[acb-hsp] Support groups: Processes

[J.Rayl]

The Minnesota Gdss Research Project: Group Support
 Systems, Group Processes, and Outcomes*
by Gerardine Desanctis , Marshall Scott Poole , Ilze Zigurs , George Desharnais ,
Marianne D'Onofrio , Brent Gallupe , Michael Holmes , Brad Jackson , Michele Jackson
, Howard Lewis , Moez Limayem , Jooeng Lee-Partridge , Fred Niederman , V. Sambamurthy
, Chelley Vician , Richard Watson , Julie Billingsley , Laurie Kirsch , Rebecca Lind
, Dale Shannon
1.0 Introduction
In 1985 a group of scholars at the University of Minnesota undertook an extended
program of research on group decision support
 systems (GDSSs), also known as group
support
 systems (GSSs).1 A GDSS is a computer-based technology designed to help committees,
project teams, and other small
groups
 with activities such as problem identification and analysis, decision making, planning,
creativity, conflict management, negotiation, and meeting management (Gray, 1987;
Bostrom, Watson, and Kinney, 1992; Jessup and Valacich, 1992; Poole, 2002). GDSSs
combine communication, information, and decision
support
 technologies in an integrated environment. Depending on their specific features,
GDSSs can
support face-to-face or distributed groups and single session or long-term groups
 (DeSanctis and Gallupe, 1987).
In the late 1980s, driven by advances in computer and telecommunications technology,
there was an upsurge of interest in groupware, ranging from presentation software
to computer conferencing, to GDSSs (Johansen, 1987). The rise of interest in teams
and networked organizations led academics and managers alike to search for novel
tools that could make group processes more effective. Today this phenomenon is evidenced
in interest in collaboration technologies, virtual teams, and online communities.
GDSSs are an important part of the mix of collaborative technologies, as noted in
a Gartner report that argued such systems will be critical to the effectiveness of
web-based conferencing and collaboration tools (Austin, Drakos and Mann, 2006).
This article analyzes and describes the evolution and results of the Minnesota GDSS
Research Project 20 years after its conception. During this period, the project produced
29 refereed articles, 34 book chapters and proceedings publications, 11 doctoral
dissertations, and 19 unpublished reports (see http://hdl.handle.net/2142/5350 for
a complete listing of these). The project influenced subsequent research not only
as doctoral students graduated and established their own lines of inquiry but in
other venues as researchers built on the ideas that came out of the project. It is
useful to integrate findings across the various studies, because these findings provide
an in-depth exploration of a theoretical model that specifies key dimensions of GDSS
design and effects and the group processes that mediate these effects. Many of the
findings included in this summary have not been previously published, offering additional
illumination to the published studies. The process by which the project evolved is
also worthy of examination, as an example of how the interplay among people and ideas
in an interdisciplinary team drives a programmatic effort.
Several extensive reviews of research on GDSS and groupware exist, most notably Fjermestad
and Hiltz (1998-99); see also McLeod (1996) and Scott (1999). This article is not
a broad review in those traditions, but rather a deep consideration of one line of
research organized around a unified theoretical framework, similar and comparable
GDSS technology, and a uniform experimental protocol. The studies in the Minnesota
GDSS program built on one another, and their results were cumulative because constructs
were consistently defined and operationalized. The program developed through an extended
"conversation" among researchers in which questions raised by earlier studies became
the focus of subsequent research. GDSSs are complex technologies, and using the same
system over numerous studies in the lab and field also facilitated studying many
aspects of the system and replication of various features, which enhanced the likelihood
of meaningful results that informed theory and practice. In advocating the advantages
of a unified theoretical model and a consistent technology, it is also important
to acknowledge their limitations, which we consider in the discussion.
The article is structured as follows: Section 2 lays out the conceptual framework
for the project, recounting the motivation for the project and the theoretical model
of Adaptive Structuration Theory that emerged from it. Section 3 describes the GDSS
laboratory and the GDSS technology used in the studies, Software Aided Meeting Management.
Section 4 then summarizes a series of laboratory investigations that tested and developed
the model. Section 5 reviews two major field studies undertaken to test the lab results
in business settings and to further develop the theoretical model.
The article concludes with a discussion of the project, a consideration of its strengths
and weaknesses, and an outline of future directions for research. The article is
organized primarily around the theoretical model and the key questions driving the
research. For a more detailed account of the history of this project, how it was
organized and developed over time, factors that contributed to its success, and challenges
it faced, see Poole and DeSanctis (in press).
2.0 Conceptual Framework
2.1 Motivation for the Project
DeSanctis and Poole met as assistant professors at the University of Minnesota in
1986. Though they were from different disciplines, they realized that they had complementary
interests and trajectories of research. Along with Gary Dickson, Gallupe, Watson,
Zigurs, and Sambamurthy, DeSanctis and Poole began to formulate a theoretical and
empirical framework.
At the time, information systems scholars were trying to develop decision support
 systems to structure group problem solving, with the goal of improving decision
efficiency and effectiveness. Classical behavioral decision theory posited that if
decision makers were given tools and techniques to overcome known biases and dysfunctions
in human decision processes-such as the tendency to make decisions on the basis of
anecdotal examples rather than thorough analysis-decision making might be improved
(Simon, 1997). In the group context, this leap was proving extraordinarily difficult
due to (a) the challenge of developing technology that could accommodate multiparty
participation in the decision process and (b) the inadequacy of existing theory to
predict or explain technology effects. In many arenas, information technology (IT)
was not providing the advantages for which technologists hoped.
Starting in the early 1980s, DeSanctis, Gallupe, and Dickson (DeSanctis and Gallupe,
1985; 1987) began developing a theoretical basis for group decision
support. A key issue they faced was that early on in the study of group decision
support
, researchers realized that GDSS technology would not necessarily bring about the
advantages intended by designers. Fresh theoretical perspectives were needed to spur
innovative technology design, as well as to understand the technology implementation
process. They had already explored the literature on
groups in social psychology and communication and were incorporating some of its
insights into their conceptualization of GDSSs.
Poole had moved to the University of Minnesota in 1985 and began working with DeSanctis
and her information systems colleagues in 1986. A central question of Poole's previous
research was how to explain variations in group decision processes and resultant
outcomes, even when
groups
 had similar resources and attempted to utilize the same decision techniques. Poole's
early research had shown that the then-dominant "stage theories" of group decision
making, in which task
groups
 were said to pass through a set series of stages in making decisions-for example,
orientation, problem definition, solution generation, and choice-generally did not
apply, even in controlled laboratory settings. More complex and dynamic theoretical
approaches were needed to advance understanding of basic group communication processes.
Structuration theory, developed in sociology to explain the evolution of large-scale
social phenomena (Giddens, 1979), was being developed and refined by Poole, Seibold,
McPhee (1985, 1986) and their colleagues in order to illuminate small group phenomena,
such as the development of decisions and argumentation and influence in
groups
. Research conducted within this framework emphasized direct observation of group
interaction as the best way to map group processes and to determine the influences
upon it.
Effective group decision support
 was a significant issue in contemporary society in the mid-1980s. Driven by advances
in computer and telecommunications technology, there was an explosion of interest
in "groupware," ranging from presentation software to computer conferencing to group
support
 systems (Johansen, 1988). The growth of the team-based quality movement and interest
in team-oriented management practices led academics and organizational leaders alike
to search for novel tools that could make these processes more effective. The emergence
of local and wide-area network technologies and the Internet, and the need for integration
of information within and across organizational boundaries as joint ventures and
alliances proliferated, also made it possible and necessary for distributed
groups
 to work over networks and emphasized the need for tools to help these groups
 to work effectively. Thus, organizations had a significant interest in implementing
advanced IT and making it successful.
An integration of the information systems and communication approaches promised to
yield insights into several important questions, and DeSanctis and Gallupe (1987)
developed a conceptual paper crystallize some of the early concepts and helped to
set the agenda for the first phases of the research. The key questions at this point
included:
* What effects do GDSSs have on group processes and outcomes? The hypothesis that
technology would improve outcomes had received uneven
support
 in early GDSS studies. DeSanctis and Gallupe (1987) articulated several avenues
through which GDSSs should improve group outcomes. Comparisons of
groups using a GDSS to similar groups
 without a GDSS were designed to test the impacts of GDSSs on outcomes. Analysis
of whether the avenues defined by DeSanctis and Gallupe were actually taken by
groups
 as they used the GDSS attempted to identify mediating group processes that led to
effective and ineffective use of the GDSS.
* What meaningful dimensions underlie the design of GDSSs and how do these dimensions
affect group processes and outcomes? Answering this question required going beyond
lists of GDSS features to more fundamental constructs in the design of GDSSs and
groupware. DeSanctis and Gallupe had defined different levels of
support
 that could be designed into GDSSs. Research by Poole and colleagues suggested some
additional features of group procedures that might underlie GDSS effects. Studies
of GDSSs varying along these dimensions attempted to determine differences due to
their designs.
* What additional types of support
 can facilitate GDSS use? As with any information system or group procedure, using
a GDSS is a learned skill. Training, assistance from facilitators or leaders, and
other interventions may be used to assist
groups
 to learn how to use the GDSS for best effect and to encourage them to utilize available
tools. Studies of various types of interventions sought to clarify which ones made
a difference and how best to deliver them.
* How does the process of using a GDSS mediate its impacts on group processes and
outcomes? The mixed results of GDSS studies strongly implied that GDSSs did not have
deterministic effects. The researchers sought to develop a model of the emergent
impacts of GDSSs along the lines defined by Markus and Robey (1988). This theory
was designed to take human agency into account as a source of variation in
groups.
These questions reflected both scholarly and practical aspirations. The project attempted
to develop a theory and empirical evidence that addressed the four questions. The
researchers suspected that the research would bridge causal and interpretive approaches,
and so they set out to develop a theory that stretched the bounds of current social
scientific work in information systems and communication by combining causal and
social constructionist accounts in a common model. This was done by layering functional
and constitutive explanations (Poole and DeSanctis, 2004), and the research reported
here shows the layered, incremental analysis that was required to do this. The researchers
hoped to be able to get past the either-or thinking that characterized-and still
does characterize-much social scientific research. The development of Adaptive Structuration
Theory and subsequent analyses were central to bridging these perspectives.
On the practical side, the project was designed to develop tools that could help
groups function more effectively. Research on decision
support
, group processes, and information systems has much to offer practitioners, and embodying
this research in a system that directly engages
groups was an appealing way to move from scholarship to practice.
