimoff, 1995). One perspective oncommunication in computer-mediated groupscontends that transfer of information is re-stricted in a computer-mediated environmentowing to a reduction in the amount and type ofcues available to interactants. More specifically,the types of information reduced in a computer-mediated environment are largely cues of con-text—verbal and nonverbal behaviors that playa large part in defining the message that iscommunicated. In brief, the available body ofresearch suggests that (a) communication in-volves the transfer of information through mul-tiple channels, (b) computer-mediated interac-tion may filter out certain communicative cues(primarily visual and contextual) found in face-to-face interaction, and (c) this restriction incontextual cues may lead to significant changesin group communication (see Culnan &Markus, 1987).

One fundamental problem present in comput-er-mediated communications is the failure toestablish and maintain mutual knowledge(Cramton, 2001; Thompson & Coovert, 2003)or common ground (Clark & Brennan, 1991).Mutual knowledge refers to knowledge thatteam members share and know they share(Krauss & Fussell, 1990). Accordingly, a lackof mutual knowledge would be represented byteam members who hold differing informationand do not realize they do so. Whittaker andO’Conaill (1997) noted that group membersmust establish common ground in regard to thecontent of communication (mutual knowledgeregarding the content of conversation) and alsothe process of communication (mutual knowl-edge regarding who will speak, who will listen,and how transitions are made).

Mutual knowledge is established throughseveral mechanisms, including direct knowl-edge gained from shared experiences and first-hand observations of other team members andthrough the dynamics of the interaction processitself (Krauss & Fussell, 1990). Virtual teammembers are less likely to gain direct knowl-edge of other team members when they arelocated remotely from one another. Further-more, establishing mutual knowledge throughinteraction is made more difficult in a computer-mediated environment because information ex-change is less complete, is slower, and requiresmore effort. Cramton (2001) has noted that onereason communication is more difficult in com-

puter-mediated environments is the greater ef-fort required to convey nuances of speech with-out the use of expressive and paralinguisticcues.

For example, one of the most common waysof establishing whether a statement is under-stood is through acknowledgment. Acknowl-edgments are attempts to provide positive evi-dence of understanding. Acknowledgments areoften verbal but in many cases constitute whatare termed back-channel responses, includingsignals or gestures such as head nods, shrugs, orsmiles. Another basic form of acknowledgmentis sustained attention. In normal communica-tion, individuals typically monitor moment-to-moment what other interactants are doing. Wedetect cues of sustained attention from others,often through eye gaze, that indicate their un-derstanding. On the other hand, when someonelooks puzzled, breaks eye contact, or raises abrow, this provides critical information that un-derstanding has not been established. In a com-puter-mediated environment, because this typeof contextual communication is often lacking, itmay be more difficult to ground communicationand establish that a statement is understood;additionally, the costs of grounding communi-cation will be greater, as team members mustexpend more effort to ensure that information isunderstood by others.

A related problem in attempting to achievemutual knowledge in virtual teams is that re-stricted feedback may make it more difficult totransfer information and to identify and correcterrors in the transfer of information as theyoccur. Cooperative behavior in a team setting isan iterative cycle of individual performance,feedback from others, adjustment, and action.Ashford and Tsui (1991) note that in manyreal-world task settings, formal feedback mech-anisms regulate behavior only loosely. That is,the feedback that is received directly from ex-ternal sources may in some cases be of lessrelevance than that which is abstracted activelyfrom the environment by the individual. Draw-ing feedback from other team members requiresan active process of seeking and evaluatingfeedback, monitoring performance, and initiat-ing and revising corrective actions.

Models of self-regulation emphasize the im-portance of feedback that individuals generatefor themselves, rather than information that ispassively provided by others (Butler & Winne,


1995). Self-regulation refers to the active pro-cess whereby individuals set performance stan-dards, seek and evaluate feedback, detect dis-crepancies in their performance, and takeactions to reduce those discrepancies. Self-regulation is, in part, a social process. As teammembers monitor their ongoing task perfor-mance, they seek feedback from sources such asother team members. This feedback allows thelearner to establish the validity of his or her ownperceptions regarding current task performance,verify performance standards, and gather infor-mation regarding potential strategies to correctperformance.

To the extent that team members are lessimmediate and accessible, accessing peer feed-back is more difficult. Often, individuals mayseek feedback from a team member’s glance,nod, or frown—nuanced behaviors that are noteasily transmitted in computer-mediated envi-ronments. Not only is this type of feedbackmore difficult to perceive in a computer-medi-ated environment, it may be less likely to beprovided by a team member who is sitting re-motely at a computer screen rather than in thedirect presence of others. A second concern isthat the nature or valence of feedback providedby members of virtual teams may differ fromthat of face-to-face groups. Ashford and Tsui(1991) have noted that there is a general ten-dency for people to give each other positivefeedback spontaneously, while withholdingnegative feedback. However, research suggeststhat under certain circumstances, members ofcomputer-mediated groups may offer more neg-ative or extreme evaluations of others thanmembers of face-to-face groups (Walther,1997). Thus, there may be a tendency in com-puter-mediated environments for team membersto be exposed to more negative feedback, dis-approval, or censure. Finally, virtual team envi-ronments may affect how feedback is evaluated.Individuals do not simply absorb feedback butactively evaluate the extent to which feedback isan accurate reflection of their performance.Feedback that is seen as inaccurate is discount-ed; feedback that is seen as accurate is morelikely to be accepted. Some research suggeststhat feedback is more likely to be accepted fromsources that are psychologically closer to thesource than from those more distant (Ilgen,Fisher, & Taylor, 1979). Thus, we may acceptfeedback less readily from others in a virtual

team environment who are seen as more remoteor removed from the immediate social setting.

