Sei sulla pagina 1di 23

A Theory of Task/Technology Fit and Group Support Systems Effectiveness Author(s): Ilze Zigurs and Bonnie K.

Buckland Reviewed work(s): Source: MIS Quarterly, Vol. 22, No. 3 (Sep., 1998), pp. 313-334 Published by: Management Information Systems Research Center, University of Minnesota Stable URL: http://www.jstor.org/stable/249668 . Accessed: 07/11/2011 17:00
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.

Management Information Systems Research Center, University of Minnesota is collaborating with JSTOR to digitize, preserve and extend access to MIS Quarterly.

http://www.jstor.org

GSS Task/Technology Fit

of Theory Fit Task/Technology and Group Support Systems

of task complexityand theirrelationship relto evant dimensions of GSS technology. Propositions to guide further research are developed fromthe theory. Keywords: Groupsupportsystems, electronic meeting systems, group tasks, task complexity ISRL Categories: AA09, AC0402, DB05, DC02, HA0301,HA1101,IB03

Effectiveness1

By: lize Zigurs University of Colorado, Boulder College of Business and Administration Campus Box 419 Boulder, CO 80309-0419 U.S.A. zigurs@colorado.edu

Introduction
The study of task has a long history, both in the organizationalliterature(Drazin and Van de Ven 1985; Perrow 1967; Thompson 1967) and from a group process perspective (Hackman1968; McGrath1984). Manyscholars of group behavior have argued that the natureof the task plays an important role in a group's interactionprocess and performance (Poole et al. 1985; Shaw 1981). The advent of computer-basedgroupsupportsystems (GSS) provides a new environmentin which to study the role and nature of task. Specifically, the interaction grouptasks withGSS technology of is a potent area of study that has possibilities for enhancinggroupperformance. In the GSS literature, early conceptual papers emphasized the importance of tasks (DeSanctis and Gallupe 1987; Huber 1984) and broughtinto the GSS mainstreama task classification scheme that has been widely used (Hollingshead and McGrath 1995; McGrath 1984). McGrath'stask circumplex has had a significantimpacton the GSS field, and subsequent frameworks GSS research for have included task as an importantvariable (e.g., Benbasat and Lim 1993; Nunamakeret al. 1991). Recent theoretical developments have gone beyond the frameworks and triedto account for complexitiessuch as change and adaptation over time (e.g., DeSanctis and Poole 1994; McGrath and Hollingshead1994) and task plays an important in these theopart ries as well. MIS 1998 Quarterly/September 313

Bonnie K. Buckland Great Plains Regional Medical Center P.O. Box 1167 North Platte, NE 69103-1167 U.S.A. buckland@ns.nque.com

Abstract
The characteristics of a group's task have been shown to account for more than half the variation group interaction.In the context of in group supportsystems (GSS), the importance of task has been underscored by the recommendation that achieving a fit between task and technology should be a principlefor effective GSS use. Althoughthe body of groupsupport systems research has grown in recent years, and experience withdifferenttasks and technologies now exists, no generally accepted theory of task/technologyfit has emerged. Thispaper develops a theoryof task/technology fitin GSS environmentsbased on attributes
1Robert Zmudwas the acceptingsenioreditorforthis paper.

GSS Task/Technology Fit

But what do we reallyknow about the relationship between group tasks and GSS technolocombinationsof gy? Can we specify particular task and GSS technology that will enhance group performance? Recent reviews and of GSS research have meta-analyses addressed these issues to some extent. In some cases, GSS was found to be more appropriate for complex rather than simple tasks (Dennis and Gallupe 1993; Dennis et al.1990b), while in others, GSS was more for appropriate less complex tasks and singlesolution tasks (Benbasat and Lim 1993). For idea-generationtasks, GSS-supported groups performed better or no worse than non-supported groups (Dennis and Gallupe 1993; Dennis et al. 1990b; Hollingshead and McGrath1995). For negotiation and correctanswer tasks, however, GSS groups performed worse (Hollingshead and McGrath 1995). These findingsare equivocaland reveal that a considerable opportunity exists for enhancing our knowledge of the task-related circumstances under which GSS can be used most effectively. The opportunity lies in several areas. First, even though McGrath's circumplex is both useful and widely-used, it is only one of severdefinal different ways to define task. Alternate itions or categorizations need to be examined to determine whether they provide greater leverage for matching task to technology. Second, of the many characteristicsby which task has been described, complexityappears often in GSS research yet has rarely been defined much less consistently used. A more in-depth examination of complexity's role in task and GSS technology issues may provide new insights. Third,the concept of fit itself is not well developed in GSS research and more often impliedthan defined. Since the conceptualizationof fit is related to subsequent data analyses (Venkatraman1989), it is important to define fit explicitlyin GSS contexts. Finally, although general frameworksfor the study of groups and GSS have made a useful contribution by cataloguing characteristics of both tasks and technology, a detailed scheme for measuring and applying those characteristics is still missing. Such a scheme could providea consistent structure within which to conduct
314 MIS Quarterly/September 1998

and compare studies, as well as a foundation for future evolution of technology design and
use.

Given these opportunities, purpose of this the paper is to develop a theoryof task/technology fit in GSS environments that serves to enhance and supplement current knowledge. The theoryfillsa currentgap by firstseparately defining task, GSS technology, and fit. Task/ technology fit is then defined as ideal profiles of task/technologyalignment,and propositions are presented for predicting GSS effectiveness and thereforeenhanced group performancein a GSS environment. The theory is deterministic to the extent that it prescribes expectations about performance, based on characteristics of task and technology. However,the theory is seen as being embedded in the largercontext of human actors, technology, and institutional properties, where technology is only one of many elements of social context that influence et patternsof action (Barley1986; McGrath al. 1993; Orlikowski1992). Thus the theory coexists usefully with the "soft determinism"of social construction perspectives that view technology as both an objective and emergent phenomenon (Barley 1986). Ideal profiles of task/technologyfitcan be viewed as consistent of withthe perspective of a "taxonomy scripts" behavior that describe episodes of structuring to differences in technologicaleffects. leading Those episodes are larger in scope than the profiles in this paper since the episodes account for the strategic role of human actors, but the theory presented here is a useful startlayer" ing pointthat can be viewed as an "inner of the complex totalityof human and technoduringgroupwork. logicalinteraction

Task
Definingtask
The theory is not intended to encompass all tasks, but to focus on the kinds of tasks that are typically encountered in organizational decision makinggroups. Table 1 summarizes

GSS Task/Technology Fit

a select group of relevant task classifications that have been put forward since 1950. These classification schemes distinguish among tasks in various ways. These perspectives are summarized as belonging to one of

four conceptualizations of task: (1) task as behaviordescription,(2) task as abilityrequirements, (3) task qua task, and (4) task as behavior requirements(Hackman 1969). The argument,accepted here, is that the first and second perspectives are not helpful for

Table 1. Examples of Task Classifications Author(s) and Carter,Haythorn, Howell(1950) Shaw (1954) Bass, Pryer,Gaier,and Flint(1958) Hackman(1968) O'Neilland Alexander (1971) Steiner (1972) Classification Basis Groupactivities Task Categories Clerical,discussion, intellectual mechanicalassembly, construction, motorcoordination, reasoning Simplevs. complex Easy vs. difficult Production, discussion, problem solving Discussion,decision, performance vs. vs. Unitary divisible,maximizing prescribedprocess vs. optimizing, permitted process (disjunctive, additive,discretionary) conjunctive, solutionmultiplicity, intrinsic Difficulty, interest,cooperationrequirements, intellectualpopulation familiarity, manipulative requirements Amongcooperativegroups: intellective decision;among vs. competitiveor mixed-motive groups: two-person,two-choicetasks vs. and vs. bargaining negotiation coalition formation Difficulty, variability, interdependence Generate (planning creativity); vs. choose (intellective decision vs. making); negotiate (cognitiveconflict vs. mixedmotive);execute (contests/ battlesvs. performances/ psychomotor) Simple,decision,judgment,problem, fuzzy
MISQuarterly/September 1998 315

Complexity Difficulty Behaviorrequirements of intellective tasks thatyield writtenproducts Behaviorrequirements of intellective tasks tasks ultimately classified Unitary intoone of fourtypes on basis of how membercontributions are permittedtobe combined to producethe single final product Six non-exclusivecontinua

Shaw (1973)

