Group Decision Support Systems Research Paper

Groups engage in a wide variety of activities as they make decisions and solve problems, including information sharing, problem identification and formulation, problem analysis, criteria development, solution identification and development, solution evaluation and selection, and implementation planning. Other group processes, such as leadership and facilitation, planning, project management, conflict and conflict management, creativity, negotiation, role definition, socializing, and team development, also contribute to effective decision making and problem solving. Most GDSSs have been designed to support decision making, problem solving, facilitation, and project management but most also incorporate features that support other activities.

For example, GROOVE™, a popular online tool set that can be used as a GDSS, provides a shared whiteboard that can be used for group drawing, which contributes to creativity (Griffith & Sawyer, 2006) but is also for playing around with. Because they provide general support for groups, these systems have also been referred to as electronic meeting systems (EMSs) and group support systems (GSSs). In this research paper, we will focus primarily on their use in decisionmaking activities as defined above.

GDSSs are employed in a wide variety of groups for general decision making, strategic planning, product design, and quality improvement, among other uses. GDSSs can be used by groups that meet face-to-face (FTF) or by virtual teams. A typical GDSS meeting starts with an agenda outlining the meeting and indicating the features of the GDSS that will be used in various steps. For example, a group might (1) brainstorm ideas using an idea generation tool; (2) discuss the ideas; (3) consolidate similar ideas to reduce the list using a categorization tool, which puts similar ideas under a common category; (4) vote on ideas using a voting tool; and (5) make a final decision based on the vote. Note that only Steps 1, 3, and 4 require the use of the GDSS. In Steps 2 and 5, the group engages in free discussion. A GDSS is not intended to replace existing modes of group communication. Rather, it is intended to support and encourage verbal and nonverbal interaction, as well as to provide additional channels for communication and decision support. Typically, members work at the GDSS episodically, using it for a time, then interacting without it, then coming back, and so on.

The potential of GDSSs lies in their ability to augment group information processing and analysis and to facilitate communication during decision making. GDSSs may also increase the salience of procedures to group members, thus promoting their beneficial effects. For example, if a GDSS prompts members to “Enter ideas into the brainstorming tool,” member activity is synchronized and attention is focused on this step of the decision-making process. By increasing the salience of procedures, GDSSs may also help educate groups and create an awareness of the importance of systematic approaches.

Historical Development of Group Decision Support Systems

Douglas Englebart and colleagues at Xerox Parc, early developers of many interactive computer technologies, demonstrated the use of computers for collaborative work in the early 1970s. Many of the earliest applications of networked computers were motivated by a desire to support group work. Murray Turoff used the precursor of the Internet to conduct Delphi sessions involving dozens of authorities in the development of future scenarios. The development of computer conferencing was also driven in part by the goal of supporting group decision making and deliberation (Hiltz & Turoff, 2000).

The group decision support system grew out of research and development on decision support systems (DSSs). The concept of DSSs was given an early articulation in Little’s (1970) Models and Managers: The Concept of a Decision Calculus. Keen and Scott Morton (1978) provided the first broad behavioral orientation to decision support system analysis, design, implementation, evaluation, and development. The original DSS concept was most clearly defined by Gorry and Scott Morton (1989), who developed a framework based on organizational activities related to various types of decisions. They motivated DSS not on the basis of technology but on the basis of decision-making activities, viewing information systems as support tools for decision making.

Freyenfeld (1984) developed a six-category classification of DSS components. He viewed DSSs as an interactive user-friendly data-processing and display system that uses terminology familiar to the user and with its selective features helps the user avoid information overload. His six categories were (1) chief executive information systems, (2) commercial operational analysis and planning systems, (3) industrial operational analysis and planning systems, (4) preference determination systems, (5) cognitive mapping systems, and (6) expert advisory systems.

In the early 1980s, a number of researchers applied DSSs in group settings, and by 1986, the idea of developing DSSs into dedicated systems specifically designed for groups gained currency. Articles by Huber (1984) and DeSanctis and Gallupe (1987) defined the parameters of the growing field.

By the early 1990s, a number of GDSSs had been developed, mostly stand-alone systems that required special servers (Bostrom, Watson, & Kinney, 1992). In recent years, GDSSs have moved to the Internet. The number of available GDSS systems peaked in the early 1990s and has markedly decreased since about 1995. At present, ThinkTank™ and Facilitate.com™ are the two major commercially available GDSSs. However, general-purpose meeting and conference systems such as WebEx™ and GROOVE™ are sufficiently powerful and flexible so that GDSSs can be improvised within them by a knowledgeable user.

