Sie sind auf Seite 1von 22

A Foundation for the Study of Group Decision Support Systems Author(s): Gerardine DeSanctis and R.

Brent Gallupe Reviewed work(s): Source: Management Science, Vol. 33, No. 5 (May, 1987), pp. 589-609 Published by: INFORMS Stable URL: http://www.jstor.org/stable/2632288 . Accessed: 17/09/2012 00:28
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

INFORMS is collaborating with JSTOR to digitize, preserve and extend access to Management Science.

http://www.jstor.org

MANAGEMENT SCIENCE Vol. 33, No. 5, May 1987 Printed in U.S.A.

A FOUNDATION FOR THE STUDY OF GROUP DECISION SUPPORT SYSTEMS*


GERARDINEDESANCTISAND R. BRENT GALLUPE Management Sciences Department, University of Minnesota, 271 19th Avenue South, Minneapolis, Minnesota 55455 School of Business, Queen's University, Kingston, Ontario Canada K7L 3N6
Technical developments in electronic communication, computing, and decision support, coupled with new interest on the part of organizations to improve meeting effectiveness, are spurring research in the area of groupdecisionsupport systems(GDSS). A GDSS combines communication, computing, and decision support technologies to facilitate formulation and solution of unstructured problems by a group of people. This paper presents a conceptual overview of GDSS based on an information-exchange perspective of decision making. Three levels of systems are described, representing varying degrees of intervention into the decision process. Research on GDSS is conceived as evolving over time from the study of simple "shell" systems, consisting of menus of features,available for selection by a group, to consideration of sophisticated rule-based systems that enable a group to pursue highly structured and novel decision paths. A multi-dimensional taxonomy of systems is proposed as an organizing framework for research in the area. Three environmental contingencies are identified as critical to GDSS design: group size, member proximity, and the task confronting the group. Potential impacts of GDSS on group processes and outcomes are discussed, and important constructs in need of study are identified. (GROUP DECISION MAKING; DECISION SUPPORT)

1. Introduction The need to improvegroupdecisionmakingis of longstanding concernto organizational researchers, having strongpracticalas well as scientificrelevance.Within this tradition,there is growinginterestin group decision supportsystems(GDSS), which combine communication,computer, and decision technologiesto supportproblem formulationand solution in group meetings. Technologicaladvancements,such as electronicboardrooms,local area networks,teleconferencing, and decision support softwarehave spurredresearchin this area. In addition, fundamentalchangesin the in externalenvironmentof organizations encouraging are research GDSS. Our society is experiencing emergence a post-industrial the of environmentcharacterized greater by knowledge,complexity,and turbulence(Huber 1984b).One effectof this trendis that decision-related meetings are becoming more frequentand more important.At the same time, the decisionsconfronting groupsarebecomingmore complexand must be made more quickly,and with greaterparticipation than in the past (Huber 1986). As scientistsareexploring partof the transitioninto this new environment,organizational advancedtechnologieswhich mightbe employedin groupmeetings. A GDSS aims to improvethe processof groupdecision makingby removingcomdecision analysis, mon communicationbarriers, providingtechniquesfor structuring and systematically directingthe pattern,timing, or content of discussion.The more sophisticatedthe GDSS technology,the more dramaticis the interventioninto the decision process.Of course, more dramaticintervengroup'snatural(unsupported) tions are not necessarilybetter for the group. The effectivenessof the technology dependson its appropriate designand use by the group.Communicationtechnologies
*Accepted by George P. Huber; received February 1985. This paper has been with the authors 18 months for 3 revisions. 589 0025-1909/87/3305/0589$0 1.25
Copyright ? 1987, The Institute of Management Sciences

590

GERARDINEDESANCTIS AND R. BRENT GALLUPE

availablewithin a GDSS includeelectronicmessaging,local- and wide-areanetworks, teleconferencing,and store and forward facilities. Computer technologies include multi-useroperatingsystems, fourth generationlanguages,data bases, data analysis facilities, data storage and modificationcapabilities,and the like. Decision support technologiesincludeagendasetting,decisionmodelingmethods(suchas decisiontrees, risk analysis, forecastingmethods, and multiattributeutility functions), structured group methods (such as the Nominal Group and Delphi techniques),and rules for directinggroup discussion.Empiricalinvestigationsinto the use of GDSS have been reportedby Adelman(1984), Grayet al. (1981), Gallupe(1985), Lewis(1982), Turoff and Hiltz (1982), Quinn et al. (1985), and others. In orderfor researchon GDSS to generatecumulativeknowledge,a common perspective regardingthe purpose and characteristics these systems is required.In of particular,a taxonomy outlining variablescritical to GDSS design and impacts is neededto identifyresearch questionsand organizeresearch results.The purposeof this paperis to providea foundationfor the studyof decisionsupportsystemsfor groups.A conceptualoverviewof GDSS, theirpurpose,components,and anticipatedimpacts,is presented. Ourperspective basedon an information-exchange of groupdecision is view making and posits that the effects of GDSS occur due to changes in the pattern of interpersonal communicationbroughtaboutby the technology.In particular, use of the GDSS alters the nature of participationwithin the group, which, in turn, impacts decision quality and other outcomes of a meeting. Three levels of GDSS systemsare an described,each representing increasing degreeof technologicalsophistication a and more dramatic intervention into the process of group exchange. We propose that researchproceedprogressively throughthese three levels of systems,only considering morecomplexsystemsafterthe effectiveness less sophisticated of designsis understood. This paperproposesa multidimensional taxonomy of GDSSs using a contingency In approach. particular, suggestthatthe designof supportsystemsbe drivenby three we factors: size of the group,the presenceor absenceof face-to-face the and interaction, the task confrontingthe group (see Figure 1). Empiricalresearchcan proceedsystematically by examining the role of decision support technology within each cell of the taxonomy.Our paperconcludesby discussingresearch issuesrelatedto the designand impactsof GDSS technology.Generalhypothesesand importantconstructsto be used in GDSS researchare identified. 2. GroupDecisionMaking
The Decision-Making Group:A Definition

It has been well documentedthat managers spenda greatdeal of theirworkinghours in meetingsand that much of this meeting time is not for decision-making purposes (Argyris Schon 1974;Hoffman 1979;Mintzberg1973).But what we wantto focus and on herearethose situationswheregroupsmust, in fact,reacha decision,and wherethe intentionis that the decisionbe implementedfollowingthe collaborative experienceof the groupmeeting. For our purposesthen, a decision-makinggroup can be defined as two or more peoplewho arejointly responsible detectinga problem,elaborating the natureof for on the problem,generating possiblesolutions,evaluatingpotentialsolutions,or formulating strategies implementingsolutions.The membersof a groupmay or may not be for locatedin the same physicallocation, but they are awareof one anotherand perceive themselvesto be a partof the groupwhichis makingthe decision.Common examples of decision-makinggroupsinclude projectteams, governingboards,legislativecommittees, and medical teams. In most instancesthe grouphas been delegatedresponsibility for mattersconsideredto be too significantfor any single individual.Through

FOUNDATIONFOR STUDY OF GROUP DSS GROUP SIZE Smaller Dispersed Face-to-Face Larger

591

Planning Creativity

>< Intellective|

Preference Cognitive Conflict Mixed Motive


FIGURE

1. A Contingency for Perspective GDSS Research.

discussion, the group aims to utilize the resources brought to the meeting by each individual and to generate a high level of motivation among the membership to solve the problem and implement the group's decision.

Process Viewof the GroupDecision-Making An Information-Exchange


A group decision occurs as the result of interpersonal communication-the exchange of information among members. The group's approach to making a decision is exhibited in their patterns of interpersonal communication. More specifically, the decision process is revealed in the production and reproduction of positions regarding group action, which are directed toward the convergence of members on a final choice. The communication activities exhibited in a decision-related meeting include proposal exploration, opinion exploration, analysis, expressions of preference, argumentation, socializing, information seeking, information giving, proposal development and proposal negotiation (Bedau 1984; Poole 1983a). Supporting group decision making primarily involves changing, in a positive direction, the interpersonal exchange that occurs as a group proceeds through the problem solving process. In this sense the goal of GDSS is to alter the communication process within groups. The greater the degree of change in communication introduced by the technology, the more dramatic the impact on the decision process and, presumably, on the decision outcomes. For example, if the GDSS merely automates what occurs in the normal (unsupported) course of events (e.g., such as electronically soliciting and tallying votes in a legislative session) then the impact on the group's exchange processes will be minimal. On the other hand, if the technology determines who speaks when, in what order, to whom, and for how long, the impact on the decision process will be substantial. By providing a group with opportunities to speed up, change the content, or change the direction of message exchange, GDSS technology aims to improve the outcomes of meetings. Outcomes of meetings may be measured on many dimensions, including decision quality and timeliness, satisfaction with the decision, CoSt or ease of imple-

