Developmental differences between distributed and face-to-face groups in electronically supported meeting environments: An exploratory investigation

Group Decision and Negotiation, 4:213-233 (1995) �9 1995 Kluwer Academic Publishers

Developmental Differences Between Distributed and Face-to-Face Groups in Electronically Supported Meeting Environments: An Exploratory Investigation

KELLY BURKE Idaho State University, College of Business, Campus Box 8020, Pocatello, Idaho 83209

LAKU CHIDAMBARAM The University of Hawaii at Manoa, College of Business Administration, 2404 Maile Way, Honolulu, Hawaii 96822, USA

Abstract .

This longitudinal pilot study compared the developmental patterns of groups in three types of electronically supported meeting modes: face-to-face, dispersed-synchronous, and dispersed-asynchronous. The modes differed primarily in interactivity, channel capacity, and synchronicity. Comparisons were made along several behavioral and socio-technical dimensions which influence the group development process. Face-to-face groups tended to exhibit more effective leadership and coordination competence over time as compared to the distributed groups. However, along several other group process dimensions such as cohesiveness and equality of participation, dispersed groups did not differ from their face-to-face counterparts. Moreover, groups in all three modes performed equally well in terms of the quality of outputs. These results suggest that electronically distributed work groups--with adequate time and training--can become cohesive and perform effectively in the long run.

1. Introduction

The modern organization is expanding globally while flattening structurally, as layers of middle management disappear from many firms (Cash et al. 1992, p. 183). Emerging organizational designs will necessitate communication in new directions and through new channels within and between teams of workers. Workteams of tomorrow will be varied and flexible and engage primarily in "intellectual teamwork"; these teams will consist of "people working together over substantial periods of time to create information-intensive products" (Galegher and Kraut 1990, p. 65). Further, as Johansen (1990) argues, management will need to support these changing workteams with new and flexible technologies.

The convergence of these conditions assures management researchers and practitioners that dispersed groups will become more prevalent as the primary work units responsible for organizational and job design, decision making, project development, and other activities previously considered part of management's purview. Electronic communication and meeting technologies, herein referred to as electronic meeting systems (EMS)--

214 BURKE AND CHIDAMBARAM

especially those supporting distributed collaboration--can, given the right set of circumstances, facilitate team processes and improve group performance in organizations (Johansen 1990).

This preliminary study attempts to extend existing research, as called for by several researchers (e.g., Siegel et al. 1986), by investigating the impact of an EMS on various group processes. Among the many factors that EMS researchers have studied are (a) outcome variables such as time, performance, and quality relative to decision reached, number of alternatives generated, satisfaction with decision or performance, etc. (e.g., Chidambaram and Bostrom 1992; Hiltz and Johnson 1990; Jarvenpaa et al. 1988; Zigurs et al. 1990); (b) decision process characteristics such as consensus (e.g., Poole et al. 1988; Siegel et al. 1986) and equality of participation (e.g., Connolly et al. 1990; Hiltz et al. 1986; Lewis 1987; Siegel et al. 1986); (c) communication processes such as categorization or quantita- tive analysis of communicative interacts (e.g., Hiltz et al. 1986; Rice and Love 1987); and (d) perceptions of communication effectiveness (Walther and Burgoon 1992). The present study focuses specifically on factors related to group development processes.

In the next section we present a general overview of the problem and outline the perspective of this article. We define the issues relevant to the process of group development in electronic meetings and describe the theoretical foundations of our approach. Section 3 briefly reviews the research relevant to this study, while section 4 presents our research design and hypotheses. Finally, we examine the results of the study and conclude with a discussion of its implications for both practitioners and researchers.

2. Research framework

This study employs an inputs-process-outputs perspective to examine group interactions. Such an approach assumes that individual perceptions of both communication and cohesion factors serve simultaneously as indicators and determinants of group development processes. The model used in this study is represented graphically in Figure 1.

According to this model, differences in various characteristics of the meeting mode, (inputs) coupled with the influence of boundary variables such as duration of the interaction (time), will result in corresponding differences in the way meeting participants behave and perceive such interactions (processes). Differences in processes and perceptions will determine the performance of groups along a variety of measures (outputs).

2.1. Inputs

2.1.1. Meeting mode. The meeting mode refers to the general environment in which groups perform their tasks. The three modes represented here, i.e., face-to-face, dispersed- synchronous, and dispersed-asynchronous all occur in typical organizational settings. The meeting modality incorporates several characteristics.

Technology support. In this study all modes were supported by the same EMS. Thus, rather than measure differences between EMSs, we examine differences between meeting environments of which the EMS is an integral part.

DEVELOPMENTAL DIFFERENCES BETWEEN DISTRIBUTED AND FACE-TO-FACE GROUPS 215

f Research Framework

INPUTS* OUTPUTS

Meeting Mode: 0s a combination of - Technology Support - Dispersion - Synchronic~y and - Channel Capacity)

�9 Face- to -Face �9 D ispersed-

Synch ronous �9 D ispersed-

Asynch ronous

PROCESSES

Socio-Teehnical: �9 Social Presence �9 Equal Part lc ipat lor

Behavioral: �9 Cohesiveness �9 Leadersh ip

Ef fec t iveness �9 Coord inat ion

Compe tence

t Time (Boundary Variable)

Performance: Q u a l i t y o f O u t p u t

*While several inputs like group compostJon, faoflr support, group size etc affect group processes and outputs, ~,%,~ese factors were either controlled or randomized Only those inputs that were manipulated are shown here

J

Figure 1. Research framework.

Dispersion. Group members can either be collocated (in the same place) or distributed. For the purposes of this study, collocated groups met face-to-face, while dispersed group members (both synchronous and asynchronous) used separate rooms.