2.2 Theoretical Framework: Adaptive Structuration Theory
Figure 1 portrayed the basic model that ultimately guided the project. The main components
of the model are represented by the boxes. The specific dimensions or factors within
the model, listed within each box, evolved over the course of the project. The essentials
of the model were worked out in 1986, but the model did not explicitly guide research
design until 1988. Studies by Gallupe, DeSanctis and Dickson (1988) and Watson, DeSanctis
and Poole (1988) were not based on this framework, although videotapes from Watson
et al. (1988) were used in later structuration-based analyses. The model was originally
developed for a grant proposal to the National Science Foundation in 1987 and was
largely fleshed out by 1990 (Poole and DeSanctis, 1989; Poole and DeSanctis, 1990).
Its fullest statement was in DeSanctis and Poole (1994).
The model represents a specific instantiation of structuration theory in the context
of technology use in groups
 and organizations that was termed Adaptive Structuration Theory (AST) (Poole and
DeSanctis, 1990; DeSanctis and Poole, 1994). AST argues that the effects of GDSSs
on group processes and outcomes depend on the design of technology structures and
on the emergent (adaptive) structures that form in the group as members interact
with the technology over time. Thus, from the design perspective, one can identify
and develop structural capabilities that are likely to bring about desired group
interaction processes. But outcomes of GDSS use depend on how these structures are
brought into interaction, how they blend with other available structures in the work
environment, and how the structures are effectively "redesigned" by the group in
the course of their use for specific purposes. AST provides a dynamic view of GDSS
technology and group interaction, focusing on the emergence of new social orders
through active use of technology structures.
Procedures and GDSS features embody what Giddens (1979) termed structures, rules
and resources that actors use to generate, organize, and sustain social systems such
as
groups
 or organizations. A GDSS presents a group with an array of potential structures
to draw down into its work. The research (DeSanctis and Poole, 1994; Poole and DeSanctis,
1990 and 1992) distinguished two aspects of technological structures: the spirit,
the general values and attitudes the technology is built around (such as democratic
decision making) and the specific features built into the system (such as anonymous
input of ideas or a stakeholder analysis procedure).
The spirit is the principle of coherence that holds a set of rules and resources
together, the general intent of the GDSS as reflected in its design and how it is
implemented. Spirit can be described in terms of a general set of values or principles
for the system, and it suggests general patterns or strategies for using the GDSS,
how to interpret its features, and how to fill in gaps in procedure that are not
explicitly specified.
The features of a GDSS are designed to promote its spirit, but they are functionally
independent and may be used in ways contrary to the spirit. Usually the features
of an information technology like a group decision
support
 system are designed to promote its spirit, although in some cases features may be
inconsistent due to sloppy design. Information technologies also differ in the degree
to which their spirit is coherent; lack of reflection by designers, changes in designers
over the course of the technology's development, and misunderstanding of the spirit
by implementers may all result in an ambiguous or less coherent spirit.
Structuration is a process through which groups
 select, adapt, and develop their working structures from among those provided by
the GDSS. In general terms, structuration can be defined as the process by which
systems are produced and reproduced through group members' use of rules and resources
(Giddens, 1979; Poole, Seibold and McPhee, 1985). This definition is founded on several
additional assumptions and distinctions. First, structures are dualities; they are
used to produce and reproduce the group system and to do the group's work, but the
structures themselves are produced and reproduced through the group's activities.
Structures have no reality independent of the social practices they constitute. Hence,
when a group utilizes a GDSS voting procedure, it is employing the rules embodied
in the GDSS features and putting them into action-hence constituting structures in
the system of action-but it is also reminding itself that these rules exist, working
out a way of using the rules, perhaps creating a special version of them-in short,
the group is producing and reproducing its own version of the procedure for present
and future use. When the group uses the features the next time, it is likely to constitute
the structures based on its previous experience. So it is really somewhat misleading
to regard structures as static entities or to presume that the structural potential
embodied in a GDSS is equivalent to the structures that are used or usable by the
group. A voting feature does not, for all practical purposes, exist for a group if
the group never considers or employs it. A voting feature may, however, have salience
if the group considers it and deliberately chooses not to use it, since that deliberation
may affect subsequent experience.
Central to the structuration of group processes through GDSSs is the interplay between
the spirit of the technology and the specific features members use. A group develops
a reading of the spirit of the GDSS that is an important aspect of the meaning of
the GDSS to the group. Their initial take on spirit is shaped by how the system is
explained, members' prior beliefs about GDSSs and what they regard as similar systems,
and the group's sense of itself and its context. The group's reading of the spirit
of the GDSS influences its mode of appropriation of GDSS features. For example, if
the group perceives that the GDSS is intended to speed up its decision processes,
members might apply voting procedures in such a way that they rush on to their next
agenda item after taking a vote. In some cases, a group's reading of spirit may not
be consistent with how the designers and implementers of the GDSS present the GDSS.
For example, the designers and implementers may have attempted to stress the GDSS
as a vehicle for rational and careful decision making. However, some members of the
group may want the GDSS to be a tool to cut down on meeting time and add an emphasis
on speed. This new emphasis may be inconsistent with the structure of a system set
up to
support
 rational and careful decision making, setting up tensions between system capacities
and the uses to which they are put. Such cases are called ironic appropriations of
the GDSS because they turn its structures in ways contradictory to its intended spirit.
As Figure 1 shows, the group's interaction process is influenced by the group's appropriation
of structural elements embodied in the GDSS and other external sources. Three types
of inputs influence group processes: technological structures, other sources of structure,
and the group's internal interaction system. Technological structures provided by
the GDSS can be characterized in terms of feature configurations and the spirit of
the technology. Features delimit the specific structures embodied in the GDSS, such
as idea listing, voting, multi-criteria decision analysis, or note taking. The entire
ensemble or configuration of features can also be described in terms of two dimensions:
* Level of support. DeSanctis and Gallupe (1987) distinguished three levels of support
provided by GDSSs. Level 1 features provide support
 for enhanced communication among group members. These features include idea listing,
evaluation techniques such as voting or rating, and comment recording. Level 2 features
provide decision
support
 tools such as multi-criteria decision making, stakeholder analysis, and problem
formulation. A Level 2 GDSS supports activities that members could not undertake
on their own in a reasonable amount of time and enables members to utilize advanced
decision models and techniques. Level 3 features provide guidance for the group through
such tools as automated facilitation and expert systems that advise the group on
strategies and approaches for making the decision. They are intended to help members
determine which Level 1 and Level 2 procedures to use and to take the burden of learning
how to use the GDSS off members' shoulders.
* Restrictiveness of the system is the degree of freedom the user has in applying
the technology (Silver, 1988; Wheeler, 1996). Restrictiveness influences appropriation
of the GDSS: a very restrictive system must be used in a formulaic manner, which
tends to result in faithful use of the GDSS, but may be inflexible and difficult
to adapt to the situation. A less restrictive system leaves more room for users to
improvise and adapt, but it may also be used ironically.
In addition to information technology, groups
 can draw on the structural potential offered by other sources. One important source
of rules and resources is the group's task or work. Strategies for making decisions
or work procedures often must be melded with the procedures built into the GDSS in
order to adapt it to the task. Other aspects of work, such as looming deadlines or
specific types of information required to make the decision, also shape structuration
of the group decision process by placing constraints on how (and how much) the GDSS
can be used. A second source of rules and resources is the general environment of
the group, particularly the encompassing organization and members' previous experiences
with similar work, as well as the organization's culture - an important source of
norms that can be used to guide group work: General social norms such as reciprocity
or equity can also be imported into the group's interaction.
The internal system of the group, which refers to the nature of relationships among
members, typical interaction patterns, and individual and shared knowledge, is the
third influence on adaptive structuration after technological structure and other
sources of structure mentioned above. Specific aspects of the internal system that
influence appropriation include:
* Interaction styles, either those characteristics of the group as a whole, such
as group conflict management style (Kuhn and Poole, 2000), or of key members, such
as individual conflict management style, that will influence appropriations through
their impact on decisions about task strategies and employing the GDSS. For example,
a group that tends to engage in a lot of socializing will use the GDSS differently
than one that is mostly task-focused.
* Group norms that have been developed prior to using the GDSS will influence how
the system is appropriated and used. A group that has norms favoring very structured
meetings will most likely welcome the GDSS, whereas one whose norms favor low structure
may resist the GDSS. Often, preexisting norms must be melded with the structural
elements drawn from the GDSS. For example, Parliamentary Procedure might be combined
with the decision processes built into the GDSS.
* Members' degree of understanding of the technological features and structures will
affect how they use them. A member who knows the GDSS very well will use it differently
from a novice; he or she may educate the group or, alternatively, use the technology
to manipulate other members.
* Leadership will obviously influence how the GDSS is used. A directive leader will
have more influence on how the GDSS is appropriated than will a more laissez-faire
leader. In addition to the group's leadership, facilitators helping the group use
the technology will also exert an influence on GDSS use. In some cases, the facilitator
will be a critical source of leadership, especially when the group has hired an outside
facilitator to assist its deliberations.
As noted, the GDSS is appropriated into the group's interaction. Dimensions to characterize
appropriation of the GDSS include:
* Degree of appropriation - in terms of amount of use of the GDSS, where use is appropriated
and the number of structures appropriated.
* Appropriation moves - how the GDSS is appropriated, how members make sense of the
GDSS and its operations and potential, and how the GDSS is combined with other structures.
* Faithfulness of appropriation - overall, how consistent the appropriation of the
GDSS is with its spirit.
* Consensus on appropriation - the degree of agreement among members concerning how
the GDSS should be used and how it fits with the group's work.
* Instrumental use of the GDSS - the particular uses to which the group puts the
GDSS, including task, process, power, social, individualistic, and exploratory uses.
* Attitudes toward the technology - members' comfort with the GDSS, their respect
for the GDSS as useful, and the challenge the system poses for the group to work
hard and excel.
How the system is appropriated then influences the decision process, including the
number and quality of ideas generated, the level and balance in member participation,
how the group manages conflict, how members influence one another, and how the group
organizes its decision process.
The outcomes of the interaction process include results such as objective and perceived
quality of the decision, consensus on the decision, commitment to the decision, confidence
in the decision, satisfaction with the decision, and satisfaction with the process
by which the decision was made. New structures may also result, which influence subsequent
interaction. For example, following use of a brainstorming tool in a GDSS, a group
might decide to add a rule that it should generate multiple options before all decisions,
changing prior procedures. And outcomes of using the GDSS, such as a list of priorities,
can be resources that will be consulted in future decision making.