592

GERARDINE DESANCTIS AND R. BRENT GALLUPE

mentation, member commitment to implementation,or the group's willingnessto work togetherin the future.In short, althoughsupportof the cognitive processesof individualgroupmembersmay be includedin a GDSS, the primaryaim of the group exchange.It component of the system must be to alter the structureof interpersonal follows that GDSS researchmust be deeply concernedwith the impact of alternative and systemconfigurations featureson the natureof communicationin the group. Patternsof Information Exchange Earlyresearch groupbehaviorarguedfor the existenceof phasessuch as "orientaon and "control"in decision making(Babadand Amir 1978; Bales tion," "evaluation," and Strodtbeck1951;Bennisand Shepard1956;Mabry 1975;O'Connor1980;Tuckman 1965);however,more recentresearch indicatesthe presenceof multiplesequences in groups (Burlingameet al. 1984; Poole 1981; Scheidel and Crowell 1964; Poole in 1983a).In fact, the variability approaches decisionmakingacrossgroupsappears to to be even greater Thereis even substantial variationin the thanthatacrossindividuals. developmental stages experiencedby successful and unsuccessful problem-solving groups(Hirokawa1983). The only meaningfuldifferencebetweenthese two extreme grouptypesis that successful groupstend to devoteadequatetime to problemformulation and planningof meeting strategy,whereasunsuccessfulgroupstend to immediately begin to searchfor alternativesolutions(Hackmanand Kaplan 1974;Hirokawa 1983). Just as individualdecision makerstend to have unique stylesin their decision processes,dictating flexibilityin the design of decision support systems, so too do groups. in of Our understanding what transpires the initial stagesof group developmentis of somewhatbetterthan our understanding the later stagesof groupdecision making by cautious (Babadand Amir 1978). Initialmeetingtime is characterized uncertainty, interaction, concernfor integration and (O'Connor1980;Babadand Amir 1978;Bales and Strodtbeck1951). The patternof interactionsexhibitedby a group of decision At makerschangesas a meetingprogresses. different pointsin time the groupis likelyto interpretcomments differently,use decision rules differently,or rely on power in differentways (Poole et al. 1985). The exact natureof these differences variessignificantly from groupto group. The only consistentpatternobservedacrossall types of groupsis that two majorinteractionorientationsdominateas the meetingprogresses. have recognizedthat interpersonal Since the work of Bales (1950), groupresearchers exchangein groupstends to be orientedtowardeither task (gettingthe job done) or social needs (tension release, agreement/disagreement, solidarity/antagonism)(cf. Blakeand Mouton 1964).The relativeemphasison thesetwo concernstendsto shiftas a meeting progresses(Bales and Strodtbeck 1951; Hare 1962). Bales-claimed that groupsinherentlyseekequilibrium betweenthe need to completethe taskand the need to maintain the group. Measurementschemes have been developed to isolate these componentsin interpersonal exchange,althoughthe constructsare difficultto separate (Fisher1971;Mungyand Doise 1982). and follow from this discussion Severalimplicationsfor GDSS designers researchers of patternsof informationexchange.First,GDSSs must accommodatea wide rangeof for in decisionprocesses groups.A normativedecisionprocesscannotbe established all groups.Second,a largenumberof groupscan benefitfromtools thatallowthem to plan Initialsystemsmight include supportfor agendasettingand proba meetingstrategy. lem formulationin groups.Third,GDSSs shouldaim to supportboth the social needs that activities.Encouraging communications deal of groupsas well as theirtask-focused of with groupmemberfeelingsareimportantfor members'acceptance the solutionand their willingnessto work together in the future (Miner 1979). Finally, researchers studyingthe impact of GDSS on decision making must anticipatethe difficultiesof

FOUNDATIONFOR STUDY OF GROUP DSS

593

tracking decision paths and deciphering decision patterns in groups. Merely capturing general stages is inadequate for understanding how the technology is influencing the meeting; rather, longitudinal methods that capture changes in members' cognitions, actions, and feelings over time are required. With these initial guidelines in mind, we can begin to elaborate a conceptual view of GDSS. 3. Group Decision Support: Three Levels of Systems Within the information-exchange view of group decision making, three approaches to supporting the group are possible. Level 1 GDSSs provide technical features aimed at removing common communication barriers, such as large screens for instantaneous display of ideas, voting solicitation and compilation, anonymous input of ideas and preferences, and electronic message exchange between members. Level 1 features are found in meeting rooms commonly referred to as "computer-supported conference rooms"9 "electronic board rooms." Level 1 systems improve the decision process by or facilitating information exchange among members. Preliminary systems developed by Execucom (Kull 1982), Gallupe (1985), and Lewis (1982) fall into this Level 1 category. Example GDSS features within a Level 1 system are shown in Table 1. Level 2 GDSSs provide decision modeling and group decision techniques aimed at reducing uncertainty and "noise" that occur in the group's decision process. Level 2 thus represents an enhanced GDSS, as opposed to Level 1 which is a communication medium only. A Level 2 GDSS might provide automated planning tools, or other aids commonly found in individual decision support systems, for group members to work on and view simultaneously, again using a large common screen. Modeling tools to support analyses that ordinarily are performed in a qualitative fashion, such as social

TABLEI withCorresponding ExampleProblems Needsof Decision-Making and Groups Features a Level I GDSS for GroupProblemor Need Sendingand receiving information efficiently amongall partiesor specificgroupmembers Accessto personal datafilesor corporate data duringthe courseof a meeting Displayof ideas,votes,data,graphs,or tablesto all members simultaneously Reluctance some members speakdue to their of to ideas shyness,low statusor controversial to due Failureof some members participate to lazinessor "tuningout" Failureto efficiently organize analyzeideasand and votes Failureto quantifypreferences or Failureto developa meetingstrategy plan Failureto stickwith the meetingplan GDSS Feature Electronic messaging, broadcast point-to-point or Computer terminalfor eachgroupmember; to gateway a localareanetworkor central computer screen Largecommonviewingscreenor "public" at each groupmember's terminal Anonymousinputof ideasand votes Activesolicitationof ideasor votes fromeach groupmember and summary Summary displayof ideas;statistical and displayof votes Provideratingscalesand/orranking schemes; solicitand displayratingsand rankings Providea mock agendawhichthe groupcan complete Continuously displaythe agenda; providea time clock;automatically displayagendaitemsat the appropriate time

594

GERARDINE DESANCTIS AND R. BRENT GALLUPE TABLE 2

ExampleProblems Needsof Decision-Making and withCorresponding Features a Level2 GDSS Groups for
Group Problem or Need Need for problem structuring, planning, and scheduling Decision-analytic aids for uncertain future events Decision-analytic aids for resource allocation problems Decision-analytic aids for data-oriented tasks Decision-analytic aids for preference tasks Desire to use a structured decision technique but insufficient knowledge or time to use the technique GDSS Feature Planning models, e.g., PERT, CPM, Gantt Utility and probability assessment models, e.g., decision trees, risk assessment Budget allocation models Statistical methods, multi-criteria decision models Social judgment models Automate the Delphi, Nominal, or other ideagathering and compilation technique(s); provide an online tutorial for the group or a human facilitator

judgment formation, risk analysis, or multiattribute utility methods can be introduced to the group via a Level 2 GDSS (see Table 2). In addition, group structuring techniques such as those developed in the organization development literature can be efficiently administered to groups within Level 2 technology. Turoff and Hiltz (1982) have experimented with automating the Delphi method and the Nominal Group Technique, and Huber (1982) has discussed the possibility of automating similar group structuring methods that can be cumbersome and ineffective when administered manually. Group facilitators may provide an interface between the group and the technology to assist the group in using computer-based models; the Decision Techtronics Group at SUNYAlbany has combined human and technical support to groups in "automated decision conferencing" (Quinn et al. 1985). Level 2 systems also have been developed by Steeb and Johnston (1981), and Bui and Jarke (1984). Level 3 GDSSs are characterized by machine-induced group communication patterns and can include expert advice in the selecting and arranging of rules to be applied during a meeting (see Table 3). To date, very little research has been done with such high-level systems. Hiltz and Turoffs (1985) description of computer-mediated communication systems (CMCS) which actively filter and structure information exchange are a type of Level 3 system. At this level each member of the group represents a node in the communication network and deliberate communication patterns are imposed on

TABLE 3

and ExampleProblems Needsof Decision-Making withCorresponding Groups Features a Level3 GDSS for
Group Problem or Need Desire to enforce formalized decision procedures Desire to select and arrange an array of rules for discussion Uncertainty about options for meeting procedures GDSS Feature Automated Parliamentary Procedure or Robert's Rules of Order Rule base; facility for rule selection and application Automated counselor, giving advice on available rules and appropriate use

