Using a GDSS to Facilitate Group Consensus: Some ... Decision Using a GDSS to Facilitate Group...

Group Decision

Using a GDSS toFacilitate GroupConsensus: SomeIntended andUnintendedConsequences

By: Richard T. WatsonSchool of BusinessWestern Australia College of Advanced

EducationP.O. Box 217Doubleview W.A. 6018Australia

Gerardine DeSanctisManagement Sciences DepartmentUniversity of Minnesota271 19th Avenue SouthMinneapolis, MN 55455

Marshall Scott PooleSpeech-Communication DepartmentUniversity of Minnesota317 Folwell HallMinneapolis, MN 55455

AbstractA cumulative body of experimental research isemerging that examines the ability of computertechnology to support the processes and out-comes of small group meetings. For the most partthe group decision support system effort hasbeen concerned with demonstrating the useful-ness of the technology in planning and decision-making situations where the quality of themeeting’s outcomes can be objectively as-sessed. In many decision situations, however,there is no objective measure of decision qualityavailable. Rather, the group must reconcile dif-ferences in opinion, personal preference, orjudgment and achieve consensus about a partic-ular mode of act/on. As a contribution to the accu-mulating research on GDSS, the current studyexamines the effects of a GDSS in resolving con-flicts of personal preference. In a task requiringresolution of competing personal preferences,82 groups--the largest sample size in the GDSSliterature to date--were randomly assigned toone of three experiment conditions: (1) com-puter-based support system (GDSS); (2) a

ual, paper and pencil, support system; or (3) support whatsoever. Groups were either of size 3or 4 persons. Use of the GDSS was expected tofacilitate democratic participation in group dis-cussion, move group members toward agree-ment with one another, and result in a high level ofsatisfaction with the group decision process.While several of the intended effects of the tech-nology were observed, the groups experiencedsome unintended consequences as a result ofusing the GDSS. In general, the GDSS technol-ogy appeared to offer some advantage over nosupport, but little advantage over the pencil andpaper method of supporting group discussion,

Keywords:Decision support, group decision sup-port systems, problem solving

ACM Categories: K.0, K.M, H.0, H.4.2, H.4.M

IntroductionGroup decision support systems (GDSS) com-bine computer, communication, and decisionsupport technologies to support problem formula-tion and solution in group meetings. The goals of aGDSS are to reduce the "process loss" associ-ated with disorganized activity, member domi-nance, social pressure, inhibition of expression,and other difficulties commonly encountered ingroups and, at the same time, to increase the effi-ciency and quality of the resulting group decision(Turoff and Hiltz, 1982; Huber, 1984a; DeSanctisand Gallupe, 1987). These goals are similar tothose that historically have been associated withinformation technology. That is, the hope hasbeen that computers will bring greater efficiency,reliability, and quality to the nature of work.

Much of the literature on computer-supportedcommunication and decision making rests on theassumption that the addition of the electronicmedium to verbal information exchange will leadto better decisions and actions, and to higher pro-ductivity. The electronic medium should allowgreater participation in decision making and in-crease access to meeting information. Conse-quently, the resources of a group should be fullyextracted in a group discussion; a more demo-cratic decision process should emerge; and somestructure should be added to what is otherwise a"muddling through" process for groups (DeSanc-tis and Gallupe, 1987; Huber, 1984a; Gray, et al.,1981 ; Rohrbaugh, and Wehr, 1978). Siegel, et al.(1986) argue that the reduction of social contextcues in computer-mediated communicationshould "reduce normative influence relative to in-formational influence. This should reduce the im-

MIS Quarterly/September 1988 463

Group Decision

pact of implicit reference group norms and ofgroup members’ social approval of one another,and increase the importance of arguments or de-cision proposals" (p. 162). In other words, ideasshould become the object of discussion ratherthan the proponents of ideas.

These are the intended effects of a GDSS, andthere is growing evidence that at least some ofthesedesired outcomes can be achieved with thetechnology. But, as Kiesler (1986) points out, theaddition of an electronic medium for interpersonalcommunication can bring about some negative,unintended effects as well. For example, the useof keyboard input and the greater volume of infor-mation flow can add to the level of effort requiredin a group meeting, thus lowering group effi-ciency. Similarly, heightened awareness of mem-bers’ viewpoints and greater objectivity inreviewing proposed ideas or solutions to a prob-lem may raise the level of conflict in the group.These unanticipated negative effects may coun-teract or wash out the gains realized from aGDSS.

This study attempts to assess the relative balanceof intended and unintended consequences ofgroup decision support for a task that requiresachieving group consensus on a matter for whichthere are competing personal preferences. Thetheory of GDSS would argue that a GDSS shouldfoster more even participation in these kinds of sit-uations and facilitate a systematic, or structured,group decision process (DeSanctis and Gallupe,1987; Huber, 1984a), resulting in effective conflictmanagement. Consequently, group consensusshould be higher when GDSS-supported groupsare compared to groups without computer-basedsupport.

In the next section we discuss the nature of groupdecision support systems and the purpose of thisstudy within the context of available GDSS re-search. Next, the research hypotheses are pre-sented and the appropriate independent anddependent variables are described. The researchmethod is also described, followed by an exami-nation of both the quantitative and qualitative find-ings of the study. The paper concludes bydiscussing the implications of the results for futureGDSS research.

Rationale and ResearchStrategyA GDSS provides communication support and de-cision structuring to a group via a software systemthat is accessible to all members of the group. A

facilitator may assist the group in using the GDSS,or the system may be entirely user-driven, withoutthe assistance of a chauffeur (DeSanctis andGallupe, 1985). Because decision making ingroups occurs through interpersonal communica-tion (Bedau, 1984; Poole and Hirokawa, 1986),the most fundamental goal of a GDSS is to sup-port the exchange of ideas, opinions, and prefer-ences within the group (Bui and Jarke, 1984;DeSanctis and Dickson, 1987). A GDSS that pro-vides communication support features, such asanonymous messaging, idea recording, prefer-ence rating or voting, is a Leve/-1 system. Whendecision modeling or mathematical techniquesare added to the system, the GDSS becomes anenhanced, or Leve/2, system. DeSanctis andGallupe (1987) argue that GDSS research shouldbegin with the study of Level-1 systems, proceed-ing to study higher level systems only after someunderstanding of lower-level systems has beenachieved. In fact, most of the empirical GDSS re-search to date has been with Level-1 systems(Gallupe, 1985; Lewis, 1982; Siegel, et al., 1986;Turoff and Hiltz, 1982). Very few studies haveused Level-2 systems (Applegate, 1986; Steeband Johnston, 1981).