Group Decision Support Systems Configurations

Configurations for GDSSs vary in terms of the physical layout of the space in which the technology is used, in terms of the features the system incorporates, and in terms of the technological platform the GDSS runs on. The physical layout of GDSSs is a factor because human groups take up space and the arrangement of meeting areas influences how groups operate. The spatial arrangement of groups is strongly dependent on their size. Small groups of three to seven members can easily arrange themselves so that everyone can see and communicate with everyone else. The arrangement of larger groups must be more carefully designed if all members are to participate. Of course, information and communication technology means that groups need not convene FTF. Dispersed or virtual teams are quite common, and they add an additional spatial dimension.

DeSanctis and Galllupe’s (1987) typology of GDSSs distinguished physical layouts based on their size and whether group members were colocated or distributed. The local area decision network supports small distributed groups, typically in the same office building and working on the same project or task. Members typically work in their own spaces and may interact via text messages, audio links, or video links. The computer-mediated conference supports larger groups whose members are dispersed but must work on common tasks. Several modes of communication are available for computer-mediated conferences, but there is also often a need for tools or leaders to facilitate participation (e.g., to form a queue to speak). Small, FTF groups meet in the decision room. A typical design for these rooms has members seated around a U-shaped table with a common display at the front of the group and a workstation for each member. The legislative session supports larger colocated groups. Spaces for legislative sessions must be carefully designed and often include break-out rooms in addition to a large general-assembly room.

The DeSanctis and Gallupe typology was very fruitful for GDSS research and development but needs to be expanded in at least two respects. First, the colocated and fully dispersed groups represent extreme cases. Many virtual groups combine both colocated and dispersed aspects in that subsets of members may be colocated in different places (e.g., two members in one office in city Z, one member in city X, and three members in one office in city Q). Second, the spread of wireless communication has altered the nature of GDSSs (and all group support tools) in several respects. First, unlike the dispersed members in the DeSanctis-Gallupe typology, members of today’s dispersed groups need not be fixed to computers hardwired into offices or homes but can roam freely. Second, portable computers and wireless networks make decision rooms and even legislative conference centers “portable.” Even a GDSS for a colocated group no longer needs to be fixed in place.

The second aspect of GDSS configurations, after their physical layout, is their features. A wide range of procedures may be incorporated into a GDSS, including agenda setting; idea generation; commenting on ideas; categorizing ideas; evaluating ideas through voting, rating, or ranking; minute taking; and more complex procedures such as multicriteria decision analysis, stakeholder analysis, and synectics. De Vreede, Briggs, and Kolfschoten (2006) have recently described a number of “thinklets,” basic modules into which many procedures can be decomposed, which promise to be useful in GDSS design. Other features may be built into the GDSS as well, such as a “group” display screen that displays common group information (e.g, lists of ideas or vote tabulations; this supplements the traditional flip chart or chalk board); a shared whiteboard for group drawing; links to project management tools, calendars, and databases; and communication media such as instant messaging, audio links, and video links.

The final aspect of GDSS configurations is the technological platform that they run on.

Most GDSSs rely on computer workstations, fixed or portable, although some specialized tools such as “clickers” (remote push-button voting or rating devices) may be used with large groups. Early GDSSs were implemented as specialized software packages that had to be purchased and installed on a network. Today, GDSSs are also available via the Internet and can be rented as well as purchased from application service providers such as GroupSystems™, which developed ThinkTank™.

Theories of Group Decision Support Systems Impacts

Research on GDSSs has been guided by several theoretical frameworks. The initial work, similar to early work on DSSs, consisted of taxonomies and structural descriptions that defined the major dimensions of GDSSs. Particularly noteworthy is a highly cited framework by DeSanctis and Gallupe (1987) that distinguished different levels of support GDSSs could offer, four contexts of use (outlined above), and a number of possible functions GDSSs could serve in groups. Huber (1984) wrote a prescient paper on issues in the design of GDSSs, in which he argued that GDSSs should be designed around the activities of information sharing and use, particularly textual and relational information, rather than centering on numerical analysis, as most DSSs do. He also argued that decisions sufficiently significant to warrant a GDSS do not occur very often. For these reasons, he predicted that it will be more difficult to convince organizations to procure and implement specialized, dedicated GDSSs than if the GDSS were offered on a service or rental basis. Finally, Huber argued that dedicated support personnel and a degree of familiarity with the management of groups would be necessary for GDSSs to succeed.