Davis, Laughlin, and Komorita (1976) Laughlin (1980) Poole (1978) McGrath (1984)

and Groupmemberrelationships task performance processes

Dimensionsof tasks at the workunitlevel Whatthe groupor individual is to do; presented as a a "circumplex,"circledivided intofourquadrantswithtwo categories in each quadrant Variouscombinationsof four basic complexityattributes

Campbell(1988)

GSS Task/Technology Fit

advancing research about group tasks. Task as behavior description is not an instructive approachbecause task is defined by whatever the group members actually do. In other words, this approach defines the independent variable (task) in terms of the dependent variable (group performance)ratherthan in terms of propertiesof the independentvariableitself. Task as ability requirementssimilarlyfails to define a task in terms of its own properties. Instead, this perspective uses "relatively enduringaspects of the performer" (Hackman, 1969, p. 111) to describe tasks. The two remainingapproaches to definingtask, howevattention. er, deserve further The task qua task approach focuses on aspects of the actual task materials that are presented to the group (Hackman1969). Inthe literature task qua task, many studies have on focused on task complexityas the characteristic of greatest interest,for instance, differentiation between analyzable and unanalyzable problems based on their complexity (Perrow 1967). Task complexity has been defined as consisting of three components: (1) coordinative complexity (the number of non-linear sequences between components and task products),(2) componentcomplexity(the number of distinctacts and the numberof distinct information cues involved in the task), and (3) dynamic complexity (the stability of the relationships between inputs and the product) (Wood 1986). Other research focuses on task complexityand buildson these earlierideas by defining complexity in terms of solution paths and their relationshipto outcomes (Campbell 1988). That research relates task complexity "directlyto the task attributes that increase information load, diversity,or rate of change" (Campbell1988, p. 43) and furtherstates that this allows complexityto be determinedindethe pendent of the person performing task. The final view-of task as behavior requirements-also provides fruitful groundfor group research. Because required behaviors vary from task to task, it is argued that "behavior requirements can legitimately be viewed as characteristicsof tasks (ratherthan characteristics of the performer)"(Hackman 1969, p. 111). McGrath'stask circumplexis based on 1998 316 MIS Quarterly/September

task as behavior requirements to the extent that each task is categorized by its objective, i.e., what the group members are supposed to do to accomplishthe task, for example, a creativity task requires generation of ideas. However, behavior requirements for a task includenot only whatmust be accomplishedto meet stated goals, but how those goals should be accomplished,i.e., the processes by which the task should be carried out (Hackman 1969). The stimulus set for any given task (whether experimental or not, and whether explicitor not) includes physical materialsand instructions,with the instructionshaving "the functionof elicitingthe mediatingprocesses for problemsolving"(Gagne 1966, p. 135). The final view of task complexity integrates these two conceptualizationsof task: the task qua task approach and the task as behavior requirements approach (Campbell 1988). While it focuses on characteristicsof the task as presented to the group, it simultaneously acknowledgesthat those characteristicsdefine both what is to be accomplished and how it is to be accomplished. Thus, a group task is defined here as the behaviorrequirementsfor accomplishingstated goals, via some process, using given information. of This definition grouptask focuses on stated goals, that is, on the problemas it is presented to the group. If a group member has hidden goals and tries to redefine the task according to some privateagenda, or if a group support system imposes structure on the task to the extent that the task is modifiedto fit the tools or agenda enforced by the GSS, it is possible that the assigned task may not be the one actually performedby the group. Indeed, the changing natureof the task is a key potentially of structurational perspectives (Barley part 1986; DeSanctis and Poole 1994). This focus on the assigned task is an initialstep that permits reliable description of tasks in objective terms that allow comparisonwithinand across to studies. Such a view is not a barrier assessing tasks over time as they are redefined throughgroupinteraction.

GSS Task/Technology Fit

uniquetask environments Defining


The task definitionpresented in this paper is an extension and refinementof earlierdescriptions (Campbell1988; Wood 1986) and focuses on the central importanceof task complexiof ty. Complexityis the one characteristic task that has been studied the most. Many prior efforts at task classification have either focused directlyon complexityor includedit as one of several importanttask characteristics (e.g., Bass et al. 1958; Campbell1988; Herold 1978; Perrow 1967; Poole 1978; Shaw 1954, 1973). Task complexity has also been an importantpart of analyses of the GSS literature (Dennis and Gallupe 1993; Dennis et al. 1990b; Gray et al.1990; Pinsonneault and Kraemer1989). Finally,complexity relates to both process and outcomes of task performance. Thus, complexity plays a key role in distincttask environments. differentiating As noted earlier, task complexity is related "directlyto the task attributes that increase information load, diversity,or rate of change" (Campbell1988, p. 43). In addition,the levels of these three informationprocessing factors indicate the level of the cognitive demands placed on the person performingthe task. A complex task, then, places high cognitive demands on the task performer. The complexity level of a task has been defined via four dimensions (Campbell1988), each of which is importantin defining unique task environments.The first dimension is outcome multiplicity. Outcome multiplicity means that there is more than one desired outcome of a task. This is important because it increases information load and information diversity. Each outcome requiresa separate information processing stream and is essentially a criterion against which a potentialsolutionis evaluated. An example of a task with outcome multiplicity is one where there is more than one stakeholder and each stakeholder has different explicitexpectations about what the objectives of the task are, e.g., selecting a family home froma set of alternativesbased on price, size, specific features (forexample, accommodation of a handicappedfamilymember),proximity to to work,proximity good schools, etc. Note that

it does not matterwho is completingthis task; its outcome multiplicityis unaffected by the task performers. The second dimension can be called solution scheme multiplicity. Solutionscheme multiplicity means that there is more than one possible course of action to attain a goal. This dimension of the task increases informationload. The term"solution path"has been used for this idea (Campbell 1988). The term used in this paper has the same meaning but is clearer in distinguishinga solution path from a process path. If, for example, there are alternative ways to reach a goal, and those alternatives can be specified using a decision tree, this shows the presence of multiple solution schemes in that many configurations a final of solution are possible depending on which branches of the decision tree are chosen. The idea of havingto configurea set of elements in reaching a decision also helps to clarify the solutionschemes. Ifa task meaningof multiple involves the explicit consideration of implementationissues, then it becomes more comvarious elements; plex in terms of configuring such a task has solution scheme multiplicity. Solutionscheme multiplicity increases information load because one must consider multiple elements and their best configuration. conIn trast, simplychoosing one item from among a given set of alternativesis an example of a single solution scheme. A game of chess, a jigsaw puzzle, and employee scheduling are all examples of tasks with solutionscheme multiplicity.Again, it does not matterwho is completing the task, as the existence of multiple solutionschemes is inherentin the task. The thirddimension is conflictinginterdependence. Conflicting interdependencemay exist among solution schemes where adopting one scheme conflictswith adoptinganother possible solutionscheme. Inthis case, the adoption of any one scheme substantially alters the situation such that the decision makers cannot simplychange theirminds, undo that adoption, and returnto essentially the same conditions presented in the originaltask to make a new decision. Conflicting interdependence also exists in cases where outcomes are in conflict with one another, e.g., the classic quantity MIS 1998 Quarterly/September 317

GSS Task/Technology Fit

versus quality problem. Finally, it may exist where information in conflict,for example, in is intelligencereportsfromvarioussources. The fourth and final dimension can be called solution scheme/outcome uncertaintyand is defined as the extent to which there is uncertainty about whether a given solution scheme will lead to a desired outcome. Solution scheme/outcome uncertaintycan range from low to high, where high means that the relationship between a solution scheme and the desired outcome is uncertainor highly probabilistic (Campbell 1988). Many factors can enter into a determination the level of soluof tion scheme/outcome uncertainty.Uncertainty is increased in tasks where outcomes are not explicit, where the scope of the problem is large, where there is little useful historical information about solving a particular problem, or where outcomes are difficultto measure. is Uncertainty decreased, for example, in tasks where the task instructionsdo not requirethe task performerto explicitlyconsider the implications of implementation. It has been noted that other characteristics associated with complexity, such as lack of structure, ambiguity, and difficulty, can be seen as consequences of the four basic complexity attributes outlined above (Campbell 1988). If a task has one or more of the four basic attributes,it is also likelyto possess one or more of the associated characteristics(i.e., it is likely to be ill-structured, ambiguous, and/ordifficult). the other hand, a task may On be experienced as ill-structured, ambiguous,or

difficult"forreasons that are independent of the basic characteristics of the task itself" (Campbell1988, p. 45). Forexample, writinga piece of software to solve a problem may be easy for veteran programmersbut difficultfor novices, even though the basic characteristics of the task are unchanged. Different combinationsof the fourbasic dimensions of complexitywould result in 16 distinct task environments, which have been aggregated into five task categories based on similarities in the presence or absence of the four basic complexity attributes (Campbell 1988). These five task categories are shown in Table 2. Each category is defined in terms of primary attributes that contribute to its complexity. characterizedby a Simple tasks are primarily outcome and solution scheme, problem single tasks by solution scheme multiplicity, decision tasks by outcome multiplicity, judgmenttasks by conflictinginterdependenceor uncertainty, and fuzzy tasks primarily the jointpresence by of outcome multiplicity and solution scheme multiplicity.