Synchronicity. Another feature pertains to synchronicity or degree of interactivity. Face- to-face meetings are the most immediate or interactive. A second mode describes people meeting at the same time while in different places (e.g., conference calling) in what we term distributed-synchronous meetings. These meetings are somewhat less interactive than face-to-face meetings, because they permit fewer types of message transmission. Such meetings, with their delays in interaction, can restrict the nature of cueing information exchanged as compared to regular face-to-face communication. Finally, the least interactive mode is the distributed-asynchronous setting. People in this situation meet at different times and from different places and often experience extensive delays between informational exchanges.

Channel capacity. Channel capacity differs across meeting environments, in part, as a function of the type of electronic support available. Short et al. (1976) argue that media vary in their capacity to transmit amounts and types of cues. Generally, some meeting modes exhibit characteristics which allow them to carry more information than others. For example, face-to-face communication allows the expression of a variety of verbal and nonverbal messages, while telephone communication permits transmission of far fewer types of messages. The relevance of channel capacity to developmental dynamics will be explored in greater detail below.


2.2. Processes

McGrath (1984) argues that three processes in particular are critical in the determination of the development of group dynamics; they are cohesion, communication, and conform- ity. He further asserts that these three processes are interdependent, since each has been found to covary with the others.

2.2.L Behavioral issues

Cohesiveness. Group cohesion is defined as the aggregate of the interpersonal attrac- tions of individual group members to each other and to aspects of the group as a whole (McGrath 1984). Cohesiveness increases as people in a team feel that both individual and common goals can be achieved through mutual influence and group action. There is consensus in the group research literature that cohesiveness is positively correlated with performance (see for example, McGrath 1984).

Leadership effectiveness. Taking a functional approach to leadership requires looking at various activities which are essential to any group endeavor. Two assumptions underlying the functional viewpoint are important to the current study. First, most of these critical activities common to all group exercises (e.g., planning, coordination, information gathering and sharing, consensus seeking) require effective communication. Second, these actions may be performed by any group participant (Fisher and Ellis 1990). That is, functional leadership can be seen as the result of the emergence of a congruence of group activity as a whole, not the property of one or a few individuals. Thus, the effectiveness of group leadership is dependent on useful information exchange among all team members and should increase as a function of the communication capacity of the meeting environment.

Coordination competence. Collaborative work needs to be supported by communication and coordination (Lim and Benbasat 1991). It is impossible, however, to separate these two functions, because coordination is dependent on communication. For example, systematic organization and administration of task, individual, and procedural issues are all functions of the coordination process. Yet, each of these activities requires the appropriate management of information flows between group members. Horton and Biolsi (1992) found that poorly coordinated groups all exhibited extended periods with no communication between members, while well coordinated groups demonstrated relatively continuous communication. Examination of these coordination functions reveals that the single most requi- site activity for coordination is communication.

2.2.2. Socio-technical issues

Social presence. Short et al. (1976) introduced the concept of social presence. Those media which transmit greater amounts of information simultaneously (including nonverbal and "back-channel" information) are said to possess greater social presence, i.e., they are more friendly, emotional, and personal (Hiemstra 1982) and convey a more compelling sense of the presence of the communicative partner. Conversely, those environments which allow the transmission of fewer types of messages are said to exhibit less social presence and are less expressive in general, while being more depersonalized and businesslike (Hiemstra 1982).


Equality of participation. A fundamental precept of group work is that more heads, up to a point, are better than one. People view issues from different perspectives and will, therefore, produce contrasting opinions, ideas, etc. In task execution, considering a larger number of ideas, issues, perspectives, and so on is generally thought to lead to improved performance. Implicit in this assumption is that in order for the group to realize higher contribution levels, all members must participate fully. Hence, equality of participation is crucial to effective group performance (Fisher and Ellis 1990).

2.2.3. Time. Relatively few MIS studies have observed the effects on groups of EMSs over time. Many studies employ designs wherein subjects meet only one time for performing a specific task (e.g., Connolly et al. 1990; Kiesler et al. 1984; Lewis 1987; Poole and Lind 1991; Siegel et al. 1986). Groups in organizations, however, seldom meet just once. Generally, groups meet on multiple occasions over an extended period of time. The significance of time as a dimension of group behavior is that the development process is a cumulative one and groups are likely to behave differently later in their existence than in the early stages.

The key to understanding the implications of group development lies in understanding the consequences of and the changes in relational behaviors (e.g., cohesion and influence) during the developmental process. Processes underlying and comprising group development evolve over time (McGrath 1991) and can be understood fully only through longitudinal studies such as the present one.

2.3. Outputs

2.3.1. Task performance. For a technology to have practical application, managers must want it to facilitate work and increase either productivity, work effectiveness, or perceived quality of work life. Thus, it is critical to determine under what conditions a technology- supported meeting environment may have a deleterious or enhancing effect on group performance.

3. Relevant prior research

Two streams of research, group support systems (GSS) and computer-mediated communication (CMC), are related to this study. Variables relevant to group development which have been examined in both areas, can be categorized as relational and performance-oriented, respectively. Relational variables include such factors as group cohesion, social influence, and leadership skills, while performance-oriented variables include task correctness, decision quality, and degree of participation. Below is a brief review of the literature that has focused on these aspects of computer-supported group development.

3.1. Relational facwrs

Computer support for groups has generally been associated with a more even distribution of power and influence and a marked tendency to remain task focused (Zigurs et al. 1988;


Watson et al. 1988). Generally, the literature has not provided conclusive proof about the positive impact of computer support on relational factors in a group. In one instance, however, a study of 28 groups, meeting over a period of four sessions, showed that GSS groups increasingly exhibited greater levels of cohesiveness than did non-GSS groups (Chidam- baram 1989).

Results of that study, which used face-to-face groups, may not be directly applicable to dispersed groups. Little empirical evidence exists about the impact of dispersion on relational factors. Kutsko and Smith (1991) surveyed team members of a real organization on their perceptions of team interaction and effectiveness between distributed and co-located (one site) teams. The dependent measures most relevant to (though somewhat more global in nature than) those employed in the present study concerned the perceived collaborative climate and the structure. Interestingly enough, the distributed teams indicated a more positive perception of the collaborative climate of their team situation. Factors comprising a collaborative climate included trust in each other, perceived competence, and shared values. Distributed groups also perceived the structure of the group as being more results driven; i.e., they perceived the structure to facilitate more effective exchange of information and interaction and exhibited greater feelings of goal directedness than co-located teams.