Subsequently, several theories of GDSS effectiveness were advanced. Hiltz’s (1988) systems contingency approach posited that the impacts of GDSSs on productivity and performance were contingent on “characteristics of the higher-level systems within which the technology is used” (p. 1440). For instance, GDSS’s impacts depend on compatibility with the demands of the organization, which may vary among organizational subunits.

Nunamaker, Dennis, Valacich, Vogel, & George (1991; see also Dennis, George, Jessup, Nunamaker, & Vogel, 1988) developed a high-level input-process-output model that describes the major influences a GDSS (and electronic meeting systems in general) can have on group processes and outcomes. The factors influencing GDSS outcomes are group characteristics (size, member characteristics, cohesiveness, etc.), task characteristics (task type, task complexity, etc.), contextual characteristics (organizational culture, incentives and reward systems, etc.), and the specific GDSS technologies in use. These authors argued that a GDSS improved the quality of group decisions by minimizing “process losses,” such as restrictions on member participation and free riding, while maximizing “process gains,” such as the ability of members to catch each others’ errors and build on one another’s ideas.

Poole and DeSanctis (1990, 2004; see also DeSanctis & Poole, 1994) proposed adaptive structuration theory (AST), which was intended to resolve some of the conflicting results gathered from empirical research on GDSSs. AST factors the way in which groups use the GDSS into the equation. The basis of the model comes from a distinction between system and structure made by the British social theorist Anthony Giddens (1984). The system is the observable behavior of the group using the GDSS, whereas structures are unobservable (but definable by inference), “rules and resources actors use to generate and support this system” (Poole & DeSanctis, 1990, p. 179). Structures come into play in the process of structuration, members’generative use of structures in generating and maintaining the system. Poole and DeSanctis (1990) argue that GDSSs embed structures underlying the procedures that are built into them (e.g., a brainstorming module) and the resources they provide for the group (e.g., the facility for all members to enter ideas into the brainstorming model at once). In the case of brainstorming, the structures include the rules for brainstorming (e.g., no criticism of ideas) at the feature level. Another type of structural element in a GDSS is the overarching set of values it is designed to promote, such as creativity or participative decision making, which is called the “spirit” of the structural set.

Structuration of group decision making occurs as members use the GDSS, thus appropriating various structures from the GDSS to enable and constrain their work. Each group does this differently. Some groups, for example, might use a voting procedure as a straw poll to get an idea of their members’ opinions, whereas others might use the same procedure to browbeat the minority into accepting the will of the majority. The different modes in which groups develop “structures-inuse” account for the range of processes and outcomes that occur in groups using GDSSs. Unlike contingency theories and most input-process-output frameworks, which presume that a set of variables causes or accounts for outcomes, the AST advocates a “softer” determinism, in which the GDSS may enable and constrain the group but actors have considerable control over how it is used and hence over group processes and outcomes. The AST expects there to be a considerable degree of unpredictability in processes and outcomes promoted by GDSSs.

Gavish and Kalvenes (1996) advanced an economic model of GDSSs that considers GDSS use by groups to be in the format of a search problem with a substantial solution space. According to this model, which solutions are feasible depends on the cost of performing the search versus the payoff of the solution. One important parameter in their model is group size, which has complex effects on the size of the solution space and other aspects of the model.

Dimensions of Group Decision Support Systems

To characterize GDSSs for purposes of research and development, it is important to identify the basic dimensions underlying their configurations. This enables researchers to move past the very large number of possible designs and focus on fundamental distinctions. One important dimension is synchrony, the degree to which members must use the GDSS at the same time versus the capability to use the GDSS at different times. In synchronous work, members must be coordinated more tightly than in asynchronous work, placing greater demands on the group. A second dimension along which GDSSs vary is distribution, the degree to which members can use the system if they are in different locations. The synchrony and distribution dimensions represent the two dimensions of time and place in the DeSanctis and Gallupe (1987) typology of GDSS configurations.

GDSSs in general and their features may also vary in the degree to which they constrain user behavior, that is, in their restrictiveness (Silver, 1988). Some tools are highly restrictive because they require users to follow closely specified rules about how to use the tool and do not offer members much freedom to adapt them. An example of this would be a multi-attribute decisionmaking tool that requires users to enter options, enter criteria next, and then rate the options on the criteria, after which it provides a graphical display of results. In this case, the user behavior must follow the “script” provided by the GDSS. Other tools, such as a virtual whiteboard, are low in restrictiveness because they present a “blank slate” to the group that members can use as they wish.