Performance Effects

Cramton (2001) found several consequencesof the failure to establish mutual knowledge incomputer-mediated teams. Team members haddifficulty in communicating and obtaining con-textual information regarding the specific con-ditions and constraints under which remoteteammates worked. Information was distributedunevenly among team members, and widelyvarying impressions were formed regarding thetask. Team members had difficulty in commu-nicating and in understanding what was mostimportant in the information transmitted. Fi-nally, team members reported problems in re-ceiving timely feedback from other team mem-bers and in interpreting the meaning of otherteammates’ silence or lack of response. Inter-estingly, some team members interpreted oth-ers’ failure to communicate as an unwillingnessto work, reflecting the tendency to make nega-tive personal attributions when situational infor-mation is absent. Thompson and Coovert (2003)found that a lack of mutual knowledge in com-puter-mediated teams led to greater confusionand inaccuracies in recording team decisions.Finally, some research has shown that the sim-ilarity of shared mental models among teammembers predicts the quality of team processesand performance (Mathieu, Heffner, Goodwin,Salas, & Cannon-Bowers, 2000).

Moderators

Type of task. Cramton (2001) noted that theimpact of computer-mediated communicationson mutual knowledge is likely to be greater fortasks in which individual team members hold agreat amount of unique information and inwhich contextual information between remotesites differs. Furthermore, this problem is exac-erbated by high requirements for complexity,workload, and interdependence. Others have ar-gued that technological mediation, especiallythe loss of visual cues, may have a greaterimpact on communication effectiveness fortasks that have a greater interpersonal require-ment, such as social, manipulative/persuasive,or negotiation tasks (Sellen, 1995; Short, 1974;Williams, 1977).

315VIRTUAL TEAMS

Type of computer-mediated environment.The advantages of audio over text-based com-munications are substantial, leading Whittakerand O’Conaill (1997) to conclude that speech isthe critical medium for interpersonal communi-cations. Thus, the negative impact of technolog-ical mediation on communication should be re-duced when audio is available. However, thevalue of adding video to audio communicationsis less apparent. Rudman, Hertz, Marshall, andDykstra-Erickson (1997) examined the role ofthe video channel in supporting computer-me-diated communication. They proposed that onebenefit of adding video to audio and data chan-nels in a computer-mediated environment is thatit allows teams to more effectively regulate theflow of interaction and to communicate emo-tions and attitudes. Rudman et al. found thatteam members using audio-only communica-tions had difficulty knowing when others werepaying attention and could not tell when theywere understood because they could not seeothers’ nonverbal reactions. Without visualfeedback, team members delayed asking othersfor information out of fear of interrupting theirwork; further, they had problems knowing whenothers needed help, because they could not seeothers’ level of frustration or confusion. In gen-eral, Rudman et al. observed that without visualinput, team members had problems evaluatingothers’ level of attention and concentration, de-termining how positive or negative others werefeeling, determining whether others neededhelp, and knowing when to interrupt.

Nevertheless, the bulk of evidence on thevalue of adding video capabilities to computer-mediated systems is not supportive. In one ofthe earliest such studies, Chapanis, Ochsman,Parrish, and Weeks (1972) found that addingvideo to group communications did not enhancethe efficiency of group problem solving or resultin higher quality outcomes. Other studies havesimilarly shown little advantage of video overaudio-only communication in enhancing teamperformance (Anderson et al., 1997; Boyle,Anderson, & Newlands, 1994; Doherty-Sned-don et al., 1997; Sellen, 1995; Williams, 1977).In a recent review of this literature, Whittakerand O’Conaill (1997) concluded, “Laboratorystudies to demonstrate the benefits of adding avisual communication modality to voice have ingeneral shown few objective improvements”(p. 24).

Why does video apparently do so little toenhance communication in computer-mediatedsystems? First, Doherty-Sneddon et al. (1997)note that a primary focus on task outcome maybe misleading and that a more useful evaluationof communication requires consideration ofboth outcome and communicative process. Bylooking at outcome alone, their results showedlittle difference between face-to-face, audio,and audio–video communications. However,they also found that the structure of communi-cation varied across these conditions and thatthe visual channel was useful to speakers, forexample, in facilitating conversational flow.

However, one further factor underlies the ap-parent ineffectiveness of video in computer-mediated communications. Research suggeststhat people are able to make fairly accuratejudgments of others’ personality on the basis ofminimal interactions (Ambady, Hallahan, &Rosenthal, 1995) and that people are able torecognize emotions as well as deception fromdemeanor (Frank & Ekman, 1997). However,studies also show that the assessment of otherstates, moods, or motivations from nonverbalbehavior can be more difficult. For example,Jecker, Maccoby, Breitrose, and Rose (1964)found that assessing comprehension from non-verbal behavior was difficult and that teacherswere able to accurately assess student compre-hension about 30% of the time from video re-cordings. Furthermore, research that has at-tempted to improve the ability to decode non-verbal behavior suggests that training mayincrease decoding accuracy (Costanzo, 1992).However, most studies that contrast audio-onlywith audio–video communications simply makeavailable a video channel (typically a facialimage on a computer screen window) for teammembers to use. One reason that providing avideo channel to computer-mediated systemshas shown few benefits may be that individualsmake poor use of the social cues that are avail-able without specialized training to ensure ac-curacy in decoding nonverbal behavior.