The empirical literature suggests that many of thehopes for GDSS can be realized. For example,Lewis (1982) and Gallupe (1985) find that groupssupported by a GDSS make higher quality deci-sions than groups without GDSS support. Apple-gate (1986) and Steeb and Johnston (1981) demonstrated the Viability of GDSS in live plan-ning situations. Positive effects of a GDSS ongroups have also been reported by Gray, et al.(1981), Turoff and Hiltz (1982), and Siegel, et (1986). Computer support has been shown to fos-ter a democratic approach to the decision pro-cess, with more equality of participation amongmembers (Siegel, et al., 1986). Improved satisfac-tion with the decision process has been reported(Applegate, 1986), as well as a greater shift awayfrom initial individual preferences (Siegel, et al.,1986).

These intended effects of the technology havebeen demonstrated for a limited number of tasktypes. To date, positive effects of GDSS havebeen observed for idea generation (Applegate,1986; Lewis, 1982), problem finding (Gallupe,1985), intellective choice (i.e., selection of a "cor-rect" answer among a given set of alternatives)(Turoff and Hiltz, 1982), and planning tasks (Ap-plegate, 1986; Steeb and Johnston, 1981). In twoof these studies, group members were dispersedand interacted with one another via a communica-tion network (Turoff and Hiltz, 1982; Siegel et al.,1986), while in the other studies, group members

464 MIS Quarter/y/September 1988

Group Decision

met in a face-to-face conference room setting. Inall cases, each member had direct access to theGDSS, and in most of the studies the performanceof the group is compared to an objective measureof decision quality.

Of course, many organizational meetings occurwithout prior or post knowledge of the "correct"outcome of a group meeting. For this reason, thisstudy aims to build on the available knowledge ofGDSS impacts by examining the usefulness of thetechnology in situations where a group must re-solve competing personal preferences and maxi-mize agreement on a solution to a problem. Insuch situations, achieving high decision quality isnot the primary goal of the group meeting. Thetheory of GDSS would argue that the technologyshould be as useful in achieving consensus as inidentifying correct solutions. In either situation,the GDSS should foster more even participation inthe decision and a more systematic, or structured,group decision process. If the intended effects ofGDSS may be obtained for personal preferencetasks, then the potential domain of application ofthe technology would be expanded.

In addition, this study attempts to measure theconflict potential in the subject groups (i.e., differ-ences in preferences with respect to the decision)so that the effects of the GDSS for different levelsof conflict can be gauged and controlled. Severalstudies report higher levels of conflict and nega-tive emotional expression in computer-mediatedcommunication than in face-to-face communica-tion (Applegate, et al., 1986; Siegel, et al., 1986;Gallupe, 1985). It is unclear, however, whether in-creased conflict is a direct result of computer-me-diated communication itself, or whether the GDSSsimply provides a mechanism that brings out ex-isting differences among group members. If theformer interpretation holds, then whether theGDSS improves decision making depends on thebalance between the negative effects of conflictand the positive GDSS effects. If the GDSS cre-ates conflict, rather than bringing out conflicts thatare already there, then group consensus shouldbe low, even for groups with little conflict potential.However, if the GDSS merely facilitates expres-sion of existing conflicts, without adding new con-flict, then consensus should improve as groupsare able to recognize and deal with their differ-ences in the context of the GDSS.

Much of the GDSS research is being conducted inlaboratory settings where the organizational con-text and other factors can be controlled, permittingprecise assessment of the impact of the technol-ogy on group outcomes. In most previous labora-tory experiments, the approach has been to

compare groups that use a GDSS with groups thathave no support in order to identify the relative ad-vantage that these systems can offer to decision-making groups. Prior experimental work in theGDSS area is summarized in Table 1. This studyaims to build on the available GDSS research bysystematically comparing groups supported witha Level-1 GDSS with groups that have either nosupport whatsoever ("baseline" groups) or a pa-per-and-pencil ("manual") support system thatcontains the same decision structure as theGDSS (cf. Lewis, 1962). Without two controlgroups it is impossible to determine whether in-crements or decrements in outcomes are due tothe GDSS or simply due to imposing a problem-solving structure on the group. The comparison ofbaseline to manual groups permits determinationof simple structure effects. The comparison ofmanual and GDSS groups permits determinationof computer system effects. The comparison ofbaseline and GDSS groups permits determina-tion of effects due to support versus no support.Consistent with prior experimental work, thisstudy is limited to considering the usefulness of aGDSS for a small group meeting. Because thereis some evidence that group size can have an im-pact upon interaction patterns (Bales and Bor-gatta, 1965; Reynolds, 1971 ) groups of three andfour persons were chosen.

Variables and HypothesesThe major concern of this study is with the effectsof a GDSS on the level of group consensus for atask requiring resolution of conflicting personalpreferences. The major independent variable isthe level of support for the group. GDSS groupsreceived a Level-1 GDSS. Manual support sys-tem groups were supported with structure througha paper-and-pencil version of the GDSS. Baselinegroups were freely-interacting and were given nosupport system whatsoever. A second explana-tory variable is conflict potential, which is mea-sured as a continuous variable by assessing thedegree of pre-meeting consensus with themethod developed by Spillman, et al. (1980). Thestatistical methods used in this study, in effect,controlled for pre-meeting consensus in the anal-ysis of effects of level of support on post-meetingconsensus. This allows for determination of sup-port effects uncontaminated by differences in pre-meeting consensus among groups (see Cook andCampbell, 1979) and enables (or allows) us to sess the effect of pre-meeting on post-meetingconsensus.

Degree of post-meeting consensus is the majordependent variable of the study. In addition to

MIS Quarterly~September 1988 465

Group Decision

consensus, various other dependent measureshave been taken, including members’ percep-tions of the oUtcome and decision process, andequality of influence.

Individual group members’ decisions wererecorded before and after the group meeting tocompute pre- and post-meeting consensus ineach group. The Spillman, et al. (1980) techniquehas been used to calculate consensus from the in-dividual choices of each group member.

To control for possible effects of team size withineach experimental condition, half of the groupshad three members and the remaining groups hadfour members. Three- and four-person groups arevery small, and the impact of a GDSS may bemore dramatic in larger-size groups. However,the average number of people attending an orga-nizational meeting is only five (Datamation,1986). More importantly, the interest of the studyis in determining whether the effects of a GDSScan be detected in small groups. Group size in thisstudy is similar to that in previous research (Lewis,1982; Gallupe, 1985; and Siegel et al., 1986).

Using these variables, three hypotheses havebeen developed. The theory of GDSS would ar-gue that the use of the technology should facilitateresolution of conflicting preferences and, conse-quently, create a high level of consensus in the

group. Within the context of this study, groups withaccess to a GDSS were expected to achievehigher post-meeting consensus than groups with-out GDSS support for two reasons.

First, the GDSS provides a meeting structure forthe group. By providing a menu of facilities, suchas idea generation, rating, ranking, and voting, thesystem suggests that the group consider the pro-cedures they use in reaching their decisions.Group research strongly suggests that groupscan benefit from reflecting on the process of theirmeeting, regardless of the specific process thatthey eventually follow (Hackman and Kaplan,1974; Hirokawa and Pace, 1983).