GSS Technology
DefiningGSS technology
Groupsupportsystems have been defined as systems that combine communication, computer, and decision technologies to support problem formulation and solution in group

Table 2. Aggregated Task Categories (Adapted From Campbell 1988) Simple Tasks no no no not applicable Problem Tasks no yes yes or no low to high Decision Tasks yes no yes or no low to high Judgment Tasks no no yes or no low to high Fuzzy Tasks yes yes yes or no low to high

Outcome Multiplicity SolutionScheme Multiplicity Conflicting Interdependence SolutionScheme/ I Outcome Uncertainty

1998 318 MIS Quarterly/September

GSS Task/Technology Fit

meetings (DeSanctis and Gallupe 1987); as information-basedenvironments that support group meetings and include but are not limited to distributed facilities,computerhardwareand software, audio and video technology, procedures, methodologies, facilitation,and group data (Dennis et al.1988); and as the collective of computer-assistedtechnologies used to aid and addressing problems, groups in identifying opportunities,and issues (Huberet al. 1993). This is only a sample of definitionsof GSS, but common elements are evident, and the definitions tend to be all-inclusive rather than restrictive. Classificationschemes for group supportsystems are as abundant as definitions of the technology. Table 3 shows examples of GSS technology classifications. Some authors name explicit categories of systems (e.g., McGrath and Hollingshead 1994), but the of majority authors describe the technology in terms of dimensions, each of which typicallyis defined by more detailed attributes (e.g., Fjermestad, forthcoming). As Table 3 shows, the technologycan be characterized from many different perspectives, ranging from relativelybroad functionalcategories, to more detailed tool-based descriptions, to configurations defined by time and space. However, three common themes are evident: support for communication, for process structuring,and for informationprocessing. Regardless of the hardwareor physical configuration of a system, these three themes are the distinguishing characteristics of GSS that are used the most often in the classificationsshown in Table 3 and that have been studied the most in prior research. They are also consistent with the major properties of advanced technologies identifiedin priortheory: communication, decisional, and informational properties (Huber 1990). Therefore, group support systems technology is defined as a set of communication, structuring,and information processing tools that are designed to work together to support the accomplishment of group tasks. Communicationsupport relates to communicationneeds of the group (e.g., input,feedback, and display capabilities, as well as time and space configuration);

structuring support relates to setting up a process by which groups interact(e.g., providing or enforcing agendas); and information and strucprocessing relates to manipulating turinginformation (e.g., aggregating, evaluating, or structuring information). As with task, the potential gap between designer intentionsfor GSS and how it is actually used is recognized.The dualityof technology means that it is enacted by human actors as well as institutionalized in structure (Orlikowski 1992). Research has shown how GSS technology can be used in ways that differ from the original spirit of the designers (DeSanctis and Poole 1994). Again, however, it is only the intended system features and tools that can be reliablydescribed objectively, and hence the focus is on the intended system. However, as the technology is adapted through group use, its nature can be reassessed via the characterization scheme.

Defining uniqueGSS environments technology


Each of the three dimensions from the definition of GSS can be defined further. Communication support can be defined as any aspect of the technology that supports, enhances, or defines the capability of group members to communicate with each other. It includes such elements as simultaneous input, anonymous input, input feedback, and a groupdisplay.Groupand individual messaging can be considered part of input and display. The communication support dimension also includes the physical configuration of communication channels, since that configuration defines how group members can communicate. Process structuring any aspect of the techis nology that supports,enhances, or defines the process by which groups interact, including capabilities for agenda setting, agenda enforcement,facilitation,and creating a complete record of group interaction (via storing the agenda, all the input, the votes, and so on). Information processing is the capabilityto MIS 1998 Quarterly/September 319

GSS Task/Technology Fit

Table 3. Examples of GSS Technology Classifications Author(s) DeSanctis and Gallupe(1987) Pinsonneaultand Kraemer(1989) Dennis, George, Jessup, and Nunamaker, Vogel (1988) Nunamaker, Dennis, Valacich,Vogel, and George (1991) Johansen (1992) DeSanctis and Poole (1994) Classification Basis Groupsize and proximity; patternsof information exchange Type of aid providedby the system Groupprocess and outcomes, methods, and environment Process support,process structure,task support, and task structure Time and space configuration features and Structural spiritof technology GSS Technology Categories Decision room,legislativesession, local decision network, computer-mediated conference;level 1, level 2, level 3 Groupdecision supportsystems, group communication supportsystems Overallcategories are not defined;each element described of classificationbasis is further includesgroupsize, (e.g., environment location,and timingof meeting) participant Overallcategories are not defined;each element described of classificationbasis is further (e.g., process supportincludesparallel communication, groupmemory,and anonymity) Same-time/same-place,same- time/ different-time/ different-place, different-time/same-place different-place, Overallcategories are not defined;each element of classificationbasis is characterized scalable dimensions (e.g., by spiritconsists of decision process, leadership,efficiency,etc.) Overallcategories are not defined;each element of classificationbasis is further includespublic described(e.g., functionality display,anonymousvoting,private messaging, etc.) communication supportsystem, Groupinternal supportsystem, groupinformation support groupexternalcommunication supportsystem system, groupperformance Closed, closed/open, open/closed, open

DeSanctis, Snyder, and Poole (1994) and McGrath Hollingshead (1994) Vickers(1994) Rana, Turoff,and Hiltz(1997) Fjermestad (forthcoming)

Interface,functionality, and holisticattributes

functionof system Primary (communication, or information, task support) Extentto whichend users can controland/ormodify system Componentpartsof system, i.e., presence or absence of generic supportfeatures Communication mode, design, process structures, and task support

Individual support,groupprocess support, meta process support,groupmodel support Overallcategories are not defined;each element of classificationbasis is further mode described(e.g., communication be voice, text, video, etc.) might

320

1998 MISQuarterly/September

GSS Task/Technology Fit

gather, share, aggregate, structure,or evaluate information, including specialized templates such as stakeholderanalysis or multiattributeutilityanalysis. It should be noted that templates that structure the problem may shape solutions as well, since people may view possible solutions in terms of how they structured viewed the problem. or Table 4 shows these three dimensions and examples of elements of each dimension, drawn from what is commonly provided in existing group support systems. An actual GSS implementation typicallycombines several elements and dimensionstogether, although a specific tool withinthat GSS may focus on a single element or dimension. For example, both Topic Commenter and the electronic brainstorming tool in GroupSystems (Nunamakeret al. 1991) would be characterized as communication support tools, while GroupSystems' voting tool would focus on informationprocessing support. The information processing dimension in particularmight be implemented via a variety of tools. For instance, the element of structureinformation could be provided via stakeholder analysis,

multiattribute utilityanalysis, a decision tree, and so on. In addition,where this element is present, it typicallyincludesthe otherelements of gathering,aggregating,and evaluatinginformation.A multiattribute tool, for instance, utility willtypicallyincludethe other elerents as part of the steps forcarrying the analysis. out The second of SAMM generation and (Sambamurthy DeSanctis 1990) is a good example of how a specific GSS combines various tools, each of which support different dimensions. SAMM2 includes modules for allocation, stakeholder analysis, and multiattributeutilityanalysis, as well as the abilityto gather, aggregate and evaluate information, thus ratinghigh on the information processing dimension. SAMM2 also includes simultaneous input, anonymous input, input feedback, and groupdisplay,thus ratinghigh on communicationsupport.However,this particular GSS is very flexibleas to agenda configuration and facilitation options, thus ratingrelativelylow on process structuring.This example shows the importance of looking at a GSS's different tools and how they are used in orderto assess the technology'sdimensions.