Kiesler et al. (1984) and Siegel et al. (1986) (in separate reportings of the same data) report differences between computer-supported and face-to-face (non-supported) groups along several relational dimensions. Interesting as a possible indicator of interpersonal influence was their finding that computer-supported groups, as compared to non-supported groups, tended to report greater shifts in decision choice from the individuals' means. The authors suggested that disinhibition may have permitted members to more easily abandon personal choices.

Hiltz et al. (1986) compared face-to-face and computer-mediated settings for differences in communication content, emotionality, participation, and decision quality. Bales IPA coding scheme was employed for content analysis and showed computer-supported groups expressing tess communication and expressing less agreement than face-to-face groups. In general, there was more socio-emotional exchange (e.g., tension release, joking, etc.) in face-to- face groups, and more task-related exchange in computer settings. This supports the notion that CMC is a less personalized (and hence, less relational) medium than face-to-face meeting. The implication is that CMC has less potential for interpersonal influence and cohesion.

Due to the need for interpersonal influence in group development, the potential for relational communication is an issue of concern for electronically supported meetings. Walther (1992a) assessed relational communication in 32 groups engaged in either face-to-face or asynchronous electronically supported meetings over a period of time. Surprisingly, computer-supported groups reported higher perceptions of immediacy/affection than face- to-face groups. That immediacy/affection decreased over time in both conditions was also an unexpected result. Walther also found informality (a measure of relational closeness) was higher in computer-supported than face-to-face groups. This too was not the predicted trend. Finally, and also unanticipated, EMS groups exhibited higher levels of social than task communication relative to face-to-face groups. Thus, generally, computer-supported groups perceived their interactions as more personal than the face-to-face groups. These results run counter to what theory has predicted regarding communication via electronic channels and highlight the need for more research.


3.2. Social presence

Social presence has been examined theoretically in numerous articles (e.g., Short et al. 1976; Trevino et al. 1990; Walther 1992a), but little has been done to measure perceptions of media or specific social presence characteristics, especially in asynchronous settings. This study seeks to uncover information pertinent to these issues.

3.3. Participation

In several early studies of GSS (Gallupe 1985; Beauclair 1987; Jarvenpaa et al. 1988; Watson 1988; Easton 1989), no differences were found in degree of member participation between computer-supported and non-supported groups. Later studies, however, suggest that computer-supported groups may--under certain circumstances--exhibit greater equality of participation than non-supported groups. For instance, Jessup et al. (1990) found that a GSS which permits anonymity will result in greater participant deindividuation than a system which allows members' input to be identified. In this case, anonymous groups contributed significantly more critical comments, solution questioning comments, and total comments than the non-anonymous groups.

The results from the Kiesler et al. (1984) and Seigel et al. (1986) studies generally support the argument that more information is exchanged in less time in face-to-face than in EMS settings. Participation, however, was less evenly distributed in face-to-face groups than in EMS groups. There was also more flaming (i.e., more inflammatory remarks) in EMS than in face-to-face groups. The latter two findings indicate that computer-supported media may be more anonymous and, consequently, serve to disinhibit or deindividuate (Jessup et al. 1990) group members, resulting in more participation in general and more critical participation in particular. Participation in an electronic meeting may also be equal- ized as a result of the attenuated ability of potentially dominant members to control discussion.

Rice and Love (1987) sampled six weeks' worth of an ongoing computer bulletin board (BBS) conference in order to look at socio-emotional message rates. Approximately 30% of the total messages (on a professionally oriented BBS) were socio-emotional. The use of an ongoing BBS may have been a problem, since there is no real way of knowing how many times participants may have interacted with each other in the past. A larger issue, however, is the potential lack of comparability between public BBS and task-oriented interaction.

3.4. Performance

Gallupe (1985) found that with difficult tasks computer-supported groups produced more alternatives and made higher quality decisions than did groups without such support. Easton et al. (1989) found no differences in decision quality between EMS and non-EMS groups. Similarly, Chidambaram (1989) found no differences between supported and manual groups in terms of average decision quality. Contradictory performance results are the norm in


GSS research to date. One reason for this is that these studies have all used different systems and different tools. In this study, all groups used the same EMS and the same tools, albeit in different temporal and geographical settings.

Asynchronous group performance has largely been ignored in EMS research. An excep- tion, however, was a field experiment conducted by Smith and Vanecek (1989). They observed ten groups of non-simultaneous computer conference users and ten groups of face- to-face users solving a murder mystery (intellective task type). Results indicated that the distributed groups shared considerably less total information and were less comprehensive in their analysis of information than face-to-face groups. Further, the asynchronous groups exhibited a lower level of perceived progress toward their goal. This latter factor may be important to the overall feeling of group effectiveness by members.

This brief review of relevant EMS research highlights two areas of concern: (a) the pau- city of research examining asynchronous interaction over time in task focused groups, and (b) the lack of longitudinal studies comparing face-to-face and EMS environments. The present study was designed to fill these gaps in the literature.

4. Theory and hypotheses

Anthony Giddens (1979) argues that social reality is constructed at the nexus of human agency and social structures. Social reality is a result of the interaction between intentional actors and the rules which constitute society. Giddens maintains that the structures which emerge from these interactions then constrain and influence subsequent human action which, in turn, further impacts and restructures social norms.

Poole (1983, 1989) has developed a structurational theory of group development which has its roots in Giddens's social theory of structuration. Its underlying concept is that social and group structures, e.g., roles, rules, norms, facilities, media, etc., act as constraints within which groups must function, but that with repeated use, these structures are trans- formed or altered to better serve the specific needs of the group. Thus, groups will appropriate comparable structures in differing ways depending on such factors as task and group personality. Through customization, the existing structure is altered for future application, often with unintended consequences.