Second, the electronic medium associated withthe GDSS should allow the structure to be invokedand applied more quickly than is possible with amanual system. Groups with manual supportwere expected to achieve higher post-meetingconsensus than baseline groups, which had nosupport system. In addition, groups with low con-flict potential (i.e., high pre-consensus) were ex-pected to achieve higher post-meetingconsensus than groups with high conflict potentialin all conditions.

H1. The degree of post-meeting consensuswill vary as a function of the type of sup-port given to the group.

Table 1. Summary of GDSS Experimental Research in Laboratory Settings~

Researcher(s) GDSS Task TaskLevel= Context Type3

2 Foreign embassy Planningtakeover by aterrorist group

Severe financialproblems in auniversity

Steeb and Johnston(1981)

Lewis (1982) 1

Gallupe (1985) 1 A firm is losing Problemprofits at the findingsame time thatsales are rising

Current Study 1 $500,000 can be Preferenceallocated to 6 allocationcompeting projects

GroupSize

3

Idea 3generation

3

3,4

TreatmentGroups

GDSS andno support

GDSS,manual,and nosupport

GDSS andno support

GDSS,manual,and nosupport

1This summary is limited to studies of face-to-face meetings in which all group members interface with thedecision support system.

2GDSS level refers to the type of technological support for the group. A Level-1 GDSS is a communicationmedium only, whereas a Level 2 provides structured techniques, such as mathematical models, for improv-ing the group’s decision.

3This categorization of task types is based on McGrath (1984).

466 MIS Quarterly~September 1988

Group Decision

Hla. Post-meeting consensus will be higherin the GDSS groups than in the manualsupport or baseline groups, controllingfor initial level of conflict.

Hlb. Post-meeting consensus will be higherin manual support groups than in thebaseline groups, controlling for initiallevel of conflict.

This hypothesis implies that the GDSS will bringout pre-existing conflict rather than creating con-flict itself. Indeed, the design of the GDSS em-ployed in the present study (summarized below) intended to lead to lower levels of conflict thanthose used by Siegel, et al. (1986) and others, be-cause it allows both face-to-face and mediatedcommunication.

The theory of GDSS would argue that a GDSS willresult in a more democratic decision process, withmore equal participation among members. Thisoccurs because a GDSS provides a frameworkwithin which group members who are reluctant tocontribute are encouraged to participate and po-tentially influence the group discussion (Siegel, etal., 1986). The structure provided in this studysuggested that each group member should statehis or her personal preferences at a number ofpoints during a team meeting. Ranking, rating andvoting are methods for getting participants to ex-press a personal judgment. As a result, in groupssupported by a GDSS, there should be greaterequality of influence because the relative contri-bution of those less willing to participate is in-creased. For similar reasons, groups with amanually supported structure are also expectedto have a more equitable distribution of influencethan baseline groups.

H2. Equality of influence will vary as a func-tion of the type of support given to thegroup.

H2a. Influence will be more even in the GDSSgroups than in the manual supportgroups.

H2b. Influence will be more even in the man-ual support groups than in the baselinegroups.

If hypothesis H2 is supported, the findings of thisinvestigation will be in accord with prior experi-mental research. While prior studies do not exam-ine equality of influence, they do look at equality ofparticipation (Siegel, et al., 1986), a measure thatis conceptually similar. Measures of influence andparticipation are both predicated upon determin-ing the extent to which group decision-making is ademocratic process. The measure of influencewill compare each team member’s choice with the

group’s choice. In contrast, the consensus mea-sure is based upon team members’ choices afterthe meeting.

While positive effects on group outcomes havebeen observed in many studies (Applegate, 1986;Siegel, et al., 1986; Turoff and.Hiltz, 1982), thereis also some indication that the use of GDSS tech-nology can lead to negative sentiments on the partof group members. Gallupe (1985) finds that,compared to non-supported groups, GDSSgroups in a face-to-face setting report lower satis-faction with the group decision process, less con-fidence in the final solution, and a higher degree ofconflict in the course of group discussion. Giventhese results, some differences in attitudes to-ward the GDSS were anticipated.

H3. Attitudes toward the group process willbe different In the GDSS groups than inthe manual system and baseline groups.

Our approach to examining this hypothesis is veryexploratory. A battery of attitudinal questions wasgiven to each group member at the end of the ex-periment. The questionnaire consisted of a groupprocess satisfaction measure developed byGreen and Taber (1980), a decision quality mea-sure designed by Gouran, et al. (1978), and someadditional questions developed by the authors.Both Green and Taber, and Gouran, et al. reportevidence supporting the reliability and validity oftheir measures. The instrument was pretested on20 groups, and several expert informants com-mented on it. This information was used to"sharpen" and revise the instrument to its finalform.

Research Method

Subjects and group compositionForty-four three-person and 38 four-persongroups participated in the study. The groups weremade up of graduate and undergraduate studentsenrolled in introductory MIS classes at a ~arge ur-ban university. Many of the students were em-ployed full-time in business settings, and mostwere working at least part-time. On average, theparticipants were 24 years of age with slightlymore than two and a half years of work experiencein a business or related setting. Approximately 60percent of the subjects were male. All of thegroups were "live" groups in that they had beenactively working together as teams on other classassignments, in this way, the initial socialization


Group Decision

that occurs early in group formation could beavoided during the data collection.

The GDSSThe GDSS, called "Software Aided Meeting Man-agement" (SAMM), Version 1.1, was designed,coded, and tested by a research team before itwas used in the experimental setting. Approxi-mately five groups used the system prior to actualdata collection as a pre-test for the smooth opera-tion of the software. The training materials for thesoftware were also tested in this manner and thenrefined prior to actual data collection. Pre-tests in-dicated that people were comfortable using thesystem following a 20-minute training session.The software was developed rather than pur-chased because the available commercial GDSSsoftware was either chauffeur-driven (not allow-ing each group member to interface with the sys-tem), too limited in scope (message storing only,without preference rating, ranking, or voting), orhardware-dependent (running only under special-ized operating systems). Lack of standard com-mercial software in the GDSS area has led mostresearchers to develop their own systems.

The system is described in DeSanctis and Dick-son (1987) and is being used for several relatedstudies of group DSS (Poole and DeSanctis,1987; Watson, 1987; Zigurs, 1987). Basically, thesystem incorporates a rational problem-solvingagenda (Dewey, 1910). The software is similar that used by Lewis (1982) and Gallupe (1985) that it performs the basic functions of recording,storing, and displaying problem definitions, and ofcriteria for evaluating solutions, alternative solu-tions, and a final group decision. Group memberscan enter relative weights for solution criteria, and

the system will aggregate and display averagegroup weightings. In addition, the system will cu-mulate and display ratings, rankings, and votesassociated with one or more alternative solutionsto a problem. These features have been identifiedas appropriate for supporting the communicationneeds of groups (Huber, 1984b; DeSanctis andGallupe, 1987; Joyner and Tunstall, 1970).