Table 4. Examples of Elements for the Dimensions of GSS Technology Dimension Communication Support * * * * * Process Structuring Examples of Elements Simultaneousinput Anonymousinput feedback Input Groupdisplay of channels (e.g., synchronous Physicalconfiguration communication or asynchronous,proximate dispersed) or * Agenda setting * Agenda enforcement e Facilitation * Completerecordof groupinteraction * Gatherinformation * Aggregateinformation * Evaluateinformation * Structure information (e.g., allocation,stakeholderanalysis, multiattributeutility analysis, cross-impactanalysis)
MISQuarterly/September 1998 321

Information Processing

GSS Task/Technology Fit

Fit
Definingfit
Althoughthe term fit is widely used in a variety of models that deal with contingencies among variables, its precise nature and meaning are rarelystated (Joyce et al. 1982). One exception to this lack of clarityis the strategic manwhere fit (typically between agement literature, strategy and structure)has been examined in some detail. Differentdefinitionsof fit in three distinct approaches to structuralcontingency

fit theory have been identified: as congruence, fit as interaction, fit as internal and consistency (Drazinand Van de Ven 1985). These ideas were extended to identifysix unique perspectives on fit in the strategy literature: as modfit eration, as mediation, as matching, as gestalts, as profiledeviation, and as covariation (Venkatraman 1989). These perspectives vary in theirdegree of specificityof the theoretical relationship between variables,in the number of variables in the fit relationships,and in whetherthe concept of fit is anchoredto a particularcriterionvariable. Table 5 summarizes the six perspectives on fit.

Table 5. Perspectives on Fit (From Venkatraman1989) Underlying Conceptualization Description A matchbetween two Matching relatedvariablesis theoretically defined,withoutreferenceto a criterion variable. A patternof covariation 2. Fitas or Internal internal covariation consistency consistency among a set of underlying related theoretically variablesis defined,without referenceto a criterion variable. 3. Fitas Internal Gestalts are definedin terms of the degree of internal cohergestalts congruence ence among a set of theoretical attributes, involving many variables,but not specified with. variable referenceto a criterion The impactthat a predictor 4. Fitas Interaction variablehas on a criterion varimoderation able is dependenton the level of a thirdvariable,whichis the moderator. A significantintervening Intervention 5. Fitas mechanism(i.e., an indirect mediation effect) exists between an antecedentvariable and the consequent variable. related A profileof theoretically Adherence 6. Fitas profile variablesis specified and to a deviation relatedto a criterion variable. specified profile Perspective 1. Fitas matching
322 MISQuarterly/September 1998

Example Proposition (From Venkatraman) The matchbetween strategyand structure enhances administrative efficiency. The degree of internal consistency in resource allocationshas a significanteffect on performance. The natureof internal congruence among a set of strategicvariables differsacross high and low performing firms. effects of The interactive strategyand managerial characteristics have for implications performance. share is a key Market variable intervening between strategyand performance. The degree of adherence to a specified profilehas a significant effect on performance.

GSS Task/Technology Fit

The firstthree conceptualizationsof fit are criterion-free(Venkatraman 1989), i.e., they have universalapplicability are not anchored to and any particular dependent variable, such as effectiveness. Since the desire in this paper is to tie task technology fit to effective performance of groups, these three conceptualizations are not suitable (Nidumolu 1996). The fourthand fifth conceptualizations are limited in the numberof variables considered, that is, they are typically used in an association between a single predictorvariable, a single moderatingor interveningvariable, and a single dependent variable. The most promising perspective for task/technology fit in a GSS context is the idea of fit as an ideal profile.This view is consistent with approach to contingency theory presented earlier (Drazin and Van de Ven 1985), allowing for a holistic approachto examiningthe complexityinherent in organizations. From this perspective, fit is viewed as feasible sets of equally effective alternative designs. Each design should be consistent and matchedto a configuinternally rationof contingencies that face the organization. This is the idea of fit as an adherence to an externallyspecified profile(Venkatraman 1989). Testing this conceptualization in an organizational strategy context would involveidentifying distinctenvironments, specifying ideal resource deployments for each environment, and testing the performance effects of environment/strategycoalignments and (Venkatraman Prescott 1990). This conceptualization fit translateswell into of a GSS environment. Task/technology fit can be defined as ideal profiles composed of an internallyconsistent set of task contingencies and GSS elements that affect group performance. The greater the degree of adherence to an ideal profile,the better the performance of the group. Ideal profilescan be operationalized as viable alignmentsof task and technoloand Prescott gy. As suggested (Venkatraman 1990), a test of task technology fit would distincttask requirethree steps: (1) identifying environments, (2) specifying ideal technological supportfor each task environment,and (3) testing the performanceeffects of task/technology alignments.

Fitin existingGSS research


Achieving a fit between a group's task and GSS technology was suggested as a principle for effective use of groupsupportsystems in a seminal paper (DeSanctis and Gallupe 1987). McGrath'scircumplexof task types was recommended as a classification scheme, and prescriptionsfor appropriateGSS technology were providedfor three main task types, e.g., choose type tasks should be used with a GSS that aids in selection of a solution.The majority of GSS researchers have used this circumplex to classify their tasks, but DeSanctis and Gallupe's fit prescriptionshave not been systematicallytested. There is one problemwith how researchers have applied the circumplex, based on the fact that tasks typicallyhave elements of both creativity (generate type) and problem-solving (choose type). In empirical work,even if authorsclassified a task in multiple categories of McGrath's circumplex according to the task's different phases, the subsequent analysis and discussion of results were typically combined across the entire group session. This practice makes it difficult to unraveltask/technology issues. fit A theory of task/technology interaction(TTI) that defines task by the dimensions of complexity,validation,and coordinationand technology by the dimensionsof individual support, process support, and meta-process support has been proposed (Rana et al. 1997). Each dimension is furtherdefined by more detailed for attributes; instance, the complexitydimension is defined by the attributesof structure, specificityof relationships,and task uncertainty. Each task dimension has a prescribed, best-fit technology dimension associated with it. However,was noted in the study, the propositions are quite broad and a characterization of expected differences in group interaction based upontheirtask distinctions yet to be has developed. Two other recent theoretical developments are notable with respect to their attempt to account for the complexityand dynamicnature of groupinteraction. theAdaptivestructuration ory includes task as a key source of structure that combines with other sources, such as MIS 1998 Quarterly/September 323

GSS Task/Technology Fit

technology and the group's internal system, to affect social interaction (DeSanctis and Poole 1994). Inthe model of the flow of effects for computer-aidedwork groups (Hollingshead and McGrath1995), task is a key input variable that combines with other inputvariables, such as technology and group member attributes, to set up conditionsthat resultin differing These two modpatterns of group interaction. els are very similarin the role that task plays, namely as a key source of variation in the interactionprocess. Detailed characterizations of task and technology remain unspecified, however, as does the combination of characteristics that should lead to optimal performance. Only a few studies have explicitly compared different tasks in GSS environments in an attempt to test fit empirically.One compared an intellective with a decision-makingtask in manual versus face-to-face GSS versus dispersed GSS groups (Clapper et al. (1991). That study argued that task type would determine the primary kind of influence that occurred duringinteraction,and differentGSS environmentswould enhance different types of influence. Results showed no main effect for task type, althoughmanual groups workingon the judgment task reached consensus significantly faster than dispersed groups on the intellectivetask. A task/technology interaction on influence was suggested, but no clear conclusions about task/technologyfit could be drawn.A second study also examined two different tasks (generate and choose) with two differenttechnologies (Gopal et al. 1992-93). Task did not account for any differencesin the dependent variables, but the study speculated this was because a consistent procedure across experimental conditions might have reducedtheirdissimilarity. A different approach was taken in a third study, where two differentGSS technologies were used for the same task (Easton et al. 1990). The task was classified as a combination creativityand intellectivetask and predicted differenttypes of outcomes based on the type of technology. Although not all expectations were borne out, creativitywas enhanced with one type of technology, while decision
324 MISQuarterly/September 1998

quality was enhanced with another. These results were consistent with the prescriptions in the study for how differentcharacteristicsof the technologies should match to differentelements of the task. Electronic conversationprovided the focus needed in an intellectivetask, whilethe divergenceof ideas providedby electronic brainstormingsupported the creative componentof the task. Perhaps the most extensive comparison of tasks with different GSS environments has been the work of colleagues (e.g., Sia et al. 1996; Tan et al. 1994), who have compared a witha decision-making task specific intellective under a varietyof conditions.Althoughresults tend to be mixed, there is some support for task-related differences under varying conditions of the technological environment. For instance, intellectivetasks are best supported by decision aids that enhance reasoning and promote informationalinfluence, while decision-makingtasks requiresupportfor consensus and the support of normative influence (Tanet al. 1994). Several conclusions can be drawn from the existing conceptual and empiricalwork related to tasks and GSS technology. First,McGrath's circumplexis the dominanttask classification based on it have scheme, but the prescriptions not been tested in a systematic way across differenttypes of GSS technologies. It is not known whether this approach provides the best leverage in a GSS context. Second, the few theories of task/technology fit that have been presented address relatively broad to concepts that make it difficult formulatespecific fit prescriptions. For instance, the study by Rana et al. notes that levels of sophistication that define the three dimensions of technology support still need to be defined and the functions for each dimension need to be characterized. Third, the few studies that explicitlycompare task and technology reveal that there are differences, but to best understand their effects, a consistent theoretical model of fit is needed. Moreworkneeds to be done on how best to confirmexisting findings and extend them to a broaderset of tasks and technologies.