The process is recursive and can be cumulative, in that the resultant structure constrains future activities, which causes the development of even newer structural forms and so on. The changes in the forms may be subtle or radical. Evidence of this structurational phenomena has been documented in several studies (e.g., Chidambaram and Bostrom 1993; Gopal et al. 1992-1993).

We draw on Structuration Theory (Giddens 1977) and Adaptive Structuration Theory (Poole and DeSanctis 1989) to argue that, when faced with structural impediments to effective interaction, groups are likely to explore ways of adapting environmental factors (e.g., technological, procedural, or communicative structures) to compensate for such problems. Further, any number of structures may be targeted for adaptation in the effort to improve group effectiveness. Over time, adaptations which are explored and implemented may themselves become structures which influence subsequent behaviors. Consequently, we assume that the structural context of a group's activites must be assessed in order to understand the developmental processes.


4.1. Behavioral Issues

Cohesiveness. In the case of mediated interactions, there are clearly differences between the environments in terms of both temporal and spatial characteristics. Of the three conditions which this study examines, face-to-face is the most immediate in spatial and temporal proximity, while dispersed-synchronous groups are more spatially and temporally immediate than asynchronous groups. Immediacy in the present context is defined as either temporal or spatial proximity, and the degree of social influence bears an inverse relation to proximal distance.

One of the three essential components of social influence according to social-impact theory (Latan~ 1981) is immediacy. In view of the relative differences in immediacy between the modes, one would expect greater influence in face-to-face groups than in distributed- synchronous groups, and more influence in synchronous than asynchronous meetings. Since cohesion varies with influence (McGrath 1984), one would expect cohesion to behave similarly.

Adaptive structuration theory (e.g., Orlikowski 1991; Poole and DeSanctis 1989) suggests that as group members interact over time, those in less proximal environments should begin to converge with proximal groups in cohesiveness, as they explore and adapt novel ways to communicate and compensate for media, temporal, and spatial restrictions. Given the relatively short duration of the exercise in the present study, however, there should have been less opportunity for adaptation.

We sum up our hypotheses concerning cohesion as follows:

HI: Cohesiveness will differ across treatments over time. Il ia: Cohesiveness will be greater initially in face-to-face groups than in distributed groups. I-lib: Cohesiveness will increase in later sessions more in distributed groups than in face-

to-face groups.

Leadership effectiveness. Leadership effectiveness is closely linked with the ability to communicate clearly. The extent to which the members of a group share an understanding of the need for an action and assume responsibility for performing that action is very much related to the ability of that group to exchange complete and timely information. Shared understanding and commitment is developed through continuous interaction among all members and is, therefore, very much a function of the communication capacity and interactivity of the meeting environment.

Many of the functions of leadership are time sensitive. For example, planning an activity agenda usually precedes action. Thus, delays in the structuring process result in delays in sequentially dependent activities. In an environment which constraints the timeliness of interaction, we expect that structuring functions would be delayed or perhaps even pre- empted. For instance, if an asynchronous group waits until all members have attended a session and contributed their initial procedural ideas before attempting to select a course of action, we can expect normal processes to consume large amounts of meeting time. Such procedural inefficiencies should lead to declining perceptions of leadership effectiveness over time, as the handling of these functions becomes increasingly problematic.


We sum up our hypotheses concerning the interaction effects of treatment and time on leadership as follows:

1-12: Leadership effectiveness will differ across treatments over time. 112a: Leadership effectiveness will be greater in face-to-face groups initially. 112b: Leadership effectiveness will increase over time more in distributed groups than in

face-to-face groups.

Coordination competence. The fact that the primary coordination activity is communication suggests that the ability of a group to successfully manage the coordination process depends largely on the group's communication environment. More immediate and interactive conditions which facilitate communication, should exhibit more systematic and productive coordination. Similarly, less interactive groups should perform the coordinative tasks less effectively.

Coordination functions, like some leadership functions, are extremely responsive to timing. Many coordinative tasks presume prior actions, e.g., assigning activities to participants requires identification of appropriate objectives first. This would seem to imply that meeting environments which exhibit delayed interaction are more likely to experience coordination problems than would be the case under more interactive conditions.

We sum up our hypotheses concerning the interaction effects of treatment and time on coordination as follows:

1t3: Coordination competence will differ across treatments over time. 113a: Coordination competence will be greater in face-to-face groups initially. H3b: Coordination competence will increase over time more in distributed groups than

in face-to-face groups.

4.2. Socio-technical issues

Social presence. Social presence theories (e.g., Short et al. 1976) and media richness theories (e.g., Trevino et al. 1990) take similar positions in arguing that media vary in the amount and type of information which they are capable of processing. Richer media, like face-to-face meetings, enable greater reduction of ambiguity and elicit perceptions of greater social presence. Specifically, Short et al. (1976) assert that face-to-face is most efficient, because it allows transmission of many proximal, facial, and other nonverbal cues. ~ e r environments, like distributed (non-face-to-face) modes, have less capacity for reducing ambiguity because they allow few if any nonverbal cues and would be expected to result in less social presence.

Further, the immediacy of an environment depends in part on the capacity for transmission of various types of cues. Spatial proximity constrains both quantity and type of nonverbal communication, while temporal proximity determines interactivity or reciprocity. For example, increasing temporal distance (time lapse) between communicative interacts can lead to a reduced perception of the importance the message (Walther 1992a). Thus, face-to-face modes are more immediate than distributed modes and can be expected to report greater


social presence. Similarly, synchronous conditions are more temporally immediate, and ought to exhibit, therefore, greater social presence than asynchronous environments.