Desireto developrulesfor the meeting

Rule-writing facility

FOUNDATION FOR STUDY OF GROUP DSS

595

the groupby the technology.For example,specificpatternsof exchangecan be forced on the groupby alteringthe degreeof centralityof each node. Bavelas(1968) created ring, tree, and star communication patterns in groups by placing walls between membersand usingslotsin the wallsfor them to exchangewrittenmessages.Electronic mediawould not only speed up this process,but rulescould be appliedto changethe natureof communicationchannelsdependingon the content,timing, or otherfactors occurringin the group discussion.A rule base, coupled with a fast-pacedquery and feedbacksystem,would allow rapidapplicationof decision rulesto deliberately structure the content or timing of interpersonalexchange. Along these lines, Stodolsky (1981) has developed a preliminarysystem for large group conflict resolution that automatically mediatesconflictinggroupmemberrequeststo have the floor.Members signala computermediatorwhen they wishto speak;the systemstoresand respondsto requestsaccordingto predefined rules,subsequently displayingthe name of each new speakerand activatingthat person'smicrophoneor telephone. Systems that include rule bases to apply parliamentary procedure,Robert's rules of order, or discussion procedures self-designed the groupwould also be consideredLevel 3 systems.In the by long run, GDSS at this level also should provide advice to the group in selectingor buildingrulesfor groupdiscussion. The objectiveof decisionsupportsystemsforgroupsis to discoverand presentgroups with new possibilitiesand approaches makingdecisions.They do this by actingon for the group'sinformationexchange.Messageexchangecan be hastenedor smoothedby removingcommon barriers in Level 1 systems);systematictechniquescan be used (as in the decision process (as in Level 2 systems);or rules for controllingthe pattern, timing,or content of informationexchangecan be imposedon the group(as in Level3 the systems).The higherthe level of the GDSS,the more sophisticated technologymust be and the more dramaticis the interventioninto the group'snatural(unsupported) decisionprocess. 4. A Contingency Perspectivefor GDSS Research We proposethat researchinto the design and use of GDSS should proceedin an iterativemanner,beginningwith Level 1 and Level 2 systems and advancingto the of study of Level 3 systemsaftersome understanding the neededfeaturesand impacts of lower level systems has been achieved. With this approachin mind, how do researchers begin?A criticalfirst step is to constructa softwareenvironmentin which alternative GDSS designsand featurescan be comparedfor theirrelativeeffectiveness. In conceivingof this softwareenvironment,we wouldpreferto havea generalmodel for GDSS design,which could be appliedto each of the threelevels. However,we are prohibitedfrom constructingan architecture suitablefor all group decision support situations,even within a given GDSS level, for severalreasons.First,as noted earlier, there is considerablevariationin information-exchange patternsacross groups and generalizedstagesof decision makingcannot be empiricallyverified.Second, there is how structure be addedto a group'sdecision can little conclusiveknowledgeregarding processto yield better decision making.In the case of a Level 2 GDSS, the problem becomes selectingamong many models or techniquesto build into a system.Over 70 differentproblem solving techniques have been offeredfor use in groups (see Van Gundy 1981),but only a few havebeen subjectto regular studyor use. Forexample,the Delphi method and Nominal Group Techniquehave been shown to increasegroup productivity and satisfactionover freelyinteractinggroups(Van de Ven and Delbecq to 1974). In general,researchsuggeststhat addingstructure the decisionprocesspositivelyimpactsdecisionoutcomes(Smith 1973).The key effectof structured'approaches appearsto be increasingmemberparticipation (Whiteet al. 1980), focusingthe problem, avoidingconformitypressures, and keepingthe group on track(Gallupe 1985).

596

GERARDINE DESANCTIS AND R. BRENT GALLUPE

But the relative advantagesof one method of structuringover another have been difficultto demonstrate(White et al. 1980). Prescribinga particularstructuredapGDSS is thus very difficult. proachfor use in a generalized blockto creatinga generaldesignat any of the threeGDSS levelsis A finalstumbling proximity.Because variables, suchas groupsize and membership the role of situational these factorsinfluencethe patternof informationexchangeexhibitedby groups(Hare 1962;Hare et al. 1967;Hoffman 1979), we can infer that the natureof technological task supportshouldvaryas a functionof such factors.In addition,the particular that a groupaddressesmay impact a group'sdecision processesand the design of the information systemto improvedecisionmaking. in Given the variability behaviorand supportneedsacrossgroupsand groupsettings, of the and the currentlack of knowledgeregarding structuring groupdecisionmaking, from the most appropriate starting point in GDSS designis to buildan arrayof features which a group can select during the course of a meeting. We propose that GDSS to researchers begin systemsdesignby ai,ming supportgroupexchangeactivities(proposal development, opinion exploration,expressionsof preference,argumentation, socialization,etc.), and be less concernedabout the precisesequencingof these activities. Once a group'sdecisionpath becomes understood-through analysisof informacan tion exchangeoverthe courseof a meetingor seriesof meetings-GDSS features be in added or arranged a manner that encouragesthe group to adopt a more effective decisionpath. The relativeadvantageof alternativesystem features,or ordereduse of features,can be comparedto determinethe impacts of systems design on decision processesand outcomes. Within this perspective,GDSS design begins by identifyingfeaturesthat may be usefulin a varietyof groupssettingsthatcan be offeredwithina "shell"system(a menu equivalent of facilitiesto aid the group).This approachis akin to creatinga group-level of of whatSprague to Oncethe features the shell (1980) has referred as a DSS generator. systemhave been outlined-to supportthe generalor typicalneeds of groups-we can groupdecisionsettings.To do this, we proceedto explorethe specialneedsof particular ideallywould like to have a contingencytheoryof groupbehaviorfromwhich a GDSS architecture could be mapped.Althoughsome preliminary workhas been done in this contingencytheoryis available.Consequently, area(Poole 1983b),no comprehensive our approach designbeyonda GDSS shellwillbe to identifya few criticalsituational for (see variables and describehow GDSS designcan varyas a functionof these variables Figure 1). We will develop a taxonomy of GDSS systemsthat considersthe support needs of four specificgroupsettings,dependingon groupsize and memberproximity. This taxonomyis then expandedto considerthe role of a group'stask in the decision supportprocess.
The GDSS Shell

potentiallyusefulto The objectiveof the GDSS shellis to providean arrayof features a variety of decision-makinggroups. In the case of a Level 1 system, a reasonable approachto systemsdevelopmentis to identifycommon needs and problemsin decision-making groupsand then determinehow technologymightbe usefulin supporting those needsand overcomingcommon problems.In Table 1 we providesucha mapping and process.Van de Ven and Delbecq(1971; 1974),Hackmanet al. (1976), Cartwright Zander(1968) and others(e.g.,Festinger1968;Fisher1980;Janis 1972)havediscussed the inhibitinginfluenceson groupproblem-solving, their ideas form the basis for and Table 1 and the followingdiscussion. The common problemsexperienced decision-making groupsinclude:dominance by of discussionby one or more members;extremeinfluenceof high-status membersand members; toleranceof minority low lack of acknowledgement the ideasof low-status of

FOUNDATIONFOR STUDY OF GROUP DSS

597

or controversial opinions;inabilityto accessinformationthat is down the hall or in the computerduringthe courseof the groupmeeting;and undueattentionto socialactivities relativeto the taskactivitiesof the group.The GDSS shell systemcan address these needsby providingelectroniccommunication,anonymousinput and displayof ideas, solicitationand analysisof votes and preferences, accessto personalor corporate and databases. The GDSS shell can be enhancedto includeLevel 2 features,such as those given in Table 2. For example, a module might be designedto encouragegroup membersto voice dissentingopinions or play a devil's advocaterole so as to avoid "groupthink" beforea criticaldecision is made. Group activitiesmight be storedand used later for self-analysisby the group, or for preparinga public report on the group decision process.Planningaids, utility models, or forecasting tools could be made availableto the group,with resultsdisplayedon a common viewing screen. Finally, modules for usingthe Nominal Grouptechnique,Delphi method,or similarstructured approaches could be incorporated into the Level 2 shell. The technologyrequired builda Level3 GDSS shellis not yet available, this is to and one reasonto delay researchat this level. The Level 3 shell would be expert-system based,withthe abilityto applypre-setor group-selected rulesto the meetingprocedure. As suggested Table3, the systemmightenforceParliamentary in procedure Robert's or Rules of Orderin the meetingand, more importantly, might provideadviceon selecting or definingrulesfor groupdiscussion-such as who speakswhen, for how long, on what topic, etc. In developinga Level 3 shell researchers must keep in mind that the and goalof GDSS,even at this level, is to provideflexibility varietyin decisionpathsfor a group. For experimentalpurposeswe may be interestedin imposing one pre-set on procedure a group,but ultimatelythese systemsshouldleave controlof the technology in the handsof the group,not the system. In additionto task-oriented support,featuresintendedto addressthe social needs of groups should be included in GDSS shell systems. For example, Turoff and Hiltz (1982) have suggested use of an electroniccoffeebreak the wherebymemberswho are communicatinglong-distance interjectpersonalcomments,jokes etc., duringthe can course of a meeting. Similarly,a GDSS system developedat the IndianaUniversity includesa smokingcall in whichany membercan interrupt meetingto anonymously a poll othermembersto askif they arebotheredby smokein the meetingroom. Development of a shell systemthat supportsboth task and socializationneeds will requirean with the innovativesystemsdevelopmentteam, coupledwith iterativeexperimentation systemin live groupmeetings.Ideally,the shell systemat any level shouldallowgroups to modify existingfeaturesor developtheirown features(see Hiltz and Turoff 1981). A Taxonomyof Systems The GDSS shell system should be useful to groups across a variety of decision settings.However,once the shell is constructedthe system can then be enhancedto accommodate needsof specialgroupsituations.The needs and dynamicsof groups the on on varydepending a varietyof situational factors.The socialscienceliterature group on behavior,as well as limited research electroniccommunicationand groupdecision support,suggeststhat the nature of the informationexchangeand the outcomes of decisionmakingchangewhen groupsbecome extremelylarge,have irregular communication, or when face-to-facecontact is absent (Hiltz and Turoff 1978; Leduc 1979; Valeeet al. 1974).Of thesefactors,groupsize and the proximityof membersduringthe meetingappearto be the most criticalto GDSS design. Classifying groupsaccordingto size is ratherarbitrary; perhapsit is best to think of smallor relatively groupsas beingrelatively large.Althoughthe fundamental principles of groupdynamicsapply to groupswith from two to severalthousandmembers,the

