Decision development in small groups, III: A multiple sequence model of group decision development ...

22
This article was downloaded by: [Central Michigan University] On: 05 December 2014, At: 05:37 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Communication Monographs Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/rcmm20 Decision development in small groups, III: A multiple sequence model of group decision development Marshall Scott Poole a a Assistant professor of speech communication , University of Illinois , UrbanaChampaign Published online: 02 Jun 2009. To cite this article: Marshall Scott Poole (1983) Decision development in small groups, III: A multiple sequence model of group decision development , Communication Monographs, 50:4, 321-341, DOI: 10.1080/03637758309390173 To link to this article: http://dx.doi.org/10.1080/03637758309390173 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions

Transcript of Decision development in small groups, III: A multiple sequence model of group decision development ...

Page 1: Decision development in small groups, III: A multiple sequence model of group decision development               1

This article was downloaded by: [Central Michigan University]On: 05 December 2014, At: 05:37Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Communication MonographsPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/rcmm20

Decision development in small groups, III: A multiplesequence model of group decision developmentMarshall Scott Poole aa Assistant professor of speech communication , University of Illinois , Urbana‐ChampaignPublished online: 02 Jun 2009.

To cite this article: Marshall Scott Poole (1983) Decision development in small groups, III: A multiple sequence model of groupdecision development , Communication Monographs, 50:4, 321-341, DOI: 10.1080/03637758309390173

To link to this article: http://dx.doi.org/10.1080/03637758309390173

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Page 2: Decision development in small groups, III: A multiple sequence model of group decision development               1

DECISION DEVELOPMENT IN SMALL GROUPS, III:A MULTIPLE SEQUENCE MODEL OFGROUP DECISION DEVELOPMENT1

MARSHALL SCOTT POOLE

This paper argues that traditional models of group decision making as a series ofdiscrete, consecutive phases are inadequate. It advocates replacing them by a morecomplex and accurate model of continuously-developing threads of group activity. Insection I the phasic model is evaluated in view of recent research on decisiondevelopment in small groups. To overcome several problems with the phasic modelsection II develops a new model, which conceptualizes decision development as a setof continuously evolving tracks of activity, intertwining over time. Based on previousresearch, this model advances a new descriptive system, which specifies (a) threetracks of group activity, (b) several types of "breakpoints" which mark changes in thedevelopment of the strands, and (c) a structural model of task accomplishment. Theinterrelations among the three descriptive components give a more complex andaccurate picture of decision development than the phasic model. In section III, twoexplanatory factors which determine the evolution of multiple sequences areexplicated. They are used to generate a series of propositions about multiple sequencesin decision development. These propositions form the framework of a contingencytheory of decision development in small groups.

THAT groups pass through a defi-nite sequence of developmental

stages has become a truism in basic andadvanced textbooks in communication,management, social psychology, andsociology. Widespread acceptance of thisproposition stems from two sources—support garnered from dozens of studiesand, perhaps more important, repetitionof the same simple idea in dozens ofsummaries. Simplicity is one of the hall-marks of good science. However, sim-plicity and elegance can also hide impor-tant details. In the case of group develop-ment, simple summary statements haveomitted careful qualifications concerningscope-conditions and departures fromthe rule that earlier researchers placedon their results.2 A growing body of

Marshall Scott Poole is assistant professor ofspeech communication at the University of Illinoisat Urbana-Champaign.

1An earlier version of this paper was presented at theConference on Small Group Communication Research,Pennsylvania State University, University Park, Penn-sylvania, 30 Apr. 1982.

2Robert F. Bales and Fred L. Strodtbeck, "Phases in

evidence suggests developmental pro-cesses in groups are considerably morecomplex than has hitherto been sup-posed. This evidence warrants a reexam-ination of previous developmental mod-els not only in terms of their particularpropositions but in terms of the basicmodel underlying them—the analysis ofdevelopment as a series of phases.

In this paper, I will argue that thephasic model creates a blind spot incurrent group research. I will argue forreplacement of the phasic conception bya new model that better fits the complex-ities of group development and promisesa more productive perspective. The firstsection of this essay criticizes the phasicmodel of group development and ad-vances a new descriptive system. Thissystem specifies three separate strands ofgroup activity, a number of breakpointswhich interrupt the development of thesestrands, and a logical structure for devel-

Group Problem Solving," Journal of Abnormal andSocial Psychology, 46 (1951), 485-95; Warren G. Ben-nis and Herbert A. Shepard, "A Theory of GroupDevelopment," Human Relations, 9 (1956), 415-37.

COMMUNICATION MONOGRAPHS, Volume 50, December 1983

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 3: Decision development in small groups, III: A multiple sequence model of group decision development               1

322 COMMUNICATION MONOGRAPHS

opmental activity. In the second sectionthe descriptive system is linked withexplanatory factors to generate a seriesof propositions about multiple sequencesof decision development. Although thesepropositions do not in themselves consti-tute a theory, they advance us along theroad to a contingency theory of decisiondevelopment.