The theoretical model just described began as a looser framework that had all the
major components in Figure 1, but much less detail on specific constructs. As the
team applied for grants and conducted successive studies, the model was specified
and modified until it assumed its (more or less) final form.
3.0 The GDSS Laboratory
The GDSS system and lab evolved gradually over a number of years (see Dickson, Poole
and DeSanctis, 1992, for a more detailed summary).
Groups
 met around a rectangular table-and later around a specially built horseshoe-shaped
conference table-with a terminal and keyboard for each group member. Chairs swiveled
and had rolling feet, so users could move about comfortably to face one another.
A large monitor-later a projector and screen-at the front of the room displayed group
information (such as vote tallies or idea lists generated during the meeting). Two
video cameras recorded group interaction during GDSS sessions on a split screen.
Cameras were backed up by a stereo audio recording system in case the cameras malfunctioned.
Figure 2 shows the final configuration of the Software Aided Meeting Management (SAMM)
laboratory for a 10-person group.
The GDSS developed for this research, SAMM, was designed to be used by groups
 meeting synchronously in a decision room (DeSanctis, Sambamurthy, and Watson, 1987;
Dickson, Poole and DeSanctis, 1992). A menu-driven system, SAMM provided a group
with a range of procedural control options: members could control the system themselves,
or a facilitator or technician could help. SAMM used the UNIX operating system.
SAMM was designed to support participatory, democratic decision-making in three to
16 person groups
). The final set of SAMM features can be diagrammed as shown in Figure 3.
The SAMM system was purposely designed to fulfill a specific spirit, embodying the
following set of values: (a) participatory decision making guided by rational discussion;
(b) democratic, shared leadership; (c) efficient use of group resources; (d) confrontive,
constructive conflict management; and (e) an informal, safe climate for the group.
To assess whether the specific set of features that was developed did indeed reflect
these values, DeSanctis, Snyder and Poole (1994) conducted a functional evaluation
of SAMM that asked novice users to describe SAMM. Their responses indicated that
the design of SAMM and its features reflected the intended spirit to a good degree.
4.0 Laboratory Studies
Commencing in 1986, the research team conducted a series of experiments to investigate
the impacts of GDSSs on decision making, problem identification, and conflict management
processes and outcomes. These experiments manipulated features of the GDSS, facilitation,
and training and measured effects on a number of objective and subjective process
and outcome variables. The experiments utilized
groups
 of three to seven individuals in order to assess the effects of group size, which
generally were negligible. Unlike many experiments, there was no attempt to control
or tightly script the behavior of group members. Participants were granted a measure
of freedom to use the GDSS (or not use it) as they saw fit. The logic of AST proposed
differences in appropriation of the technology and procedures across
groups
, and the experimental
groups
 were given some leeway so that these differences would emerge. This approach enabled
the study of appropriation process and the factors that shaped it. All sessions were
videotaped for future analysis.
Analytical approach was as important to the project as the experimental design. The
researchers conducted three layers of analysis. First, a traditional causal analysis
focused on factors that influenced measurable outcomes. Outcomes were measured objectively-for
example, by assessing degree of consensus on the final decision through the use of
multiple coders and assessments of inter-coder reliability-and subjectively via questionnaires
that measured perceived quality of the decision and overall satisfaction with the
decision process. This analysis enabled the assessment of the causal effects of the
manipulated factors.
The second layer was analysis of the group process and assessment of how the manipulated
factors influenced processes and how processes affected outcomes. The theoretical
framework was based on the assumption that group interaction processes mediated the
effects of input factors on outputs. This assumption required the identification
of elements of group processes that specifically mediated GDSS effects. These elements
were measured objectively, by systematically coding the interaction using several
classification schemes, and subjectively via questionnaires that asked participants
for their perceptions of interaction. The coded data required secondary processing
to generate three types of meaningful information about the interaction process:
(1) distributional structure, the total number of acts in particular categories,
for example, the total number of statements that suggested or elaborated solutions;
(2) phase structure, which mapped the sequence of holistic episodes of activity in
which the group as a whole engaged, for example, orientation followed by idea generation
followed by idea evaluation; and (3) critical events, particular acts or occurrences
in the interaction that represented important points in the group process, for example,
open conflict or using the GDSS to take a vote. Each of these types of data was used
to investigate specific hypotheses or research questions about the group process
and the mediating effects of process on outcomes.
The third layer of analysis focused on the structuring process. AST assumed that
structuring processes influenced and were influenced by other aspects of the group
interaction process and also that structuration mediated the impact of the GDSS on
outcomes. To get a structuring process required identification of indicators in the
use of the technology in interaction, since structuration was not directly observable.
These processes were identified through coding appropriation of the GDSS features
with classification schemes described in DeSanctis and Poole (1994). One of these
schemes coded specific "appropriation moves," ways in which individual members applied
a feature in the group's work, and a second coded "instrumental uses," the specific
aspects of the group process (task, process, power) for which the GDSS was appropriated.
These codes were subjected to secondary analysis similar to that described in the
previous paragraph to characterize structuration of the technology. In addition,
a system was developed to code faithfulness of appropriation of the GDSS. The coding
system scored whether participants were using the features in a way consistent with
the spirit of the GDSS, yielding an overall faithfulness profile for each group.
Finally, subjective data on structuration were gathered through measuring participants'
attitudes toward the GDSS in terms of their perceived respect for, comfort with,
and challenge posed by the GDSS.
In going from the first to third layer of analysis, the focus moved to "deeper" levels
of the group process. Relationships among data gathered at each layer and between
objective and subjective data brought different domains of data together, increasing
confidence that results were not due to common method effects or experimental artifact.
4.1 Experimental Procedures
Although each experiment varied in the purpose and variables involved, the procedures
were generally consistent across the studies. Table 1 describes a typical experiment
in terms of sequence and nature of activities.
The researchers developed a script for each experiment, with detailed instructions
for each condition. The script ensured consistent administration of procedures across
experimenters and provided a documented record of the process for follow-up analysis.
All experiments were videotaped, which added video data to the questionnaire and
computer log data that was available. The set-up of the GDSS room varied from one
experiment to another as the technology and laboratory facilities evolved. One constant
in the set-up was that all participants had individual control over their access
to the GDSS, so that everyone had an equal opportunity to participate.
4.2 Tasks
A major advantage of a program of research is the ability to develop a set of experimental
tasks that can be refined and reused. Researchers developed several tasks over the
course of the laboratory studies, each of which had specific characteristics that
allowed for testing of different aspects of group process and outcomes. The tasks
can be characterized in the following broad categories: (a) problem identification
and formulation, (b) decision making with no verifiable solution; (c) decision making
with a verifiable solution; (d) creativity; (e) negotiation; and (f) planning. This
diverse array of tasks enabled us to sort out the influence of task on GDSS process
and outcomes. More detailed task descriptions are available in DeSanctis, Poole,
Limayem, and Johnson (1990).
Two different problem identification and formulation tasks were developed: the Marketing
Business Case and the Minnesota Merchandising Case. The Marketing Business Case had
two levels of difficulty, high and low. Group members had to identify problems in
the business case, and outcomes were measured in terms of quality of problem identification
and number of problem statements. The Minnesota Merchandising Case involved a company
that had a problem with the use of sensitive data, with issues ranging across legal,
ethical, strategic, personnel, and technical concerns. Group members were required
to discuss different points of view to come to a common formulation of the problem.
Outcomes measures were information search and equivocality reduction, coverage of
critical issues, group member reactions, and time - all of which formed a general
assessment of perceived problem formulation quality.
The Foundation Task was developed as a decision making task with no verifiable solution.
Groups
 were asked to allocate a given sum of money among six projects, each of which represented
different values. This task had high potential for conflict because value differences
were built into the choices, and the task had no verifiable outcome. Researchers
measured the outcome in terms of shift in consensus from pre-meeting to post-meeting
and satisfaction with decision outcome and process. This task was used repeatedly
across the studies, as it represented the most difficult type of real world task
for which the GDSS would be used.
DeSanctis and her colleagues developed two tasks as decision making tasks with a
verifiable solution: the Student Admissions Task and the Security Measures Task.
The Student Admissions Task required
groups
 to decide among different candidates for admission to a university. Each candidate
was described on certain criteria, which group members were told to use as the basis
for their decision. Prior to group discussion, participants were trained individually
on the relationship between the criteria and success of the university. Thus, the
task had a verifiable outcome, and the decision criteria were provided to group members
in advance. Performance was the key outcome measure. The second task in this category
- the Security Measures Task - required subjects to rank alternative methods for
improving campus security. The verifiable solution was based on the expert opinion
of campus security officers.
SAMM Lab researchers employed several creativity tasks, requiring brainstorming of
ideas about parking, tourism, cultural diversity, and campus security. All of the
creativity tasks had the same general format, namely, for each topic, group members
were asked to generate ideas about how that specific issue could be improved. Some
issues were in the context of a university (e.g., parking, campus security), while
others were in the context of the larger geographic region in which people lived
(e.g., tourism), but the format was consistent across issues. In all cases,
groups were prompted to develop ideas of both high quality and high quantity.
Finally, the Tidewater College Task served as the planning task, which involved stakeholder
analysis and option identification. The steps of this task required
groups
 to identify key stakeholders for a proposed project, identify their concerns and
assumptions, evaluate those concerns and assumptions in terms of their importance
to stakeholders and project success, analyze the concerns and assumptions to identify
the most important issues for planning, and then select an appropriate strategic
option. A planning task is especially relevant for Level 2 functionality within a
GDSS, since these types of functions can perform the complex calculations that are
required to combine individual members' evaluations in representations of group opinion.
4.3 Measurement
Two types of data were gathered for the lab studies: (1) objective and subjective
measures and (2) observational data through interaction coding. Details and scales
for many of these measures can be found in DeSanctis et al. (1990).
Constructs and Scales. Researchers employed a core set of scales across most studies,
adding special measures when appropriate for a given study. The constructs and how
they were measured are listed in Table 2, along with the source from which they were
developed.
Observation. Videotapes of experimental sessions allowed analysis of group interaction
processes. These analyses were useful in that they provided insight into how the
GDSS produced observed effects. Group process is the vehicle through which members
enact decisions, engage in conflicts, plan, and identify problems, among other things,
and observing how members used the GDSS and manual procedures gave insight into how
the technology was influencing group operations.