Group members can enter individual ideas, mes-sages, weights, ranks, or votes at their private ter-minals, and the public screen is used to displaygroup ideas, messages and aggregated (aver-age) values of weights, ranks, and votes. As notedearlier, our interest was in studying the impacts ofa Level-1 system on group consensus. Conse-quently, no modelingtechniques were included inthe system. The system is easy-to-use and menu-driven. Figure 1 shows the main menu for the sys-tem. Between one and three sub-menus underlieeach of the items on the main menu.

Manual and baseline conditionsIn the case of the manual groups, subjects wereprovided with an 11-page handout outlining thesame agenda that was on the GDSS. Each pageof the handout, which corresponded to a screenwithin the GDSS, explained an agenda item, giv-ing details parallel to those in the sub-menus ofthe GDSS on how to accomplish the item. Manualgroups were given only a flip chart to display ideaspublicly. Every effort was made to ensure thatmanual groups had the same structural aids asthe GDSS groups, the only difference being thatthe manual groups operated without a computersystem. GDSS groups were provided with a 20-minute training session on use of the system;manual groups were also trained in how to use the

Agenda

1. Define/View Problem2. Define/View Selection Criteria3. General Discussion4. Define/View Alternatives5, Rate Alternatives6. Rank Alternatives7. Vote or Straw Poll or Alternatives8. Define/View Decision9. Conclude Meeting

Figure 1, Main Menu for Software Aided Meeting Management, Version 1.1

468 MIS Quarterly/September 1988

Group Decision

meeting structure. Baseline groups were given nostructure, flip chart, or training. They were told tooperate with their own resources. Both the man-ual and baseline procedures had been pre-testedon a small number of groups prior to actual datacollection for the experiment.

Task setting and vafidationThe research task, or the "foundation task," (Wat-son, 1987) requires subjects to allocate a givensum of money among six competing projects thathave requested funds from a philanthropic foun-dation. Conflict arises because the team mem-bers have varying preference structures, whichresults in different allocation patterns. Theprojects that subjects can fund are based uponthe personality components scheme described bySpranger (1928), who asserts that there are sixbasic interests or motives in personality: the theo-retical, economic, aesthetic, social, political, andreligious. The six projects that can be funded cor-respond to Spranger’s six personality traits, or val-ues. Table 2 shows the six projects and thepersonality value to which each is designed to ap-peal. (See Appendix A for the task.) The task waspre-tested and iteratively refined over the courseof six months, and over 100 students participatedin these pretests of the experimental task. Corre-lation analysis based on the 300 experimentalsubjects was used to check that the amount allo-cated to a project by an individual was significantlycorrelated with that person’s values as measuredby the study of values instrument (AIIport, et al.,1970).

Experimental proceduresExperimental groups were assigned to one ofthree rooms. Two of the rooms were used inter.

changeably for baseline and manually supportedgroups; these were conference-style rooms, eachcontaining a round table with comfortable seatingfor four people. In the "decision rooms" that wereset up for the GDSS groups (cf. DeSanctis andGallupe, 1985; Gray, et al., 1981), each subjecthad a terminal and was able to readily view thepublic screen.

The experimental procedure was as follows:

1. Subjects listened to a standard introductoryscript read by the administrator of the experi-ment and then read a background statement.

2. Subjects completed a consent form, a back-ground questionnaire, and the study of valuesinstrument.

3. Subjects individually allocated funds to the sixprojects requesting support from the philan-thropic trust. (These measures were used tocalculate pre-meeting consensus.) Subjects.individually allocated funds to five sets of sixprojects each. One of these allocation taskswas later used in the group meeting and the in-dividual scores were used to calculate pre-meeting consensus.

4. Groups allocated funds to the six projects re-questing support from the philanthropic trust.

5. Subjects completed a post-meeting question-naire for measuring an individual’s perceptionof the group’s decision-making process, andindividually allocated funds to the 6 projects re-questing support from the philanthropic trust.(These were used to calculate post-meetingconsensus.)

6. The administrator conducted a debriefing ofthe subjects.

During step 4 of the experiment, the group deci-sion-making phase, teams were placed into one

Table 2. Personality Value and Corresponding Project

Personality Projectvalue

Theoretical

Economic

Aesthetic

Social

Political

Religious

To purchase additional volumes for the community’s library system.

To create a tourist bureau to develop advertising and other methods of attracting tourisminto the community.

To purchase art for display in. the community’s art gallery.

To establish a community arts program featuring art, music, and dance programs for chil-dren and adults.

To purchase a new computer system for the county government in order to hold localtaxes constant.

To establish an additional shelter for the homeless in the community.

¯ MISQuarterly/September 1988 469

Group Decision

of the three treatment groups discussed previ-ously. Instructions for the GDSS [and Manual]groups were as follows:

In your group meeting you will follow an agenda,because this is the established practice for most or-ganizational meetings. You are expected to useyour personal judgment in following the agenda. Ingeneral, a group will follow the agenda; however, itmay vary the order in which topics are handled anddecide not to follow the strict sequence of theagenda..The agenda appears as the initial screen[page] of your terminal [handout]. There are one ormore screens [pages] underlying each agendaitem that explain its use ....

You are required to use the computer system[handout]. You may vary the sequence of theitems, and you do not have to use every method ofassessing the group’s collective opinion of alterna-tives. However, you must use the computer system[handout].

Baseline groups were simply told to use their ownresources as a group in making decisions. All ex-perimental conditions were instructed: "You willwork as a group. It is important that you make de-cisions as a group. You should make a decisionthat is acceptable to all members of the team."

Results

Hypothesis testingAnalysis of the data indicates that there were nodifferences in pre-consensus, post-consensus,equality of influence, or attitudes for three andfour-person groups. Therefore, the data for bothgroup sizes has been combined to test the three -hypotheses of the study.

H1: Effect on Consensus

Consensus was measured using an adaption ofthe method developed by Spillman, et al. (1980).

This method produces a scale in the range 0 to 1,where 1 implies complete agreement in the group.Table 3 provides descriptive statistics for the threeexperimental conditions on measures of pre-meeting and post-meeting consensus.

For each of the three conditions, the average levelof consensus improved and the variance in con-sensus was reduced following the group meeting.Post-meeting consensus was highest in groupsreceiving the manual decision aid. Table 4 showsthe results of the regression analysis for effects ofdecision aid, pre-meeting consensus, and the in-teraction of these variables on post-meeting con-sensus. While analysis of covariance might havebeen the preferred approach to testing Hypothe-sis 1, that technique could not be used because ofthe presence of an interaction between the experi-mental treatment variable and pre-meeting con-sensus (the covariate). The regression resultsindicate that, while post-meeting consensus issignificantly related to pre-meeting consensus,there are no significant differences due to the de-cision aid treatment. Hypothesis 1, as formulated,is not supported.