DESCRIPTION OF DEVELOPMENTAL

SEQUENCES

Phasic Descriptions and Their Problems

The "phase" or "stage" is the tradi-tional unit of analysis in the study ofgroup development. The dictionarydefines stage as "a presumably naturalor non-arbitrary division of a changingprocess" and phase as "a recurrent statein something that exhibits a series ofchanges; for example, the several phasesof the moon."3 Both terms connote aperiod of more or less unified activitythat fulfills some function necessary tocompletion of the group's task—forexample, the orientation phase of Balesand Strodtbeck's model, in which thegroup defines its problems, decides howto attack it, and shares relevant informa-tion.4

The phasic model is the dominantarchetype for developmental studies ofgroup communication: as such, it exertsa subtle pressure on researchers to utilizea certain type of description and conformto a single theoretical form. In the idealcase, a phasic model would be composedof a small, manageable number of phaseswhich can be easily identified viaempirical operations and which occur ina definite order (or a small set of dif-ferent orderings). The phases would beapplicable to a wide range of groups and,in the best of all possible worlds, wouldsupport generalizations about all group

3Webster's Third New International Dictionary(Springfield, MA: G. C. Merriam, 1961).

4Bales and Strodtbeck.

experiences. The model generally ex-plains group behavior in terms of neces-sary structural conditions that require acertain set of phases to occur in a definitetemporal order, such as Bales andStrodtbeck's logical sequence for prob-lem-solving, Fisher's four-step conflictresolution pattern, or Lacoursiere's life-cycle model.5 While contingencies mayrequire limited departures from thesequence, phases are presumed to existand to occur in a definite order due to thepractical prerequisites of completing atask, coming to consensus, living in agroup, etc. With a few exceptions, pre-vious research has operated within thisideal and has strained to realize it at theexpense of attention to other forms ofdescription or explanation.

Without question, the phasic modelhas led to great advances in our under-standing of group processes. Researchemploying this model has generated astrong foundation of knowledge. How-ever, progress in the social sciencesrequires a continuing critique of preva-lent ideas with an eye to improving ourrepresentations and explanations ofsocial processes. Several developments inrecent research suggest that it is time tomove beyond the phasic model.

We should begin with the observationthat the great majority of studies in thephasic tradition identify developmentalphases ex post facto. This processusually consists of tracking trends inindicators of group activity and deducingphases from observed activity. Fisher,6

for example, identified decision-makingphases by dividing discussions into fourparts and comparing the relative levels ofoccurrence of various acts across parts.He labeled his first segment the orienta-tion phase because it contained more acts

5Bales and Strodtbeck; B. Aubrey Fisher, "DecisionEmergence: Phases in Group Decision-Making," Com-munication Monographs, 37 (1970), 53-66; R. Lacour-siere, The Life Cycle of Groups (New York: HumanSciences Press, 1980).

6Fisher, "Decision Emergence."

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 4: Decision development in small groups, III: A multiple sequence model of group decision development               1

DECISION DEVELOPMENT IN SMALL GROUPS, III 323

and interacts characteristic of orientation(e.g., ambiguity, clarification, tentativeevaluation of opinions) than the otherthree parts. Other studies using thismethod include those by Bennis andShepard, Poole, Mintzberg et al., andHirokawa.7 Summaries that attempt topresent "general" models, such as Tuck-man's widely-cited review and Lacour-siere's book, rely on ex post facto studiesas their primary source of evidence.8

There is, of course, nothing inherentlywrong with ex post facto methods. Theyserve the valuable function of disclosingnew phenomena and offer patterns forinterpretive understanding which areoften superior to that permitted bydeductive designs. However, when expost facto patterns become commonknowledge, there is a danger they willlead to the assimilation of observations toexpectations, to the sharpening of someoccurrences and the neglect of others inservice of a coherent story. They canquite easily become self-fulfilling andblind the researcher to the "marvelousparticularity" of the object of study.When most support for a theory stemsfrom ex post facto studies, we should bewary of accepting it wholeheartedly. AsTudor9 has observed, discovery of pat-terns is only one aspect of research;researchers must also confirm their exis-tence via empirical tests.

Several studies have explicitly at-

7Bennis and Shepard; Marshall Scott Poole, "Deci-sion Development in Small Groups, II: A Study ofMultiple Sequences in Group Development," Commu-nication Monographs, 50 (1983), 206-32; HenryMintzberg, Duru Raisinghani, and Andre Theoret,"The Structure of 'Unstructured' Decision Processes,"Administrative Science Quarterly, 21 (1976), 246-75;Randy Y. Hirokawa, "Group Communication andProblem-Solving Effectiveness: An Investigation ofGroup Phases," Human Communication Research, 9(1983), 291-305. See also Lacoursiere's excellent reviewfor several dozen other studies employing this method ofphasic identification.

8Bruce Tuckman, "Developmental Sequence inSmall Groups," Psychological Bulletin, 63 (1965), 384-99; Lacoursiere.