Temporal context. It is likely that commu-nication difficulties, and especially problems inestablishing mutual knowledge, will be less forpreexisting teams that have a history of interac-tion, team members who are more experiencedusing the communications media, and more ma-ture teams that can function with less rich in-formation exchanges (McGrath & Hollings-


head, 1994). Moreover, the value of video toenhancing communication in virtual teams maybe greater as team members gain familiaritywith one another. In ad hoc or newly formedgroups, expressions and gestures are idiosyn-cratic. In other words, when we have notworked with someone before, that person’s ges-tures, posture, and expressions may hold littlemeaning for us. It is only over time that theseidiosyncratic gestures gain meaning (e.g., welearn that when Alice looks away, she is think-ing, but when Bob looks away, he is bored).Therefore, the use of video in computer-medi-ated systems may be of less value for newlyformed teams, because the social informationprovided may hold less meaning.

Discussion

Some see a future in which virtual teammembers interact seamlessly over advancedcommunications systems. Others are more cau-tious and argue that virtual teams differ in sig-nificant ways from teams that work face-to-face. For each of the topics we have discussed,we have examined research that has comparedface-to-face teams and technology-mediated orvirtual teams. The existing evidence leads us toconclude that overall, distance seems to mat-ter—that being mediated by technology canhave a significant impact on how teams per-form. These performance effects may stem fromchanges in cohesiveness, status structure, coun-ternormative behavior, and communication.

Technological mediation can have a negativeeffect on cohesiveness, affecting interpersonalbonds (interpersonal attraction), normativebonds (group pride), and instrumental bonds(task commitment) among team members.Team performance is likely to be affected espe-cially to the extent that task commitment islowered.

Technological mediation may affect the sta-tus structure of the team by blocking the trans-mission of status information, dampening theeffects of status cues, or weakening the norma-tive process that supports status-based behavior.To the extent that more competent persons oc-cupy higher positions in the status hierarchy ofthe team, performance can be degraded if statusdifferentials are reduced.

Although counternormative behavior is notnecessarily a feature of computer-mediated

teams, technological mediation can lead to agreater prevalence of behaviors, both desirableand undesirable, that deviate from the norm.Negative counternormative behavior may stemfrom a number of causes, including deindividu-ation, frustration, lack of accountability, andconfusion, and can disrupt the smooth interper-sonal relations required for effective teamperformance.

Technological mediation can make it moredifficult to communicate information to othersand to interpret the communications of others.The relative loss of contextual information incomputer-mediated communications can resultin greater difficulty in establishing mutualknowledge and can lead to greater confusionamong team members and inaccuracies inperformance.

However, the effects of technological medi-ation on these team processes are moderated bythe type of computer-mediated environment, thetype of task, and the temporal context of theteam. We described the effects of these threemoderators as overarching, and the evidencereviewed suggests they are indeed important.For example, there are few generalizations thatcan be made about virtual teams without con-sidering the nature of the communications en-vironment. Overall, less rich communicationsmedia serve to exacerbate the effects of techno-logical mediation. Thus, to the extent that tech-nological mediation impairs interaction, it im-pairs interaction more so for text-only commu-nications than for text–audio and for audio–video communications. However, there alsoseems to be something about the physical pres-ence or immediacy of another that differentiatesface-to-face interaction from even high-qualityaudio–video. Sellen (1995) concluded,

There appears to be something critically differentabout sharing the same physical space that needs to beexamined more carefully. What aspects of interactionare fundamentally altered when people no longer sharethe same physical space? Why does the presence of avideo channel fail to compensate? (p. 440)

At present, there is no satisfactory explicationof this issue.

Similarly, few firm statements can be maderegarding team interaction in virtual environ-ments without consideration of the type of taskthat the team is performing. This variable issomewhat more difficult to examine, as thereare any number of dimensions on which tasks

317VIRTUAL TEAMS

may be classified. We have attempted to focuson the behavioral requirements of the task; thedistinction that seems to be most useful is thatbetween tasks that have high interpersonal/so-cial requirements (i.e., social and persuasivetasks) and tasks that have greater instrumentalrequirements (i.e., technical or logical tasks).The comparison is not a simple one, but tech-nological mediation may be more disruptive fortasks that have high demands for either social orinstrumental interdependence than for tasks thatcan be accomplished with less social or instru-mental interaction.

Finally, McGrath (1990) has noted, “Groupsdevelop and exist in a temporal context” (p. 23),an admonition that is all but ignored in mostresearch on virtual teams. Overall, time andexperience tend to brace teams against the del-eterious effects of technological mediation.Thus, some of the negative effects of techno-logical mediation observed in ad hoc, short-term teams may not be evident in intact orexperienced teams. Further, some effects oftechnological mediation that are observed in theshort term may be temporary and subside overtime.

To draw firm conclusions from a model ofteam performance such as that presented in Fig-ure 1, a research literature is needed that clearlyand systematically varies important team andtask characteristics. However, as Hollingsheadand McGrath (1995) noted, this may be theideal, but the reality is that the body of literatureon computer-mediated teams virtually ignoresthe operation of key team and task variables.Accordingly, in many cases, our analyses arespeculative rather than conclusive, and moreresearch is needed to further elucidate the spe-cific conditions under which technological me-diation impacts team interaction. However, as isoften the case, there is value in noting what wedo not know at this point. Specifically, furtherresearch is needed to examine the effects oftechnological mediation on other processes,other outcomes, and other moderators.