To further explore the combined effect of decisionaid and pre-consensus, the correlation betweenpost-meeting and pre-meeting consensus wasexamined. Table 5 shows that while there is no re-lationship between pre-meeting and post-meet-ing consensus in baseline groups, there is asignificant correlation for manually supported andcomputer assisted groups. Figure 2 illustrates thedifferent relationships between pre- and post-meeting consensus for the three treatmentgroups.

Figure 2 suggests that, in both manual and GDSSconditions, post-meeting consensus is related topre-meeting consensus. In the baseline groups,however, there was little relationship betweenpre- and post-meeting consensus. In the manualand GDSS conditions, group members appar-ently became aware of their pre-meeting agree-ment and differenceS, whereas the baseline

Table 3. Pre-Meeting and Post-Meeting Consensus (Mean and Standard Deviation) for ThreeExperimental Conditions

Decision AidConsensus Baseline Manual Computer Total

(N = 27) (N = 26) (N = 28)

Pre-meeting mean .27 .25 .24 .25Std. dev. 1.00 .87 .82 .90

Post-meeting mean .55 .62 .51 .56Std. dev. .19 .21 .19 .20


Group Decision

Table 4. Regression Results for Effect of Decision Aid and Pre-Meeting Consensuson Post-Meeting Consensus

Source of Variation SS df F Sig.

Decision Aid 150.7 2

Pre-Meeting Consensus 212.8 1

Decision Aid × PreoMeetingConsensus 190.6 2

Model 597.4 5

Error 262.1 75

R Square = .186

2.16 .123

6.09 .016

2.73 .072

3.42 .008

Table 5. Correlation Between Pre-Meeting and Post-Meeting Consensus by Decision Aid

Decision AidBaseline Manual Computer

Correlation coefficient - 0.022 0.439 0.457

Significance 0.913 0.025 0.015

1.0

0.00.0

Manual

Computer

Baseline

0.2 0.4 0.6 0.8 1.0Pre-meeting consensus

Figure 2. Post-Meeting Consensus as a Function of Pre-Meeting Consensus by Type of Decision Aid


Group Decision

condition did not encourage members to revealtheir pre-meeting opinions. The implication is thatstructure, in either manual form or in a GDSS, fa-cilitated expression of agreement or conflict in thegroup.

H2: Effect on Equality of Influence

Equality of influence was measured in a two stageprocess. In the first stage, the influence of eachperson in a group was measured by comparingthe relative distance of that person’s individual so-lution.from the group’s solution and from a "fair"decision. A fair decision is considered one that re-sults from each member having equal influenceupon the group’s decision. The second stage ofthe method was to combine the influence mea-sure for each group member into a measure thatindicates how equal the influence was in thegroup. This measure was computed by calculat-ing influence scores for all subjects in the experi-ment and then combining the scores of eachgroup to arrive at a measure of equality of influ-ence in that group. The lower the value for thescale, the more equal the influence.

Table 6 presents descriptive statistics on equalityof influence for the three experimental conditions.The trend in the means is in the expected direc-tion, with equality of influence being most equal inthe GDSS groups and most unequal in the base-line groups. Notice that the variance for the equal-ity of influence measure is extremely high in thebaseline condition. In fact, the variances acrossdecision aid conditions are significantly different(F = 19.08, p<.001 ). The variances are so ex-treme that no transformation of the data is suc-cessful-in achieving the equality of variancenecessary for analysis of variance. Analysis ofvariance using the regression approach suggeststhat the differences across experimental condi-tions in their influence scores are not significantand that pre-meeting consensus is not a meaning-ful predictor of the dependent variable. (See Table7.) A KruskaI-Wallis analysis on the ranks on theinfluence measure confirms the finding of no sig-nificant difference among the three experimentalconditions (chi square = 1.05, p = .59).

Hypotheses 2a and 2b are not supported. How-ever, the difference in variances across the condi-tions is interesting in itself. Lower variances in themanual and GDSS conditions would suggest thatthe imposition of structure on the meetings re-duces the dispersion across groups with respectto member influence.

H3: Effects on Group Attitudes

An exploratory approach is used to examine theimpact of various decision support systems upongroup attitudes. The subjects’ answers for each ofthe items in the post-meeting instrument wereused in a linear discriminant analysis, a usefulmethod of exploratory data analysis when causalrelationships are not well understood (Johnsonand Wichern, 1982). The decision aid received byeach group is used as the classification variable.A stepwise discriminant analysis with a signifi-cance level of 5% selected seven variables for alinear discriminant function. The results aredisplayed in Table 8, which lists each questioncontributing a variable to the discriminant func-tion, the width of the scale used to collect re-sponses, the significance of the variable, and themean score of the variable for each of the threetreatments.

The perception of the issues explored in a group’sdiscussion is the most important variable in dis-criminating between .~he three treatments. GDSSgroups reported that their discussion was lesssubstantial than the other two forms of decisionsupport. In addition, GDSS groups described theirproblem-solving process as less understandable.The manual groups were the most confident thattheir solution was correct. The least confidentwere the baseline groups, and the members ofthese groups also reported that they made thefewest suggestions about doing the task. Thebaseline groupsindicated a higher emergence ofinformal leadership than the other two treatments.GDSS groups perceived a higher level of inputinto the group’s final solution. The group task wasfound easiest by the non-supported groups, withthe other two groups reporting about the samelevel of difficulty.

The results of the linear discriminant analysis indi-cate that the attitudes of the three groups did dif-fer. Because of the exploratory nature of thisanalysis, no formal methods have been used totest for significant differences in attitudes.

Some quafitative findingsFormal hypothesis testing was the major purposeof this study, and the quantitative analyses pre-sented above must form the primary basis of anyconclusions drawn from the study. Nevertheless,there were certain Consistencies in behavior of thegroups during the experimental sessions that areworth noting. These observations are based onthe 40 hours of video and audio tape recordings ofthe experimental sessions. These observations


Group Decision

Table 6. Equality of Influence (Mean and Standard Deviation) for Three Experimental Conditions

Decision AidBaseline Manual Computer Total(N = 27) (N = 26) (N = 28)

Equality of .596 ,563 .557 .572influence mean

Std, dev. .126 .051 .041 .07

Table 7, Analysis of Variance Using the Regression Approach for Effect of Decision Aidon Equality of Influence

Source of Variation SS df F Sig,

Pre-meeting consensus .328 1 1.025 .31

Decision aid .645 2 1.006 .37

Error 24.34 76

Total 25.~,5 79

Table 8. Discriminant Analysis: Post-Meeting Questionnaire

Question Scale Width Significance Mean ScoreBaseline Manual Computer

How would you describe 7the issues explored inyour discussion?

How would you describe 52your group’s problem-solving process?

To what extent are you 53confident that the group’ssolution is correct?