Adaptive structuration theory (AST) asserts that as people interact with technologies, they will often use the technology in novel and innovative ways (e.g., Orlikowski 1991; Poole and DeSanctis 1989). Extending this argument, the present study assumes that as participants find that a meeting mode is inappropriate for their specific style or for a particular type of task, they will often adapt the modalities in novel ways which compensate for communicative shortcomings of the environment. We also expect to observe more extensive and effective adaptation to occur over time. In conditions with less time available, we expect less effective adaptation to occur, thereby resulting in less perceived social presence.

We sum up our hypotheses concerning the interaction effects of treatment and time on perceptions of social presence as follows:

1-14: Perceived social presence will differ across treatments over time. I-I4a: Social presence will be perceived higher initially in face-to-face groups than in

distributed groups. H4b: Perceived social presence will increase more over time in the distributed groups than

in the face-to-face groups.

Equality of participation. A factor which appears critical to effective interaction is the extent to which an EMS--being an important characteristic of meeting mode--fosters participation. Analysis of EMS research (e.g., Siegel et al. 1986; Jessup et al. 1990) suggests that environmental immediacy and presence appear to be related positively to dominance and inversely to participation. Thus, it would seem that the two distributed conditions would demonstrate greater equality of participation than the more immediate face-to-face mode.

AST takes the position that people are likely to appropriate a technology so as to overcome deficiencies in communication capacity relative to their personal as well as group needs. This raises the possibility that individuals with a strong need to control the group process may become creative in devising methods to assert their dominance in spite of the inherent neutrality of the EMS. For example, using all capital letters might indicate a self-declared position of authority. Since distributed interaction is more anonymous than face-to-face interaction, there may be more opportunity for distributed group members to employ controlling techniques unique to the EMS, thus reducing participatory equality over time.

We sum up our hypotheses concerning the interaction effects of treatment and time on participatory equality as follows:

H5: Equality of participation will differ across treatments over time. I-ISa: Equality of participation will be greater initially in distributed groups than in face-

to-face groups. I-I5b: Equality of participation will decrease more over time in distributed groups than

in face-to-face groups.


4.3. Performance

The earlier discussion concerning media cueing capacity suggests that some meeting modalities will be more effective for the communication exchange necessary for the joint creation of a complex document. We would expect, then, that the conditions providing the greatest channel capacity and the greatest opportunity for effective and immediate interaction should result in the best task performance overall. However, as the distributed groups adapt their environmental structures to their needs, all groups should converge in performance.

Our hypothesis concerning the effects of treatment on task performance is as follows:

H6: Task performance will not differ across treatments over time.

5. Research design

The present study utilized a repeated-observations design capturing data from groups during each of three one-hour meetings. Such a longitudinal design should increase the exter- nal validity of the results relative to results obtained from single meeting studies. A single factor (meeting mode) design with the following three levels was used:

1. face-to-face with EMS (face-to-face); 2. distributed-synchronous with EMS (synchronous); 3. distributed-asynchronous with EMS (asynchronous).

5.2. Treatments

Participants were randomly assigned to groups, and groups were randomly assigned to meeting modes. Each group worked on the same task and was required to collaboratively write a document using an electronic group-writing tool. The system used was a networked, distributed EMS developed by Ventana Corporation, called GroupLink. Groups were trained to use two tools for communication. The first was an "annotation" feature which allowed them to embed comments in sections of the document without disrupting the flow of the document. The second was a "news" feature which allowed them to post notices to teammates at the main menu level. All participants were asked to avoid discussing the exercise outside of the scheduled session times. The experimenter was present during all sessions for technical advice and to preserve procedural integrity.

Face-to-face EMS. Members of face-to-face groups met at the same time in an Elec- tronic Meeting Room similar to the one described in Dennis et al. (1988). Terminals were organized in a semicircle, with each participant having full view of the other group members. Face-to-face participants were instructed that they should make task-specific comments using the annotation feature, but that they could talk freely with their teammates.


Synchronous EMS. Groups in the distributed synchronous condition met at the same time, but in four different rooms containing PCs throughout the building. No single group member could start a session before the others had arrived and were ready to begin. No face-to-face interaction between members was permitted during meeting times.

Asynchronous EMS. Asynchronous group members each logged onto the system and "attended" the meeting at times when other members were not at their computers. Except for this, all other procedures were the same as described above.

5. 3. Experimental procedures

Subjects. The study used nine groups of undergraduate business students. Each group consisted of four members. Three groups were randomly assigned to each of the treatments, i.e., face-to-face, synchronous, and asynchronous. Data from one of the face-to-face groups were dropped, because the group had only two members at the end of the study.

Task. The current study employed an unstructured type of task in a setting which has generally been ignored by EMS researchers. The task required groups to develop a policy manual to be used to train new recruits hired by a multinational consulting company.

Dependent variables. As discussed earlier, all dependent measures (except performance) were gathered three times. Identical questionnaires, employing semantic differential scales, were completed by participants at the end of each session. Factor analysis confirmed the robustness of the five perceptual variables measured in this study--namely, cohesiveness, leadership effectiveness, coordination competence, social presence, and equality of participation. Finally, performance was measured using a panel of expert judges who evaluated the quality of the final document based on such criteria as scope and depth of issues covered, potential effectiveness of the solution, and the organization and clarity of the document.

6. Results and discussion

MANOVA was used to test for differences in effects over time in our repeated observations. A summary of the MANOVA results for all variables is presented in Table 1.

Table 1. Analysis of variance results.

Variable Plllais Value df F Statistic p Value

Cohesiveness 0.162 4,50 1.099 0.367

Leadership effectiveness 0.356 4,52 2.815 .034*

Coordination competence 0.524 4,50 4.620 .003*

Social presence 0.314 4,50 2.062 0.153

Equality of participation 0.014 4,50 0.234 0.917

Document quality 2,4 1.460 0.316

*Significant at p = <. 10


Where interaction effects were detected, follow-up, one-way ANOVAs were performed to identify the nature of the differences. Where the ANOVAs detected differences, Duncan's range test, a frequently used multiple comparison procedure, was used to determine the precise nature of the differences. Those results are presented in Table 4, later in this section.