Other Processes

We have discussed in some detail the effectsof technological mediation on cohesiveness,status processes, counternormative behavior,and communication. There are a number ofother team processes of interest such as leader-

ship, decision making, cooperation, and confor-mity that have not been fully examined in thisliterature. For example, one process that may beparticularly relevant given the nature of thecomputer-mediated environment is social loaf-ing. Social loafing refers to the tendency forindividuals to become less motivated to exertfull effort when working on a collective grouptask than when working independently (Latane,Williams, & Harkins, 1979). In the early part ofthis century, Ringelmann (1913) found thatwhen research participants performed a rope-pulling task as a group, less force was exertedthan would be expected by combining the forcesof each individual. Furthermore, as the size ofthe group increased, the difference between ex-pected and actual productivity became greater.Subsequent research has shown that social loaf-ing is most likely to occur when the individualtask performer is immersed in a larger group,when the task contributions of others cannot beeasily identified, when the work group is com-posed of strangers or lacks cohesiveness, andwhen group members are less likely to perceivethe direct contributions of their own efforts tothe group outcome (Karau & Williams, 1993;Mullen & Baumeister, 1987). These conditionsreflect those that occur in a virtual team envi-ronment. Members of virtual teams work to-gether but individual outputs are often difficultto distinguish or monitor. Members of virtualteams work with others who are dispersed overa wide geographic area and with whom theyhave little direct contact, and they may be one“node” of a network that may contain a largegroup of members. Thus, a virtual team mem-ber, who may see five other group membersrepresented on a computer screen (conferencingsoftware typically places an image of each othermember in a separate window on the screen),may become lost in the crowd and devote lesseffort to the task. Although a virtual team set-ting may be a prime environment for socialloafing to occur, very little research has exam-ined this topic.

Other Outcomes

We limited our interest to the effects of tech-nological mediation on team performance,broadly defined. There are clearly more specificindicators of team performance that should beexamined, including performance accuracy,


speed, and quality of solutions, among othermeasures. There are also important outcomevariables other than performance, includingteam member satisfaction, the longevity or via-bility of the team, and organizational outcomes.

Other Moderators

There are a number of other potential mod-erators that warrant further scrutiny, includingthe size of the team. Steiner (1972) noted thatteam size may impact performance in severalways. As a group gets larger, the diversity of itsmembers increases, thus increasing the varietyof viewpoints and potentially increasing thenumber of problem solutions available to theteam. However, larger teams do not alwaysperform better than smaller ones. Increasingteam size makes coordination requirementsmore difficult, and as groups increase in sizeand complexity, more group structure (role dif-ferentiation, etc.) is required to coordinategroup activity. It is well established thatwhereas total group performance may increaseas the size of the group increases, group mem-ber performance (i.e., performance per person)and satisfaction decrease as a function of groupsize (e.g., Mullen, 1991; Mullen & Baumeister,1987).

Separating team members from one another,as in a virtual team environment, may serve to“reduce” the size of the group in a perceptualsense. In other words, we may be able to alle-viate the negative consequences of being part ofa larger team by having team members worktogether but remotely over a communicationsnetwork. Thus, it is possible that technologicalmediation may increase team member satisfac-tion in large teams by making a large group feelsmaller, yet it may decrease satisfaction in smallteams by disrupting the closer interpersonalbonds forged by smaller groups. However, therehas been little research conducted to examinepotential social advantages of technologicalmediation.

In summary, some differences that have beenobserved between face-to-face and virtualteams are likely to be temporary. For example,the mechanical friction and user frustrationsinherent in using a relatively crude audio–visualinterface will dissipate as advances in hardwareand software are achieved. Some differencesthat have been noted may be illusory: The ob-

servation that members of virtual teams ver-bally disagree more than members of face-to-face teams may simply reflect the fact that vir-tual team members are less able to disagreenonverbally. However, some differences aremore fundamental. We believe that one suchdifference is the distinction between the imme-diacy of a physically present team member andthe remoteness of a virtual team member. How-ever, a social psychological analysis of the con-cept of presence is lacking.

One feature that characterizes much of theresearch on virtual teams is an emphasis ondeveloping advanced technological environ-ments for virtual team interaction. One disad-vantage of this technology-focused approach isthat key social and psychological variables maybe overlooked or ignored, as Hollingshead andMcGrath (1995) observed. If ever there was anessential need and role for social and psycho-logical research, this is it. Although research onvirtual teams may span the disciplines of humanfactors, communication, and human computerinteraction, the knowledge of group dynamics iscentral to understanding performance in virtualteams.

References

Ambady, N., Hallahan, M., & Rosenthal, R. (1995).On judging and being judged accurately in zero-acquaintance situations. Journal of Personalityand Social Psychology, 69, 518–529.

Ambady, N., & Rosenthal, R. (1992). Thin slices ofexpressive behavior as predictors of interpersonalconsequences: A meta-analysis. PsychologicalBulletin, 111, 256–274.

Anderson, A. H., O’Malley, C., Doherty-Sneddon,G., Langton, S., Newlands, A., Mullin, J., et al.(1997). The impact of VMC on collaborative prob-lem solving: An analysis of task performance,communicative process, and user satisfaction. InK. E. Finn, A. J. Sellen, & S. B. Wilbur (Eds.),Video-mediated communication (pp. 133–155).Mahwah, NJ: Erlbaum.

Ashford, S. J., & Tsui, A. S. (1991). Self-regulationfor managerial effectiveness: The role of activefeedback seeking. Academy of Management Jour-nal, 34, 251–280.