9Andrew Tudor, "Misunderstanding EverydayLife," Sociological Review, 24 (1976) 479-503.

tempted to fulfill this requirement bytesting a priori hypotheses about devel-opmental phases. Unfortunately, theimprint of the phasic model of thesestudies is so strong that their results,though favorable, are inconclusive andequivocal. In some cases researchershave posited vague and indefinite hy-potheses, which are partially-confirmed,then rescued and revamped in the discus-sion.10 Those studies which carefullyspecify clear, definite hypotheses arehindered by problems of design andanalysis that bias them in favor of find-ing a unitary sequence of phases.11 Theseproblems include (a) dividing discussionsinto the same number of segments asexpected phases, thereby hiding develop-mental complexities that might refutethe hypotheses and (b) combining dataacross groups, thereby eliding between-group developmental differences.12

Moreover, the a priori tests, as well asa large proportion of the ex post factoanalyses, use only a few types of behav-ior categories to identify phases or con-firm phasic sequences. Bales and Strodt-beck, for example, relied on trends infour categories—information, opinions,solutions, and socioemotional acts—totest their model. Focusing on such asmall sample of functions may ignoreothers which would qualify conclusionsabout phases. An ambiguous and aimlessperiod at the beginning of discussion

10See, for example, Charles Morris, "Changes inGroup Interaction During Problem Solving," Journal ofSocial Psychology, 81 (1970), 157-65; John M. Ivance-vich, "A Study of a Cognitive Training Program:Trainer Styles and Group Development," Academy ofManagement Journal, 17 (1974), 428-39.

11Examples of such studies include: Bales and Strodt-beck; Edward A. Mabry, "Exploratory Analysis of aDevelopmental Model for Task-Oriented SmallGroups," Human Communication Research, 2 (1975),66-74; Philip J. Runkell, Marilyn Lawrence, ShirleyOldfield, Mimi Rider, and Candee Clark, "Stages ofGroup Development: An Empirical Test of Tuckman'sHypothesis," Journal of Applied Behavioral Science, 1(1971), 180-89.

12See Marshall Scott Poole, "Decision Developmentin Small Groups, I: A Comparison of Two Models,"Communication Monographs, 48 (1981), 1-24.

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 5: Decision development in small groups, III: A multiple sequence model of group decision development               1

324 COMMUNICATION MONOGRAPHS

seems to be quite different from a quick,well-planned orientation by the leader,yet they would both be classified asorientation phases if information-givingand -asking were the functions used toidentify periods. This problem is likelyto be even worse in cases where only aselective sample of acts is used, i.e.,where the coding system has an "other"category or only a few categories from asystem are used. Those acts not sampledmay play a critical role in the discussionand lend a flavor to group activities notindicated by the restricted sample ofphasic markers.

Studies examining the detailed localstructure of group activities haverevealed a much more complex picture ofgroup behavior than simple phasicschemes can encompass. Poole13 isolatedclusters of associated behaviors from 30-interact segments of discussions for asample of student groups and a sample ofphysician groups. He found 19 distinctactivity clusters for students and 17 forphysicians, suggesting considerable di-versity of behavior throughout the dis-cussions. In a study of member-to-leaderbehavior in classroom groups Mann14

found six dimensions that combined incomplex ways as the group interactionunfolded. Considering that both studiesfocused on only a small range of groupbehaviors, they reveal a high degree ofcomplexity and suggest things are not assimple as they may seem.

The results of previous research alsosuggest that decision development is con-siderably more complex than the phasicmodel envisions. Although reviews gen-erally stress convergence in the findingsof ex post facto studies, a closer lookuncovers notable differences. For exam-ple, five studies of task-oriented groupsby Bales and Strodtbeck, Cronin andThomas, Fisher, Tuckman, and Zurcher

13Poole, "Decision Development, I."14Richard D. Mann, "The Development of Member-

Trainer Relationships," Human Relations, 19 (1966),84-117.

report developmental sequences that fol-low the form of the unitary phasic model.When we look at specifics, however, eachstudy reports a different set of phasesand posits somewhat different explana-tory mechanisms for movement throughthe sequence.15 Gustafson and Cooperhave noted a similar diversity in phasicanalyses of therapy and traininggroups.16 Of course, these differences canbe interpreted as the result of "noise"from different research designs andgroup situations. This is the route syn-thetic articles have traditionally taken,and it leads to summary schemes thatemphasize commonalities across the var-ious findings. However, given the con-ceptual and methodological biases raisedearlier, another interpretation is possi-ble: differences in findings may reflectactual differences in development, differ-ences which are attenuated by efforts toaccommodate our observations to anoverly-restrictive conception of develop-mental processes.

Recent studies of multiple sequencesand cycles in group decisions support thelatter interpretation. Several studieshave tested the assumption of a simpleunitary sequence of phases and found ituntenable. This and other researchindicate that phases can occur in manypossible orders, depending on variousfactors.18 One such factor is the nature ofthe group's task: studies of phasic devel-opment in bargaining have shown anordering of phases quite different from

15Bales and Strodtbeck; Fisher; Tuckman; T. E.Cronin and N. C. Thomas, "Federal Advisory Pro-cesses: Advice and Discontent," Science, 171 (1971),771-79; Louis A. Zurcher, "Stages of Development inPoverty Program Neighborhood Action Committees,"Journal of Applied Behavioral Science, 15 (1969), 223-58.