A liberal level of significance (p = . 10) was used to test the hypotheses. Such a level is appropriate in the context of this study, because there have been very few longitudinal experiments of EMS use in distributed environments, suggesting that the present effort should be viewed as exploratory in nature. In addition, since this exploration was a pilot study, the sample size was comparatively small and may have impacted the potential for detecting true effects.

6.1. Behavioral issues

6.1.L Cohesiveness

HI: Cohesiveness will differ across treatments over time. (Not supported)

Contrary to what we expected, there were no differences in cohesion across meeting modes over time. Within the framework of this article, three possible explanations exist for the absence of hypothesized differences. First, the small sample size may have made it difficult to detect any effect. Second, perhaps functions determined by physical and spatial distance (e.g., immediacy, influence, and reciprocity) are not factors in task-focused group interaction. This is an unappealing conclusion, however, because we know from experience that physical distances, as well as other types of barriers, often reduce the feeling of immediate influence (for example, in the case of being physically threatened).

Finally, a more compelling explanation is that, as AST predicts, over time, groups in the less immediate conditions were successful in adapting the technology to serve their influence requirements. We believe that three meetings might have been too short a time frame for appropriation to occur however, since the task used in this study may have induced relatively little conflict. With relatively low levels of conflict, interpersonal influence would have been relatively unnecessary, and perhaps a minimal level of adaptation was sufficient.

6.1.2. Leadership effectiveness

H2: Leadership effectiveness will differ across treatments over time. (Supported) H2a: Leadership will be more effective in face-to-face groups initially. (Not supported) H2b: Leadership effectiveness will increase over time more in distributed groups than in

face-to-face groups. (Not supported)

We argued that the functions of leadership are better served in environments which facilitate timely and productive exchange of information. Appropriate information exchange is fundamental to a shared understanding of process requirements and to notions of group togetherness. As can be seen from the statistical analysis (Figure 2 and Table 2), when we compare leadership effectiveness among treatments during each of the sessions, the differences generally support the hypothesis.


r~

7

6

5

4

3

2

- -@--F- t -F

--[:]--s~ - - & - ~

r

A

1 I I

Session 1 Session2 Session3

Figure 2. Leadership effectiveness.

Table 2. Leadership effectiveness means.

Session 1 Session 2 Session 3

F-t-F 4.21 5.67 5.46

Sync 3.96 4.26 4.52

Async 5.67 4.22 3.72

In the first session, as predicted, there were no significant differences among treatments, possibly because there had been little time up until that point for leadership to emerge in any of the conditions. In the second and third sessions, the face-to-face groups reported greater leadership effectiveness than the asynchronous groups, and, in session 2, face-to- face groups also rated leadership better than the synchronous groups. This appears to support our contention that the meeting modality is a critical dimension influencing the leadership effectiveness in groups.

When comparing the treatments over the three sessions, only the face-to-face mode improved as expected, reporting that leadership was better in session 2 than session 3, and better in both 2 and 3 than in session 1. The structural constraints may have been too difficult for distributed groups to overcome in three sessions.


6.1.3. Coordination competence

H3: Coordination competence will differ across treatments over time. (Supported) H3a: Coordination will be greater in face-to-face groups initially than in distributed groups.

(Not supported) H3b: Coordination competence will improve more in distributed groups over time than

in face-to-face groups. (Not supported)

Coordination is a requirement in all group projects, and the extent to which a group is well coordinated determines performance (Horton and Biolsi 1992). The current study found that there is an interaction effect of meeting mode and time on coordinative competence (Figure 3 and Table 3).

In session 1 the synchronous groups unexpectedly reported better coordination than the face-to-face groups. However, by session 2, the face-to-face groups began to show greater

7

6

5

4

3

2

1 I I

kssion 1 kssion 2 Session 3

Figure 3. Coordination competence.

Table 3. Coordination competence means.

Session 1 Session 2 Session 3

F-t-F 3.31 4.75 5.75

Sync 4.72 3.06 4.89

Async 4.17 3.92 4.46


Table 4. One-way ANOVA results.

Treatment Session No.

F-t-F 2 >> 3 > 1

Sync

Async

Treatment Session No.

F-t-F 3>2>1

Sync 1 >3 > 2

Async

Leadership Effectiveness

Session No.

1

2

3

Treatment

m

F>>S>A

F>A

Coordination Competence

Session No.

1

2

3

Treatment

S>F

F>S*

F>>S>A*

p = <.05 *p = <.10 > Greater than next >> Greater than both

coordinative ability than the synchronous groups, and, in session 3, more than both synchronous and asynchronous conditions. Apparently the more immediate and interactive face-to-face environment afforded greater opportunity for coordination. Contrary to our expectations, this effect appears to have held more over t ime for the face-to-face groups than for the distributed groups.

62. Socio-technical issues

6.2.1. Social presence

H4: Perceived social presence will differ across treatments over time. (Not supported)

Counter to what is suggested by theories of social presence and media richness, there was no significant difference in perceived social presence among treatments over sessions. These results provide evidence that social presence may not be as media dependent as has been assumed. In this study, there were clearly differences in communication capacity between the co-located and the distributed conditions, yet the face-to-face groups reported their environment to be no more personal and engaging than did dispersed groups.

One explanation may be that the participants perceived the technology and the meeting environment as separate issues, and only considered the technology as a communication medium when responding to the survey. Anticipating this possibility, survey instructions in future iterations of the study will emphasize more clearly that the survey items concern the entire environment, i .e., technology and meeting mode.


Since the task was the primary focus, it is also likely that time constraints precluded extensive non-task related communication. If this was the case, there may have been only a minimal need for any socially oriented communication. Thus, it is possible that the participants took a purely functional perspective in judging the social presence of the mode (i.e., the environment allowed communication sufficient for the task). Perhaps, then, the participants attached no significance to any criteria of social presence.