GDSSs differ in terms of the level of support their features offer. DeSanctis and Gallupe (1987) distinguished three levels of GDSS support, each of which incorporates the level below it and adds additional capabilities. Level 1 GDSS features support more efficient and effective communication among group members. Examples of Level 1 features are instant messaging, idea generation, evaluation techniques such as voting or rating, and minute taking. Idea generation, for instance, enables members to simultaneously enter ideas into the system, which are then displayed for the group to view, greatly increasing the speed with which the group can generate ideas and the pool of ideas they generate. The group could also list ideas “manually” by going around the group and having members state their ideas, which are then written on a blackboard. Level 1 features essentially enhance and improve traditional modes of communication and add the ability to communicate across distance.

Level 2 GDSS features provide decision support tools, such as multicriteria decision making, stakeholder analysis, and creativity techniques. These more complex tools are added on top of the Level 1 tools. A Level 2 GDSS supports activities that members could not accomplish unaided and brings advanced decision models into meetings. A multicriteria decision-analysis tool, for instance, would enable members to define evaluation criteria for a set of options, then enter weights reflecting the importance of the criteria to them, and then rate each option on the criteria. The GDSS would then calculate weighted averages of the ratings for each option, in a mathematical process that would take considerable time if done manually. The system might then provide numerical and graphical displays of the results for the group as a whole, with indications of the variability in ratings and weights for each item. This would enable the group to determine points of disagreement, which they could then discuss and resolve. Level 2 features fulfill the promise of the GDSS to improve group decision making through more sophisticated decision analysis.

Level 3 features seek to overcome a major barrier to the effective utilization of Level 1 and Level 2 GDSS features, the members’ lack of knowledge about procedures and how to employ them. It takes time to learn procedures, organize meeting agendas that incorporate them, and facilitate the group’s use of them. In the absence of a trained facilitator or consultant, a GDSS can present a steep learning curve to busy leaders and group members. Level 3 features attempt to flatten out the learning curve by providing guidance for the group through tools such as automated facilitation and expert systems that advise the group on strategies and approaches for making the decision. Some Level 3 systems are currently being developed, but this level of support remains some time away from general use. A final dimension on which GDSS features vary is parallelism, the degree to which members can communicate or work with the system simultaneously. Electronic brainstorming, for instance, is high in parallelism because all members can enter ideas at the same time. A public display of results of a stakeholder analysis is typically lower in parallelism because members must view it and discuss it together. Parallelism enables GDSSs to overcome the human limitations of discussion in FTF groups, in which only one or two members can hold the floor at the same time. It promotes gains in efficiency for groups in tasks that require individual inputs and not simultaneous discussion.

Key Findings of Research on Group Decision Support Systems

There have been a number of reviews and meta-analyses of GDSS research (Dennis & Wixom, 2001–2002; DeSanctis, Poole, Zigurs, & Associates, 2008; Fjermestad & Hiltz, 1998–1999, 2001; McLeod, 1992, 1996; Rains, 2005; Scott, 1999). The results of these reviews are for the most part consistent, so specific works will not be cited in this discussion unless there is a unique finding.

More than 50 studies have compared GDSS groups with FTF groups that did not use a GDSS (though some of them did use procedures manually). In terms of group outcomes, GDSS groups attain higher levels of group effectiveness than do FTF groups, when effectiveness is measured in terms of decision quality, quantity of ideas generated and considered, quality of ideas, confidence in the decision, depth of evaluation, and commitment to results. On the other hand, GDSS groups are less efficient than FTF groups in terms of time to decision and ease of decision making but more efficient on tasks that could be done in parallel (such as idea generation) or complex decision-making tasks (such as multicriteria decision analysis). Members of GDSS groups are less satisfied with their decision and decision process than are members of FTF groups. Members often reported that the GDSSs were difficult to use. Finally, GDSS groups have more trouble coming to a consensus and face more conflict than do FTF groups.

There are several moderating variables for results on outcomes. When the level of GDSS features is taken into consideration, most studies find that FTF groups are no different from groups using Level 1 GDSS features in terms of effectiveness but that Level 2 GDSS groups outperform both Level 1 and FTF groups. Groups with a Level 2 GDSS also attain a consensus better and manage conflict more effectively than do groups with a Level 1 GDSS; Level 2 GDSS groups showed no differences from FTF groups on these variables. However, FTF groups are still more satisfied with their group and decision process than were users of either level of GDSS.

Distribution of the group is a second important moderator. Outcomes are markedly better for groups meeting FTF in a decision room than in the small number of studies that have thus far been conducted on distributed GDSS groups. This is a concern in view of the migration of these technologies to the Internet.