These studies of group process took both functional and constitutive approaches.
Some analyses focused on the functions of group interaction such as the task functions
involved in decision making (e.g., defining the decision problem, analyzing the problem,
or defining criteria). The results of these studies fit in with mainstream studies
of decision making, problem solving, and conflict management processes. A second
group of analyses focused on adaptive structuration in the constitution of decisions,
plans, and other group activities. These studies required methods for studying structuration
processes in interaction. The researchers developed methods to analyze structuration
at the micro level (act-to-act), the meso level (interaction episodes, phases), and
the global level (entire meetings, series of meetings). These methods were somewhat
involved, and we leave description of them for the section on structuration studies
below.
Due to the intensive work involved in interaction analysis, we were only able to
analyze a subset of groups
 from each experiment. We selected 10-15
groups
 per condition in a way that reflected the mean and range of outcomes in the original
samples for each condition in the design. We prepared and coded transcripts of the
meetings in multiple passes with various coding systems. What was coded was determined
by expectations regarding the impacts of GDSSs on group processes. For example, GDSSs
were expected to foster more organized decision processes, greater equality of influence,
and the consideration of more ideas. We developed a list of more than 15 possible
effects that the GDSS might have on the group interaction process and worked out
ways to identify them through interaction analysis. In some cases, we used established
coding systems, such as Putnam's (1981) Procedural Order coding system and Poole's
(1981) Decision Functions Coding System. In other cases, special procedures for coding
indicators of possible GDSS effects were developed. For example, to assess whether
GDSSs helped
groups
 generate and consider a greater number of ideas than manual systems, we developed
rules for counting ideas. Table 3 lists the various indicators of group process.
4.4 Results
Previous sections identified experimental procedures, tasks, and measurement that
were used for the laboratory studies. This section summarizes general findings of
the laboratory studies. We group results according to the core questions introduced
previously and present them in synoptic form. Of necessity, the results are summarized
at a high level of abstraction. The Appendices display more specific findings in
tabular format, and a detailed narrative of the studies and findings can be found
online at https://www.ideals.uiuc.edu/handle/2142/5349.
4.4.1 What Effects Do GDSSs Have on Group Processes and Outcomes?
The initial studies focused on differences between groups using the GDSS and groups
 employing more traditional modes of operation. Normatively, the studies were focused
on the question of whether there was any net improvement in decision outcomes such
as quality, satisfaction, and commitment due to GDSSs. To address these questions,
we compared three conditions: (1)
groups with no support
 that were given a task and left to their own devices (Baseline
groups), (2) groups
 with a manual version of the procedures built into the GDSS (Manual
groups), and (3) groups with a GDSS (GDSS groups
). The contrast of conditions 1 and 2 with condition 3 identified the effects due
to computerization, while the contrast of condition 1 with conditions 2 and 3 identified
the effects due to structured procedures, whether automated or not. These comparisons
enabled the sorting out of impacts due to procedures, which could be employed manually
as well as with the GDSS, from impacts due to computerization.
Laboratory experiments by Gallupe (1985), Watson, DeSanctis and Poole (1988), Zigurs
(1987; Zigurs, Poole, and DeSanctis, 1988), Sambamurthy and DeSanctis (1990), and
Niderman and DeSanctis (1995), and a field experiment by Niederman and Bryson (1998)
compared GDSS with Manual and Baseline
groups
 in terms of various outcome variables that included objective quality, consensus
change, satisfaction with the solution and the decision process, and confidence in
and commitment to the decision. Studies by Zigurs et al. (1988), Poole, Holmes and
DeSanctis (1991), Sambamurthy and Poole (1992), Poole, Holmes, Watson, and DeSanctis
(1993), and Poole and Holmes (1995) analyzed the interaction in subsets of
groups
 drawn from the three conditions in the Watson et al. (1988), Zigurs et al. (1988),
and Sambamurthy and DeSanctis (1990) studies. These follow-up studies compared GDSS,
Manual, and Baseline
groups
 in terms of amount and types of communication, nature of the decision process, quality
of discussion and analysis, and conflict management.
Table 1 in the Appendix presents the results of studies that compared GDSS groups
to traditional groups. Key findings include:
* Groups using a Level 1 GDSS generated higher quality solutions than Manual groups
 on a task with an objective performance measure when the task was high in complexity,
but not necessarily when the task was low in complexity.
* Groups using a Level 1 GDSS generated more ideas than Manual groups
 when the task was open-ended, but not when the task was closed and most options
were already on the table.
* Computerization made procedures easier to apply when the procedure did not easily
map onto the task. Manual groups
 had difficulty carrying out a procedure that was not a good fit to the task, whereas
GDSS-supported
groups
 were able to carry out the procedure and work around it to achieve higher quality
results.
* Both Level 1 GDSS and Manual groups attained higher levels of consensus change
than Baseline groups. GDSS groups
 achieved higher consensus than Manual
groups when preexisting disagreement was high, but not when it was low.
* In general, groups
 using a GDSS had lower levels of perceived quality, satisfaction with the decision
process, and (to some degree) confidence in the decision than
groups
 using Manual procedures. This finding held for two studies conducted in the lab,
but not for a field experiment (Niederman, 1990), which also found that
groups
 using a Level 2 GDSS had greater confidence and commitment to their problem formulation
than did Manual
groups.
* Groups
 using a GDSS devoted a large proportion of their time to procedural messages, suggesting
that understanding the system, deciding how to use it, addressing problems, and coordinating
use occupied a great deal of members' time. This represented "friction" that detracted
from immediate focus on the task. There was also evidence that use of a GDSS created
a higher level of understanding of procedures than was attained in Baseline or Manual
groups.
* Groups
 using Manual procedures devoted less time to discussing and organizing procedures
and more time to goal emphasis and substantive and critical discussion of ideas than
GDSS
groups.
* There were mixed results in terms of perceptions of control over the process and
influence in GDSS compared to Manual and Baseline
groups.
* Groups using a procedure had more organized and less complex decision processes
than did Baseline groups.
* Groups using a Level 1 GDSS deviated more from a normative problem-solution decision
sequence than did groups
 using the same agenda manually.
* Members of GDSS groups reported higher levels of conflict than did members of groups
 using manual procedures. Consistent with this perception,
groups using a GDSS engaged in more open conflict than groups using Manual procedures.
Level 2 GDSS groups
 confronted conflict and managed it in a more integrative fashion than did Manual
groups.
* Groups differed in how they used the GDSS. When groups
 used the GDSS to promote open discussion and participation, confronted conflict
in a constructive manner, and actively adapted it to the task,
groups
 could achieve levels of consensus, perceived quality, and decision scheme satisfaction
comparable to or better than effective Manual and Baseline
groups
. When they let the GDSS drive their activities and used it mechanically or when
they had problems with the GDSS, they achieved lower levels of consensus, perceived
quality, and decision scheme satisfaction.
Zigurs, DeSanctis and Billingsley (1991) conducted a longitudinal study of user acceptance
of the SAMM GDSS that shed some light on overall reactions to this system and to
GDSSs as a whole. They measured users' perceptions across eight meetings of eight
groups
 conducting class projects that had the potential to impact how a university department
marketed itself to students. The study measured users' perceptions of the quality
of the meeting, level of personal participation, negative socio-emotional behavior,
and informal leadership, as well as overall evaluation of group behavior, satisfaction
with the meeting, and satisfaction with the decision process. Three of these
groups
 (adopters) enthusiastically adopted the GDSS, while four (discarders) initially
adopted but greatly reduced their use after the third or fourth meeting, and one
group rejected the GDSS from the outset. There did not seem to be any differences
in the patterns of attitudes over time between the adopter and discarder
groups, but the rejecter group had less positive attitudes overall.
These findings suggested an interesting irony: GDSSs can improve group decision making,
but groups
 often do not like to use them. This was suggested in the first studies by Gallupe
(1985) and Watson et al. (1988), and this possibility was explored in several subsequent
studies. That GDSS
groups had more friction and less critical examination of ideas than Manual groups
 offered an explanation for the finding of lower levels of satisfaction with the
decision process for GDSS
groups compared to Manual groups.
It seemed possible that some of these results were due to the fact that the GDSS
required members to relearn how to conduct procedures that are natural to them-such
as idea listing and voting-via the computer, making the operations seem awkward and
imposing a learning curve. There was also more conflict in the GDSS
groups than the Manual groups
, which was probably a result of the display of ratings, which tends to highlight
differences among members, thereby steering the discussion toward points of disagreement.
The investigators speculated that GDSSs might not evoke such a negative reaction
when they embody procedures that would be difficult for
groups to do manually or with a Level 1 system, such as multi-criteria decision analysis.
In such a case, the value added by the GDSS support
 should be more evident. Hence, for this type of Level 2 procedure the GDSS might
outperform Manual procedures. This observation motivated a set of studies dedicated
to addressing a second question.
4.4.2 What Meaningful Dimensions Underlie the Design of GDSSs, and How Do These Dimensions
Affect Group Processes and Outcomes?
To address this question, researchers conducted studies that compared groups
 using three Level 2 procedures-Multi-criteria Decision Analysis, Stakeholder Analysis,
and Problem Formulation based on principles from Synectics-to
groups
 working on the same task using the problem solving agenda employed in the first
set of studies. The
groups
 worked on tasks appropriate to the procedures. The one exception was that they did
not include a manual control group, because it would have taken members too long
to conduct the same operations, and it seemed to be an inappropriate comparison.
Sambamurthy and DeSanctis (1990; Sambamurthy and Chin, 1994) studied Stakeholder
Analysis implemented in a Level 2 GDSS; Dickson, DeSanctis, Poole, and Limayem, (1991)
focused on Multicriteria Decision Analysis, and Niederman (1990; Niederman and DeSanctis,
1995) investigated Level 2 Problem Formulation procedures. Follow-up studies by Sambamurthy,
Poole, and Kelly (1993) and Sambamurthy and Poole (1992) examined the interaction
in Level 1 and Level 2 Stakeholder Analysis
groups
, while additional observations regarding group interaction were reported by Niederman
and Bryson (1998) and Dickson et al. (1990).
Table 2 in the Appendix summarizes results of these studies. Key findings include:
* Groups using Level 2 GDSS tools could achieve higher levels of consensus on decisionmaking
tasks than groups
 using a Level 1 GDSS when the
groups
 had a high degree of initial disagreement. This effect did not hold when there was
a low degree of initial disagreement among members.