6.2.2. Equality of participation

H5: Equality of participation will differ across treatments over time. (Not supported)

Previous studies indicated that EMSs offer group members anonymity (Jessup et al. 1988) and prevent conveyance of status messages. These factors combine to help "level the play- ing field" sometimes leading to increased equality of participation (e.g., Siegel et al. 1986). The present study found no support for such a position.

A possible reason for the contrast of our findings with earlier studies may lie with the nature of the dependent measure. Some prior studies measured participation objectively (e.g., Siegel et al. 1986), with number of ideas generated or number of statements made as a basis for comparison of participatory levels. The current study used perceptual measures. This approach was taken in recognition of the importance of members' perceptions to group development. Future studies should examine both subjective and objective equality of members' participation to get an accurate picture.

6.3. Performance

1-16: Task performance will not differ across treatments. (Supported)

Whether groups met in co-located or dispersed environments, there was no effect on the quality of the document produced. Both media richness and social presence theories suggest that performance should decline as the environment permits fewer channels of communication relative to task and user requirements. In light of these theories and given the barriers to effective interaction and communication inherent in distributed environments, at first glance, it appears difficult to understand why and how distributed groups performed as well as face-to-face groups. The fact that this study found no differences between treatments in terms of performance suggests that media richness theory in particular may need to be reevaluated. If one considers that disadvantaged groups may alter environmental structures in ways which permit more effective interaction, as predicted by AST, the answer to "why?" may be explained. The answer to "how?" remains elusive and is a subject for future research.

7. Conclusion

Overall, this study provides some support for our basic model of group development. Specifically, we found that the meeting environment within which groups operate has a


significant impact along certain developmental dimensions, namely, leadership and coordination. In general, this study suggests that the greater the capacity exhibited by an environment for complete and timely exchange of information, the better served the functions of leadership and coordination will be.

The fact that this investigation did not detect effects of mode and time on cohesiveness has implications along two important dimensions. First, our results suggest that media richness theory may not explain the true nature of behavior in a task setting. Media richness theory implies that groups working on a task like the one in this study, i.e., complex and high in equivocality, will presumably find leaner environments (e.g., asynchronous) less suitable for interaction and for high quality performance. However, our results indicate that neither are there significant differences between the environments in perceived social presence (a characteristic important to influence and cohesion) nor are there differences in quality of output.

Secondly, our findings may offer support for Adaptive Structuration Theory. In other words, when faced initially with a restrictive environment, people may alter and adapt the structures to use them in ways which better suit their objectives. In this case, the meeting environment and invidiual behavior patterns were potential targets for adaptation. Our results indicate tentative support for the notion that groups can adapt the environment to suit their needs. Thus, dispersed meeting modes, despite restrictions in channel capacity, can be used to convey--in certain contexts--information rich enough to facilitate task-oriented communication among group members. However, more evidence needs to be gathered to further clarify the nature of the effects detected in this study.

Given the importance of effective coordination in organizations, especially in complex group projects, these results have important implications for managers. For example, it may be that coordination and leadership functions become less critical as projects develop. This would suggest that a more versatile (and probably more expensive) communication environment may be necessary early in a project, but that more limited (and less costly) systems might be adequate to support later project stages. Further, the fact that our study found no differences in cohesiveness, social presence, or performance among the various meeting modes suggests that management can expect groups, with adequate training and time, to perform effectively in distributed environments.

As with any exploratory undertaking, the results of this study must be considered somewhat tentative and subject to further validation. Future replications of this project should utilize larger sample sizes over longer periods. Such design adjustments can help validate the findings from this study and may reveal effects that the current study was not able to detect.

Finally, this was a tightly controlled experiment, using student subjects and two specific system tools. Imposing such controls precludes applying the results widely to other settings. In order to provide a richer picture of group processes in organizational settings, future studies should employ systems and environments with greater functionality and, where possible, validate experimental results in real organizations. The use of EMSs for conduct- ing asynchronous group activities is likely to become widespread. This should provide researchers with ample opportunity to further investigate developmental processes in the context of organizational settings.


References

Beauclair, R.A. (1987). '9,n Experimental Study of the Effects of Group Decision Support System Process Sup- port Applications on Small Group Decision Making?' Unpublished doctoral dissertation, Indiana University.

Chidambaram, L., and R.P. Bostrom. (1993). "Evolution of Group Performance Over Time: A Repeated Measures Study of GSS Effects," The Journal of Organizational Computing.

Chldambaram, L., R.P. Bostrom, and B.E. Wynne. (1990-1991). "A Longitudinal Study of the Impact of Group Decision Support Systems on Group Development:' Journal of Management Information Systems 7 (Winter), 7-25.

Connolly, T., L. Jessup, and J.S. Valacich. (1990). "Effects of Anonymity and Evaluative Tone on Idea Genera- tion in Computer Mediated Workgroups," Management Science 39(6), 689-703.

Daft, R.L., and R.H. Lengel. (1986). "Organizational Information Requirements, Media Richness, and Struc- tural Design" Management Science 32(5), 554-571.

DeSanctis, G., and R.B. Gallupe. (1987). ' ~ Foundation for the Study of Group Decision Support Systems," Management Science 33(5), 589-609.

Fisher, B.A., and D.G. Ellis. (1990). Small Group Decision Making: Communication and the Group Process, 3rd ed. New York: McGraw Hill Publishing Company.

Galegher, J., and R.E. Kraut. (1990). "Computer Mediated Communication for Intellectual Teamwork: A Field Experiment in Group Writing?' Proceedings of the Conference on Computer-Supported Cooperative Work, pp. 65-78.

Gallupe, R.B. (1985). The Impact of Task Difficulty on the Use of a Group Decision Support System. Unpub- lished doctoral dissertation, University of Minnesota.