598

GERARDINE DESANCTIS AND R. BRENT GALLUPE

predominant activitiesof a groupdifferbetweensmalland largegroups(Cartwright and Zander 1968). As membershipincreases,the number of potential information exchangesrisesgeometrically, the frequency,durationand intimacyof information and exchangeall decline. Consensusbecomes harderto achieve, and affectionalties and satisfaction with the groupdecline (Cartwright Zander 1968;Hare 1962;Thomas and and Fink 1963; Hoffman 1979). There is greaterinterestin giving informationand suggestionand less interestin askingopinion, giving opinion or showingagreement. Smaller groups are more likely to actively attempt to resolve opinion differences, whereaslargergroupstend to use humoras a tension-reducing mechanism(Balesand Borgatta1955;Hare et al. 1967).Thesedifferences imply that GDSS designmust vary somewhat across large and small groups. For example, in the Level 1 system vote tallyingand displaymay be of more use in largegroupsthan in small ones, whereas anonymousmessageexchangemay be of more use in small groupsthan in largeones. Becauselargegroupsexperiencemore dramaticcommunicationdifficulties,decision supportsystemsmay have a more positiveimpactin largegroups. In addition to'group size, the design of a GDSS, as well as its impacts, will be somewhatdifferentwhen group membersare remote from one another ratherthan Remotegroupcommunicationmay be usefulin situationswhere meetingface-to-face. but a face-to-face meetingis preferred is inconvenient,or wherea face-to-face meeting is possiblebut is avoidedbecauseindividualsarebest left to workalone on portionsof the group task. The best example of the latter case is where creative solutions to a known problemmust be generated. Here, individualsworkbetteralone than together tend to inhibitcreativity developmentof quality and since socialcomparisonprocesses solutions(Allport1920;Hare 1962;Mungyand Doise 1982;Van de Ven and Delbecq 1974).Decision supporttechnologycan be used to pool, organizeand feed back ideas collectedfrom individualsworkingremotely.Althoughelectroniccommunicationin some ways allows separated groupmembersto interactas if they are in close physical in proximity,thereare some distinctdifferences the exchangeactivitiesof remoteand face-to-facegroups. Groups using computer-mediated communication, when comparedwith face-to-facegroups,participatemore equally within the group, are more uninhibited,and reachdecisionswhich deviate furtherfrom initial individualpreferences (Siegelet al. 1986).However,while dispersionof memberscan improvedecision conflict tends to be greater,communication accuracyfor some tasks, interpersonal becomeslessefficient,and satisfaction withthe groupprocesstendsto decline(Bojeand Murninghan1982;Siegelet al. 1986). The taxonomy shown in Figure 2 presentsfour environmentalsettingsbased on differencesin group size and dispersionof group memnbers. These settings are not mutuallyexclusive,but representbipolarextremesthat will affectthe dynamicsof a groupmeetingand the selectionof decisionsupportfeatures. Althoughothercategories for illustrating how GDSS design arepossible,thesefourprovidea usefulstarting point
GROUP SIZE Smaller Decision Room Local Area Decision Larger Legislative Session Computer-Mediated Conference

MEMBER PROXIMITY

Face-to-face

Dispersed

Network
FIGURE

Cneec

2. A Taxonomy of GDSS Settings Representing the End Points of Scales for Group Size and

Proximity GroupMembers. of

FOUNDATIONFOR STUDY OF GROUP DSS

599

can be made compatiblewith the decisionsetting,and how the impactsof the technolare to ogy can varyas a functionof the setting.GDSS researchers encouraged consider the particular settingwhen includingGDSS featuresand evaluatingtheir impact. The Decision Room (smaller group,face-to-face meeting). In this setting group membersarein a singlephysicallocationand meet for a fixedblockof time to addressa particular problemor set of problems.This may be thoughtof as the electronicequivalent to the traditionalmeeting. The organizationsets up a room (much like a boardtablefacinga largescreen,with room)with specialfacilities,such as a horseshoe-shaped Groupmember a displaymonitorand data input device availablefor each participant. communicationsmay be transmittedverballyor via electronicmessaging.The public screen is used to display ideas, show analyses of group preferencesand votes, and display decision aid tools (spreadsheets, graphs,charts, etc.). Most availableGDSS productsfall into this category(see Kraemerand King 1984).For furtherdiscussionof the Decision Room see Grayet al. (1981) or Huber(1982, 1984a). The LegislativeSession (largergroup,face-to-facemeeting). If the group membership becomes too largeto be accommodatedin a Decision Room, a largerfacilitywill will be needed.Eachparticipant need an input device and monitor,althoughaccessto these may be shared by groups of two or three people. In a Decision Room each participantis able to send a messageto all other membersand to the public screen, or may be permittedto Sessiononly a facilitator chairperson whereasin the Legislative send informationto the largescreen.Also, member-to-member communicationmay be hierarchically establishedso that memberscan send messagesto only their fellow party members or party chairpersonbut not to the meeting chairpersonor to the usefulin this setting, may be particularly meetingmembership large.Level 3 software at can enforcementof the meetingprotocol.Meetingproceedings be recorded, facilitating who are affectedby the group'sdecisions. analyzedand summarizedfor constituencies in meetings,and othergatherings which Governmentlegislativemeetings,stockholder formalized are procedures used to involve a largenumberof peoplein decisionmaking could benefitfrom such computer-based support. The Local Area Decision Network(smallergroup,dispersed). The GDSS will take on a differentconfigurationif groupsdo not meet face-to-face.Projectteams, sales groups,and committeesmay benefitfrom electroniccommunicationand Level 1 aids that particularlyaddressthe needs of groups who are physicallydispersed.Several A are systemconfigurations possiblewithinthis GDSS category. local areanetworkcan in electronicallyconnect group memberslocated at workstations the privacyof their offices.Long distancenetworkscan link memberswho are at home or travelingaway can from the office,or teleconferencing be used to link two or more Decision Rooms together.Groupmeetingsmightextendover a periodof days,and may not requirethat all group membersbe online simultaneously.An electronicor human facilitatorcan for promptparticipants specificinputsto the meetingor guidethe sessionaccordingto a predefined agendaor decisionprocedures. The Computer-Mediated Conference (largergroup,dispersed). If a largenumberof people who are physicallydistant from one another must participatein a decisionmaking session, a somewhat differentGDSS configurationis required.Instancesin which large, dispersedgroups are responsiblefor decisions are rare, but this type of systemmay become more frequentin the future(Valleeet al. 1974).The technological basis for this categorywill be long distancetelecommunications networksand group decisionsupportsoftware. Althoughit is conceivablethat a largenumberof employees in the same organization may use electroniccommunicationas a medium for decision but making,more likely the memberswill be from differentorganizations sharesome common affiliation,such as political party membership,or membershipin a labor union or professional group.