Did you make suggestions 53about doing the task?

Did anyone emerge as an 5informal leader?

To what extent does the 5final solution reflectyour inputs?

How easy or difficult didyou find the task in agroup setting?

0.0001 5.28 5.47 4.30

0.0039 1.68 1.86 2.15

0.0138 3,64 3.96 3.77

0.0252 3.55 3,84 3.68

0.0363 3.37 2.94 2.91

0.0380 3.63 3.68 3.74

0.0449 2.11 2.39 2.35

1Scale anchored on 1 (trivial) to 7 (substantial).2Scale anchored on 1 (understandable) to 5 (confusing).3Scale anchored on 1 (not at all) to 5 (to a very great extent).4Scale anchored on 1 (extremely easy) to 5 (extremely hard).


Group Decision

have not yet been confirmed in any quantitativeway, but they may be useful for explaining the lackof increase in consensus for the GDSS treatmentgroups vis-a-vis manual groups.

1. Use of the GDSS tended to reduce face-to-face interpersonal communication in thegroup.

This was suggested by several observations.First, group members appeared to interface morewith the system than each other. Even thoughthey were seated around a conference table, theyfrequently turned to their individual computerscreens and keyboards or faced the large screenwhile talking to one another. Second, even thoughthe members met face-to-face around a confer-ence table, the electronic communication still hadan asynchronous quality in that there was a delaybetween the presenting of an idea, vote, or com-ment and the other group members’ viewing ofthat idea, vote, or comment. This was reflected inthe verbal "quiet time" spent by the group as itread screens or entered text or data into the sys-tem, either as individuals or as a group. Ratherthan consisting of continuous verbal exchange,the meeting had periods of quiet (work) time in themidst of verbal discussion. This resulted in re-duced eye contact between group members and acorresponding increase in social distance amongmembers.

2. Use of the GDSS presented a challenge tothe groups, thus making their meeting taskmore difficult than groups without theGDSS.

The GDSS groups appeared to struggle with theproblem of how to effectively use the technology.This struggle was evidenced on two levels. First,each group member had to become comfortablewith the user interface provided by the GDSS soff-ware. The 20-minute training period was sufficienttime for group members to know how to use thesystem without asking questions or referring tothe user manual. However, despite the fact thatthe system is menu-driven, simple in design, andeasy-to-use, each user still had to think aboutwhat keys to push during the group meeting. Thesystem was novel to them, and some mental effortwas required to use the system.

In addition to becoming facile with the system in-terface, GDSS users faced a second challenge.They had to learn how to incorporate the featuresof the system into the group meeting. Whereasthe groups using the manual system treated thesystem features as structural cues from whichthey would quickly develop an approach to meet

their particular needs, the GDSS groups let thetechnology drive the group meeting and were lesscreative in their use of the structural cues providedby the system. With a few exceptions, GDSSusers seemed to want to follow the features of thesystem in a very precise way, hoping that the sys-tem would somehow lead them into the "best" de-cision. In contrast, the manual system usersappeared to utilize the support methodology as atool, not as a "group director" who would guidethem through each meeting activity. In this sense,the GDSS created a set of cognitive challengesthat went beyond the task itself. Questions suchas, "What should we do first?" and, "OK, whatscreens do we want to use now?" arose muchmore frequently in the GDSS groups than in themanual system groups, despite the fact that thefeatures provided by the two approaches wereidentical.

3. Groups using the GDSS appeared to be-come very procedure-oriented, rather thanIssue-oriented, in their discussions.

The preference task used in this study requiredgroups to detect differences in individual valuesand then resolve these value differences throughdiscussion. Research on groups indicates that insuch circumstances, groups rarely attend to theprocess they employ; rather, they jump directlyinto discussion of issues (Hirokawa and Pace,1983). The introduction of structure thus shouldimprove the quality of the meeting outcomes byforcing the group to consider their decision pro-cess. The manual system had this positive effectin this study. In the case of the GDSS groups,however, the groups became overly concernedwith procedural matters, attending more to doingrating and voting than to the issues underlying theratings and votes. Comments such as, "What doyou want to do next?" or, "What items do you wantto add to this list?" or, "How do you want to vote onthese items?" were much more frequent in theGDSS groups than in groups supported with themanual system.

DiscussionThis study has produced several noteworthyfindings:

The main statistical test indicates no difference inmean values of post-meeting consensus amongthe GDSS, manual, and unsupported groups,However, secondary analysis indicates that post-meeting consensus is positively related to pre-meeting consensus in the manual and GDSSgroups, whereas there is no relationship in the


Group Decision

baseline groups. This implies that GDSS andmanual groups deal with potential conflicts in away that improves group consensus, while base-line groups do not. A possible explanation is thatbaseline groups have no framework to assistthem in discovering pre-meeting level of agree-ment in the group. As a result, they start the con-sensus-building process from scratch. Incontrast, the structure provided to manual andGDSS groups enables them to find their initiallevel of agreement and build upon it to create post-meeting consensus. The finding that the GDSSimproved group consensus, coupled with the find-ing of no overall differences in consensus, cou-pled with the finding of no overall differences inconsensus level among the conditions, suggeststhat the GDSS do not, in itself, create conflict. In-stead, by providing a structure, it seems to facili-tate conflict management in the groups.

In studies of computer conferencing, Kiesler(1986) notes that the electronic medium of com-munication results in a loss of social context cues,which results in more negative, or "flaming," com-munication among members. There is some qual-itative evidence that the same effect hashappened in the GDSS groups in the currentstudy. Even though the groups met face-to-faceand had non-verbal and other cues available tothem, heightened interaction through the elec-tronic medium, coupled with greater human-com-puter interaction relative to human-humaninteraction, seemed to increase the sense of dis-tance among members. At times the groups-sought context information. For example, in re-sponse to a controversial idea presented on thelarge screen, there might be comments such as,"Who said that?" followed by people turning to ob-serve each other’s faces. The group’s cognitivestruggle and procedural orientation may implythat chauffeurs will be necessary in GDSS set-tings. Or it may be that repeated use of a GDSSwill reduce these problems. Notwithstanding, inthis case the GDSS does not seem to have pro-duced conflicts that have a negative impact ongroup outcomes. The positive intended outcomesof imposing a structure seem to outweigh the neg-ative impacts of more aggressive communication.

No differences are found in equality of influenceamong the three conditions. This is inconsistentwith the findings of previous studies. However,there is less variation in influence scores in theGDSS and manual conditions than in the baselinegroups. This suggests that imposing structureregularizes influence patterns in groups.

Overall, the results on consensus and equality ofinfluence for the GDSS and manual conditions

tend to be similar, showing different patterns thanthe results for the baseline condition. It may bethat a finding of "no difference" is the best wecould expect, since only a Level-1 GDSS was em-ployed. Enhancements provided by a Level-2 sys-tem may be required in order to detect notableimprovements from the use of a GDSS. Firm con-clusions on this point, however, must await furtherresearch, particularly interaction analysis thatcompares the pattern and style of group pro-cesses across baseline, manual, and computer-support conditions.