16James P. Gustafson and Lowell Cooper, "Uncon-scious Planning Small Groups," Human Relations, 32(1979), 1039-64.

17Poole, "Decision Development I" and "DecisionDevelopment II": Teresa A. Chandler, "Decision-Making in Small Groups: A Comparison of Two Mod-els," master's thesis, Cleveland State University 1981.

18Hirokawa; Mintzberg et al.

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 6: Decision development in small groups, III: A multiple sequence model of group decision development               1

DECISION DEVELOPMENT IN SMALL GROUPS, III 325

that obtained for problem-solvinggroups.19 Mintzberg et al. suggest thatthe degree of uncertainty about the situa-tion and the degree of consensus onvalues govern the complexity of decisionpaths.2 Several studies, including Schei-del and CrowelPs classic work, havefound complex cycles in group decisions,cycles so complicated that they may notreduce to a small set of unified phasicunits.21

The complex nature of group activitytypes and the extreme variability ofphases and phasic sequences suggeststhat there is a much wider range ofgroup activities than has formerly beenassumed and reinforces the notion thatthe traditional phases may be too restric-tive. The difficulties inherent in cram-ming diverse findings into a single,restricted set of phases are illustrated bythe following passages from Lacour-siere's otherwise excellent summary ofgroup development research:

The more the conditions of a new experienceapproach those that maximize the development ofGDSs (Group Developmental Stages), the morelikely it is that these will occur, and the moreeasily they will be observed.... When conditionsseem suitable for the development of GDSs andthey do not seem to occur, a more careful examina-tion should show either that GDSs are present, butperhaps partly obscured, that they are not veryintense, that they have taken the form of one of theresistive variations, or that conditions are notreally appropriate.22

This observation was based on anexhaustive examination of over ahundred developmental studies in whichLacoursiere attempts to reconcile a trulydiverse set of findings with his GDS

19Linda L. Putnam and Tricia S. Jones, "The Role ofCommunication in Bargaining," Human Communica-tion Research, 8 (1982), 262-80.

20Mintzberg et al.21Thomas Scheidel and Laura Crowell, "Idea Devel-

opment in Small Groups," Quarterly Journal of Speech,50 (1964), 140-45; Mintzberg et al.; Poole, "DecisionDevelopment II"; Uma A. Segal, "The Cyclical Natureof Decision Making: An Exploratory Empirical Investi-gation," Small Group Behavior, 13 (1982), 333-48.

22Lacoursiere (emphasis mine).

model. At its conclusion he includes these"escape clauses," some of which cu-riously resemble the Ptolemaic effort to"save the appearances," to account forthe diverse departures from his centralformula.

Extensive qualifications are necessaryto maintain the phasic model because itcan no longer encompass what we knowabout decision development. Recentfindings sketch the phasic model pastreasonable bounds and suggest that it istime to move beyond it. This certainlydoes not mean that group research is nowin a period of "crisis." Rather it is whatshould be expected in an advancing disci-pline: existing concepts and models areused as stepping stones toward moreaccurate and encompassing positions.23

Nor does it mean that findings of pastresearch are not valuable. Phasicresearchers have spelled out necessarystructural sequences for decision makingwhich can serve as explanatory mecha-nisms in future models. Dewey's prob-lem-solving sequence, adapted by Balesand Strodtbeck among others, is proba-bly the most widely-accepted of these.24

Fisher's conflict management sequenceis another promising effort.25 Bothschemes attempt to set out a logically-required sequence for moving to a deci-sion: they advance minimum necessarystructures for group convergence andaction. Although their authors equatelogical order with temporal order, thereis no reason this is empirically neces-sary.26 The schemes certainly representpaths of least resistance for group work,but various contingencies, such as con-flict or task difficulty, may preventgroups from following them and diver-sify group progressions considerably.The power and value of these schemes

23Stephen Toulmin, Human Understanding (Prince-ton, NJ: Princeton University Press, 1972).

24John Dewey, How We Think (Boston: D. C.Heath, 1910).

25Fisher.26Poole, "Decision Development, II."

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 7: Decision development in small groups, III: A multiple sequence model of group decision development               1

326 COMMUNICATION MONOGRAPHS

lies not in their temporal ordering but intheir specification of logical priorities forgroup decision making, priorities thatshape, but do not determine, group activ-ities. They can still serve as one explana-tion of decision development, just not asthe sole explanation. In addition to itsgenerative mechanisms, phasic researchprovides us with a wealth of particularfindings that set bounds and parametersfor future models. As the next sectionshows, these findings are the source ofmany ideas in the more advanced model.

To summarize the thrust of the argu-ment thus far, traditional phasic defini-tions—which emphasize isolating a sim-ple set of periods of unified activity—aretoo general and too vague to encompassthe diversity of group activities. Studiesof group development show a far greatervariety of activities than can be coveredby any simple set of phases; they alsosuggest that prolonged, unified periodsof activity may not always occur ingroups. These conclusions imply that weshould move away from the traditionalconception of phases toward a moredynamic and flexible model.