600

GERARDINE DESANCTIS AND R. BRENT GALLUPE

be As in the LocalAreaDecision Network,all groupmemberswould not necessarily required be online simultaneously orderfor a meetingto take place;an electronic to in can for to facilitator promptparticipants specificinputsor guidethe sessionaccording a in predefined agenda.However,the conferencingsettingwill call for some differences GDSS shell featuresand system use when comparedto the local networksetting.In groupwill contrastto the smalldispersed group,communicationin the largedispersed tend to be more structured and formal;informalmessageswill be less frequent;and usage levels will vary greatlyacross members,with some membersusing the system with associated only on a veryoccasionalbasisand othersusingit heavily.Complexities the large group size may requirea hierarchyof shell features,with certain menus but availableto some subgroups not to others.In the local areadecision environment thereis usuallythe potentialfor a face-to-face meeting,but in the computer-mediated conference there is not. Consequently,while the GDSS in the local area decision networkmay concentrateon supportingpre-meetingactivities, such as establishing agendaitems, the GDSS in the conferencesetting must be capableof leadinggroup membersto problemsolutionswithoutextended,informaldialogueand usuallywithout the possibilityof face-to-faceinteraction.A significantprogramof researchon computer-mediated conferencinghas been ongoing for nearly ten years at the New JerseyInstituteof Technology(Hiltz and Turoff 1978, 1981, 1985;Hiltz et al. 1985). Theirresearchindicatesthat the supportneeds of a dispersedgroupchangesdramatifeaturesthat filterand becomes very large.More specifically, cally as the membership in communicationare necessary orderto avoid informationoverloadduring structure with Level3 softwareassociated groupdiscussions(Hiltz and Turoff1985).Rule-based GDSS is neededto provideshape,continuity,and socialorderto thesegroupmeetings. TheRole of Task task that confrontsa Beyond the situationalfactorsjust considered,the particular groupmust be a drivingforce in GDSS design.As Poole et al. (1985) point out, "the often generalvariable'grouptasktype' is emergingas an especiallyimportantvariable, (p. accountingfor as much as 50%of the variancein group performance" 88). Task attributesdetermine the need for informationand the consequent communication of practices the decision-making group(Poole 1978).A grouptaskcan be characterized its goals, criteriafor completion, rules and roles that must be followed, imposed by stressor time limits, or consequencesof successor failure(Hare 1962; McGrathand to However, Altman 1966).Everytaskwill be uniquewith regard these characteristics. in a generalcategorization scheme is importantfor progress GDSS research. schemeshave been proposedin the groupliterature Severaltask categorization (e.g., Hackman1976;Hackmanand Morris1975;McGrathand Altman 1966;Shaw 1973). Recently McGrath (1984) integratedthese various approachesinto a "circumplex model"of grouptask types.Accordingto this view, taskscan be categorized according to whatthe groupmust accomplishduringthe courseof its meeting.Majorgroupgoals in decision-related meetingsinclude:(1) GENERATINGideas and actions.Planning tasksrequiregenerationof action-oriented tasksrequiregeneration of plans.Creativity novel ideas. (2) CHOOSINGalternatives.Intellectivetasks require selection of the tasksrequireselectionof an alternative whichthereis for correctalternative. Preference no objectivecriterionof correctness.And (3) NEGOTIATINGsolutions. Cognitive conflict tasks involve resolution of conffictingviewpoints, and mixed-motivetasks involve resolutionof conflictingmotives or interests. Dependingon the type of task confrontingthe group,the featuresselectedfromthe shell GDSS will vary, and the added featuresor enhancementsto the system may change(see Table 4). The contributionof GDSS to GENERATEtasks should be to facilitateinput and displayof ideasfromall groupmembers,to speedup the processof

FOUNDATION FOR STUDY OF GROUP DSS TABLE 4 Example GDSS Features to Support Six Task Types Task Purpose GENERATE Task Type Planning GDSS Level Level 1 Level 2 Possible Support Features

601

Creativity

Level 1

Level 2 CHOOSE Intellective Level 1 Level 2

Large screen display, graphical aids; Planning tools (e.g., PERT) Risk assessment, subjective probability estimation for alternative plans Anonymous input of ideas; pooling and display of ideas; search facilities to identify common ideas, eliminate duplicates; NGT, Brainstorming Data access and display; synthesis and display of rationales for choices; Aids to finding the correct answer, e.g., forecasting models, multiattribute utility models; Rule-based discussion emphasizing thorough explanation of logic Preference weighting and ranking with various schemes for determining the most favored alternative; voting schemes; Social judgment models; automated Delphi Rule-based discussion emphasizing equal time to present opinion Summary and display of members' opinions; Using social judgment analysis (SJA), each member's judgments are analyzed by the system and then used as feedback to the individual member or the group Automatic mediation; automate Robert's Rules Voting solicitation and summary; Stakeholder analysis Rule base for controlling opinion expression; automatic mediation; automate Parliamentary procedure

Level 3 Preference Level 1

Level 2 Level 3 NEGOTIATE Cognitive conflict Level 1 Level 2

Level 3 Mixed-motive Level 1 Level 2 Level 3

idea (or plan) evaluation,and to providetechnicalsupportfor creativethinkingtechniques.For example,in planningtasks,wherethe goals of a projectare known but the action plans for achievinggoals must be developed,the group might benefit from a or onto whichall memberscan write largescreenthat servesas a blackboard scratchpad as their planningprocessproceeds(Level 1 GDSS). Computergraphicsor planning models,whichnormallyare usedby individualusers,can be displayedon the common screen,with additionsor modificationsto the model enteredby any group member (Level 2 GDSS). In creativitytasks, wherethe objectiveis to generateas many novel ideas as possible, a Level 1 GDSS could aid in collection, synthesis, filtering,and displayof ideas that were enteredanonymouslyby groupmembers.With the aid of a human facilitator,a Level 2 GDSS could guide the group through Brainstorming, Nominal Groupor otheridea-generation procedures. In the case of CHOOSINGtasks, where alternativesare known and in need of

602

GERARDINE DESANCTIS AND R. BRENT GALLUPE

evaluationby groupmembers,the objectivesof GDSS technologyshouldbe to aid in the selectionof eitherthe correctsolution(in intellectivetasks)or the sociallypreferred solution(forpreference tasks).Facilitiesthataid in datacapturing display(Level 1), and as well as computation, mathematicalmodeling, and statisticalanalysis (Level 2), would be supportiveof intellectivetasks.In preference tasks,wheresocial comparison processesare important,featuresthat promptmemberson theirpreferences, combine these preferencesand then display the results to members for discussion might be includedin Level 1 systems.AutomatedDelphi or otherconsensus-seeking techniques in mightbe incorporated Level 2 systems.In the long run, Level 3 systemsmightaid in applyingdiscussionrules emphasizingpresentationof logic, or rationalefor choices, (for intellectivetasks) or even participation membersin expressionof preferences of (for preference tasks). In NEGOTIATION tasksmembers'interestsare pittedagainsteach otherand what is most effectivefor the group is not necessarilybest for the individualmembers.In addition to voting solicitationand summary(Level 1), stakeholderanalysisand resourceallocationmodels (Level 2) may be of use to the group.But a more significant electronicfacilitator. who has the floor featuremay be a rule-based Decisionsregarding might be based on a preprogrammed, mutuallyagreeableprocedure(Level 3) rather than on who speaksmost or loudest.Prompted,structured questionsthat aim to get at issuesof conflictingpartiesmay be usefulin improvinginter-party commuunderlying nication.Conflictresolutionmodels proposedby game theoristsmay be helpful.Mulin ticriteriadecision models (Bui and Jarke 1984) may also be useful, particularly cognitiveconflicttasks. 5. ResearchDirections In orderto study decision supportsystems for specificenvironmentalsettingsand grouptasks, a common conceptualscheme for organizingthese variablesis required. The taxonomy of GDSS settingsoutlined in Figure2, togetherwith McGrath'scircumplex model of tasks in Table 4 providesa usefulstartingpoint from which GDSS featurescan be identified.Figure1 combinesthesetwo taxonomiesinto a three-dimensional framework GDSS research.We have presentedsome preliminary for ideas regardingthe natureof GDSS supportrequiredfor each environmentalsettingand task type. Furtherdevelopmentand refinementof the ideas presentedhere should yield specificguidelinesfor GDSS design, as well as testablehypothesesfor study within a programof researchon group decision support.The most appropriate tactic for researchers may be to selectone level of GDSS (e.g., from Tables 1, 2, or 3), one category of GDSS (e.g., from Figure 2), and considerthe design issues of that categoryfor a task type (e.g., from Table 4). The impacts of variousfeaturesor levels of particular GDSS on decisionprocessesand outcomescan be comparedwithin and betweencells of the three-dimensional framework. In examiningdecisionprocessesand outcomes relatedto GDSS use, an astounding must come to gripswith what numberof issues become worthyof study. Researchers issues,among the many of relevance,they should address.Put anotherway, the question becomes, "whatdo we study?"
in Socialinteraction decision-making and groupsis characterized suchvariety,complexity, by that disorder it seemsto defyneatanalysis. The keydifficulty apparent seemsto be choosingan appropriate aspectof the group'sbehaviorfor observation (Kerr1982,p. 62).