Back, K. W. (1951). Influence through social com-munication. Journal of Abnormal and Social Psy-chology, 46, 9–23.

Baron, R. M., & Kenny, D. A. (1986). The modera-tor–mediator variable distinction in social psycho-logical research: Conceptual, strategic, and statis-

319VIRTUAL TEAMS

tical considerations. Journal of Personality andSocial Psychology, 51, 1173–1182.

Berger, J. (1992). Expectations, theory, and groupprocesses. Social Psychology Quarterly, 55, 3–11.

Berger, J., Webster, M., Ridgeway, C., & Rosen-holtz, S. J. (1986). Status cues, expectations, andbehavior. In E. Lawler (Ed.), Advances in groupprocesses: Theory and research (Vol. 3, pp. 1–22).Greenwich, CT: JAI Press.

Blanton, H., & Christie, C. (2003). Deviance regula-tion: A theory of action and identity. Review ofGeneral Psychology, 7, 115–149.

Bos, N., Olson, J. S., Gergle, D., Olson, G. M., &Wright, Z. (2002). Effects of four computer-medi-ated communications channels on trust develop-ment. In Proceedings of CHI 2002 (pp. 135–140).New York: ACM Press.

Boyle, E., Anderson, A., & Newlands, A. (1994). Theeffects of visibility on dialogue and performance ina cooperative problem solving task. Language andSpeech, 37, 1–20.

Burgelman, J. C. (2000). Traveling with communica-tion technologies in space, time and everyday life:An exploration of their impact. Retrieved from http://www.firstmonday.dk/issues/issue5_3/burgelman/

Butler, D. L., & Winne, P. H. (1995). Feedback andself-regulated learning: A theoretical synthesis.Review of Educational Research, 65, 245–281.

Cairncross, F. (1997). The death of distance: Thetrendspotter’s guide to new communications. Bos-ton: Harvard Business School Press.

Cappel, J. J., & Windsor, J. C. (2000). Ethical deci-sion making: A comparison of computer-supportedand face-to-face groups. Journal of Business Eth-ics, 28, 95–107.

Carver, C. S., & Scheier, M. F. (1981). Attention andself-regulation: A control-theory approach to hu-man behavior. New York: Springer-Verlag.

Chapanis, A., Ochsman, R. B., Parrish, R. N., &Weeks, G. D. (1972). Studies in interactive com-munication: I. The effects of four communicationmodes on the behavior of teams during cooperativeproblem-solving. Human Factors, 14, 487–509.

Clark, H. H., & Brennan, S. E. (1991). Grounding incommunication. In L. B. Resnick, J. M. Levine, &S. D. Teasley (Eds.), Perspectives on sociallyshared cognition (pp. 127–149). Washington, DC:American Psychological Association.

Connolly, T., Jessup, L. M., & Valacich, J. S. (1990).Effects of anonymity and evaluative tone on ideageneration in computer-mediated groups. Manage-ment Science, 36, 97–120.

Costanzo, M. (1992). Training students to decodeverbal and nonverbal cues: Effects on confidenceand performance. Journal of Educational Psychol-ogy, 84, 303–313.

Craig, T. Y., & Kelly, J. R. (1999). The effects oftask and interpersonal cohesiveness on group cre-ativity. Group Dynamics: Theory, Research, andPractice, 3, 243–256.

Cramton, C. D. (2001). The mutual knowledge prob-lem and its consequences for dispersed collabora-tion. Organizational Science, 12, 346–371.

Culnan, M. J., & Markus, M. L. (1987). Informationtechnologies. In F. M. Jablin, L. L. Putnam, K. H.Roberts, & L. W. Porter (Eds.), Handbook of or-ganizational communication: An interdisciplinaryperspective (pp. 420–443). Newbury Park, CA:Sage.

Curtis, R. C., & Miller, K. (1986). Believing anotherlikes or dislikes you: Behaviors making the beliefscome true. Journal of Personality and Social Psy-chology, 51, 284–290.

Daft, R. L., & Lengel, R. H. (1984). Informationrichness: A new approach to managerial behaviorand organization design. Research in Organiza-tional Behavior, 6, 191–233.

Daft, R. L., & Lengel, R. H. (1986). Organizationalinformation requirements, media richness, andstructural design. Management Science, 32, 554–571.

DePaulo, B. M., Rosenthal, R., Eisenstat, R. A.,Rogers, P. L., & Finkelstein, S. (1976). Decodingdiscrepant nonverbal cues. Journal of Personalityand Social Psychology, 36, 313–323.

Devine, D. J. (2002). A review and integration ofclassification systems relevant to teams in organi-zations. Group Dynamics, 6, 291–310.

Doherty-Sneddon, G., Anderson, A., O’Malley, C.,Langton, S., Garrod, S., & Bruce, V. (1997). Face-to-face and video-mediated communication: Acomparison of dialogue structure and task perfor-mance. Journal of Experimental Psychology: Ap-plied, 3, 105–125.

Driskell, J. E., Hogan, R., & Salas, E. (1987). Per-sonality and group performance. In C. Hendrick(Ed.), Review of personality and social psychology(Vol. 9, pp. 91–112). Newbury Park, CA: Sage.

Driskell, J. E., & Mullen, B. (1990). Status, expec-tations, and behavior: A meta-analytic review andtest of the theory. Personality and Social Psychol-ogy Bulletin, 16, 541–553.