* Level 2 GDSS procedures were more complex and challenging to use than Level 1 procedures
and could require additional external support
 such as training and facilitation to achieve their benefits.
* There were mixed results in terms of subjective reactions to the Level 2 GDSS.
In some cases reactions were more favorable than those to the Level 1 GDSS, but in
other cases, there were no differences.
* Level 2 GDSS groups engaged in deeper and more thorough analysis of solution options
than Level 1 groups.
* Groups using a Level 2 GDSS were more effective in managing conflict than groups
 using a Level 1 GDSS. Both types of GDSSs surfaced disagreements, but groups using
the Level 2 GDSS were able to resolve disagreements more effectively than groups
with a Level 1 GDSS.
* Groups using a Level 2 GDSS managed conflict differently from groups using manual
procedures. A Level 2 GDSS enabled groups to surface disagreements and manage them
effectively. Groups using Manual procedures tended to avoid open conflict; they used
low-key critical discussion to work out disagreements and make a final decision.
* Groups using Manual procedures managed conflict more effectively than the average
group with a Level 1 GDSS, but groups that employed the GDSS to foster discussion
of ideas and explore options could be just as effective as Manual groups in conflict
management.
* Control over and management of procedures tended to be less equal in GDSS groups
than in groups using Manual procedures.
The interesting result that Level 2 GDSS may lead to superior outcomes-but only under
the right conditions-suggested exploration of several measures that might be used
to create these "right conditions." This led to the third set of studies.
4.4.3 What Additional Types of Support Facilitate GDSS Use?
Various avenues of additional support
 were a logical concern with a complex technology like GDSS. The project investigated
the impact of altering the internal group system by utilizing external interventions
of heuristics, role training, facilitation, and Level 3 guidance. The studies in
this series parallel in an interesting way Dennis, Wixom, and Vandenberg's (2001)
appropriation
support tools.
DeSanctis, D'Onofrio, Sambamurthy, and Poole (1989) investigated the impact of heuristics-guides
to making decisions-on GDSS use and outcomes. Vician (1994, Vician and DeSanctis,
2000) studied the delivery of GDSS role training within self-managed groups for repetitive
and changed task situations. Lee-Partridge (1992; Dickson, Lee-Partridge and Robinson,
1993) focused on the effects of facilitation styles for level 2 GDSS sessions. Limayem
(1992, Limayem and DeSanctis, 2000) investigated the automation of facilitation within
the GDSS itself, leading to the development of a Level 3 GDSS.
Table 3 in the Appendix summarizes results of these studies. Key findings include:
* Groups required training and guidance to use the GDSS effectively. If this was
provided, members understood the GDSS better, and better outcomes ensued.
* Heuristics to guide use of GDSSs led to better outcomes if the heuristics were
not overly complex and gave groups a set of guidelines organized around a small set
of consistent principles.
* Heuristics that described a general approach to decision making (such as the consensus
approach) led to better outcomes than more specific heuristics that described the
use of the system in detail or that combined the general approach with specific descriptions
of how to carry it out. (Specific instructions on how to use the system were useful
in terms of implementation, however.)
* There was no difference due to training members of GDSS groups in fixed or rotating
roles vs. training in simply using the GDSS.
* Facilitated Level 2 GDSS groups had superior outcomes to Level 2 GDSS groups without
facilitators.
* Flexible Facilitation, which gave members some control over how the GDSS was used,
was more effective than Firm Facilitation, which compelled members to use the GDSS
as the facilitator specified. Both types of facilitation yielded better results than
no facilitation when the GDSS features were complex.
* A Level 3 GDSS enabled groups to achieve outcomes superior to those obtained with
a Level 2 GDSS through increasing members' understanding of the GDSS, which had a
positive impact on outcomes.
A key premise of the Minnesota GDSS project was that the impacts of GDSSs-positive
and negative- would be mediated by the ways in which they were structured by the
groups that used them. Hence, the next step was to study structuration in the GDSS
sessions.
4.4.4 How Does the Process of Using a GDSS Mediate Its Impacts on Group Processes
and Outcomes?: Adaptive Structuration of GDSS
A central aspect of the experimental design in these studies was that groups were
not marched "lockstep" through the procedures, but instead were given some leeway
to use the GDSS as they saw best. Even in the studies of guidance and facilitation,
groups were given room to vary their use of the GDSS. This practice created variation
in how the GDSS was appropriated by the groups and enabled the study of how groups
structured their process using the GDSS and how, in turn, outcomes were affected.
A negative side effect of this variation was that the positive impacts of the GDSS
on group process and outcomes were likely to be understated due to the fact that
some groups were left free to "misuse" or "underutilize" the system. However, the
benefits of being able to study how groups structured the system on their own outweighed
this loss.
Structuration processes were studied in several of the experiments summarized in
previous sections. The Minnesota researcher employed to analyze structuration: (1)
tests of within-cell variation in processes and outcomes; (2) direct analysis of
structuration; and (3) analysis of user attitudes related to structuration (challenge,
comfort, respect). We discuss each of these strategies in turn.
4.4.4.1 WithinCell Variation
If the impacts of the GDSS (and of manual procedures) on group outcomes were mediated
by their structuration during the group decision process, then groups should vary
in how they appropriated the GDSS, and these variations should relate to outcomes.
In terms of Analysis of Variance, this involved a test for within-cell variation
in effects, with the inference that these would be due to different appropriations.
These analyses were conducted by Poole, Holmes and DeSanctis (1991), Zigurs et al.
(1989), Sambamurthy and Poole (1992), Armstrong, Perez and Sambamurthy (1993), and
Limayem and DeSanctis (2000).
4.4.4.2 Direct Analysis of Structuration
Methods to study structuration directly focused primarily on group interaction, since
structuration is a collective process in which members work out appropriations as
they interact. Thus, it is a public process in which members must share. Structuration
was investigated at two levels. First, microlevel structuration was tracked through
analysis of interaction moves that appropriated, produced, and reproduced structures
(see, e.g., Poole and DeSanctis, 1992). Second, global patterns of structuration
that characterized the process across an entire decision or series of decisions were
identified through analysis of the phase sequences or general patterns of structuration
(e.g., Poole, DeSanctis, Kirsch, and Jackson, 1994). It is important to note also
that at least one more level influences structuration: macrolevel organizational
and societal discourse concerning appropriation of various structures that are commonly
shared across a population of groups or organizations. This level could not be investigated
in the laboratory studies, but some insights emerged from later field studies. The
researchers expected all three levels to interpenetrate and influence each other.
DeSanctis and her colleague developed three schemes for coding micro-level appropriations,
each of which yielded coded data that could be processed to generate global characterization
of patterns of structuration. The first scheme was an Appropriation Checklist of
actions that groups had to complete to carry out the agenda for a given experiment.
This checklist consisted of a list of the GDSS features that corresponded to steps
in the agenda and the correct way to carry out steps to enact the spirit of the SAMM
GDSS. For example, for the step "Evaluate Ideas," members would have to rate, rank,
or vote on the ideas as prescribed by the SAMM system and then discuss the results
of the evaluation prior to moving toward a decision. A checklist of the use of Level
1 SAMM for the Reflective Thinking Process is shown in Figure 4. Groups were scored
on how many of the correct actions they took for each feature of SAMM they used.
These scores were then converted to a proportion of correct steps they took over
the total number of correct actions they should have taken to yield a score for faithfulness
of appropriation that ranged from 0.0 to 1.0. This numerical appropriation score
could then be used in quantitative analyses. This checklist was employed in studies
by Poole, Lind, Watson and DeSanctis (1992) and Sambamurthy and Poole (1992). Limayem
et al. (2006) used the Chin, Gopal and Salisbury (1997) measure of faithfulness of
appropriation in a complementary study.
The second scheme drew on the systems of tropes used by rhetorical scholars to devise
an Appropriation Move Coding System (Poole and DeSanctis, 1992; DeSanctis and Poole,
1994). Tropes describe ways in which speakers or writers can use the structures of
language to achieve various effects and so suggest a variety of structuring "moves"
through which technology can be employed by users and groups. Based on lists of tropes
and analysis of the groups, researchers developed a typology of 37 appropriation
moves, organized into nine general categories. These categories were based on the
following distinctions: (1) Did the move involve a single structure or more than
one structure? and (2) Did the move consist of an active use of the structure, an
attempt to understand or clarify the structure, or a response to another member's
appropriation move? The category system is shown in Table 4.
Two categories coded moves that involved a single structure: direct appropriation
represented active use of the structure, while constraint represented an attempt
to interpret and understand the structure. For example, a direct appropriation of
a GDSS involved simply using some feature of the GDSS, while explaining how to use
the GDSS would be a constraint move, because the explanation would focus members'
attention on a particular interpretation of what the feature meant and how it should
be used. Four categories coded moves involving more than one structure. For example,
combination coded active uses of two or more structures. In a combination move, two
structures were melded in various ways, as when a group decided to use parliamentary
procedure to run meetings in which it used the GDSS; in this case, the rules of parliamentary
procedure were combined with those for using the GDSS. The enlargement category was
used to code moves in which two structures were likened to each other. In one study,
members of quality teams likened the GDSS they were using to a "secretary" and "coach,"
which added to the meaning of the GDSS and probably created expectations in members'
minds about what the GDSS could do for them.
Each of these appropriation moves had to be accepted by other members to influence
structuration of the GDSS, so responses to the moves were also coded. Affirmation
and negation represented the positive and negative modes of response to others' appropriations,
while ambiguity represented uncertainty and confusion in response to some structure.
Ironic appropriations could occur in constraint, enlargement, or contrast moves that
imposed definitions on the structure that were inconsistent with its features (e.g.,
"the voting procedure in this GDSS can be used to determine who agrees with the leader"),
or substitution or combination moves that put structures inconsistent with the spirit
together (e.g., the secretary metaphor created false expectations that the GDSS had
artificial intelligence, leading some members to reject the system when it did not
provide active suggestions).
In addition to coding the specific appropriation moves, the scheme also identified
the sources of structures, including the technology, the task, and outside norms.
So a technology structure might be combined with a task structure, resulting in an
appropriation of the GDSS that served to adapt the system to the group's task.
This scheme identified basic structuring moves and responses in group interaction.
Several analyses were generated from these codings: (1) profiles of the general types
of appropriations made by groups as well as which members made and controlled them,
(2) the phases of appropriation that occurred in the groups, (3) critical junctures
at which appropriation of the GDSS changed, (4) conflicts in the structuring process,
and (5) ironic (nonfaithful) uses of the GDSS.