Task also moderates outcomes. Outcomes for GDSS groups are superior to those of FTF groups for idea generation tasks and for complex decision and planning tasks. These types of tasks can benefit from the technology regarding idea generation because the GDSS allows fast parallel entry of ideas by all members and complex tasks since GDSSs can help groups carry out complicated decision analysis and decision support procedures. For simpler, more straightforward decision or judgment tasks, the technology does not provide as much benefit.

In terms of the decision-making process, the reviews suggest that GDSS and FTF groups are about equal in terms of process gains and losses, with both process gains and losses for both technologies being shown across a wide range of studies. Nevertheless, the GDSS has positive effects on structuration processes in comparison with FTF groups (DeSanctis et al., in press; Fjermestad & Hiltz, 1998–1999). The guidance provided by the GDSS appears to help the groups use procedures more faithfully and effectively. Rains’s (2005) meta-analysis found that GDSS groups also have greater equality of participation, influence equality, and experience less member dominance than do FTF groups (though, see Fjermestad & Hiltz, 1998– 1999, who report no difference on the basis of a simple count of results across studies). Level of GDSS also appears to moderate their effects on group process, with Level 2 GDSS features leading to more positive group processes than do Level 1 GDSS features. Using a GDSS also increases the degree of task-oriented behavior compared with FTF groups.

Facilitation and leadership have strong effects on GDSS process and outcomes. GDSS groups that are facilitated have a much greater likelihood of positive outcomes than do groups with no facilitator. Dennis and Wixom (2001–2002) found that facilitated groups have higher decision quality and greater satisfaction with their process than groups working without a facilitator. There is some evidence that a flexible form of facilitation is more effective than rigid, tightly scripted facilitation (DeSanctis et al., in press).

Reviews that compared field studies of GDSS with laboratory studies report more favorable results for the field implementations compared with the lab for some outcomes. Field studies regularly report that GDSS groups are more efficient than regular FTF groups, a finding inconsistent with the laboratory studies. This may be because the groups in most field studies are engaged in longer-term, more open tasks, whereas most laboratory tasks are kept relatively simple to fit into the time allotted for the experimental session. Hence, there is more room for improvement over common meeting practices in the field than in the lab. Field studies also typically report higher levels of user satisfaction with the GDSS than is measured in laboratory studies. This may be due to the care with which GDSS groups in the field are facilitated and supported. In most field studies, the researchers have a vested interest in having the GDSS succeed because if the group does not do well, GDSS use is discontinued and the study stops. So researchers take special care to ensure that the groups using the GDSS have the best possible experience. In those field studies that report unsuccessful implementations of GDSSs, the factors leading to negative outcomes include lack of facilitation or poor facilitation, little training or support, inflexibility in applying the GDSS, and a domineering or unenthusiastic manager or group leader.

Finally, we should note that the effects of GDSS procedures depend on how they are implemented in ongoing groups. Differences in results for field and lab studies, for instance, appear to stem from the greater level of auxiliary support given to groups in the field than in the lab. So, too, are there differences in how individual groups use the GDSS. Many GDSS features are complex and require some guidance or practice to use properly. Members do not always understand the rationale behind GDSS procedures and existing group norms, or organizational culture may conflict with the spirit of the GDSS. As groups adapt the GDSS to their own goals and norms, they may change the procedure in subtle yet important ways. For example, when a GDSS designed for U.S. users was employed in an experiment in Singapore, several groups did not use the voting procedure to indicate individual preferences but instead had the leader vote first and only then entered all their votes (in agreement with the leader). This was consistent with the collectivistic culture of Singapore but led to considerable changes in the impacts of the GDSS. The AST was developed to track and account for the impacts of appropriations by individuals and groups on the effects of GDSSs.

Diffusion and Implementation

The reviews cited in the previous section identified several critical success factors in implementing GDSSs. One important factor is training. In addition to “nuts and bolts” training in using GDSS features, users must be trained in the general philosophy behind the GDSS, the tasks it is best suited for, and how to design agendas. Without the “big picture,” users cannot understand what the features are for. A second critical success factor is facilitation. As noted in the previous section, effective facilitation increases the effectiveness of GDSS use. The facilitator not only serves as a resource for the group but also becomes a technology advocate and often identifies and cultivates “champions” for the system among members.