GSS Task/Technology Fit

The theory of task/technology fit presented here builds on and extends this prior work. Existing fit prescriptions and empirical research provideda startingpoint for assessing key constructs and their potentialrelationships. The detailed, operational definitionsof task, technology, and fit in this paper providea level of detail that goes beyond the relatively broad nature of priorwork. The next section presents the specifics of the theory and a set of propositionsderivedfromit.

The Theory of Task/ Technology Fit


An appropriate task/technologyfit should result in higher performinggroups; in other words, the criterion variable is group performance. construct Groupperformanceis a multifaceted that has been defined by such concepts as efficiency, outcome quality, process quality, satisfaction, and consensus (Fjermestadand Hiltz 1997; Hollingsheadand McGrath1995). For a given task and group, the relevantperformance measure may vary, whether this be in real-life organizational tasks or in experimentally controlledsituations. Therefore, it is recognized that group performance will be operationalized in different ways, and in the discussion of ideal task/technologyalignments, the term "groupperformance" used generiis cally. Figure 1 shows the general model of task/technologyfit.

Table 6 summarizes the prescriptions of the task/technology fit theory, i.e., the table shows the fit profiles that are expected to enhance group performance.The terms low, medium, and high have been used as an approximation for the level of support prescribed. The discussion that follows reveals that several of the individual elements of Campbell's task categories (Table 2) imply particulartechnology support, i.e., outcome implies a need for processing supmultiplicity port;solution scheme multiplicity implies information processing support; and conflicting interdependence implies communicationsupport. However, as will be seen, where these elements occur in combination, implication the is that the GSS must also provide multiple dimensionsin varyingconfigurations. The remainderof this section provides justification for and propositionsabout task/technology fitfor each of the five task types. Following each proposition,selected examples are presented from existing research to support the The analysis of the research is by proposition. no means comprehensiveand focuses on specific examples that relate directlyto the type of task being discussed. In some cases, there is very little,if any, priorresearch that is relevant to the fitpropositions.

Simple tasks
Simpletasks have a single desired outcome, a single solution scheme, and no conflicting interdependenceor solution scheme/outcome

Task

Fit Profile

Group Performance

GSS Technology Figure 1. General Model of Task/Technology Fit


MISQuarterly/September 1998 325

GSS Task/Technology Fit

Table 6. Fit Profiles of Task Categories and Technology Dimensions Communication Support Dimension High Low Low High High Process Structuring Dimension Low Low High Low Medium Information Processing Dimension Low High High High High

SimpleTasks ProblemTasks Decision Tasks JudgmentTasks FuzzyTasks

uncertainty.Therefore, a GSS should provide communication supportso that group primarily members can communicate their ideas about the solutionto one another.The importanceof communication elements for tasks requiring information exchange and greater examination of alternativesolutions has been argued (Rao and Jarvenpaa (1991). Too much structureor focus on information processing could interfere with simple communicationneeds and detract from group performance and/or satisfaction (Gallupe et al. 1988). Therefore,the focus for simple tasks should be on the communication supportdimension. P1:Simpletasks should result in the best (as group performance definedfor the specific task) when done using a GSSroem that configuration emphasizescommunicationsupport. Good examples of simple tasks in existing GSS research are the so-called brainstorming tasks, where the focus is on the single out-

come of generating the greatest number of ideas. A series of studies that used the elecmodule of GroupSystems tronic brainstorming found that the GSS-supported groups performed better on a variety of measures than non-supported groups (Gallupe et al. 1990, 1991, 1992). These studies support Proposition 1 because the GSS emphasized communicationsupport via providingsimultaneous input,anonymousinput,inputfeedback, and group display of ideas. Figure 2 shows what the general model might look like for a simple task; a similarmodel could be drawnfor the othertask types.

Prblm

task

Forproblemtasks, the mainfocus is on finding the best solutionscheme from among multiple possible schemes, which satisfies a single, well-defined desired outcome. The presence of multiple solution schemes increases the

Simple Task

High Communication Low Process Structure Low Info. Processing

Group Performance

High Communication GSS Tool Figure 2. Model of Task/Technology Fit for Simple Task
326 1998 MISQuarterly/September

GSS Task/Technology Fit

informationprocessing requirements for the task (Campbell 1988). Group members have to be able to configurethe problemin various ways in order to achieve the best outcome. Increased information load can be handled by a GSS with elements of information processing. For example, human information processing theory was used to argue that a GSS with computational supportwould be more effective for tasks with a high proportionof externalizable information processing activities(Rao and Jarvenpaa 1991), which would be expected in problemtasks. It has been shown that a task that required consolidating different solution proposals and deciding on the best one had the highest performance witha GSS configuration that emphasized informationprocessing within a single tool (Easton et al. 1990). Information processing is the key support needed for this type of task, to help group members deal with the increased information load. Communication supportshould be lowsufficient to provide minimal capabilities for communicating information essential to the problem while avoiding communicationoverload. Process structuringshould also be low, since problem tasks do not imply extensive steps that might need agenda support or enforcementand too much structurecould distractfromanalyzinginformation. Therefore,the focus for problemtasks should be on the informationprocessing dimension. P2:Problem tasks should result in the best group performance(as defined for the specific task) when done using a GSS configurationthat emphasizes information processing. A research and development projectplanning task was used with the Capture Lab GSS in two studies that compared GSS-supported groups with non-supportedgroups (Losada et al. 1990; McLeod and Liker 1992). The task requiredgroup members to reach consensus on the appropriatesequence of management activities for the project, thus the task meets this study's definition of a problem task. No significant differences were found between groups in the first study, a finding consistent with the prediction, since the GSS was very low on the information processing dimension.

For the second study, an information structuring tool was added (an outlining tool) that was information specificallydesigned for structuring in that task, and GSS groups performedbetter than non-supportedgroups, again consistent withthe prediction.

Decisiontasks
The focus for decision tasks is on crafting a solutionthat best satisfies multiple(and potentiallyconflicting)outcomes. Each desired outcome involvesa separate information processing stream (Campbell 1988), implying both high information load and high information diversity.Each desired outcome is a criterion against whichthe proposedsolutionis evaluated. This type of task requires heavy information processing supportand, in particular, supProcess strucport for evaluatinginformation. turing is also important to ensure that the groupcarriesout all the steps of criteriaidentification and evaluation against alternatives. Multicriteria utility analysis is a classic template for structuring this type of problem. Communication supportshould be low to avoid communicationoverload for this type of task. The focus for decision tasks should be on the two dimensions of information processing and process structuring. P3:Decision tasks should result in the best group performance(as defined for the specific task)when done using a GSS configuration that emphasizes informationprocessing and process structuring. A decision task was used witha GSS that was relatively high on informationprocessing but quite low on process structuring (Zigurset al. 1988). No positive effect on outcome quality was found for the GSS, as would be expected since there was insufficientprocess structure to help devise the best solution from among outcomes. multiple
MISQuarterly/September 1998 327

Fit GSS Task/Technology

tasks Judgment
Forjudgmenttasks, the emphasis is on resolving the conflict and uncertaintyin information associated withthe task. Thus, again, information processing is important,but communication support is importantas well. Where conflict and uncertaintyare paramount,rich communicationis needed (Daft and Lengel 1986). Studies of preference-type tasks, which involve conflicts of values, have shown that elements of the communication supportdimension help to promote consensus (Sia et al. 1996; Tan et al. 1994). But communication support alone is not enough, as shown by studies of conferencing systems that provide only communicationsupport and no information processing. For instance, two different tasks withgroups using a conferencingsystem were compared, and greater dissatisfaction and inefficiency were found where the task dealt with attitudes and values (Straus and McGrath 1994). Althoughthis comparisonwas kinds withface-to-face groups and not different of technologies, the study nevertheless suggests that communication elements alone were insufficient for dealing with information diversity. The process structuringdimension for this type of task should be low, to avoid an overemphasis on agenda setting and enforcement, which mighttend to limitthe kindof free exchange of information that is important. Therefore,the focus for judgmenttasks should be on the communication supportand information processing dimensions. P4:Judgment tasks should result in the best groupperformance definedfor (as the specific task) when done using a GSS configuration that emphasizes communicationsupport and information processing. An earlier study used a task that required group members to generate potentialsolutions for a sales territory problem, a task that associated exhibitsthe conflictand uncertainty with a judgment task (Dennis et al. 1990a). The GSS tools emphasized communication processsupport,but were low on information ing. The comparison was of ad hoc versus established groups and the expected differ328 1998 MISQuarterly/September

ences in quality of outcome did not occur. It may be that the lack of fit in task and technology masked the expected differences, although this is speculative since the comparison was not directlyof GSS withnon-GSS groups.