Poole and DeSanctis (1992) conducted a follow-up study in which they used the Appropriation
Move Coding System to study structuration in 18 groups drawn from the Watson and
D'Onofrio studies. They sampled groups that varied in level of Restrictiveness of
structures and in terms of effectiveness, as measured by consensus change during
the discussion. Armstrong, Perez and Sambamurthy (1993) applied the system to study
appropriation in Sambamurthy's (1989) groups.
The third coding scheme, an Instrumental Use Coding Scheme identified the function
that the GDSS was appropriated to serve, divided into task, process, power, social,
individualistic, and exploratory uses. These uses were coded as a second track along
with the appropriation moves. This system was used to some degree in the lab research,
but served a more prominent role in the field studies that are summarized in the
next major section.
4.4.4.3 User Attitudes Related to Structuration
Researchers also developed scales for measuring attitudes related to appropriation.
These scales measured the group members' level of comfort with the technology, their
respect for the GDSS as a useful technology, and the degree to which they felt challenged
to do their best by the GDSS. Studies by Zigurs, Poole and DeSanctis (1988), Lee-Partridge
(1992), Limayem (1992), Vician (1994), DeSanctis et al. (1994), and Sambamurthy,
DeSanctis, and Poole (1995) utilized these scales.
Table 4 in the Appendix presents results from the various studies and modes of analysis.
Key findings include:
* There were differences in how groups appropriated the GDSS. Groups varied in terms
of degree of appropriation, how faithfully they appropriated the GDSS, the degree
of consensus on appropriation, how many members guided appropriation, and their attitudes
toward the technology.
* Faithful appropriation of the GDSS was positively related to consensus change.
The relationship between faithfulness and consensus change was mediated by group
interaction; to the extent that appropriation fostered interaction consistent with
the requirements of the task, it had a positive effect on consensus change.
* There was mixed evidence on the impact of appropriation on perceived quality, decision
scheme satisfaction, and confidence in the decision.
* On average, Level 2 groups appropriated the GDSS with greater faithfulness and
had less conflict over appropriations than Level 1 groups, though there was variability
of appropriation within each set of groups.
* A High Restrictiveness GDSS led to greater faithfulness than did a Low Restrictiveness
GDSS. A High Restrictiveness GDSS increased appropriations of the GDSS related to
substantive discussion. Groups using a Low Restrictive GDSS focused more on organizing
their decision process and use of the GDSS than groups with restrictive GDSSs.
* A group that effectively appropriated the GDSS:
(a) Used the GDSS only for components of the task that fit GDSS structures embodied
in its features and procedures;
(b) Organized ill-defined tasks around the GDSS structures, provided this did not
entail "force fitting" the task to the GDSS in such a way that goals or requirements
of the task were compromised;
(c) Clearly delegated a few members or a facilitator to guide appropriation of the
GDSS and use of its features and procedures;
(d) Focused on critical discussion and sense-making related to GDSS outputs rather
than simply accepting them as final results;
(e) Focused less on understanding and interpreting GDSS structures than on GDSS outputs
and the task; and
(f) Cultivated positive attitudes (comfort, respect, challenge) toward the GDSS among
members, but also encouraged members to take a critical approach to application of
the GDSS.
* Key junctures such as problems with the GDSS, conflicts, and transitions between
tasks or steps within tasks were particularly important occasions for structuration,
resulting in changes in the appropriation of the technology or confirmation of current
appropriations.
* Level 3 support
 helped groups deal with junctures and manage breakpoints. Groups with Level 3 GDSSs
typically had more positive attitudes toward the GDSS than did groups using Level
2 GDSSs.
These results complemented and enlarged upon the studies of group process functions.
In particular, the results suggest that how the group appropriated the GDSS influenced
outcomes independent of group interactions that occurred while using the GDSS. These
group interactions had a strong direct influence on outcomes, but appropriation influenced
the nature of group interactions and had modest direct effects on outcomes.
These lab studies divulged some important generalizations about the impacts that
GDSSs could have. However, the Minnesota researchers believed it was necessary to
examine GDSS use in organizations to determine which of these impacts actually held
in practice and how strongly they held.
5.0 Field Studies
Field studies began in the third and fourth years of the program. Two major field
sites were involved: the Internal Revenue Service (IRS) and Texaco Inc. Collaborators
at the field sites were interested in applying GDSSs in their organizations, which
provided the opportunity for longitudinal analysis of a variety of team processes
and activities. In both cases, the field sites provided the hardware and room facilities,
while the Minnesota group provided licensing and free
support
 for the SAMM software. In return, the field site gave researchers significant access
to the teams, with the opportunity to videotape team meetings and assess team member
perceptions through interviews and questionnaires.
The primary interest was in observing natural use of a GDSS over time, that is, to
examining the extent and nature of voluntary use of a GDSS in different types of
task and organizational contexts. Researchers worked with the field sites to implement
SAMM in areas where they identified a need. To the extent possible, they also made
improvements to the system based on on-going feedback from participants.
The initial study at the IRS consisted of an analysis of team member attitudes and
uses of the system, based on questionnaires and interviews (DeSanctis, Poole, Desharnais,
and Lewis, 1991). Subsequently, the research team analyzed videotapes to identify
structuration moves and patterns in the teams (DeSanctis, Poole, Lewis, and Desharnais,
1991-1992) and conducted an in-depth study of four teams over an extended period
of time (Poole, DeSanctis, Kirsch, and Jackson, 1994). Also examined were differences
in brainstorming sessions in teams that used the technology vs. those that did not
(Jackson and Poole, 2003).
The Texaco study also varied the type of analysis to bring out differences in the
teams' processes and use of technology. Appropriation analysis of one group of teams
identified differences in how well SAMM supported team processes (DeSanctis, Poole,
Dickson, and Jackson, 1993). An in-depth study of a single team that had surprising
success with SAMM provided insight into the role of the team leader, a continuous
learning process, and the use of different functions of the system (Vician, DeSanctis,
Poole, and Jackson, 1992). Finally, a longitudinal study of a larger number of teams
showed how teams and their technology use changed as the organization changed (DeSanctis,
Poole, and Dickson, 2000; DeSanctis and Jackson, 1994).
Finally, the researchers conducted some analysis across the two field sites that
resulted in the identification of global appropriation types and patterns of alignment
of task with technology (Poole, Jackson, Kirsch, and DeSanctis, 1998).
The sections that follow describe the field studies as a whole in terms of data gathering
procedures used, the different task contexts, the measures used, and the overall
results. In line with the research program, we applied a complex but consistent analytic
method throughout the field studies across a variety of teams and situations.
5.1 Data Gathering Procedures
researchers collected four main types of data: video, computer system use log, survey,
and interview data. Most team meetings were videotaped, and the videotapes were used
for analysis of appropriation moves. For example, to analyze structuration from the
videotapes of the team meetings, the researchers viewed a sample of the tapes for
the teams being studied. For each tape, they created a protocol of the sequence of
events, including a detailed description of meeting activities and observations about
specific interaction processes followed by the team. These protocols were used in
conjunction with the other data to develop an analysis of appropriation.
Computer system log data captured the features of SAMM that were used in each meeting.
This data allowed for a global measure of system use, as well as a characterization
of the level of use as Level 1 or Level 2 (DeSanctis and Gallupe, 1987).
Team members also agreed to fill out surveys on a periodic basis. In most cases,
team members filled out two surveys at the end of a meeting. The first survey measured
team members' perception of systematic process, openness of communication, and sense
of accomplishment. The second survey measured comfort with the GDSS technology and
perceived impact of the technology.
In addition, researchers conducted unstructured interviews with team leaders and
selected individual team members. The interviews typically included such topics as
individual roles and responsibilities, the nature of the team's tasks, perceptions
of the team's decision processes and overall progress, and reactions to the SAMM
technology.
5.2 Task Context
The IRS site was designed as a natural experiment in which teams would meet in the
GDSS room for all their meetings, but they could choose to use or not use the software
as they felt appropriate. The SAMM researchers studied 10 teams involved in organizational
quality processes over time, with team size varying from five to 10 members. Each
team had a facilitator who assisted in the application of both the quality process
and the GDSS. The task for these teams was to identify and solve problems that reduced
the quality of the agency's functioning and services. The researchers provided a
specialized agenda of quality techniques to the teams, including formats to
support
 problem definition, cause-effect analysis, and solution development. Multiple meetings
of the 10 teams were recorded. In addition, the sample included a number of staff
teams who conducted one to three meetings with the system.
The Texaco site consisted of two different sets of teams. The first set was three
teams that met over a period of eight months. The teams varied widely in their composition,
organizational level, and task. One team was composed of 14 high-level managers charged
with organizational planning. The second team had eight medium-level personnel who
provided
support
 for computer users. The third team had seven lower-level personnel who were designing
a procedure for automating database operations at night.
The second set of teams at Texaco consisted of 47 technical and administrative teams
who participated for a three-year period and were on-going teams involved in a variety
of tasks. Technical teams were likely to have more focus and less diffuse work tasks
than the administrative teams.
5.3 Measurement
Appropriation was assessed in some cases via the coding scheme that DeSanctis and
Poole (1994) had developed previously for use in the laboratory studies and that
continued to evolve as the studies were conducted. In other cases, appropriation
was assessed by survey questions. Table 5 shows appropriation measures that were
used in the field studies.
Outcome measures also varied across studies. Table 6 shows the outcome measures that
were used across the field studies, along with the source and validation of each
measure.
5. 4 Results
Key findings from the field studies include:
* There was more and better use of the GDSS when it was introduced in a newly formed
group, rather than an already established one. Existing problems or conflicts in
a group tended to carry over into its use of the GDSS, lessening the benefits groups
could derive from the GDSS.
* Use of Level 2 GDSS tools was higher among groups that had the GDSS introduced
early on than in those where it was introduced midstream. A Level 2 GDSS increased
group effectiveness when the group faced complex tasks and when there were disagreements
among members.
* Effective use of the GDSS depended on effective alignment among the system, the
group's tasks, and group norms and other structures. Alignment required a continuous
process of adjustment.
* Too much emphasis on internal group processes to the exclusion of work could lead
to ineffective appropriation of the GDSS.
* The GDSS facilitated more balanced participation by members, especially by quiet
or low power members.