To addressthis problem,the finalsection of this paperconsidersmajorresearch issues relatedto GDSS. We emphasizethat these are generalresearchareas,not a complete listingof all questionsrelevantto GDSS. Also, these areasare not necessarily independent of one another.We highlightgeneralhypothesesand suggesta research approach

FOUNDATION FOR STUDY OF GROUP DSS

603

in each area. The reader should note the nonspecific and tentative nature of this discussion; constructsmust be more preciselydefinedand the measurement the issues in addressed greaterdetailbeforeempiricalstudiescan be undertaken. 1. GDSS Design An abundanceof research neededrelatedto hardware softwareconfigurations is and of GDSS. Since most researchers today are building systems from the ground up, a wealthof knowledgeis emergingregarding technicaland social difficulties the involved in implementingthis technology,as well as possibilitiesfor overcomingthese difficulties. For example, Bui and Jarke (1986) have suggestedhow data communications mightbe managedwithinGDSS design,and Grayet al. (1981) and Huber(1984a) have consideredthe layout of decisionrooms. But much remainsto be done. in One of the key difficulties designingsystemsfor groupdecisionsupportis that user participationcannot be relied upon as the primaryinput for systems analysis. In buildinga computerized conferencing systemHiltz and Turoff(1981)foundthat "users cannot tell you what they need priorto using this technology."Consequently,users must have extendedexperiencewithGDSS beforethe effectiveness ineffectiveness or of systems design can be fully assessed. Johansen (1976) has observedin the case of teleconferencing systemsthat initialuses of the systemserveas a poor basisfor generalizing about futureuses. Similarly,Hiltz and Turoffreport"an evolution or patternof change towardsgreatercomplexityand specializationand diversityof user behavior over time" in computer-mediated conferencingsettings (Hiltz and Turoff 1981, p. 740). Similar learningeffects are likely in decision rooms, legislative,and network settings, particularly given the relative novelty of GDSS technology in most organizations. 2. Patternsof Information Exchange To the extent that electronic communicationchannels formalize interpersonal communicationwill exchange(Huber 1984b),we might anticipatethat task-oriented dominatesocial-oriented communication GDSS-supported in groups,and that negative social communication dominatepositivesocial exchanges.This hypothesisis conwill sistentwith Turoffand Hiltz'sconclusion(1982) that thereis more task-focused communicationand lessjoking and laughingin GDSS-supported groups,and Siegelet al.'s (1986) findingthat people are more criticalof each other'sideaswhen they communiIn cate electronically. a seriesof studiescomparingelectronicand face-to-face communication modalities, Rutter and Robinson (1981) conclude that social cues are lost and when people do not meet in the same room; they referto this as "cuelessness" reportthat there is increasedattentionto the task, more impersonalcommunication, in has and greater task-orientation physically task-orientation disparate groups.Greater been shown to improve decision quality in some settings(Gallupe 1985; Turoffand Hiltz 1982), but the applicabilityof this result across GDSS settings is unknown. the communicationpatternsrelatedto changesin decisionqualMoreover, underlying ity have yet to be examined. The information-exchange view arguesthat groupprocessesare both determinants and resultingexpressionsof groupdecisionmaking.The impactsof GDSS technology will be revealedin the patternof interpersonal communicationwithin a group,which, in turn,will inflencedecisionqualityand otheroutcomes.Therefore,researchers must is to the interaction discover impactthe technology havingon studytheflowof members' between the natureof the group'scognitions,actions,andfeelings, and the relationship theseattributes information of exchangeand decisionoutcomes.Analysisof audiotapes, to videotapes,and retrospective protocolsof the group'sdiscussionare all approaches capturingthe processof group interaction(Poole et al. 1985). A numberof different

604

GERARDINE DESANCTIS AND R. BRENT GALLUPE

schemes are availableto analyze group interactions.Some are based only on verbal content (Conradand Conrad1956;Mabry 1975;Poole 1980) while othersratewords, or et gestures, any activitythat indicatesa mentalstate(Freedman al. 1951).The more popularsystemsare those described Balesand Cohen (1979) and Fisher(1970). All by requiretrainedratersto assurereliabilityand are difficultto validate.Despite these GDSS designersand researchers encouragedto adopt systematic,reliare difficulties, able methodsto studyinformationexchangeassociated with a particular GDSS design, setting,or task. 3. MediatingEffectsof Participation The results of using decision support technology in groups will occur primarily throughthe effects these systems have on memberparticipationin discussion. For example,anonymousinput methodsand active solicitationof groupmembersto con' in tributeideas should encouragegreaterequalityof memberparticipation groupdisof cussions.Changein the otherwise natural(unsupported) distribution memberparticipation in group discussionmay yield both positive and negativeeffectson decision qualityand otheroutcomes. On the one hand, greateror more even memberparticipation may yield desirable effectsfor the group. Irving(1976) and Turoffand Hiltz (1982) have found the anoto nymityof electroniccommunication increasethe numberof interpersonal exchanges of and reducethe probability any one memberdominatingthe meeting.Full participation of members allows extraction of resourcesfrom a group and promotes error checking,thus makingbetterdecision makingpossible(Hackmanand Kaplan 1974; Holloman and Hendrick 1972). Existing group researchalso suggeststhat greater for memberparticipationshould increaseacceptanceand a sense of responsibility a decision (Bedau 1984; Block 1974; Hackman and Kaplan 1974), as well as greater with the group(Hare 1962). cohesivenessand satisfaction On the other hand, problemsmay arise when naturalpatternsof participation are alteredby technology.For example,suchfeaturesas roll callingfor votes or solicitation by the systemfor input of ideasby all memberswill pressure those who otherwise might be passivein the groupto take an activerole, perhapsincreasing psychological tension in the group (see Nemeth 1982). The result may be that individualswho otherwise might not speak, now speak but remain guardedin their comments so as to avoid sociallyunacceptable conflict.Relatedto this problemis the possibilityof an audience effect. That is, some membersmay fear that, althoughtheir comments are entered anonymously,the computertechnology has the capacityto store opinions for later viewingby someone.Again,psychological tension may increaseand qualityof participationdecline.Some groupresearch that addedmemberparticipation suggests tendsto increasedecisiontime and decreasesatisfaction withthe decisionprocessunlessthereis a highlevel of cohesiveness the group(Nemiroffand King 1975).Controversial in views may surfacethat could havebeen subduedin situationswhereparticipation purely was voluntaryand sentiments could not be expressedanonymously, making consensus more difficultto achieve (Gallupe, 1985;Turoffand Hiltz 1982). In conclusion,both the intendedand actualeffectsof decisionsupportsystemson patternsof participation in a groupare in need of close examinationby researchers. 4. Effectson Perceived PhysicalProximity,Interpersonal Attraction, and GroupCohesion The electroniccommunicationcomponentwithin a GDSS will influenceperceived attraction and physicaldistanceamongmemberswhich,in turn,willaffectinterpersonal group cohesion. The degree of physical distance between people is known to exert considerable influenceon organizational behavior(Hare 1962).Numerousstudieshave

FOUNDATION FOR STUDY OF GROUP DSS

605

attracbetweenphysicalproximityand interpersonal reporteda negativerelationship tion, or liking for one another(Festinger,Schacterand Back 1950; Hare 1962;Newcomb 1961).And degreeof liking,in turn,has been shownto affectgroupcohesiveness and norm development(Hollander 1964). Monge et al. (1985) recentlyarguedthat perceivedphysical distance is more critical to group behavior than actual physical distance.The opportunityand obligationfor communicationthat people perceiveby exchangeand beingphysicallyclose or distantimpactsthe natureof theirinterpersonal feelingsof cohesiveness. if Groupcohesivenessmay be irrelevant decisionqualityis the primaryobjectiveof the group;however,the variablebecomesextremelyimportantif strongmorale,long(Festerm cooperationand conformityto groupnormsare criticalto the organization
tinger 1968; Hollander 1964; Shaw 1976). We can anticipate that electronic communi-

cation will impact groups in differentways, depending on the degree of perceived physical distance that they would experiencewithout the GDSS. For example, in face-to-face meetingsthe GDSS should increaseperceiveddistanceas electroniccomfor municationis substituted at least some directverbalcommunication.On the other hand,if groupmembersarelocatedremotelyfromone another,the GDSS may reduce for perceiveddistanceas opportunities interactionare increasedand richnessis added to the verbal exchangesthat otherwisewould occur in telephone conversationsor shouldbe to examinethe One directionof GDSS research simpleelectronicmessaging. effectsof variouscategoriesof systemson perceivedphysicalproximity.Of particular betweenmemberproximityand the group'sability interestis the long-termrelationship to perceiveitself as an effective,workinggroup. 5. Effectson Powerand Influence The usualpoliticaldynamicsof the groupwill changeonce technologyis introduced to the process.To the extentthatGDSStechnologyencourages equalityof participation and discourages dominanceby an individualmemberor subgroup, perceivedmember power and influence should become more distributedand decision quality should improve.The loss of social cues due to electroniccommunicationcan be expectedto encourage open input of creativeideas,discoveryof optimalsolutions,and selectionof an alternative based on its meritsratherthan on compromise(Boje and Murninghan 1982;Rutterand Robinson 1981). The resultshouldbe improveddecision qualityin the case of creativity,decision-making,cognitive conflict, and mixed-motivetasks. is influenceby one memberor subgroup less of a concernin intellective Extraordinary and planningtasks.Here memberdominanceis only damagingif the influentialparty of an encourages incorrectansweror recommendsplanswith a low probability success. Conceivably,data analysistools, planningaids and other GDSS featureswill prevent poorchoices.In short,the effectsof the GDSSon influential membersfromencouraging member decisionquality-will varyasfunctionof powerand influence-and subsequent the grouptask. To the extent that GDSS technologyis used to detachideas from their proponents, ideas will become the object of discussionratherthan the proponentsof those ideas. While encouragingbetter decision quality, this will also preventmemberswith high perceivedpowerfrom exercisinginfluence,and prohibitthe emergenceof new power fromthose seekingit. If ideasand preferences exchangedanonymously,individual are memberswill not know who favorswhat,or wherespecificmembersstandon an issue. Evenif namesareattachedto ideas,ideaswill stilltend to receivegreater attentionthan based(Rutterand Robinson 1981). proponentswhen communicationis electronically The point is that some criticalpolitical information that is often obtainedin meetings Will the lunchroomor hallwaythen will be lost whenGDSS technologyis introduced. be used to find out this informationonce the meeting is concluded?Acquiringand