Fuzzytasks
Fuzzy tasks have very littlefocus, and group membersexpend most of theirefforton understanding and structuring the problem. load, information Information diversity,conflict, and uncertainty are all part of fuzzy tasks (Campbell 1988). Where complexity is high, enhanced information processing is important (Rana et al. 1997). The greatest emphasis, among therefore,should be on communication relagroup members, gatheringof information tive to the problem,and using problemstructuring templates to understand the problem. Process structuring is also importantto the the extentthat it helps groupsstructure process whichthey accomplishtheirwork.The strucby ture, however,should be flexibleand adaptable to group needs; a strictlyenforced agenda, for instance,could lead to biasingthe problemdefinitionin the directionprovidedby the structure rather than permitting the discovery of the free of imposed struc"true" problemdefinition ture (Gopal et al. 1992-93). Therefore, the focus for fuzzy tasks should be on communication support,information processing, and on a mediumlevel of process structuring. P5:Fuzzytasks should result in the best (as groupperformance definedfor the specific task) when done using a GSS that configuration emphasizescommunicationsupportand information processing, and includes some process structuring. A widely used fuzzy task in the experimental is task, in which GSS literature the Foundation members allocate funds for a philangroup thropicfoundationto competing organizations that reflect conflicting values and goals (Watson et al. 1988). The Foundation task of load, exhibitsthe characteristics information information diversity,conflict, and uncertainty. task with Several studies pairedthe Foundation

Fit GSS Task/Technology

a GSS that had good communicationsupport but low information processing, and these studies found no positive effect for GSS (Ho and Raman 1991; Poole et al. 1991; Watson et al. 1988). A laterversion of the GSS implemented more information processing and foundpositive and results on outcome quality(Sambamurthy DeSanctis 1990; Sambamurthy and Poole 1992). These examples lend support to the on proposition fuzzy tasks.

Conclusions and Directions for Future Research


The task/technology fit theory defined in this paper has several advantages. First,the theoof ry is based on an explicitconceptualization task complexity-a fundamentally important relevantin a aspect of task that is particularly GSS environmentand that takes into account an important of task attributes.Second, the set characterization technology is based on key of dimensions from priorwork and is specifically linkedto task demands. Third,the concept of fit has been explicitlydefined and linkedto the criterion variable of group performance. Fourth,the definitionsof task types and technology dimensions lend themselves to being operationalized via coding instruments that can be used to compare and accumulatefindings. Finally,the theory can be embedded as one partof a largercomplex of contextualvariables, such as social and culturalissues, as well as group development and system use and adaptationover time. The theory provides ideal profiles of group tasks and GSS technology. The next step for testing the theory is to operationalizeand test the profiles. This requires that measures be developed for specific task types and technology dimensions. A coding scheme for task types would include an operationalizationof each of the dimensions in Table 2 (outcome and solution scheme multiplicity, multiplicity, on so on). A task could be rated "yes"or "no" each dimension and thus assigned to the appropriatetask type. Similarly,each dimension of GSS technology would be operational-

ized and rated; a suggestion for rating the has or dimensions as "low," "medium," "high" been presented (Table 6). Existingtasks and GSS technologies can be coded on these dimensions and tested for the extent to which prescribed patterns correlate with positive group performance.Such an analysis is conceptuallybased and follows from the theoretical expectations.Anotherapproachis to derive the profile empiricallyby using a calibration groups (Sabherwal sample of high performing and Kirs 1994; Venkatraman 1989; Venkatraman and Prescott 1990). In either case, measures of the dimensions can be developed and validated. A problemwithearlierresearch was the lack of dealing with multiparttasks in a meaningful way. To some extent, the task dimensionspresented here alleviate this problem because they are not tied to outcomes from subtasks thatwouldconstitutephases of an overalltask, from the idea generation sube.g., the "idea" fromthe choose subtask. task, or the "choice" Instead,the task dimensionsand typologyrecommended here are based on the underlying complexitydimensions of the task as a whole. But, just as the McGrathcircumplex can be used on subtasks, so can this typology, althoughresearchers need to take that step. A usable coding scheme for task types willmake it easier to do so. Future research could address a variety of issues, ranging from testing specific fit prescriptionsto extendingthe theory to a broader context. Withrespect to specific organizational is one opportunity to test the role prescriptions, of differentelements of GSS dimensions. For instance, both problemtasks and fuzzy tasks are expected to benefit fromappropriate problem structuringtemplates (in the information processing dimension), and existing research confirmsthis is true in isolated cases. But in most cases, only a single template has been imposed on the task, which itself is often matched to that specific template. What is the best way to provide groups with a choice of problem structuringtemplates that they can explore and use in the most appropriate way? it to Similarly, is important test the prescription that process structuringbe flexible in some
MISQuarterly/September 1998 329

GSS Task/Technology Fit

cases (fuzzy tasks) and more rigidlyenforced in others (decision tasks). How should the prescribed flexibility in process structuring for fuzzy tasks be designed into the technology to make it maximallyavailable to and useful for the groups? Another opportunityis to examine the extent and type of flexibility that a GSS configuration should have in order to make it manageable for groups. The propositionswere stated in a general and positive way, namely that positive effects are expected from prescribedlevels of different dimensions of support. But to what extent is performanceaffected where unneeded functionalityis present? For instance, too much structureor information processing was to interfere with communication expected needs in simple tasks (Gallupe et al. 1988), butthis has yet to be tested empirically. is The question of flexibility particularly important for users, facilitators, and designers of GSS in organizations.Groupsupportsystems are part of the general trend toward development environmentsand extensible toolkitsthat of allow customizationand integration a variety of capabilities. But there are barriersto wider use of GSS technology in organizations, due to a lack of understanding of the circumstances when the technology should be used, use as well as the inappropriate of a "onesize fits all"approach. Even where differenttools are carefully chosen for different parts of a to meeting, it is important choose wisely how those tools are combined. For instance, low or high levels of process structuringshould be chosen to match the group'stask. Groupscan chafe under too rigidan agenda (an inapprofor priatelyhigh level of process structuring a or lose focus under too loose an fuzzy task), agenda (an inappropriately low level of process structuringfor a decision task). With the abilityto classify tasks and GSS technologies on the dimensions developed in this GSS can paper, facilitators choose appropriate tools for specific tasks. Continuing field research in a variety of settings will also help to evolve the theory, e.g., adding contextual factorsthat mightimpactfit issues.
330 1998 MISQuarterly/September

The group performancecriterionis also fertile ground for furtherelaboration. As a starting point, the theory presented here specified group performance inclusively and as an aggregate concept, recognizing that performance is likelyto be defined differently difin ferent situations. How mightthe theory be different if other (or more differentiated) outcome variables were considered, e.g., group cohesiveness or satisfaction? Next to performance-for tasks where performancecan be reasonably defined-satisfaction is probably the most-studiedoutcome variable.Yet results are mixedas to whetherGSS groups are more satisfied than traditionalgroups (Dennis and Gallupe1993). Satisfactionmightbe increased where GSS technology is matched to task, since a good fit is the most efficient way of dealing withthe cognitive load of a given task. Group climate factors such as cohesiveness and conflictresolutionmightalso be enhanced in situations of good task/technology fit. For instance, the effective resolution of judgment tasks requiresgroups to deal with the conflict and uncertaintyinherent in informationabout the task. If a GSS does not providethe requisite communicationsupport, group members are unlikelyto be able to resolve task issues, whichmay lead to a negative groupclimate. Of course these are speculations, but they proelaborationof the depenvide ideas for further dent variableof groupperformance. This theory was introducedwith the assertion that it is a startingpoint and only one layer of context. The the largertotalityof organizational fit prescriptionsneed to be studied in varying contexts and over time. To what extent, in what ways, and in what kindsof organizational environments do the different dimensions of GSS technology presented here exhibit"inter1992)? How does (Orlikowski pretiveflexibility" interactwithtask perthis interpretive flexibility formance?Whatis the role of humanintervention where goals are other than task performance oriented?These are questions that can addressed with field studies be appropriately of longer durationthan the typicalexperiment. Not surprisingly,multipleapproaches will be needed.