Multiple Sequence Model of DecisionDevelopment

The model developed here attempts toavoid the oversimplifications of thephasic conception, yet at the same timeaccount for the observable unity of muchgroup activity. Rather than picturinggroup decision making as a series ofphasic "blocks" dropped one afteranother into sequence, it portrays devel-opment as a set of parallel strands ortracks of activity which evolve simulta-neously and interlock in different pat-terns over time.27 Each track represents aseparate aspect or mode of group activi-ty—for example, task process or topic

27This model is presaged in Scheidel and Crowell's"spiral model" of idea development, which has beenelaborated considerably in B. Aubrey Fisher, SmallGroup Decision-Making, 2nd ed. (New York:McGraw-Hill, 1980).

focus—and the various tracks areassumed to develop unevenly. For thisreason, coherent, unified phases do notexist at all points in a discussion. Whenthe development of the tracks convergesin a coherent pattern, phases similar tothose in the classic research may befound. However, at other points theremay be no relationship among the tracksand therefore no recognizable phases. Inthis case each track is interpretable in itsown right, but the entire ensemble oftracks does not yield a meaningful analy-sis. This model provides a richer andmore complete description of develop-mental processes than does the phasicconception. It is also capable of encom-passing and reconciling the diverse find-ings catalogued above. The model con-sists of three components: a set of activitytracks, breakpoints signalling changes ingroup activity, and a structure for taskaccomplishment. Each of these will bediscussed in turn.

Activity Tracks. The group activitytracks form the core of the multiplesequence model. A wide range of activi-ties could be incorporated into the model,extending from traditional problem-solving functions and bargaining movesto relational definition and identitymaintenance strategies. Previous re-search suggests that a minimal descrip-tion of group decision making requires atleast three activity tracks:

(1) Task process activities: those activities thegroup enacts to manage its task.

(2) Relational activities: those activities thatreflect or manage relationships among groupmembers as these relate to the group's work.

(3) Topical focus: the substantive issues andarguments of concern to the group at a givenpoint in the discussion.28

28Before proceeding any further, it is necessary toacknowledge that all descriptions are theory-laden (al-though the quality and explicitness of the theory differsfrom case to case). Rather than being a weakness, this isan advantage because it provides a link between presentdescription and previous theoretical advances. Thedescriptive matrix derived here draws on the work ofdozens of previous researchers and would not be possible

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 8: Decision development in small groups, III: A multiple sequence model of group decision development               1

DECISION DEVELOPMENT IN SMALL GROUPS, III 327

TABLE 1

CLASSIFICATIONS FOR TWO ACTIVITY TYPES"

Task ProcessActivities

Problem ActivityTl . Problem AnalysisExecutive ActivityT2a. OrientationT2b. Process ReflectionSolution ActivityT3a. Establish Solution GuidelinesT3b. Solution DesignT3c. Solution EvaluationT3d. Solution Confirmation and Selection

RelationalActivities

Work-focused RelationshipsRl. Focused Work (no criticism; extended

idea development and analysis)R2. Critical Work (idea development

through criticism and repartee)ConflictR3a. OppositionR3b. Resolution-AccommodationR3c. Resolution-Avoidance/SmoothingR3d. Resolution-Integration (Bargaining/

Consensus Building/Problem-Solving)IntegrationR4. IntegrationAmbiguous RelationshipsR5. Expression of Ambiguity

"Both classifications represent the major distinctions in coding systems presently under development. The distinctionsare based on previous research, notably Fisher, "Decision Emergence," Hirokawa, Mintzberg et al., and Poole,"Decision Development, II." Finer distinctions within some classifications are also being developed.

Table 1 displays examples of categoriesfor task process and relational activi-ties.29 Methods for identifying topicalfocus are outlined in several studies. °

These particular types were chosenfor two reasons. First, they embody thenecessary behavioral requirements fororganized work. March and Simon andJames D. Thompson, among others,have argued that any organization must

without it. It is also necessary to acknowledge that a"full description" is impossible if by this we mean atimeless, absolutely adequate picture. The best we cando is to incorporate what we know of the culture'scategories for understanding and acting in interaction.These categories emerge in participants' attempts tochannel their discussions and in their reflections abouttheir own activities, as well as in researcher's reports;over the years they appear repeatedly in studies frommany different areas of research and form a bedrock forunderstanding group processes..

29Coding systems for the first two activity strands arecurrently being developed.