Third, it is important to tailor the GDSS and how it is used to the local context. The GDSS must be used in ways acceptable to its users and consistent with the overall culture of the organization. Nevertheless, it is also important to recognize that GDSSs are often introduced to change current organizational practices. There is a fine balance between fit with the organization and realizing the benefits of the GDSS. Finally, it is important that members experience success with the GDSS. If the group can use the GDSS effectively early on and see concrete benefits, it is more likely to continue to use it and be willing to put the requisite effort into learning the system. For this reason, it is important to employ the GDSS only for tasks it is well suited for and to ensure that members are trained and that there is an effective facilitator. Members who have had good experiences with the GDSS are likely to become ambassadors to the rest of the organization.

Despite these suggestions, GDSSs have achieved at most a moderate level of diffusion and utilization. Many organizations installed GDSSs in decision rooms during the late 1980s and early 1990s, when they received a good deal of attention in the business press, and there were notable success cases. For example, the Internal Revenue Service used a GDSS to implement a quality improvement program in one of its regional offices (DeSanctis, Poole, Desharnais, & Lewis, 1991). The oil company Texaco built several elaborate decision rooms equipped with multisite video conferencing and GDSSs to enable distributed meetings (DeSanctis, Poole, Dickson, & Jackson, 1993). Nevertheless, over the long term, GDSSs have tended to be underused, and many decision rooms have since been repurposed or dismantled.

There were several reasons for this. As Huber (1984) presciently observed, organizing and managing GDSSs takes a good deal of time and energy, and therefore they are used primarily for special-purpose, important tasks. Hence, decision rooms often remained unused for considerable periods of time, creating a temptation to appropriate them for other purposes. Second, most users did not have the requisite expertise in working in groups or in using procedures that the GDSS provided. The steep learning curve for the technology, mentioned earlier, served as a barrier to utilization. Third, because of this barrier, facilitators were necessary to promote utilization of the GDSS. Nevertheless, facilitators were expensive and were often the first to go when the organization experienced a downturn.

The result was a dip in demand for GDSSs. Only a few commercial GDSSs are available at the time of writing, and most meeting support or conferencing systems offer only Level 1 tools. This does not mean, however, that the GDSS should be written off. Videoconferencing endured two previous periods of experimentation and failure before its successful diffusion in Web 2.0. The next section discusses new developments that may improve prospects for the GDSS.

Conclusion and Future Directions

The GDSS began as an extension of the decision support system and has developed into a distinctive technology organized around the demands of group processes. Supporting group communication, problem solving, and deliberation requires not only decision models but also a design that supports social interaction processes, promoting those that contribute to group effectiveness and counteracting those that degrade decision making. Over the past 20 years, distinctive configurations of GDSSs have evolved around different arrangements of groups in time and space, and other dimensions underlying GDSSs have been defined, including restrictiveness, level of support, and parallelism. Research has shown that GDSSs can improve group processes and outcomes, but mixed results suggest that their effectiveness depends on a number of other factors. These include level of GDSS employed, distribution, task, and facilitation. There is no simple deterministic model of GDSS impacts. Rather, GDSS impacts depend on how the system is used and on how groups appropriate it as they work.

At least two trends seem likely to influence the future of the GDSS. First, the growth of Internet applications for groups and the concomitant increase in virtual groups has created a demand for distributed GDSSs, and such systems are now available commercially. Indeed, although GDSSs in decision rooms have not been a great success, a second generation of distributed GDSSs, which can be leased or purchased from application service providers, offer a promising future. Nevertheless, much less is known about distributed GDSSs than about GDSS configurations designed for FTF meetings, and current research suggests a disadvantage for distributed GDSSs compared with those used in colocated groups. Whereas Level 1 features such as idea generation and voting are straightforward to implement and have been incorporated into many conferencing tools, higher-level features represent a challenge. Procedures such as multi-attribute utility analysis and stakeholder analysis are typically run in colocated groups with facilitators, and there are questions as to how effectively they can be implemented at a distance. The communication problems, conflicts, and coordination difficulties faced by virtual teams pose severe challenges for distributed GDSSs.

As Fjermestad and Hiltz’s (1998–1999, 2001) review shows, studies of FTF GDSSs outnumber those of distributed GDSSs by a large margin. It is not clear to what extent results from the former apply to the latter. Will distributed GDSSs be more efficient than FTF groups with or without a GDSS? Will Level 2 features of distributed GDSSs garner the gains observed with Level 2 features in FTF groups using GDSSs? How satisfied will users of distributed GDSSs be?