Fit GSS Task/Technology

The issues discussed here illustratesome of the challenges associated with developing GSS technologies that are flexible enough to support,process strucprovidecommunication and informationprocessing support at turing, appropriatetimes for appropriatetasks. The theory provides a framework within which some of these issues can be investigated,and guidance can be provided for more effective design and use of the next generationof group supportsystems.

Acknowledgments
We thank Professors GerardineDeSanctis and KennethKozarfor helpfulcomments on earlier versions of this paper, and the MIS Quarterly editors and reviewers for supporting and insightful comments during the revision funded by process. This research was partially the Graduate School of Business and Administration, University of Colorado, Boulder.

References
Barley, S. R. 'Technology as an Occasion for Evidence from Observationsof Structuring: CT Scanners and the Social Order of Radiology Departments," Administrative Science Quarterly (31:1), 1986, pp. 78-108. Bass, B. M., Pryer, M. W., Gaier, E. L., and Effects of Control Flint, A. W. "Interacting Motivation,Group Practice and Problems Journal on Difficulty AttemptedLeadership," of Abnormaland Social Psychology (56:5), 1958, pp. 352-358. Benbasat, I., and Lim, L. "The Effects of Group, Task, Context, and Technology Variables on the Usefulness of Group Support Systems: A Meta-analysis of Experimental Studies," Small Group Research (24:4), 1993, pp. 430-462. Campbell, D. J. "Task Complexity:A Review and Analysis," Academy of Management Review (13:1), 1988, pp. 40-52. W. Carter,L. F., Haythorn, W., and Howell,M. of of A. "AFurtherInvestigation the Criteria

Leadership," Journal of Abnormal and Social 1950, (46:6), Psychology pp. 589-595. Clapper,D. L., McLean,E. R., and Watson, R. T. "An Experimental Investigation of the Effectof a GroupDecision SupportSystem in Influencein Small Groups," on Normative Proceedings of the Twelfth International Conference on InformationSystems, J. I. DeGross, I. Benbasat, G. DeSanctis, and C. M. Beath (eds.), New York, 1991, pp. 273-282. Daft, R. L., and Lengel, R. H. "Organizational Information Requirements,Media Richness and Structural Design," Management Science (32:5), 1986, pp. 554-571. S. P. Davis, J. H., Laughlin, R., and Komorita, S. 'The Social Psychologyof Small Groups: Interaction," Cooperativeand Mixed-motive Annual Review of Psychology (27), 1976, pp. 501-541. Dennis, A. R., and Gallupe, R. B. "AHistoryof Group Support Systems Empirical Research: Lessons Learned and Future Directions," in Group Support Systems: New Perspectives, L. M. Jessup and J. S. Valacich (eds.), Macmillan Publishing Company,New York,1993, pp. 59-77. Dennis, A. R., George, J. F., Jessup, L. M., Nunamaker, Jr., J. F., and Vogel, D. R. "Information Technology to Support ElectronicMeetings,"MIS Quarterly (12:4), 1988, pp. 591-624. Dennis, A. R., Easton, A. C., Easton, G. K., George, J. F., and Nunamaker, Jr., J. F. "AdHoc Versus Established Groups in an in Electronic MeetingSystem Environment," Annual Proceedings of the Twenty-Third Conference on System HawaiiInternational Sciences, Volume III,J. F. Nunamaker,Jr. (ed.), IEEE Computer Society Press, Los Alamitos,CA, 1990a, pp. 23-29. Dennis, A. R., Nunamaker,J., and Vogel, D. "A Comparison of Laboratory and Field Research in the Studyof Electronic Meeting Systems," Journal of Management Information Systems (7:3), 1990b, pp. 107-135. DeSanctis, G., and Gallupe, R. B. "A for Foundation the Study of GroupDecision Support Systems," Management Science (33:5), 1987, pp. 589-609. MIS 1998 Quarterly/September 331

GSS Task/Technology Fit

DeSanctis, G., and Poole, M. S. "Capturing the Complexity in Advanced Technology Use: Adaptive Structuration Theory," Organization Science (5:2), -1994, pp. 121-147. DeSanctis, G., Snyder, J. R., and Poole, M. S. 'The Meaningof the Interface: Functional A and Holistic Evaluation of a Meeting Software System," Decision Support Systems (11:4), 1994, pp. 319-335. Drazin,R., and Van de Ven, A. H. "Alternative Forms of Fit in Contingency Theory," Administrative Science Quarterly (30:4), 1985, pp. 514-539. Easton, G. K., George, J. F., Nunamaker,Jr., J. F., and Pendergast, M. O. "Using Two DifferentElectronic Meeting System Tools for the Same Task: An Experimental Comparison," Journal of Management Information Systems (7:1), 1990, pp. 85-100. Fjermestad, J. "An Integrated Frameworkfor Group Support Systems," Journal of Organizational Computing and Electronic Commerce,forthcoming. Fjermestad,J., and Hiltz,S. R. "Experimental Studies of Group Decision Support Systems: An Assessment of Variables Studied and Methodology," Proceedings of the ThirtiethAnnual Hawaii International Conference on System Sciences, Volume II,1997, pp. 45-65. Gagne, R. M. "Human Problem Solving: Internaland External Events," in Problem Solving: Research, Methodand Theory,B. Kleinmuntz (ed.), Wiley,New York,1966. Gallupe, R. B., DeSanctis, G., and Dickson,G. "Computer-based Support for Group An Experimental Problem-finding: MIS Quarterly(12:2), 1988, Investigation," pp. 277-296. L. Gallupe, R. B., Cooper, W., and Bastianutti, "Why is Electronic Brainstorming More Productive Than Traditional Brainstorming?" Proceedings of the 1990 Administrative Sciences Association of Canada Conference,Whistler,BC, Canada, 1990, pp. 82-92. Gallupe, R. B., Bastianutti,L. M., and Cooper, Journalof W. H. "Unblocking Brainstorms," (76:1), 1991, Applied Psychology pp. 137-142. 332 MISQuarterly/September 1998

Gallupe, B., Dennis, A. R., Cooper, W. H., Valacich, J. S., Bastianutti, L. M., and J. Nunamaker, F. "Electronic Brainstorming and GroupSize," Academy of Management Journal,(35:2), 1992, pp. 350-369. Gopal, A., Bostrom, R. P., and Chin, W. W. "Applying AdaptiveStructuration Theory to Investigate the Process of Group Support System Use," Journal of Management Information Systems (9:3), 1992-93, pp. 45-69. Gray, P., Vogel, D., and Beauclair, R. "Assessing GDSS Empirical Research," EuropeanJournalof OperationalResearch (46:2), 1990, pp. 162-176. Hackman, J. R. "Effects of Task Characteristics on Group Products,"Journal of Experimental Social Psychology (4:2), 1968, pp. 162-187. Hackman, J. R. "TowardUnderstanding the Role of Tasks in Behavioral Research," Acta Psychologica 1969, (31:2), pp. 97-128. Herold, D. M. "Improvingthe Performance Effectiveness of Groups Through a Taskcontingent Selection of Intervention Strategies," Academy of Management Review(3:2), 1978, pp. 315-325. Ho, T. H., and Raman, K. S. "The Effect of GDSS and Elected Leadership on Small Group Meetings,"Journal of Management Information Systems (8:2), 1991, pp. 109-133. Hollingshead, A. B., and McGrath, J. E. "Computer-assisted Groups: A Critical Review of the Empirical Research," in Team Effectivenessand Decision Makingin Organizations,R. A. Guzzo, E. Salas and San Associates (eds.), Jossey-Bass, Francisco,1995, pp. 46-78. Huber, G. P. "Issues in the Design of Group Decision SupportSystems," MIS Quarterly (8:3), 1984, pp. 195-204. Huber, G. P. "A Theory of the Effects of Advanced Information Technologies on Organizational Design, Intelligence, and Decision Making," Academy of Management Review (15:1), 1990, pp. 47-71. Huber,G. P., Valacich, J. S., and Jessup, L. M. "A Theory of the Effects of Group Support Systems on an Organization's