Giddens, A. (1977). Studies in Social and Political Theory. New York: Basic Books. Hiemstra, G. (1982). "Teleconferencing, Concern for Face, and Organizational Culture" In M. Burgoon (ed.),

Communication Yearbook, Vol. 6. Beverly Hills, CA: Sage Publications, pp. 874-904. Hiltz, S.R., and Johnson, K. (1990). "User Satisfaction With Computer Mediated Communication Systems,"

Management Science 36(5), 739-764. Hiltz, S.R., K. Johnson, and M. Turoff. (1986). "Experiments in Group Decision Making: Communication Proc-

ess and Outcome in Face-to-Face Versus Computerized Conferences" Human Communication Research 13(2), 225-252.

Jarvenpaa, S.L., V.S. Rao, and G.p. Huber. (1988). "Computer Support for Meetings of Groups Working on Unstructured Problems: A Field Experiment," MIS Quarterly (December), 645-665.

Jessup, L.M., T. Connolly and J. Galegher. (1990). "The Effects of Anonymity on GSS Group Process With an Idea Generating Task," MIS Quarterly (September).

Johansen, R. (1991). "Teams for Tomorrow." Proceedings of the Hawaii International Conference on System Sciences. IEEE Computer Society Press, pp. 521-534.

Kiesler, S., J. Siegel, and T.W. McGuire. (1984). "Social Psychological Aspects of Computer-Mediated Commu- nication," American Psychologist 39(10), 1123-1134.

Kraut, R.E., J. Galegher, and C. Egido. ( 1987-1988). "Relationships and Tasks in Scientific Research Collabora- tion:' Human Computer Interaction, 3, 31-58.

Kutsko, J., and J.Y. Smith. (1991). "Effectiveness Measures for Distributed Teams Using Electronic Meeting Technology: The Larson/Lafasto Instrument?' In Proceedings of the Twenty-Fourth Hawaii International Con- ference on System Sciences, Vol. 3. IEEE Computer Society Press, pp. 458-470.

Latan6, B. (1981). "Psychology of Social Impact:' American Psychologist 36, 343-356. Lewis, L.F. (1987). "~ Decision Support System for Face-to-Face Groups" Journal oflnformation Science 13,

211-219. Lim, EJ., and I. Benbasat. (1991). ' ~ Communication Based Framework for Group Interfaces in Computer

Supported Collaboration." In Proceedings of the Hawaii International Conference on System Sciences, Vol. 3. IEEE Computer Society Press, pp. 610-620.

Losada, M., P. Sanchez, and E. Noble. (1990). "Collaborative Technology and Group Process Feedback: Their Impact on Interactive Sequences in Meetings?' In Proceedings of the Conference on Computer-Supported Cooperative Work, pp. 53-64.

McGrath, J.E. (1991). "Time, Interaction, and Performance (TIP): A Theory of Groups?' Small Group Research 22(2), 147-174.


McGrath, J.E. (1984). Groups: Interaction and Performance. Englewood Cliffs, NJ: Prentice Hall. Orlikowski, W.J. ( 1991 ). "The Duality of Technology: Rethinking the Concept of Technology in Organizations.'"

Center for Information Systems Research Working Paper No. 219. Poole, M.S., and G. DeSanctis (1989). "Understanding the Use of Group Decision Support Systems: The Theory

of Adaptive Structuration" In J. Fulk and C. Steinfield (eds.), Organizations and Communication Technology. Newbury Park, CA: Sage Publications, pp. 173-193.

Poole, M.S., and R. Lind. (1991). "The Process of Using Group Decision Support Systems as a Key Influence on Communication Processes and Outcomes." Paper submitted to the Human Technology Interest Group for presentation at the International Communication Association conference.

Rice, R.E., and G. Love. (1987). "Electronic Emotion: Socio-Emotional Content in a Computer-Mediated Com- munication Network," Communication Research 14(1), 85-108.

Seashore, S.E. (1954). Group Cohesiveness in the Industrial Work Group. Ann Arbor, MI: University of Michigan Press.

Siegel, J., V. Dubrovsky, S. Kiesler, and T.W. McGuire. (1986). "Group Processes in Computer-Mediated Com- munication," Organizational Behavior and Human Decision Processes 37, 157-187.

Short, J., E. Williams, and B. Christie. (1976). The Social Psychology of Telecommunications. London: John Wiley & Sons.

Smith, J., and M.T. Vanecek. (1989). ' ~ Nonsimultaneous Computer Conference as a Component of Group Decision Support Systems." In Proceedings of the Twenty-Second Hawaii International Conference on System Sciences, Vol. 3. IEEE Computer Society Press, pp. 370-377.

Trevino, L.K., R.L. Daft, and R.H. Lengel. (1990). "Understanding Managers' Media Choices: A Symbolic Interactionist Perspective." In J. Fulk and C. Steinfield (eds.), Organizations and Communication Technology. Newbury Park, CA: Sage Publications. pp. 173-193.

Walther, J.B. (1992a). ' ~ Longitudinal Experiment on Relational Tone in Computer-Mediated and Face-to-Face Interaeton." In Proceedings of the Hawaii International Conference on System Sciences, Vol. 4. IEEE Com- puter Society Press, pp. 220-231.

Walther, J.B. (1992b). "Interpersonal Effects in Computer-Mediated Interaction: A Relational Perspective" Com- munication Research 19(1), 52-90.

Walther, J.B., and J.K. Burgoon. (1992). "Relational Communication in Computer-Mediated Interaction," Human Communication Research 19(1), 50-88.

Watson, R.T., G. DeSanctis, and M.S. Poole. (1988). "Using a GSS to Facilitate Group Consensus: Some Intended and Unintended Consequences." MIS Quarterly 12(3), 463-478.

Zigurs, I., M.S. Poole, and G.L. DeSanctis. (1988). ' ~ Study of Influence in Computer-Mediated Group Deci- sion Making" MIS Quarterly (December), 625-644.

Documents

Developmental differences between distributed and face-to-face groups in electronically supported meeting environments: An exploratory investigation