606

GERARDINEDESANCTIS AND R. BRENT GALLUPE

maintainingpower has some positive valence to all individuals,and meetings are a forum for achievingand maintainingpowerin organizations (Lippitet al. 1968). The impactof decisionsupportsystemson powerand influenceprocesses,both within and outsideof meetings,shouldbe a majorareaof studyfor GDSS researchers. is difficult It to anticipatethe precise natureof the effectsof GDSS on power relationships within organizations generally. However,we can broadlyspeculatethat use of GDSS technology will influencehow individualsthink afterthe meeting,the mannerin which they interactwith one anotheroutside of the meeting,and futureexpectationsabout how decisionsin the organization shouldbe made. 6. ThePerformance/Satisfaction Tradeoff Perhapsthe most difficultissuethat researchers organizations and must addresswith regard GDSSis the desiredbalancebetweendecisionqualityand satisfaction to withthe groupprocess.Greatertask-orientation been shown to improvea group'sdecision has quality, and one of the objectivesof GDSS is to add rationalityand a systematic approachto the decisionprocess.But acceptanceof a group'ssolution frequently a has low correlationwith the objective quality of the solution generated(Hoffman and Maier 1961). In reality,quality is often sacrificedto gain acceptanceand vice versa. Much of the justification participative for decisionmakingin organizations been to has gain acceptanceof ideas, ratherthan gain betterideas per se. Group decision support systemscannotbe expectedto resolvethis problem.If anything,the issue may become more poignantin the case of GDSS. Designers who attemptto build the technology, researchers who lay out plans to study the impactsof GDSS, and users who consider purchaseof these systems all face the same question,"whatis the purposeof a group decisionsupport system?"Is the objectiveto facilitatehigh qualitydecisions,or a high sense of satisfaction with the decision?Turoffand Hiltz (1982) arguethat high satisfaction and high decision quality cannot be simultaneouslyachieved.These may be, by theirnature,conflictinggoals,and the groupmust choosewhichgoal is moreimportant beforemakinguse of electronicsupport.Perhapsthe answerto the performance/satisfactiontradeoffvariesdependingon the organization, setting,and the task. For the the sakeof generalargument, proposethat the long-termobjectiveof GDSS technology we should be to improvethe efficiencyand effectiveness group decision making.Furof with the processis a necessary not sufficientcondibut thermore,membersatisfaction tion for a successfulGDSS. Researchdevoted to GDSS should be deeply concerned with both performance and satisfactionvariables,keepingin the mind that the longterm purposeof decision supportsystems is to improve the quality and efficiencyof organizational meetings. 6. Conclusion We have identifiedsix generalareas for GDSS researchand identifiedmajor constructsfor study in each area. Specifichypothesesmust be more carefullyarticulated and the constructsmore clearlydefinedbeforeresearch any of these areascan begin. in We leave this researchstep to the interestsand creativityof individualinvestigators. The contingencyperspective outlinedin this paperprovidesa foundationfor the study of GDSS. If the work of individualinvestigators conductedwithin this perspective, is cumulativeknowledgeon groupdecision supportin organizations emerge.' can
'This projectwas supported grantsfrom the Associationfor ComputingMachinery, McKnight the by Foundation the Graduate and Schoolof the Universityof Minnesota. The authorswish to thank one of the refereesfr workingwith us across multiple revisionsand for consistently providing with excellentsuggestions improving qualityof the paper. us for the

FOUNDATION FOR STUDY OF GROUP DSS

607

References
ADELMAN, L., "Real-Time Computer Support for Decision Analysis in a Group Setting: Another Class of

Decision Support Systems," Interfaces, 14, 2 (March-April 1984), 75-83.


ALLPORT, F., "The Influence of the Group upon Association and Thought," J. Experimental Psychology, 3

(1920), 159-182. ARGYRIS, ANDD. SCHON,Theory Practice, C. in Jossey-Bass, San Francisco,1974. BABAD, E. Y. AND L. AMIR, "Bennis and Shepard's Theory of Group Development: An Empirical Examina-

9 tion,"Small Group Behavior, (1978), 477-492.


BALES, Interaction R., Process Analysis: MethodfortheStudyof Small Groups, A Addison-Wesley, Reading, MA, 1950. ANDE. F. BORGATTA, "Size of a Group as a Factor in the Interaction Profile," In E. F. Borgatta and R. F. Bales (Eds.), TheSmall Group, Alfred A. Knopf, New York, 1955, 396-413. AND S. P. COHEN, SYMLOG: Systemfor the MultilevelObservation Groups, A Free Press,New of York, 1979. ANDF. L. STRODTBECK, 46 "Phases in Group Problem-Solving," J. Abnormal SocialPsychology, and (1951), 485-495. BAVELAS, "Communication Patterns in Task-Oriented Groups," In D. Cartwright and A. Zander (Eds.), A., Researchand Theory, ed., Harper and Row, New York, 1968. 3rd Group Dynamics: BEDAU,H., "Ethical Aspects of Group Decision Making," In W. C. Swap and Associates (Eds.), Group DecisionMaking,Sage Publications, Beverly Hills, 1984, 115-150. "A BENNIS,W. G. AND H. A. SHEPARD, Theory of Group Development," HumanRelations,9 (Summer 1956), 415-437. R. BLAKE, R. ANDJ. S. MOUTON, TheManagerial Grid,Gulf Publishing Co., Houston, 1964. M. BLOCK, W., "Member Commitment to Group Decisions: A Study of Individual and Group Determinants," Unpublished doctoral dissertation, University of Chicago, 1974. in BOJE, D. M. AND K. MURNINGHAN, "Group Confidence Pressures Iterative Decisions," Management Sci., 28, 10 (1982), 1187-1196. BuI, X. T. AND M. JARKE, "A DSS for Cooperative Multiple Criteria Group Decision Making," Proc. 5th

Internat. Conf Information Systems,1984, 101-113.


AND , "Communications Requirements for Group Decision Support Systems," Proc. 19th AnnualHawaiiInternat. Conf SystemsSciences,1986, 524-533. BURLINGAME, G., A. FUHRIMAN AND S. DRESCHER, "Scientific Inquiry into Group Process: A Multidimen-

sional Approach," Small Group Behavior, 19, 4 (1984), 441-470. D. CARTWRIGHT, AND A. ZANDER (Eds.), Group Dynamics,(3rd Ed.), Row, Peterson, Evanston, IL, 1968. D. CONRAD, C. AND R. CONRAD,"The Use of Personal Pronouns as Categories for Studying Small Group Interaction," J. Abnormal and Social Psychology, 52 (1956), 277-279. FESTINGER, L., "Informal Social Communication," In D. Cartwright and A. Zander, Group Dynamics: Research and Theory (3rd edition), Harper and Row, New York, 1968, 182-191. AND S. SCHACHTER K. W. BACK,SocialPressures Informal in A Groups: Studyof HumanFactorsin Housing, Harper Bros., New York, 1950. 37 FiSHER, B. A., "Decision Emergence: Phases in Group Decision Making," SpeechMonographs, (1970), 53-66. 21 , "Communication Research and the Task-Oriented Small Group," J. Communication, (1971), 136-149. , Small Group Decision Making (2nd Edition), McGraw-Hill, New York, 1980. AND H. S. COFFEY, "The Interpersonal Dimension of FREEDMAN, M. B., T. F. LEARY,A. B. OSSORIO Personality," J. Personality, 20 (1951), 143-16 1. GALLUPE, R. B., "The Impact of Task Difficulty on the Use of a Group Decision Support System," Ph.D. dissertation, University of Minnesota, 1985. Trans. FirstInternat. GRAY,P. ET AL., "TheSMU DecisionRoom Project," Conf DecisionSupport Systems, Atlanta, June 1981, 122-129. HACKMAN, J. R., "Group Influences on Individuals," In M. Dunnette (Ed.), Handbook of Industrial and Organizational Psychology, Rand McNally, Chicago, 1976, 1455-1525. , K. R. BROUSSEAU AND J. A. WEISS, "The Interaction of Task Design and Group Performance Strategies in Determining Group Effectiveness," Organizational Behavior and Human Performance, 16 (1976), 350-365. AND R. E. KAPLAN, "Interventions into Group Process: An Approach to Improving the Effectiveness of Groups," Decision Sci., 5 (1974), 459-480. AND C. G. MORRIS, "Group Tasks, Group Interaction Process, and Group Performance Effectiveness: A Review and Proposed Integration," In L. Berkowitz (Ed.), Advances in Experimental Social Psychol-

ogy, 8, AcademicPress,New York, 1975.