GSS Task/Technology Fit

Nature and Decisions," in Group Support Systems: New Perspectives, L. M. Jessup and J. S. Valacich (eds.), Macmillan Co., 1993, London, Publishing pp. 255-269. Johansen, R. "An Introductionto Computeraugmented Teamwork," in ComputerA Augmented Teamwork: Guided Tour,R. P. Bostrom,R. T. Watson, and S. T. Kinney (eds.), Van Nostrand Reinhold, New York, 1992, pp. 5-15. Joyce, W., Slocum, Jr., J. W., and Glinow,M. "Person-situation Interaction: Competing Models of Fit," Journal of Occupational Behaviour(3:4), 1982, pp. 265-280. P. Processes Laughlin, R. "SocialCombination of Cooperative Problem-solving Groups on Verbal Intellective Tasks," in Progress in Social Psychology, Volume 1, M. Fishbein (ed.), Lawrence Erlbaum Associates, Hillsdale,NJ, 1980, pp. 127-155. Losada, M., Sanchez, P., and Nobel, E. E. "Collaborative Technology and Group Process Feedback. Their Impact on Interactive Sequences in Meetings," in Proceedings of the Conference on ComputerSupportedCooperativeWork,D. Tatar (ed.), Association for Computing Machinery, Baltimore, MD, 1990, pp. 53-64. McGrath, J. E. Groups: Interaction and Performance, Prentice-Hall, Englewood Cliffs,NJ, 1984. McGrath, J. E., and Hollingshead, A. B. Groups Interactingwith Technology:Ideas, Evidence, Issues, and an Agenda, Sage ThousandOaks, CA, 1994. Publications, McGrath,J. E., Arrow, H., Gruenfeld, D. H., Hollingshead, A. B., and O'Connor, K. M. "Groups, Tasks, and Technology: The Effects of Experience and Change," Small GroupResearch (24:3), 1993, pp. 406-420. McLeod, P. L., and Liker, J. K. "Electronic Meeting Systems: Evidence from a Low Structure Environment," Information Research Systems 1992, (3:3), pp. 195-223. Nidumolu, S. R. "A Comparison of the Structural Contingency and Risk-based Perspectives on Coordinationin Softwaredevelopment Projects," Journal of

Management InformationSystems (13:2), 1996, pp. 77-113. Nunamaker,J. F., Dennis, A. R., Valacich, J. S., Vogel, D. R., and George, J. "Electronic MeetingSystems to SupportGroupWork," Communications the ACM(34:7), 1991, of pp. 40-61. O'Neill, M. S., and Alexander, J. F. "Family InteractionPatterns as a Functionof Task Journal of Applied Social Characteristics," Psychology(1:2), 1971, pp. 163-172. W. Orlikowski, J. 'The Dualityof Technology: Rethinkingthe Concept of Technology in Science (3:3), Organizations," Organization 1992, pp. 398-427. for Perrow,C. "AFramework the Comparative Analysis of Organizations," American Review (32:2), 1967, Sociological pp. 194-168. K. Pinsonneault,A., and Kraemer, 'The Impact of Technological Support on Groups: An Assessment of the Empirical Research," Decision Support Systems (5:2), 1989, pp. 197-216. Poole, M. S. "AnInformation-task Approachto OrganizationalCommunication," Academy of Management Review (3:4), 1978, pp. 493-504. Poole, M. S., Siebold, D. R., and McPhee, R. D. "Group Decision-making as a Structurational Journal Process," Quarterly of Speech (71:1), 1985, pp. 74-102. Poole, M. S., Holmes, M., and DeSanctis, G. "Conflict Management in a Computer-supported MeetingEnvironment," Management Science (37:8), 1991, pp. 926-953. Rana, A.., Turoff, M., and Hiltz, S. R. "Task and Technology Interaction A (TTI): Theory of TechnologicalSupportfor GroupTasks," Annual Hawaii Proceedings of the Thirtieth International Conference on System Sciences, VolumeII,1997, pp. 66-76. Rao, V. S., and Jarvenpaa, S. L. "Computer Support of Groups: Theory-based Models for GDSS Research,"ManagementScience (37:10), 1991, pp. 1347-1362. Sabherwal, R., and Kirs. P. "The Alignment Between Organizational Critical Success Factors and Information Technology Capability in Academic Institutions," Decision Sciences 1994, (25:2), pp. 301-330. MIS 1998 Quarterly/September 333

GSS Task/Technology Fit

Sambamurthy, V., and DeSanctis, G. "An ExperimentalEvaluation of GDSS Effects on Group PerformanceDuringStakeholder Analysis," in Proceedings of the TwentyThird Annual Hawaii International Conference on System Sciences, Volume III, J. F. Nunamaker Jr. (ed.), IEEE ComputerSociety Press, Los Alamitos,CA, 1990, pp. 79-88. Sambamurthy, V., and Poole, M. S. "The Effects of Variations in Capabilities of GDSS Designs on Management of Cognitive Conflict in Groups," Information 1992, Research, (3:3), Systems pp. 224-251. Shaw, M. E. "Some Effects of Problem Complexity upon Problem Solution Efficiency in Different Communication Nets," Journal of ExperimentalPsychology (48:3), 1954, pp. 211-217. Shaw, M. E. Scaling Group Tasks: A Method for DimensionalAnalysis,"JSAS Catalog of Selected Documents in Psychology (3), 1973, p. 8. Shaw, M. E. GroupDynamics:The Psychology of Small Group Behavior (3rd ed.), McGraw-HillBook Company, New York, 1981. Sia, C., Tan, B. C. Y., and Wei, K. "Exploring the Effects of Some Display and Task Factors on GSS User Groups,"Information & Management(30:1), 1996, pp. 35-41. Steiner, I. D. Group Process and Productivity, Academic Press, New York,1972. Straus, S. G., and McGrath,J. E. "Does the Medium Matter? The Interaction of Task on Group Type and Technology Performance and Member Reactions," Journalof AppliedPsychology (79:1), 1994, pp. 87-97. Tan, B. C. Y., Raman, K. S., and Wei, K. "An EmpiricalStudy of the Task Dimension of GroupSupportSystem,"IEEETransactions on Systems, Man, and Cybernetics(24:7), 1994, pp. 1054-1060. Thompson, J. D. Organizations in Action, McGraw-Hill Book Company, New York 1967. Venkatraman, N. "The Concept of Fit in Strategy Research: Toward Verbal and Statistical Correspondence," Academy of
334 1998 MISQuarterly/September

Management Review (14:3), 1989, pp. 423-444. Venkatraman, N., and Prescott, J. E. "Environment-strategy Coalignment: An Empirical Test of Its Performance Implications," Strategic Management Journal(11:1),1990, pp. 1-23. Vickers, B. "Designing Layered Functionality WithinGroup Decision Support Systems," Decision Support Systems (11:1), 1994, pp. 83-99. Watson, R., DeSanctis, G., and Poole, M. S. "Using a GDSS to Facilitate Group Intended and Some Consensus: UnintendedConsequences," MIS Quarterly (12:3), 1988, pp. 463-478. Wood, R. E. "Task Complexity: Definitionof Behaviorand the Construct," Organizational Human Decision Processes (37:1), 1986, pp. 60-82. Zigurs, I., Poole, M. S., and DeSanctis, G. "A Study of Influence in Computer-mediated Group Decision Making,"MIS Quarterly (12:4), 1988, pp. 625-644.

About the Authors


lize Zigurs is an associate professor of informationsystems in the College of Business and at Administration the Universityof Colorado, Boulder. Her research interests focus on the design and impacts of computer-based systems for supporting collaborative work. Professor Zigurs'work has been published in Journalof such journalsas the MIS Quarterly, Information Systems, and Management Decision SupportSystems. Bonnie K. Buckland earned her Ph.D. in information systems from the University of Colorado,Boulder.Her research interests are techin the small business use of information the use of technology in education, nology, technoloand the social impactsof information officer the She is currently chief information gy. at Great Plains Regional Medical Center in NorthPlatte,Nebraska.

Potrebbero piacerti anche