Driskell, J. E., Olmstead, B., & Salas, E. (1993). Taskcues, dominance cues, and influence in taskgroups. Journal of Applied Psychology, 78, 51–60.

Driskell, J. E., & Radtke, P. H. (in press). The effectof gesture on speech production and comprehen-sion. Human Factors.

Dubrovsky, V., Kiesler, S., & Sethna, B. (1991). Theequalization phenomenon: Status effects in com-puter-mediated and face-to-face decision-makinggroups. Human-Computer Interaction, 6, 119–146.

Erickson, B., Lind, E. A., Johnson, B. C., & O’Barr,W. M. (1978). Speech style and impression forma-


tion in a court setting: The effects of “powerful”and “powerless” speech. Journal of ExperimentalSocial Psychology, 14, 266–279.

Evans, C. R., & Dion, K. L. (1991). Group cohesionand performance: A meta-analysis. Small GroupResearch, 22, 175–186.

Festinger, L. (1950). Informal social communication.Psychological Review, 57, 271–282.

Frank, M. G., & Ekman, P. (1997). The ability todetect deceit generalizes across different types ofhigh-stake lies. Journal of Personality and SocialPsychology, 72, 1429–1439.

Fussell, S. R., & Benimoff, N. I. (1995). Social andcognitive processes in interpersonal communica-tion: Implications for advanced telecommunica-tions technologies. Human Factors, 37, 228–250.

Golembiewski, R. T. (1962). The small group: Ananalysis of research concepts and operations. Chi-cago: University of Chicago Press.

Griffith, T., & Neale, M. A. (1999). Informationprocessing and performance in traditional and vir-tual teams: The role of transactive memory (Re-search Paper No. 1613). Stanford, CA: StanfordUniversity Graduate School of Business.

Hackman, J. R., & Morris, C. G. (1975). Group tasks,group interaction process, and group performanceeffectiveness: A review and proposed integration.Advances in Experimental Social Psychology, 8,45–99.

Heath, C., & Luff, P. (1992). Media space and com-municative asymmetries: Preliminary observationsof video-mediated interaction. Human-ComputerInteraction, 7, 315–346.

Hiltz, S. R., Johnson, K., & Turoff, M. (1986). Ex-periments in group decision making: Communica-tion process and outcome in face-to-face versuscomputerized conferences. Human Communica-tion Research, 13, 225–252.

Hiltz, S. R., Turoff, M., & Johnson, K. (1989). Ex-periments in group decision making: 3. Disinhibi-tion, deindividuation, and group process in pen andreal name computer conferences. Decision SupportSystems, 5, 217–232.

Hollingshead, A. B., & McGrath, J. E. (1995). Com-puter-assisted groups: A critical review of the em-pirical research. In R. A. Guzzo, E. Salas, I. L.Goldstein, et al. (Eds.), Team effectiveness anddecision making in organizations (pp. 46–78). SanFrancisco: Jossey-Bass.

Hollingshead, A. B., McGrath, J. E., & O’Conner,K. M. (1993). Group task performance and com-munication technology: A longitudinal study ofcomputer-mediated versus face-to-face workgroups. Small Group Research, 24, 307–334.

Ilgen, D. R., Fisher, C. D., & Taylor, M. S. (1979).Consequences of individual feedback on behaviorin organizations. Journal of Applied Psychol-ogy, 64, 349–371.

Jarvenpaa, S. L., & Leidner, D. E. (1999). Commu-nication and trust in virtual teams. OrganizationalScience, 10, 791–815.

Jecker, J., Maccoby, N., Breitrose, H. S., & Rose,E. D. (1964). Teacher accuracy in assessing cog-nitive visual feedback from students. Journal ofApplied Psychology, 48, 393–397.

Karau, S. J., & Williams, K. D. (1993). Social loaf-ing: A meta-analytic review and theoretical inte-gration. Journal of Personality and Social Psy-chology, 65, 681–706.

Kiesler, S., Zubrow, D., Moses, A., & Geller, V.(1985). Affect in computer-mediated communica-tion: An experiment on synchronous terminal-to-terminal discussion. Human-Computer Interac-tion, 1, 77–104.

Krauss, R., & Fussell, S. (1990). Mutual knowledgeand communicative effectiveness. In J. Galegher,R. E. Kraut, & C. Egido (Eds.), Intellectual team-work: Social and technical bases of collaborativework (pp. 111–146). Hillsdale, NJ: Erlbaum.

Latane, B., Williams, K., & Harkins, S. (1979). Manyhands make light work: The causes and conse-quences of social loafing. Journal of Personalityand Social Psychology, 37, 822–832.

Lea, M., O’Shea, T., Fung, P., & Spears, R. (1992).“Flaming” in computer-mediated communication:Observations, explanations, and implications. InM. Lea (Ed.), Contexts of computer-mediated com-munication (pp. 89–112). Hertfordshire, England:Harvester Wheatsheaf.

Levine, J. M., & Moreland, R. L. (1990). Progress insmall group research. Annual Review of Psychol-ogy, 41, 585–634.

Linville, J. M., Liebhaber, M. J., Obermayer, A. H.,& Fallesen, J. J. (1991). Computer-mediated groupprocesses in distributed command and control sys-tems: Supervised shared work (Tech. Rep. 926).Alexandria, VA: U.S. Army Research Institute forthe Behavioral and Social Sciences.

Lott, A. J., & Lott, B. E. (1961). Group cohesiveness,communication level, and conformity. Journal ofAbnormal and Social Psychology, 62, 408–412.