30Robert D. McPhee, Marshall Scott Poole, andDavid R. Seibold, "The Valence Model Unveiled: ACritique and Reformulation," in Communication Year-book 5, ed. M. Burgoon (New Brunswick, NJ: Interna-tional Communication Association—Transaction Press,1981); David M. Berg, "A Descriptive Analysis of theDistribution and Duration of Themes Discussed byTask-Oriented Small Groups," Speech Monographs, 34(1967), 172-75; B. Aubrey Fisher, "The Process ofDecision Modification in Small Discussion Groups,"Journal of Communication, 20 (1970), 51-64.

attend to two basic requirements in orderto operate: (a) it must transform materi-als or ideas (the technical aspect) and (b)it must coordinate the individuals whowork together to accomplish its task (thesocial aspect).31 Group activities ad-dressed to the technical aspect arereflected in the topical focus and taskprocess categories. Activities addressed tothe social aspect of work are reflected inboth relational and task process catego-ries. At root, this description is based onthe assumption that groups are goal- orinstrumentally-oriented. The centralgenerative mechanism for the multiplesequence model is the task or goal thegroup attempts to carry out, in this caseto make a decision. Groups are assumedto be groups organized to do somethingand what they do is assumed to constraintheir developmental processes.

31James G. March and Herbert A. Simon, Organiza-tions (New York: John Wiley, 1958); and James D.Thompson, Organizations in Action (New York:McGraw-Hill, 1967). These components also play aprominent role in leadership theory and research; see,for example, David G. Bowers and Stanley E. Seashore,"Predicting Organizational Effectiveness With a FourFactor Theory of Leadership," Administrative ScienceQuarterly, 11 (1966), 238-63.

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 9: Decision development in small groups, III: A multiple sequence model of group decision development               1

328 COMMUNICATION MONOGRAPHS

The three activity types were alsoselected because they reflect basic dis-tinctions among levels of group work.Topical focus can be distinguished fromtask process and relational activities onthe basis of the content-process division32

while task process and relational activi-ties are split by the classic task-socioemo-tional or task-integration divisions.33

Previous developmental research hashonored these distinctions: one group ofphasic studies has concentrated pri-marily on task process activities,34 whileanother has focused mainly on relationalactivity,35 and a third has emphasizedcontent.36 Curiously, although a fewstudies have incorporated two genres ofactivities,37 none has spanned all three.The inclusion of all three tracks in themultiple sequence model permits muchfiner resolution than previous studieshave enjoyed and recognizes the com-plexities of group processes.

Even a cursory look at Table 1 indi-cates that other activity tracks could havebeen included in the model. Especiallypromising are the control strategies out-lined by relational researchers38 and var-ious conflict-management activities.39

Clearly, these could be incorporated intothe model in future refinements. The

32Robert F. Bales, Interaction Process Analysis: AMethod for the Study of Small Groups (Cambridge,MA: Addison-Wesley, 1950).

33Bales.34Bales.35See, for example, Cronin and Thomas; Fisher,

"Decision Emergence"; Hirokawa; Mintzberg et al.35See, for example, Bennis and Shepard; Mabry;

Poole, "Decision Development, I."36Berg; Fisher, "Process of Decision Modification."37Bales and Strodtbeck.38See, for example, L. Edna Rogers and Richard V.

Farace, "Analysis of Relational Communication inDyads: New Measurement Procedures, "Human Com-munication Research, 1 (1975), 222-39; and Joseph P.Folger and Marshall Scott Poole, "Relational CodingSchemes: The Question of Validity" in CommunicationYearbook, 5 ed. Michael Burgoon (New Brunswick, NJ:International Communication Association—Transac-tion Press, 1981), pp. 235-47.

39Alan Sillars, "Attributions and Communication inRoommate Conflicts," Communication Monographs, 47(1980), 180-200.

three tracks outlined here represent aminimal model that clearly reflects therole of the decision task as a generativemechanism for developmental processes.

The three activity tracks enter into themultiple sequence model differently thanthey would in the phasic model. Phaseswould require coordinated sets of thethree activity types, as when Tuckmancharacterizes his "forming" stage as aperiod in which the group is concernedwith problem definition on the task leveland defining relationships on the socioe-motional level.40 This implies that therewould be a fairly restricted set of combi-nations available and when one combi-nation (phase) came to a conclusion allelements in it would terminate—a giventype of relational behavior could notcontinue once its corresponding taskactivity had ceased.

The multiple sequence model, exem-plified in Figure 1, is much more flexiblethan the phasic model. It presents apicture of three strands of activity devel-oping simultaneously, but at unevenrates. Topic shifts may occur much moreoften than shifts in relational orientation;the group may undergo several shifts inrelationships while engaged in a singletask process, such as problem analysis.Although two or three specific activitytypes may co-occur regularly, they do notcoordinate rigidly. The development ofthe intertwining threads is broken atvarious points that represent transitionsbetween activities—e.g., topic shifts,moving on to the next step of a decisionprocess, the start of an argument. Some-times these breakpoints interrupt onlyone thread, sometimes two or all three.The end effect is halting development,with the strands pushing forward atdifferent rates depending on the numberof breakpoints and with differing degreesof association and coordination betweentracks as the discussion unfolds.

40Tuckman.