There was a difference in attitudes among thethree treatments. The manually supported groupsseemed to have the most "positive" attitudes. Itappears that the novelty of the GDSS may haveproduced some unintended sentiments in com-puter-supported groups. The newness of thetechnology could explain why GDSS groups re-ported less substantial discussion of the issuesand a less understandable problem-solving pro-cess. Group members expected the computer toproduce the solution for them. They focused onusing the system very mechanically and weresometimes dismayed when the system did notmagically give the "right" answer. These effectsmay be transient and disappear with sustainedusage. In many groups the system became anend in itself. Practice, or use of a facilitator orchauffeur, may lessen this effect. In the case ofthe small group meeting, it is desirable to have auser-driven system without the expensive cost ofa facilitator. More work on GDSS design andgroup instructions (learning) is needed beforeabandoning the idea of having groups use aGDSS on their own without technical or other as-sistance. Negative effects point to the importanceof further study of a GDSS in the laboratory beforemoving into the field. Only by use of control groupscan the relative added value of a GDSS to groupmeetings be assessed.

This study has several important limitations thatprohibit generalization of the findings to other pop-ulations or settings. The groups in this study weresmall, and further research might examinewhether the same results will occur in largergroups. In addition, there is the common problemof current GDSS research in that the software sys-tem used was home-grown by the researchers.While the research of Siegel, et al. (1986) on com-puter support of remote group decision makingsuggests that the impacts of the technology are"relatively robust with respect to software varia-tions" (p. 177), there is always the question whether a differently designed Level-1 GDSS, orthe same GDSS with Level-2 features, wouldchange the nature of the findings of this study. Fi-


Group Decision

nally, there is the usual concern with the use ofstudents as subjects and the appropriateness ofsuch a sample for understanding the nature ofmanagerial meetings.

This investigation has identified some intendedand unintended effects of using a decision sup-port system for groups. As intended, the presenceof a suggested structure for the group meeting im-proved the degree of post-meeting consensus.Also, in contrast to the baseline and manual sys-tem group meetings, users of the GDSS reportedmore input into the group’s solution and were lesslikely to perceive that there was a leader in thegroup. The relationship between pre-meeting andpost-meeting consensus was similar in GDSSand manual groups, but post-meeting consensuswas not significantly higher in the GDSS groupsthan in the baseline or manual groups. Althoughthe st~:ucture provided in the GDSS and manualconditions reduced the variance across groups ontheir equality of influence, use of the GDSS did notresult in more equal influence of group memberson the final solution. The most surprising unin-tended effect was that GDSS users, compared tothe other experimental groups, perceived the is-sues discussed in the group meeting to be moretrivial and the group’s problem-solving process tobe less understandable. In the future, GDSS re-search should press further to sort out whatKiesler (1986) calls "intended technological ef-fects" (faster processing, fewer errors, moreequal participation), "unintended social effects"(heightened conflict), and "transient effects" (ef-fects that will diminish with group experience withthe system) of the technology on groups. Re-search on group behavior indicates the impor-tance of using groups with a meaningful historyand future whenever possible.

AcknowledgementThis project was funded by NCR Corporation, theMIS Research Center, and the graduate school ofthe University of Minnesota. The authors thankMichael Beck, Gary Dickson, and Ilze Zigurs fortheir contributions to the design of the GDSSsoftware.

ReferencesAIIport, G.W., Vernon, P.E. and Lindzey, G. The

Study of Values, 3rd edition, Riverside,Chicago, IL, 1970.

Applegate, L.M. "Idea Generation in Organiza-tional Planning," unpublished Ph.D. disserta-tion, University of Arizona, 1986.

Applegate, L.M., Konsynski, B.R. and Nuna-maker, J.F. "A Group Decision Support Systemfor Idea Generation and Issue Analysis in Orga-nizational Planning," Proceedings of the Con-ference on Computer-Supported CooperativeWork, Austin, TX, December 1986, pp. 16-34.

Bales, R.F. and Borgatta, E.F. "Size of Group As aFactor in the Interaction Profile," in SmallGroups: Studies in Social Interaction, reviseded., A.P. Hare, E.E. Borgatta and R.F. Bales(eds.), Knopf, New York, 1965, pp. 495-512.

Bedau, H. "Ethical Aspects of Group DecisionMaking," in Group Decision Making, W.C.Swap and Associates (eds.), Sage, BeverlyHills, 1984, pp. 115.-150.

Bui, T. and Jarke, M. "A DSS for Cooperative Mul-tiple Criteria Group Decision Making," Pro-ceedings of the 5th International Conferenceon Information Systems, Tucson, AZ, Decem-ber 1984.

Cook, T. and Campbell, D.T. Quasi-Experimenta.tion, Houghton-Mifflin, New York, 1979.

Datamation, "Hardware: Offline," (32:10), May15, 1986, p.109.

DeSanctis, G. and Dickson, G.W. "GDSS Soft-ware: a ’Shell’ System in Support of a Programof Research," Proceedings of the 19th AnnualHawaii International Conference on Systems

¯. Sciences, Honolulu,~January 1987.DeSanctis, G. and Gallupe, R.B. "A Foundation

for the Study of Group Decision Support Sys-tems," Management Science (33:5), 1987, pp.589-609.

DeSanctis, G. and Gallupe, R.B. "Group DecisionSupport Systems: A New Frontier," Data Base(16:2), 1985, pp. 2:10.

Dewey, J. How We Think, D. C. Heath, Boston,1910.

Gallupe, R.B. "The Impact of Task Difficulty on theUse of a Group Decision Support Meeting," un-published Ph.D. dissertation, University of Min-nesota, 1985.

Gouran, D.S., Brown, C. and Henry, D.R. "Behav-ioral Correlates of :Perceptions of Quality in De-cision-Making Discussions," CommunicationMonographs (45), 1978, pp. 51-63.

Gray, P., Berry, N.W., Aronofsky, J.S., Helmer,O., Kane, G.R. and Perkins, T.E. "The SMU De-cision Room Project," Transactions on the FirstInternational Conference on Decision SupportSystems, Atlanta; June 1981, pp. 122-129.

Green, S.G. and Taber, T.D. "The Effects ofThree Social Decision Schemes on DecisionGroup Processes," Organizational Behaviorand Human Performance (25), 1980, pp. 97-106.


Group Decision

Hackman, J.R. and Kaplan, R.E. "Interventionsinto Group Processes: An Approach to Improv-ing the Effectiveness of Groups," Decision Sci-ences (5), 1974, pp. 459-480.

Hirokawa, R.Y. and Pace, R. "A Descriptive In-vestigation of the Possible E;ommunication-Based Reasons for Effective and IneffectiveGroup Decision Making," CommunicationMonographs (50), 1983, pp. 363-379.