Another set of questions revolves around whether GDSSs can enable virtual teams to meet their challenges. Virtual teams often suffer due to a lack of common context, which fosters misunderstanding and negative attributions and can set harmful social identity dynamics into motion (Poole & Zhang, 2005). A distributed GDSS provides an online space that would give virtual teams a context to interact within. Once established, this common context could help virtual team members to communicate more effectively and understand one another, reducing negative attributions and conflict. In essence, the GDSS could function as a boundary object linking members in different locations. Virtual teams also seem to benefit from an initial focus on their task, which is often the most important thing that binds members into a group. A distributed GDSS provides an organized environment for working on the group’s task and thus should promote early development of the group. Social relationships tend to follow task accomplishment in virtual teams, and a GDSS that enables the virtual team to make clear progress can also be a vehicle for relationship development. The second trend that has the potential to influence GDSSs is the rapid advance of information technologies. As we have noted, a major barrier to effective implementation and use of GDSSs is a lack of skill and knowledge on the part of the potential user. As a result, facilitators are needed to help plan and run sessions at present. Facilitation is particularly important for more complex procedures such as multi-attribute utility analysis. Nevertheless, a facilitator represents a considerable overhead for the organization that employs him or her and for the meeting convenor or leader, who must work with her or him to plan the session. One way of minimizing this problem is to develop Level 3 GDSSs that automate guidance and facilitation. These systems would bootstrap meeting planning and facilitation, making GDSSs more accessible to users. At present, Level 3 GDSSs primarily operate in terms of help systems and agendas that describe what the group should do (e.g., Limayem & DeSanctis, 2000). Nevertheless, advances in agent-based and intelligent systems could enable the implementation of Level 3 systems that are considerably more active and responsive than current models. A Level 3 GDSS, for example, might also learn the characteristics and tendencies of a group, allowing it to tailor its advice to the users’ level of sophistication. A more active Level 3 system might also give advice “on the fly” in response to events in the system or group and interact with the group.

Another way in which advances in information technology may shape GDSSs is through enabling development of novel, more natural interfaces for distributed meetings. Current GDSSs, especially distributed GDSSs, offer a fairly sterile and restrictive mode of interacting, quite unlike the immediacy of FTF meetings. New technologies coming online, including improved videoconferencing, tele-immersive environments that bring virtual reality to telecommunication, and haptic interfaces that operate through touch and motion, may enable the construction of meeting environments that are less intrusive and are easier to use, and promote a sense of presence far exceeding that yielded by current configurations. This may mitigate some of the challenges distributed GDSSs face. Web 2.0 technologies may also enhance the GDSS experience. Rather than meeting with other disembodied members via text, audio, or video conferencing, groups may meet in venues such as Second Life™ and interact via avatars. These and other new technologies may enable new, more compelling forms of distributed meetings.

The next 10 years will be a critical time for GDSSs. New and developing technologies and applications will determine whether they remain a useful but somewhat marginal tool or gain widespread acceptance and use.

Bibliography:

1. Alter, S. L. (1980). Decision support systems: Current practices and continuing challenges. Reading, MA: Addison-Wesley.
2. Anthony, R. N. (1965). Planning and control systems: A framework for analysis. Boston: Harvard University Graduate School of Business Administration.
3. Bostrom, R., Watson, R. T., & Kinney, S. (Eds.). (1992). Computer augmented teamwork: A guided tour. New York: Van Nostrand Reinhold.
4. Dennis, A. R., George, J. F., Jessup, L. M., Nunamaker, J. F., & Vogel, D. R. (1988). Information technology to support electronic meetings. MIS Quarterly, XX, 591–624.
5. Dennis, A. R., & Wixom, B. H. (2001–2002). Investigating moderators of group support systems use with metaanalysis. Journal of Management Information Systems, 18, 235–257.
6. DeSanctis, G., & Gallupe, R. B. (1987). A foundation for the study of group decision support systems. Management Science, 33, 589–609.
7. DeSanctis, G., & Poole, M. S. (1994). Capturing the complexity in advanced technology use: Adaptive structuration theory. Organization Science, 5(2), 121–147.
8. DeSanctis, G., Poole, M. S., Desharnais, G., & Lewis, H. (1991). Using computing to facilitate the quality improvement process: The IRS-Minnesota Project. Interfaces, 21(6), 23–36.
9. DeSanctis, G., Poole, M. S., Dickson, G. W., & Jackson, B. M. (1993). An interpretive analysis of team use of group technologies. Journal of Organizational Computing, 3(1), 1–29.
10. DeSanctis, G., Poole, M. S., Zigurs, I., & Associates. (2008). The Minnesota GDSS research project: Group support systems, group processes, and group outcomes. Journal of the Association of Information Systems, 9, 551–608.
11. Fjermestad, J., & Hiltz, S. R. (1998–1999). An assessment of group support systems experimental research: Methodology and results. Journal of Management Information Systems, 15, 7–150.
12. Fjermestad, J., & Hiltz, S. R. (2001). Group support systems: A descriptive evaluation of case and field studies. Journal of Management Information Systems, 17, 115–160.
13. Freyenfeld, W. A. (1984). Decision support systems. Manchester, UK: NCC.
14. Gavish, B., & Kalvenes, J. (1996). Economic issues in group decision support systems. In H. Pirkul & M. Shaw (Eds.), Proceedings of thefirstINFORMSconferenceoninformationsystemsandtechnology (pp. 18–27). Washington, DC: INFORMS.
15. Giddens, A. (1984). The constitution of society. Berkeley: University of California Press.
16. Gorry, G. A., & Morton, G. A.(1989). A framework for management systems. MIT Sloan Management Review, 30(3), 49–61.
17. Gorry, G. A., & Scott-Morton, M. S. (1971). A framework for management information systems. Sloan Management Review, 13(1), 55–70.
18. Griffith, T. L., & Sawyer, J. E. (2006). Supporting technologies and organizational practices for the transfer of knowledge in virtual environments. Group Decision and Negotiation, 15(4), 407–423.
19. Hiltz, S. R. (1988). Productivity enhancement from computermediated communications: A systems contingency approach. Communications of the ACM, 31, 1438–1454.
20. Hiltz, S. R., & Turoff, M. (2000). The network nation: Human communication via computer (2nd ed.). Cambridge: MIT Press.
21. Huber, G. P. (1984). Issues in the design of group decision support systems. MIS Quarterly, 8, 195–204.
22. Keen,P.G.,&ScottMorton,M.S.(1978).Decisionsupportsystems: An organizational perspective. Reading, MA:Addison-Wesley.
23. Limayem, M., & DeSanctis, G. (2000). Providing decisional guidance for multicriteria decision making in groups. Information Systems Research, 11(4), 386–401.
24. Little, J. D. (1970). Models and managers: The concept of a decision calculus. Management Science, 16, 466–485.
25. McLeod, P. L. (1992). An assessment of the experimental literature on electronic support of group work: Results of a metaanalysis. Human-Computer Interaction, 7, 257–280.
26. McLeod, P. L. (1996). New communication technologies for group decision making: Toward an integrative framework. In R. Y. Hirokawa & M. S. Poole (Eds.), Communication and group decision making (2nd ed., pp. 426–461). Thousand Oaks, CA: Sage.
27. Nunamaker, J. F., Dennis, A. R., Valacich, J. S., Vogel, D. R., & George, J. F. (1991). Electronic meeting systems to support group work. Communications of the ACM, 34(7), 40–61.
28. Poole,M.S.,&DeSanctis,G.(1990).Understandingtheuseofgroup decision support systems: The theory of adaptive structuration. In J. Fulk & C. Steinfield (Eds.), Organizations and communication technology (pp. 175–195). Newbury Park, CA: Sage.
29. Poole, M. S., & DeSanctis, G. (2004). Structuration theory in information systems research: Methods and controversies. In M. E. Whitman & A. B. Woszczynski (Eds.), The handbook of information systems research (pp. 206–249). Hershey, PA: Idea Group.
30. Poole, M. S. & DeSanctis, G. (in press). Applied research on group decision support systems: The Minnesota GDSS project. In L. Frey & K. Cissna (Eds.), Handbook of applied communication research. Thousand Oaks, CA: Sage.
31. Poole, M. S., & Zhang, H. (2005).Virtual teams. In S.A. Wheelan (Ed.), The handbook of group research and practice (pp. 363–384). Thousand Oaks, CA: Sage.
32. Rains, C. R. (2005). Leveling the organizational playing field— virtually: A meta-analysis of experimental research assessing the impact of group support system use on member influence behaviors. Communication Research, 32, 193–234.
33. Scott, C. R. (1999). Communication technology and group communication. In L. Frey, D. Gouran, & M. S. Poole (Eds.), The handbook of group communication theory and research (pp. 432–472). Thousand Oaks, CA: Sage.
34. Silver, M. S. (1988). User perceptions of decision support system restrictiveness: An experiment. Journal of Management Information Systems, 5(1), 51–65.
35. Simon, H. A. (1960). The new science of management decision. New York: Harper & Row.
36. de Vreede, G. J., Briggs, R. O., & Kolfschoten, G. L. (2006). ThinkLets:A pattern language for facilitated and practitionerguided collaboration processes. International Journal of Computer Applications in Technology, 25, 140–154.