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 10: Decision development in small groups, III: A multiple sequence model of group decision development               1

*ask , 2a . l a , 3a . 3b . l a , 3a-3b , 3c , 3d , 2a , 3a-3"d , 3d i 3e i 3dProcess . 4—— j A j 1 j | | ] [

T T D D T T

Relational , R6 . Rl .R5| R2 , R3 , R2 , Rl ,R5| R2 . R3 , R2 • R4 ,R6 , R2 | R51 I ~" p""*~^~"~^~ [ I I I I I I t ""IT C T D C D T

Topical T1 _2 T3 ( T2 T3 x4T3.T2T3.T4, T2 , T3-T2 T3 , T2 , T3 ,T5 , T3 , T2 , T3 , T2 ,T3 , T3 , T6rZZr | Tl | T 2 | T3 | T 2 | T 3 ffjffT T T T T T TT •£ T T T T T T T T T T T T T

I 1 1 I I I I 1 I I I I I I I I I I 1 I I I I I 1 I I I 1 I 1 I 1 I I I I I 1 1 1 I 1 I 1 I I I 1 I I I 1 I I I I I I I I 1 I I

1. See Table 1 for a key for task process and relational activities .2. Topics are numbered: Tl i s topic 13. Key to breakpoints:

T : Topic ShiftA : Adjournment/BreaksD : DelaysC : ConflictF : Failure

FIGURE 1A SAMPLE CHART SHOWING THE THREE INTERRELATED ACTIVITY THREADS

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 11: Decision development in small groups, III: A multiple sequence model of group decision development               1

330 COMMUNICATION MONOGRAPHS

Because it is more flexible, the multi-ple sequence model allows us to identifywhen the traditional phases occur, aswell as when more complex patternshold. As we have noted, the group'sdevelopment at any point in time can berepresented by an ordered triple, such as(1) problem analysis, conflict, topic x or(2) solution elaboration, focused work,topic y. In this scheme, phases would beindicated by certain coherent patternsamong the elements of the triple. Forexample, an orientation phase might beindicated by triples with "orientation"and "problem analysis" as their firstterm, "ambiguity" and "focused work"as the second term, and rapid shifts oftopic in the third term. Coherent pat-terns would not always occur, however.Some triples or alternatively, periodswhere no triple consistently dominatesmay indicate "transition periods"—times when no coherent theme or con-cern characterizes group activity. Any-one familiar with groups has observedperiods when a group seems confusedand incoherent or when there seems to bea mixture of concerns, with none clearlydominant. Such periods are not phases inthe traditional sense of the term, and it isuseful to be able to distinguish them fromsegments where the elements of the triple"fit together."

Breakpoints. Breakpoints are of keyimportance to an understanding of thedevelopmental process because they rep-resent developmental transitions. Break-points govern the group's pacing andserve as a good indicator of linkagesamong threads of activities. When break-points only interrupt a single track, thissuggests that the track is operating some-what independently of the others. When,on the other hand, the breakpoint inter-rupts all three tracks, the tracks arelikely to be coordinated and the rupturemay presage major, recognizable shiftsin activity.

At least three types of breakpoints,each with very different implications for

developmental processes, can be distin-guished: (1) Normal breakpoints relateto the pacing of the discussion. The mostcommon normal breakpoint is the topicshift, which may occur dozens of times asa group repeatedly doubles back to vari-ous topics.41 Two other normal break-points are natural breaks in activitiessuch as adjournment or quitting time,and planning periods, in which one ormore members attempt to organize thegroup. Normal breakpoints are more orless expected or sanctioned by the group;they are nothing unusual and the groupcopes with them as a matter of normalprocedure. (2) Delays are a second typeof breakpoint. Mintzberg et al. observed"comprehension cycles" in organiza-tional decisions—periods when the deci-sion process halted as the group workedout a problem by recycling through thesame analysis or development processrepeatedly.42 The point at which a groupenters such a cycle is an importantbreakpoint because it signals a shift inthe tenor of discussion. For all practicalpurposes the group is in a holding pat-tern until it makes a transition to anothertype of activity. Members do not ordi-narily anticipate delays; depending onthe nature of the delay and the mood ofthe group, this breakpoint can signal thestart of a difficult or a highly creativeperiod. (3) Disruptions are the thirdgenre of breakpoints. At least two sorts ofdisruptions can be distinguished. First, amajor disagreement or conflict mayoccur. When this happens in a verydefinite and salient manner, the group'sactivity is redirected and disrupted, and amajor reorientation may be required forthe group to proceed. Second, the groupmay fail—the strategies or solutionsemployed by the group may not be suffi-cient to its task. The importance of thisbreakpoint should be evident as shouldthe fact that both disruptive breakpoints

41Scheidel and Crowell.42Mintzberg et al.

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 12: Decision development in small groups, III: A multiple sequence model of group decision development               1

DECISION DEVELOPMENT IN SMALL GROUPS, III 331

require considerable adjustment on thepart of the group.

The breakpoints illustrate the impor-tance of single critical events or juncturesin group development. They pace discus-sion, and they also introduce perturba-tions into smoothly developing trendswhich require reactions on the part ofthe group. In the case of normal breaks,the reactions may be smooth and to someextent preprogrammed. For delays anddisruptions major periods of repairshould be necessary. As we will notelater on, certain contingencies will alsoresult in different group reactions tobreakpoints.