Huber, G.P. "Issues in the Design of Group Deci-sion Support Systems," MIS Quarterly (8:3),September 1984a, pp. 195-204.

Huber, G.P. "The Nature and Design of Post-In-dustrial Organizations," Management Science(30:8), August 1984b, pp. 928-951.

Johnson, R.A. and Wichern, D.W. Appfied Multi-variate Statistical Analysis, Prentice-Hall, En-glewood Cliffs, N J, 1982.

Joyner, R. and Tunstall, K. "Computer Aug-mented Problem Solving," Management Sci-ence (17:4), March 1970, pp. B212-B225.

Kiesler, S. "The Hidden Messages in ComputerNetworks," Harvard Business Review (64:1),January-February 1986, pp. 46-60.

Lewis, F.L. "Facilitator: A Micro-Computer Deci-sion Support System for Small Groups," unpub-lished Ph.D. dissertation, University ofLouisville, 1982.

McGrath, J.E. Groups: Interaction and Perfor-mance, Prentice-Hall, Englewood Cliffs, N J,1984.

Poole, M.S. and DeSanctis, G. "Group DecisionMaking and Group Decision Support Systems,"working paper, MISRC-WP-88-02, Universityof Minnesota, Minneapolis, MN, 1987.

Poole, M.S. and Hirokawa, R.V. "Communicationand Group Decision-Making: A Critical Assess-ment," in Communication and Group-DecisionMaking, R. V. Hirokawa and M.S. Poole (eds.)Sage, Beverly Hills, CA, 1986, pp.15-34.

Reynolds, P.D. "Comment on ’The Distribution ofParticipation in Group Discussions’ As Relatedto Group Size," American Sociological Review(36:4), 1971, pp. 704-706.

Rohrbaugh, J. and Wehr, P. "Judgment Analysisin Policy Formation: A New Method for Improv-ing Public Participation," Public Opinion Quar-terly (42), 1978, pp. 521-532.

Siegel, J., Dubrovsky, V., Kiesler, S. and,McGuire, T.W. "Group Processes in Computer-Mediated Communication," Organizational Be-havior and Human Processes (37), 1986, pp.157-187.

Spillman, B., Spillman, R. and Bezdek, J. "AFuzzy Analysis of Consensus in SmallGroups," in Fuzzy Sets: Theory and Applica-

tion to Poficy Analysis and Information Sys-tems, P.P. Wang and S.K. Chang (eds.),Plenum, New York, NY, 1980, pp. 291-308.

Spranger, E. Types of Men, 5th German ed.,P.W.J. Pigors (translator), Max Niemeyer Ver-lag, Halle, Germany, 1928.

Steeb, R. and Johnston, S.C. "A Computer-Based Interactive System for Group DecisionMaking," IEEE Transactions on Systems, Man,and Cybernetics (SMCol 1:8), August 1981, pp.544-552.

Turoff, M. and Hiltz, S.R. "Computer Support forGroup Versus Individual Decisions," IEEETransactions on Communications (COM-30:1),January 1982, pp. 82-90.

Watson, R.T. "A Study of Group Decision SupportSystem Use in Three and Four-Person Groupsfor a Preference Allocation Decision," unpub-lished Ph.D. dissertation, University of Minne-sota, Minneapolis, MN, 1987.

Zigurs, I. "The Effect of Computer-Based Supporton Leader Behaviors in Small Group Decision-Making," unpublished Ph.D. proposal, Univer-sity of Minnesota, Minneapolis, MN, 1987.

About the AuthorsRichard T. Watson received his Ph.D in manage-ment information systems from the University ofMinnesota in 1987. He is interested in group deci-sion support, end-user computing and manage-ment of computer resources. He has beenpublished in Large Sca/e Systems and DSS-88Transactions.

Gerardine DeSanctis is associate professor ofmanagement information systems at the Univer-sity of Minnesota. She received her Ph.D. fromTexas Tech University. Her research interestsconcern the design of systems to support man-agerial decision making and the effects of infor-mation systems use on individuals and groupswithin organizational settings.

Marshall Scott Poole is associate professor inthe Department of Speech-Communication at theUniversity of Minnesota. He received his Ph.D.from the University of Wisconsin. His research in-terests include social impacts of technology,group and organizational communication, conflictmanagement, and communication researchmethods. He has autl~ored or edited four booksa’nd numerous articles and is a member of the edi-torial boards of Human Communication Re-search, Communication Monographs, andCommunication Research.


Group Decision

AppendixThe Research Task:

Personal Trust Foundation

BackgroundUncle Sylvester is dead. He had lived a most productive life and accumulated a small fortune. As you werehis favorite, you have been made the sole trustee of your uncle’s estate. Your uncle had considerable re-spect for your personal values, and he had directed that, upon his death, you should decide how to allocatethe funds that he had placed into the Personal Trust Foundation. This is the opportunity of a lifetime. Youalone will decide how to spend the money. Just imagine, you can help those people or projects that are clos-est to your heart.

The Foundation’s goal is to fund projects in the community in which your uncle worked and lived most of hislife. You are required to select programs that you consider to be deserving of a donation from your uncle’sfoundation. Because he was a man who never sought publicity, your uncle has specified that you should notbe influenced by the extent to which a program will honor him or preserve the public’s memory of him. Al-though many factors may influence the decisions regarding which programs to fund or not to fund, the mostcritical factor is the degree to which a program agrees with your personal values.

DirectionsYou must evaluate competing requests for funding and make judgements about their relative merit. Manyprograms have merit, but limited resources require that you select the programs that you prefer to fund. Youhave discretionary funds available, and a number of projects are requesting access to these funds. Your jobis to select those that should receive support. Your goal in selecting projects for funding is to choose thoseprograms that agree with your personal values.

Begin when you are ready.

Personal Trust Foundation: $500,000You have $500,000 to allocate from this fund. Proposals received from various organizations for projects arelisted below. Each project is in need of $500,000 but can benefit from any contribution that you might make.The greater the contribution that you make to a particular project, the more likely it is that the chosen projectwill succeed.

Proposed project Recommendedlevel of Funding

1. To purchase a new computer system for the county government in orderto hold local taxes constant. $

2. To purchase additional volumes for the community’s library system. $

3. To create a tourist bureau to develop advertising and other methods ofattracting tourism into the community. $

4. To establish a community arts program featuring art, music, and danceprograms for children and adults. $

5. To establish an additional shelter for the homeless in the community. $

6. To purchase art for display in the community’s art gallery. $

TOTAL FUNDS ALLOCATED $500,000


Using a GDSS to Facilitate Group Consensus: Some ... Decision Using a GDSS to Facilitate Group...

Documents

Transcript of Using a GDSS to Facilitate Group Consensus: Some ... Decision Using a GDSS to Facilitate Group...