* Groups using the GDSS did not generate more ideas than non-supported groups using
either a flipchart or paper to record ideas. All groups in the field, GDSS and non-GDSS,
generated fewer ideas than did laboratory groups. Idea generation was more limited
in the field groups in part because some topics naturally had limits on the number
of ideas that could be generated and also because the meeting served other purposes
than generating ideas, e.g., as a ritual signaling that the group was being creative.
* Conflict over the use of the GDSS did not necessarily reduce team effectiveness
or positive impacts of the GDSS on group processes and outcomes.
* Effective appropriation of the GDSS depended on emphasizing task and process uses
and on constraining power-related uses of the system to those that moved the group
toward its goals.
* Facilitation improved the effectiveness of GDSS use at all levels.
* Effective appropriation of the GDSS depended on a continuous learning process on
the part of all (or most) members. The leader could play an important role in guiding
the GDSS and encouraging members to use the system during the learning process.
* Effective use of the GDSS was most likely if the group became independent in the
use of the system, either functionally autonomous and able to manage the system itself,
or able to determine the procedures it wanted to use and to direct the facilitator
or resident expert as to how it wished to use them. The GDSS was used more and more
effectively when members initiated use themselves, rather than relying heavily on
the facilitator.
* It was important to maintain a balance between task and process orientation. Too
much emphasis on internal group process led to less effective use of the GDSS.
* Just as in the lab studies, key junctures in system use were very important. These
critical events could be negative, such as a group crisis or technology failure,
or they could be positive, such as success with the GDSS. At these junctures, members
engaged in interchanges that determined subsequent appropriation.
One example of a key juncture was provided by a facilitator (summarized):
A team was floundering for direction and decided to do a stakeholder analysis. There
was no facilitator who knew how to use the system. They followed instructions from
the user's guide and had no problems employing the procedure. The facilitator, who
had been tacitly anti- SAMM up to that point, told us that SAMM had turned the meeting
around.
Most of these conclusions are consistent with findings from the laboratory studies,
but at least one was not. Whereas the GDSS did not equalize participation in lab
studies, it did in the field studies.
This difference probably stems from measurement. In the lab, measures were based
on coded behaviors and tended to treat every act as though it had equal weight in
tapping participation, whereas the field studies tended to yield critical incidents
in which participation was increased.
5.5 GDSSs and Other Collaborative Technologies in the Organizational Context
Studies of SAMM use in organizational settings stimulated interest in the impacts
of groupware in the larger organizational context. The emergence of new forms of
organization, such as networked organizations and post-bureaucratic organizations,
had been described by many scholars (e.g., Fulk and DeSanctis, 1999). These forms
have evolved in response to rapid technological and economic changes that require
organizations to adapt and innovate much more quickly than they had to in the past.
DeSanctis and Poole (1997) hypothesized changes in the nature of teams in networked
organizations. They argued that in networked organizations there would be more teams
and that these teams would be more geographically dispersed, more diverse, and exhibit
a greater variety of structures than would teams in hierarchical organizations. Moreover,
teams in networked organizations would have more open boundaries, and their structures
would change more rapidly than those in hierarchical organizations. These teams would
be very communication intensive, relying less on formal structures and more on information
retrieval and information sharing systems in doing their work. DeSanctis and Poole
(1997) further hypothesized that participation would replace hierarchy as the key
mode of decision making, planning, and management. They believed that processes in
teams in networked organizations would also be much more dynamic and changeable than
group processes in hierarchical organizations.
DeSanctis and Colleagues found that information and communication technologies (ICTs)
are key enablers of the changes in teams that networked organization fosters. Dispersed,
diverse, open, participative, virtual teams must supplement face-to-face communications
with technologies such as email, teleconferencing, instant messaging, GDSSs, and
computer conferencing to maintain cohesion and manage their work. Utilizing a mix
of these technologies with face-to-face communication enables networked teams to
respond to conflicting pressures for integration of information flow and member inputs,
on the one hand, and fragmentation due to increased workload, diverse perspectives,
complex team structures, and multiple team memberships on the other. The more complex
the team's work and structure, the greater the benefit from "higher end" ICTs such
as GDSSs, which structure the work and facilitate negotiation and conflict management,
compared to "lower end" technologies such as e-mail or paper memos.
DeSanctis, Poole, and Dickson (2000) conducted a longitudinal study of 47 teams in
Texaco. Texaco was moving toward a hybrid organizational structure in which the networked
form was overlaid on the organization's existing hierarchy. DeSanctis et al.'s study
focused on how teams changed as the organization changed by assessing team use of
technologies and team characteristics at three points in time over a year and a half.
They found that use of groupware ("higher end" ICTs, including teleconferencing,
computer conferencing, and GDSSs) increased over the course of the study, while email
and traditional modes of communication (face-to-face meetings, telephone, fax, memos)
remained constant (see also DeSanctis and Jackson, 1994). This finding is consistent
with the hypothesis that teams in networked organizations will be more communication
intensive. They assessed the impact of structural variables-team size, geographical
dispersion-on appropriation of the technology-measured by use of ICTs, comfort with
the technology, adaptation of the technology to the team's work, and use of ICTs
for power/control purposes. They also assessed the impact of structural and appropriation
variables on two outcomes: perceived group effectiveness and perceived strengths
of the group, a measure of coordination quality.
DeSanctis et al. found that smaller teams that met often preferred advanced ICTs
more than did larger teams, which was surprising. The structural variables did not
predict technology use over time, but prior technology use did, suggesting that use
builds on itself. Appropriation variables, particularly adaptation and power uses,
were stronger predictors of coordination quality than the structural variables. For
technically-oriented teams, at time 2, both adaptation and power uses were negatively
related to coordination quality, while at time 3, adaptation was positively related
to coordination quality; and power uses were negatively related. This change in sign
suggests a learning effect; at first, adapting to ICTs is a burden, but as teams
get used to ICTs, adapting to them is easier and helps the group. None of the variables
predicted perceived effectiveness.
6.0 Discussion
The Minnesota GDSS Project was a complicated tapestry of laboratory and field studies
conducted by an interdisciplinary team of researchers. The goal of the project was
to use a common, robust technological platform to conduct an interlocking set of
lab and field studies that would develop and test theoretical explanations for the
impacts of GDSSs.
The very complexity of the phenomenon shaped the project. GDSSs are quite complex
systems and group interaction is a many-layered phenomenon. Add in the impacts of
organizational and societal context, and the possible effects are truly multifaceted.
This context necessitated a graduated, programmatic approach to the research. For
each line of research in the project, the initial focus was on effects studies, with
the rationale that only if interesting effects were found would further research
be warranted. The focus then shifted to studies of interaction processes to ascertain
how the GDSS had its impacts and the mediating role of interaction process on GDSS
effects. Once, the researchers elucidated the interaction processes, the focus shifted
to articulating the nature of structuration processes in GDSS use. The initial studies
were conducted in a lab environment that provided a high degree of experimental control.
Subsequent field studies enabled assessment of the degree to which lab findings transferred
to settings of organizational use. The result was a "layering" of overlapping studies
each of which took a smaller part of the complex whole and, together, gave some idea
of the larger picture.
Several tensions helped to drive this research program. First was the tension between
theory and empirical research. An important principle of this project has been the
importance of theory. Developing a detailed theoretical framework at the outset provided
a reference point that clarified the larger implications of individual studies and
helped structure a systematic program. In the beginning, the framework was rather
general, and the empirical studies informed it both by showing where expectations
were wrong and by suggesting additions. The theoretical framework was articulated
over a 10-year period, in response to successive waves of findings that supported
some aspects and challenged others.
The interplay of lab and field research also shaped the project. GDSS research has
been more academically driven than much research in information systems, which tends
to pick its subjects from current practice and often lags somewhat behind industry.
By contrast, GDSSs were largely designed in university settings and so were idealized
versions built around what academics believed would be useful for practitioners.
Practitioners interested in groupware looked to universities for ideas and extrapolated
the results of laboratory research to the field based on their sense of its potential.
Results from the field served as essential tests of concepts and ideas developed
in the bell jar of academia. The field research, in turn, spurred further development
of the SAMM system and suggested additional laboratory experiments. For example,
the studies on Level 3 systems were motivated in part by feedback from the field
studies that suggested that complex systems required effective facilitation, but
that good facilitators were hard to find. Automated guidance built into the GDSS
was one way of overcoming this barrier.
Finally, the interplay of theory and practice greatly enriched the project. The impetus
to develop theories that could explain how the GDSS could help groups improve their
functioning was an important normative influence on this project. Such theories could
guide design of GDSSs and other groupware and would, thus, be eminently practical.
However, much of the theory for this project was developed "for theory's sake," and
initially failed to inform practice (though it was gratifying to academics!). Practitioners'
questions often brought the researchers "up shot" and encouraged articulation of
theories so that they had traction for organizational groups.
A few observations about the findings are in order. First, the results of lab and
field studies point to the benefits of higher-order Level 2 features such as stakeholder
analysis or problem formulation procedures, which may be substantially greater than
those resulting from Level 1 tools. However, the same studies also show that Level
2 features are often somewhat difficult for groups to understand and use properly
and may take time to master. Facilitation, training, and other
support
 are necessary for groups to capitalize on Level 2 features. It is also important
to stress that while learning to use and interpret the results of Level 2 features
requires an investment of time and resources, over the long term it can reduce the
time and effort required to conduct a sound analysis and come to consensus.
The benefits of Level 2 GDSSs underscore an irony in current information systems,
particularly Web 2.0 applications, which reflect a new generation of the Web that
provides for greater social interaction and collaboration. Most systems currently
being offered to
support
 groups and teams, especially virtual teams, primarily provide Level 1 features,
such as threaded messaging, videoconferencing, file sharing, and shared applications
of drawing or word processing tools. However, the lab studies showed clearly that
groups using Level 1 tools tend to underperform or at best equal groups using Manual
versions of the same tools, have higher levels of conflict than Manual groups, and
have difficulty managing these conflicts. It is only through use of Level 2 features
that GDSS groups have more constructive conflict management and better outcomes than
groups using Manual procedures. A key challenge to those wishing to benefit from
GDSSs is finding systems that incorporate Level 2 tools and then motivating groups
to use them. The results from these studies help to inform how one might provide
better process structure in dispersed team environments to address these issues.
Recall that the report by Gartner cited earlier predicted that GDSS tools would need
to be incorporated into conferencing and other Internet spaces in the near future.