608

GERARDINE DESANCTIS AND R. BRENT GALLUPE

HARE,A. P., Handbook of Small Group Research, Free Press, New York, 1962 (1st ed.), 1976 (2nd ed.). , E. F. BORGATTA AND R. F. BALES(EDS.), Small Groups: Studies in Social Interaction, Alfred A. Knopf, New York, 1967. HILTZ,S. R. AND M. TUROFF, "Structuring Computer-Mediated Communication Systems to Avoid Information Overload," Comm. ACM, 28, 7 (1985), 680-689. AND , The Network Nation: Human Communication via Computer, Addison-Wesley, Reading, MA, 1978. , "The Evolution of User Behavior in a Computerized Conferencing System," Comm. AND ACM, 24, 11 (1981), 739-751. AND K. JOHNSON, "Mode of Communication and the "Risky Shift": A Controlled Experiment with Computerized Conferencing and Anonymity in a Large Corporation," Research report #21, January 1985. HIROKAWA, Y. AND R. A. PACE,"A Descriptive Investigation of the Possible Communication-based R. Reasons for Effective and Ineffective Group Decision Making," Communication Monographs, 50 (1983), 363-379. HOFFMAN, R., "Applying Experimental Research on Group Problem Solving to Organizations," J. Appl. L. Behavioral Sci., 15 (1979), 375-391. The Group Problem-Solving Process, Praeger Publishers, New York, 1979. AND N. R. F. MAIER,"Quality and Acceptance of Problem Solutions by Members of Homogeneous and Heterogeneous Groups," J. Abnormal and Social Psychology, 62 (1961), 401-407. HOLLANDER, E. P., Leaders, Groups, and Influence, Oxford University Press, New York, 1964. HOLLOMAN, C. R. AND H. W. HENDRICK, "Adequacy of Group Decisions as a Function of the DecisionMaking Process," Acad. Management J., 15 (1972), 175-184. HUBER,G. P., "Group Decision Support Systems as Aids in the Use of Structured Group Management Techniques," DDS-82 Conf: Proc., 1982, 96-108. ,"Issues in the Design of Group Decision Support Systems," MIS Quart., 8, 3 (1984a), 195-204. , "The Nature and Design of Post-Industrial Organizations," Management Sci., 30, 8 (August 1984b), 928-951. AND R. R. MCDANIEL, "The Decision-Making Paradigm of Organizational Design," Management Sci., 32, 5 (1986), 572-589. IRVING, R. H., "Computer-Assisted Communication in a Directorate of the Canadian Federal Government: A Pilot Study," Paper prepared for the Non-medical Use of Drugs Directorate, Ottawa, Canada, 1976. JANIS, I. L., Victims of Groupthink: A Psychological Study of Foreign Policy Decisions and Fiascoes, Houghton-Mifflin, Boston, 1972. KERR, N. L., "Social Transition Schemes: Model, Method, and Applications," In H. Brandstatter, J. H. Davis, and G. Stocker-Kreichgauer, Group Decision Making, Academic Press, New York, 1982, 59-79. KRAEMER, K. L. AND J. L. KING, "Computer-Based Systems for Group Decision Support," Unpublished paper, University of California at Irvine, August 1984. KULL,D., "Group Decisions: Can a Computer Help?" Computer Decisions, 14, 5 (May 1982), 64-70 and 160. N. LEDUC, F., "Communicating Through Computers: Impact on a Small Business Group," Telecommunications Policy, (1979), 235-244. L. LEWIS, FLOYD, "Facilitator: A Microcomputer Decision Support System for Small Groups," Ph.D. dissertation, University of Louisville, 1982. LIPPET ET AL., in D. Cartwrightand A. Zander (Eds.), Group Dynamics, Row, Peterson, Evanston, IL, 1968. MABRY, E. A., "Exploratory Analysis of a Developmental Model for Task-Oriented Small Groups," Human Communication Res., 2 (1975), 66. MCGRATH, E., Groups:Interaction and Performance, Prentice Hall, Englewood Cliffs, NJ, 1984. J. AND I. ALTMAN, Small Group Research: A Synthesis and Critique of the Field, Holt, Rinehart and Winston, New York, 1966. MINER, F. C., "A Comparative Analysis of Three Diverse Decision Making Approaches," Acad. Management J., 22 (1979), 81-93. MINTZBERG, H., The Nature of Managerial Work, Harper and Row, New York, 1973. MONGE, P. R., L. W. ROTHMAN, E. M. EISENBERG, K. I. MILLER AND K. K. KIRSTE, "The Dynamicsof Organizational Proximity," Management Sci., 31, 9 (1985), 1129-1141. MUNGY, G. AND W. DOISE, "Social Interactions in Cognitive Development," In H. Brandstatter,J. H. Davis, and G. Stocker-Kreichgauer (Eds.), Group Decision Making, Academic Press, New York, 1982, 429-451. NEMETH, C., "Stability of Faction Position and Influence," In H. Brandstatter, J. H. Davis, and G. StockerKreichgauer (Eds.), Group Decision Making, Academic Press, New York, 1982, 59-94.
9

FOUNDATION FOR STUDY OF GROUP DSS

609

NEMIROFF, P. M. AND D. C. KING, "Group Decision-Making Performance as Influenced by Consensus and

Self-Orientation," HumanRelations,28 (1975), 1-21. NEWCOMB, M., TheAcquaintance T. Process,Holt, Rinehart, and Winston, New York, 1961. G. O'CONNOR, G., "Small Groups: A General System Model," Small Group Behavior,11 (1980), 145-174. POOLE, M. S., "An Information-Task Approach to Organizational Communication," Acad. Management Rev., 30 (1978), 493-504. , "A Test of Four Models of Decision Development in Small Groups," Ph.D. dissertation, University of Wisconsin-Madison, 1980. , "Decision Development in Small Groups: 1. A Comparison of Two Models, Communication Monographs, 48 (1981), 1-24. , "Decision Development in Small Groups: II. A Study of Multiple Sequences in Decision Making," Communication Monographs, 50, 3 (1983a), 206-232. "Decision Development in Small Groups: III. A Multiple Sequence Model of Group Decision Development," Communication Monographs, 50, 4 (1983b), 321-341. J. , D. R. SEIBOLDAND R. D. MCPHEE, "Group Decision-Making as a Structurational Process," Quart. Speech, 71 (1985), 74-102. QUINN, R. E., J. ROHRBAUGH AND M. R. McGRATH,"Automated Decision Conferencing: How it Works," Personnel, 11 (1985), 49-55. 62, D. "An Experimental Analysis of Teachingby Telephone: Theoretical and RUTTER, R. AND B. ROBINSON, Practical Implications for Social Psychology," In G. M. Stephenson and J. H. Davis (Eds.), Progress in Applied Social Psychology, Wiley, New York, 1981. SCHEIDEL, T. M. AND L. CROWELL, "Feedback in Small Group Communication," Quart.J. Speech,50 (1964), 140-145. SHAW, M. E., "Scaling Group Tasks: A Method for Dimensional Analysis," JSAS Catalog of Selected

in Documents Psychology, 8 (1973). 3, 2nd The Dynamics: Psychology SmallGroup of Behavior, ed., McGraw-Hill, New York, 1976. , Group
AND T. W. McGuIRE, "Group Processes in Computer-Mediated SIEGEL, V. DUBROVSKY, KIESLER J. S.

Behavior HumanDecisionProcesses,37 (1986), 157-187. and Communication," Organizational


SMITH, P., "Social Facilitation", Chapter 5 of Groups Within Organizations, Smith (Ed.), Harper and Row, P.

New York, 1973. R. SPRAGUE, H., "A Framework for the Development of Decision Support Systems," MIS Quart., 4,4 (1980), 1-26. "A R. STEEB, ANDS. C. JOHNSTON, Computer-Based Interactive System for Group Decision Making," IEEE Trans. Systems, Man, Cybernet., 11, 8 (August 1981), 544-552. STODOLSKY, "Automatic Mediation in Group Problem Solving," Behavior Research Methods and InD., strumentation, 2 (1981), 235-242. 13, E. THOMAS, J. ANDC. F. FINK,"Effect of Group Size," Psychological Bulletin, 60, 4 (1963), 371-384. B. TUCKMAN, W., "Developmental Sequence in Small Groups," Psychological Bulletin, 63 (1965), 384-399. M. TUROFF, AND S. R. HILTZ,"Computer Support for Group versus Individual Decisions," IEEE Trans. Communications, 1 (January 1982), 82-90. 30, VALLEE, R. JOHANSEN, H. RANDOLPH J., R. AND A. C. HASTINGS, "Group Communication through Computers, Volume 2: A Study of Social Effects," Report R-33, Institute for the Future, November 1974. VAN DE VEN, A. H. AND A. L. DELBECQ, "Nominal versus Interacting Group Processes for Committee Decision Making," Acad.Management 14 (1971), 203-213. J., , "The Effectiveness of Nominal, Delphi, and Interacting Group Decision Making ProAND cesses," Acad.Management 17 (1974), 605-621. J., VAN GUNDY,A. B., Techniques Structured of Problem-Solving, Nostrand Reinhold, New York, 1981. Van S. ANDJ. R. LANG,"The Effects of Group Decision Making Process and Problem WHITE, E., J. E. DITTRICH Situation Complexity on Implementation Attempts," Admin. Sci. Quart., 25, 3 (1980), 428-440.

Das könnte Ihnen auch gefallen