Mathieu, J. E., Heffner, T. S., Goodwin, G. F., Salas,E., & Cannon-Bowers, J. A. (2000). The influenceof shared mental models on team process andperformance. Journal of Applied Psychology, 85,273–283.

McGrath, J. E. (1984). Groups: Interaction and per-formance. Englewood Cliffs, NJ: Prentice-Hall.

McGrath, J. E. (1990). Time matters in groups. In J.Galegher, R. E. Kraut, & C. Egido (Eds.), Intel-lectual teamwork: Social and technical bases ofcollaborative work (pp. 23–61). Hillsdale, NJ:Erlbaum.

McGrath, J. E. (1997). Small group research, thatonce and future field: An interpretation of the past

321VIRTUAL TEAMS

with an eye to the future. Group Dynamics, 1,7–27.

McGrath, J. E., & Hollingshead, A. B. (1994).Groups interacting with technology: Ideas, issues,evidence, and an agenda. Newbury Park, CA:Sage.

McLeod, P. L. (1992). An assessment of the experi-mental literature on electronic support of groupwork: Results of a meta-analysis. Human-Com-puter Interaction, 7, 257–280.

Meyerson, D., Weick, K. E., & Kramer, R. M.(1996). Swift trust and temporary groups. In R. M.Kramer & T. R. Tyler (Eds.), Trust in organiza-tions: Frontiers of theory and research (pp. 166–195). Thousand Oaks, CA: Sage.

Mullen, B. (1987). Self-attention theory: The effectsof group composition on the individual. In B.Mullen & G. R. Goethals (Eds.), Theories of groupbehavior (pp. 125–146). New York: Springer-Ver-lag.

Mullen, B. (1991). Group composition, salience, andcognitive representations: The phenomenology ofbeing in a group. Journal of Experimental SocialPsychology, 27, 297–323.

Mullen, B., Anthony, T., Salas, E., & Driskell, J. E.(1994). Group cohesiveness and quality of deci-sion making: An integration of tests of thegroupthink hypothesis. Small Group Research, 25,189–204.

Mullen, B., & Baumeister, R. F. (1987). Group ef-fects on self-attention and performance: Socialloafing, social facilitation, and social impairment.In C. Hendrick (Ed.), Review of personality andsocial psychology (pp. 189–206). Beverly Hills,CA: Sage.

Mullen, B., & Copper, C. (1994). The relation be-tween group cohesiveness and performance: Anintegration. Psychological Bulletin, 115, 210–227.

Mullen, B., Salas, E., & Driskell, J. E. (1989). Sa-lience, motivation, and artifact as contributions tothe relation between participation rate and leader-ship. Journal of Experimental Social Psychol-ogy, 25, 545–559.

Nixon, H. L. (1979). The small group. EnglewoodCliffs, NJ: Prentice-Hall.

Oliver, L. W., Harman, J., Hoover, E., Hayes, S. M.,& Pandhi, N. A. (1999). A quantitative integrationof the military cohesion literature. Military Psy-chology, 11, 57–83.

Olson, G. M., & Olson, J. S. (2000). Distance mat-ters. Human-Computer Interaction, 15, 139–179.

Poole, M. S., Holmes, M., Watson, R., & DeSanctis,G. (1993). Group decision support systems andgroup communication: A comparison of decisionmaking in computer-supported and nonsupportedgroups. Communication Research, 20, 176–213.

Postmes, T., & Spears, R. (1998). Deindividuationand antinormative behavior: A meta-analysis. Psy-chological Bulletin, 123, 238–259.

Postmes, T., Spears, R., & Cihangir, S. (2001). Qual-ity of decision making and group norms. Journalof Personality and Social Psychology, 80, 918–930.

Postmes, T., Spears, R., & Lea, M. (1998). Breachingor building social boundaries? SIDE-effects ofcomputer-mediated communications. Communica-tion Research, 25, 689–715.

Postmes, T., Spears, R., & Lea, M. (2002). Intergroupdifferentiation in computer-mediated communica-tion: Effects of depersonalization. Group Dynam-ics, 6, 3–16.

Ridgeway, C. L. (1987). Nonverbal behavior, domi-nance, and the basis of status in task groups. Amer-ican Sociological Review, 52, 683–694.

Ridgeway, C. L., & Diekema, D. (1989). Dominanceand collective hierarchy formation in male andfemale task groups. American Sociological Re-view, 54, 79–93.

Ringelmann, M. (1913). Research on animatesources of power: The work of man. Annales del’Institut National Agronomique (2e serie, TomeXII), 1–40.

Rocco, E. (1998). Trust breaks down in electroniccontexts but can be repaired by some initial face-to-face contact. In Proceedings of the Conferenceon Human Factors in Computing Systems (pp.496–502). Los Angeles: ACM Press.

Rudman, C., Hertz, R., Marshall, C., & Dykstra-Erickson, E. (1997). Channel overload as a driverfor adoption of desktop video for distributed groupwork. In K. E. Finn, A. J. Sellen, & S. B. Wilbur(Eds.), Video-mediated communication (pp. 199–223). Mahwah, NJ: Erlbaum.

Saunders, C., Robey, D., & Vaverek, K. (1994). Thepersistence of status differentials in computer con-ferencing. Human Communication Research, 20,443–472.

Schachter, S., Ellertson, N., McBride, D., & Gregory,D. (1951). An experimental study of cohesivenessand productivity. Human Relations, 4, 229–238.