Second, the importance of facilitation-particularly flexible facilitation-and guidance
in the effectiveness of Level 2 systems was noteworthy. As a sidebar of the field
study-and in subsequent discussions with industry users-one of the barriers to the
adoption of GDSSs is the overhead connected with facilitation. Organizations often
do not have-or are not willing to commit-the resources to employ specially trained
facilitators. Even when facilitators are available, team leaders must spend time
working with them to select procedures and plan meetings, as well as spend "social
capital" convincing reluctant members to use the GDSS. Members must learn to use
the system. This overhead, combined with the dissatisfaction that GDSS users typically
experience until they get accustomed to the system and have some successes, is a
disincentive to use GDSSs. To successfully implement GDSSs, organizations must be
willing to commit the necessary resources and encourage employees to take the time
to master the technology and its application. Level 3 GDSSs, which build guidance
into the system, can potentially reduce the overhead involved in implementing GDSSs.
The growth in sophistication of agent-based advisory systems promises Level 3 systems
that can learn and adapt to the particular requirements of teams.
A major theme of the Minnesota GDSS research project was to encourage faithful use
of the system, that is, use in line with the spirit built into the system. A second
theme was the need to create groups that are capable of taking charge of the GDSS
themselves. The presumption has been that the GDSS should be both a tool and an opportunity
for the group to learn how to function better, and that achieving the latter means
that groups should become functionally independent in using the GDSS. However, as
groups become more independent and achieve greater facility in adapting the GDSS
to their work (and vice versa), they also are likely to embark on ironic uses of
the GDSS that violate the spirit of the technology. Ironic appropriations are not
necessarily a bad thing; many represent creative new ways to apply the system that
go beyond what designers and implementers envision. However, some ironic appropriations
do work against the values that designers and implementers of GDSSs want to promote.
Use of a voting feature to impose the majority's will on the group, for example,
is inconsistent both with rational deliberation and with participative decision making,
two norms most GDSS experts wish to promote. And if the resident experts in the GDSS
choose to use it in ways that manipulate the outcome, the group will be "shortchanged"
by the system, getting a predetermined result rather than benefiting from true collaboration
among members. Insuring that the values and processes built into the system are honored
is a continuous process. This is particularly the case because GDSSs must be melded
with existing organizational norms, which are often at odds with norms underlying
GDSS procedures.
Finally, the findings of the studies suggest that it is important to put some thought
into the selection of groups that will use the GDSS. The groups best situated to
benefit from a GDSS are those with complex tasks, fair or good relationships among
members, open communication, and some degree of comfort with information technology.
Bringing in a GDSS to solve problems that a group already has is unlikely to work.
The studies suggest that the group will simply transfer its problems into its appropriation
of the GDSS. GDSSs may benefit "troubled" groups when combined with other interventions,
such as strong facilitation that helps the group address its problems.
Adaptive Structuration Theory continues to develop, with more than 250 citations
found in a recent search. It has been used by researchers in several other fields,
including geography (Nyerges and Jankowski, 1997), management (Browning, Beyer, and
Shetler, 1995), applied psychology (Kahai, Sosik, and Avolio, (1997), and communication
(Sunwolf and Seibold, 1998). In addition to information systems it has been used
to study interorganizational ventures, development of industries and communities,
leadership, and implementation of innovations. However, it has not been without controversy,
as several critiques suggest (Jones, 1999; Orlikowski, 2000).
6.1 Strengths and Limitations
The approach taken in the Minnesota program of research has several advantages. The
studies utilized a consistent technological platform built around a coherent spirit
with a consistent look and feel, which enabled us to rule out effects due to different
technologies or interfaces. As a result, the results of different studies were cumulative
over the entire program. Utilizing multiple tasks that were often employed across
multiple studies also has advantages. Comparing results across multiple tasks, as
in the studies of Level 2
support
, tested generalizability. It also facilitated discovery of inconsistencies in results
that led to additional research on facilitation and guidance. Because the Minnesota
researchers used the same task across several studies that varied level of
support
, heuristics, and other factors, we were able to compare processes and outcomes and,
hence, to tease out the impacts of technological and procedural variations. The laboratory
and field studies provided useful complements; consistencies in results across venues
provided reassurance that substantive results and inconsistencies posed interesting
puzzles. Finally, we used a consistent theoretical model throughout the program,
which provided a compass to guide the planning of studies and interpretation of results.
The GDSS research program also has its limitations. Using a single technological
platform opens the project to the charge that results are idiosyncratic to that platform.
The same can be said for the experimental strategy and design and for the instruments
employed to measure key constructs. While these objections cannot be ruled out on
the basis of the evidence provided by the project alone, there are some mitigating
observations. First, a number of results are consistent with those from research
using other GDSSs, tasks, and designs, as indicated by a comprehensive review of
the GDSS literature by Fjermestad and Hiltz (1998-1999). Second, there is reason
to believe that, although the SAMM design is only one of many possible designs, it
was a reasonably effective technology. Many subjects in the lab and users in the
field expressed positive reactions to SAMM. In the case of the lab studies that purposely
left many groups to their own resources and allowed them to flounder and fail in
order to get variation in processes and outcomes, the researchers were surprised
by the number of subjects who were very positive about SAMM and understood its potential.
In the studies that offered
support
 and guidance, satisfaction with the system was high. In the field, groups using
SAMM for multiple sessions had about a 50 percent success rate, which is respectable
for IS implementations.
Another limitation stems from an inherent tension in research on experimental IS
in the field. For the purposes of research, it is important to observe unsuccessful
as well as successful uses of the system. No information technology works perfectly
when it is first installed, and the researcher and the system developer learn as
much when problems and issues arise or when people use it wrongly as when the system
works perfectly. This is particularly true for research informed by AST, because
testing this theory requires observations of ironic as well as faithful uses of the
GDSS. Helping the groups use the system perfectly would reduce the type of variation
needed to study its impacts on group processes and outcomes. However, on the practical
side, it was critical that SAMM work well and generate clear benefits for participating
organizations. If the system had too many problems and bugs or if it did not deliver
benefits, there would be no incentive for the organization to continue to use it.
There was in some cases, then, a potential conflict between the requirements for
good research and what would foster a good long-term collaboration. The researchers
walked a fine line between these issues throughout the project.
6.2 Future Directions
One promising set of directions extends along the trajectory already laid out by
the project. It includes research on systems for automated guidance of GDSSs, study
of processes such as group argumentation that play a major role in the impacts of
GDSSs, and studies of the social construction of group technologies.
A second direction points toward connecting the study of GDSS more closely to collaboration
technology generally. GDSS is just one of a suite of such technologies, and enthusiasm
for newer collaboration technologies such as portals, avatars, and blogs has eclipsed
GDSSs in some venues. However, research on GDSSs holds important insights for newer
collaboration technologies. Newer technologies also promise to enrich GDSSs. While
research shows clear benefits from GDSSs, they have not been very successful applications.
It may indeed be the case that newer, more successful collaboration technologies
can inform the design and implementation of GDSSs, making them more palatable to
the large mass of users.
A third direction moves beyond GDSSs and extends AST and the research strategy employed
in this project to new venues. DeSanctis and colleagues pursued study of virtual
organizations and online communities using many of the concepts and techniques developed
in this project (Fulk and DeSanctis, 1999).
7.0 Conclusion
In its 20 years, the Minnesota GDSS Research Project has made several significant
contributions. Adaptive Structuration Theory offers a general model of information
and communication technology use and impacts that have proven its utility beyond
the GDSS context. The theory has become recognized as a central theory for the information
systems domain and it has been applied in a variety of areas. Within the GDSS context,
research guided by AST has clarified the processes by which users incorporate information
technology into their work and how they realize and restructure it in so doing, which
has implications for the implementation of IS. The project's systematic study of
levels of group decision
support
 has clarified the impacts and contributions of the different levels, which has implications
for the design and utilization of a wide variety of collaboration technologies. The
project also has yielded significant findings relevant to facilitation and the use
of heuristics and other structures in concert with GDSSs. These fundamental topics
continue to be important as new developments in technology create opportunities for
new ways to
support
 groups. The theory, findings, instruments, tasks, and techniques that issued from
this project continue to find application in new areas. Though its participants have
for the most part gone on to pursue other lines of research, further analyses drawing
on its data continue to appear, providing evidence of the solid value of longterm
programs of research.
Acknowledgements
Support
 for this project was provided by the National Science Foundation, NCR, IBM, GE,
Texaco Inc., the IRS, the Hewlett Foundation, and the MIS Research Center, Department
of Information and Decision Sciences and the Department of Communication Studies
at the University of Minnesota. We gratefully acknowledge the many contributions
of Gary Dickson to the research projects. We would also like to express our gratitude
to our research assistants and system developers, without whom this project would
not have been possible. Thanks to Mike Beck, Sarah Belknap, Derek Bolt, Carol Broullette,
David Chang, Aditya Gupta, Aditya Gurajada, Janet Kelly, Russ Littlefield, Dan Luitjens,
D. J. McBride, Surya Prasad-Koneru, Usha Rani, Randy Snyder, Jonathan Trower, and
V. Venkatraj. Some of these students have gone on to distinguished careers in academia
and others are now in the private and public sectors.
Note
Jonathon Cummings served as editor for this manuscript and handled the entire process,
making the final decision on publication.
-1-
Questia, a part of Gale, Cengage Learning. www.questia.com
Publication Information:
Article Title: The Minnesota Gdss Research Project: Group Support Systems, Group
Processes, and Outcomes. Contributors: Gerardine Desanctis - author, Marshall Scott
Poole - author, Ilze Zigurs - author, George Desharnais - author, Marianne D'Onofrio
- author, Brent Gallupe - author, Michael Holmes - author, Brad Jackson - author,
Michele Jackson - author, Howard Lewis - author, Moez Limayem - author, Jooeng Lee-Partridge
- author, Fred Niederman - author, V. Sambamurthy - author, Chelley Vician - author,
Richard Watson - author, Julie Billingsley - author, Laurie Kirsch - author, Rebecca
Lind - author, Dale Shannon - author. Journal Title: Journal of the Association for
Information Systems. Volume: 9. Issue: 10/11. Publication Year: 2008. Page Number:
551+. © 2008 Association for Information Systems. Provided by ProQuest LLC. All Rights
Reserved.
Next Page

Jessie Rayl
thedogmom63 at frontier.com
www.facebook.com/Eaglewings10
www.pathtogrowth.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.acb.org/pipermail/acb-hsp/attachments/20120712/e49c5664/attachment-0001.html>