Selfe, C. L., & Meyer, P. R. (1991). Testing claimsfor online conferences. Written Communication, 8,163–192.

Sellen, A. J. (1995). Remote conversations: The ef-fects of mediating talk with technology. HumanComputer Interaction, 10, 401–444.

Shaw, M. E. (1973). Group dynamics: The psychol-ogy of small group behavior. New York: McGraw-Hill.

Short, J. (1974). Effects of medium of communica-tion on experimental negotiation. Human Rela-tions, 27, 225–243.


Short, J., Williams, E., & Christie, B. (1976). Thesocial psychology of telecommunications. NewYork: Wiley.

Siegel, J., Dubrovsky, V., Kiesler, S., & McGuire, T.(1986). Group process in computer-mediated com-munication. Organizational Behavior & HumanDecision Processes, 37, 157–187.

Silver, S. D., Cohen, B. P., & Crutchfield, J. H.(1994). Status differentiation and information ex-change in face-to-face and computer-mediatedidea generation. Social Psychology Quarterly, 57,108–123.

Spears, R., & Lea, M. (1992). Social influence andthe influence of the “social” in computer-mediatedcommunication. In M. Lea (Ed.), Contexts of com-puter-mediated communication (pp. 30–65). Hert-fordshire, England: Harvester Wheatsheaf.

Sproull, L., & Kiesler, S. (1986). Reducing socialcontext cues: Electronic mail in organizationalcommunication. Management Science, 32, 1492–1512.

Sproull, L., & Kiesler, S. (1991). Connections—Newways of working in the networked organization.Cambridge, MA: MIT Press.

Steiner, I. D. (1972). Group process and productiv-ity. New York: Academic Press.

Straus, S. G. (1996). Getting a clue: The effects ofcommunication media and information distributionon participation and performance in computer-me-diated and face-to-face groups. Small Group Re-search, 27, 115–142.

Straus, S. G. (1997). Technology, group process, andgroup outcomes: Testing the connections in com-puter-mediated and face-to-face groups. Human-Computer Interaction, 12, 227–266.

Straus, S. G., & McGrath, J. E. (1994). Does themedium matter? The interaction of task type andtechnology on group performance and memberreactions. Journal of Applied Psychology, 79, 87–97.

Terborg, J. R., Castore, C., & DeNinno, J. A. (1976).A longitudinal investigation of the impact of groupcomposition on group performance and cohesion.Journal of Personality and Social Psychology, 34,782–790.

Thompson, L. F., & Coovert, M. D. (2003). Team-work online: The effects of computer conferencingon perceived confusion, satisfaction, and postdis-cussion accuracy. Group Dynamics, 7, 135–151.

Tziner, A., & Vardi, Y. (1982). Effects of commandstyle and group cohesiveness on the performanceof self-selected tank crews. Journal of AppliedPsychology, 67, 769–775.

Wagner, D. G., & Berger, J. (1993). Status charac-teristics theory: The growth of a program. In J.Berger & M. Zelditch (Eds.), Theoretical researchprograms: Studies in the growth of theory (pp.23–63). Stanford, CA: Stanford University Press.

Walther, J. B. (1997). Group and interpersonal ef-fects in international computer-mediated collabo-ration. Human Communication Research, 23, 342–369.

Walther, J. B., Anderson, J. F., & Park, D. W. (1994).Interpersonal effects in computer-mediated inter-action: A meta-analysis of social and antisocialcommunication. Communication Research, 21,460–487.

Walther, J. B., & Burgoon, J. K. (1992). Relationalcommunication in computer-mediated communi-cation. Human Communications Research, 19,50–83.

Warkentin, M. E., Sayeed, L., & Hightower, R.(1997). Virtual teams versus face-to-face teams:An exploratory study of a web-based conferencesystem. Decision Science, 28, 975–996.

Webster, M., & Driskell, J. E. (1983). Processes ofstatus generalization. In H. Blumberg, A. P. Hare,V. Kent, & M. Davies (Eds.), Small groups andsocial interaction (pp. 57–67). New York: Wiley.

Weisband, S., & Atwater, L. (1999). Evaluating selfand others in electronic and face-to-face groups.Journal of Applied Psychology, 84, 632–639.

Weisband, S. P., Schneider, S. K., & Connolly, T.(1995). Computer-mediated communication andsocial information: Status salience and status dif-ferences. Academy of Management Journal, 38,1124–1151.

Whittaker, S., & O’Conaill, B. (1997). The role ofvision in face-to-face and mediated communica-tion. In K. E. Finn, A. J. Sellen, & S. B. Wilbur(Eds.), Video-mediated communication (pp. 23–49). Mahwah, NJ: Erlbaum.

Williams, E. (1977). Experimental comparisons offace-to-face and video mediated communication.Psychological Bulletin, 84, 963–976.

Zaccaro, S., & Lowe, C. (1986). Cohesiveness andperformance on an additive task: Evidence formultidimensionality. Journal of Social Psychol-ogy, 128, 547–558.

Zaccaro, S., & McCoy, M. C. (1988). The effects oftask and interpersonal cohesiveness on perfor-mance of a disjunctive group task. Journal of Ap-plied Social Psychology, 18, 837–851.

Received September 16, 2002Revision received August 27, 2003

Accepted September 2, 2003 �

323VIRTUAL TEAMS

Virtual Teams: Effects of Technological Mediation on Team ... · proliferation of teams that do not...

Documents

Transcript of Virtual Teams: Effects of Technological Mediation on Team ... · proliferation of teams that do not...