Structural Requirements for Decision.The activity types and breakpoints aretwo critical elements of a description ofdevelopment. But another issue remainsto be addressed: What is the object of thisactivity? If we consider that groups existfor a purpose—to accomplish work orattain some goal for members—theobject of group activity is the accumula-tion of components for task accomplish-ment. The group's action, extended overtime, gradually fills out the prerequisitesfor the task (which may be a part of alarger mission) and, when members havea sense of completion (which itself is anegotiated accomplishment), the groupmoves on to a new concern. Componentsof task accomplishment will vary fromtask to task. For decision making theseinclude recognition of the need for adecision, definition of the decision prob-lem, diagnosis of the problem, search andgeneration of solutions, adaptation ofsolutions to the group's circumstances,consensus on criteria for an acceptablesolution, selection of a solution, andimplementation planning.43 These com-ponents make up a structure for decision

43Requirements for task completion will vary fromtask to task. For a therapy group, for example, they mayinclude discovering members' needs, working out anagenda for the group, working out leader-member rela-tions, resolving differences among subgroups, and help-ing members identify and carry out changes.

that sets forth the logical priorities foraccomplishing choice.

In many cases a group will begin thedecision process with some of the prereq-uisites already satisfied. In these in-stances the group's efforts should bedirected toward working out those ele-ments of the structure it does not possess.For example, one group may alreadyhave a shared sense of the problem andalso a commitment to decide and acttogether; in this case, the major tack is anagreed upon and effective solution, andthe group's interaction should bedirected toward developing it. A dif-ferent group may have no more than avague sense of a problem; this groupwould have to fill out nearly the wholestructure of components, and its develop-ment should be much more complicatedthan that of the first group.

Summary: Advantages of the MultipleSequence Model

The description of developmental pro-cesses elaborated here is based on a viewof group process as a set of interlockingtracks of activities oriented toward taskor goal accomplishment. The elements ofthe description are a set of three activitytypes, breakpoints signalling changes ingroup activity, and an objective for deci-sion activity—the accumulation of struc-tural components for task accomplish-ment. This description presents severaladvantages and also raises several ques-tions.

One advantage of the model is thegreater accuracy and finer resolution itsdescriptions promise. The model com-bined several genres of description whichhave generally been separate—the taskprocess, relational, and topical (or the-matic) approaches—in a frame that rec-ognizes their concurrent development. Italso capitalizes on a number of studiesthat have discovered an uneven, halting,"spiralling" progress in group work: itproposes an intertwining, multiple-sequence developmental process which is

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 13: Decision development in small groups, III: A multiple sequence model of group decision development               1

332 COMMUNICATION MONOGRAPHS

disrupted at a number of breakpointsthat require reorganization of groupeffort. The role of the group's own his-tory in its development is incorporatedthrough the emphasis on accumulationof structural components for task accom-plishment. Together these three ele-ments provide a description more accu-rately and finely attuned to the picture ofgroup work emerging in recent yearsthan does the phasic model.

A second advantage of the model is itsgreater flexibility relative to the phasicmodel. More and more varied combina-tions of activities can be observed, andthis permits researchers to distinguishperiods of unified activity (i.e., periodswhere phases exist) from periods whichlack coherency or represent transitions ingroup activity. However, there is nomultiplication just for multiplication'ssake—the proposed categories all emergefrom the tradition of research in thegroup.

A third advantage of the model stemsfrom its inclusion of critical events in thedevelopmental description. Throughidentifying breakpoints such as compre-hension cycles or disruptions, the modelenables the researcher to pinpoint animportant class of critical incidents orturning points in group activity. Thisprovides a valuable counterbalance to thestandard methods of interaction analysis,which tend to assign equal weight toevery act in group discussions.

In addition to its advantages, themodel suggests three key questions. Thefirst is addressed in the remainder of thisessay. Given the assumptions of thisdescription, what factors are responsiblefor observed developmental sequences,what factors constrain or cause differen-tiation in developmental patterns? Thesecond question must be left for thefuture but is both interesting and impor-tant: How does the group's interactionaccomplish its development? It is com-monly assumed that macro-level pro-gressions are the product of micro-level

interactions among members. To under-stand how this occurs, one must investi-gate how members manage and coordi-nate their activities, how they move fromone activity to another, and how theymaintain or repair coherence of activitiesin the face of various breakpoints. Thedescriptive system provides a macro-levelmap to be explored in terms of micro-interactional principles. The third ques-tion deals with measurement of the mod-el's elements and also cannot be dealtwith in detail here. Previous research hasprovided measurement strategies for thethree activity-types. The breakpointsdescribed here could also be identifiedfrom the interaction itself. The compo-nents for task accomplishment are stick-ier; at least two strategies are possible.On the one hand, it may be possible tospecify a priori the structure for taskaccomplishment; in the case of decisionmaking enough has been written to makean informed interpretation. On the otherhand, it may be more desirable to assessmembers' perceptions of the requiredcomponents. Often groups redefine orreinterpret their tasks as they work, andthe best way to discern these changes is tomonitor the group's conception of itstask.

CONTINGENCIES GOVERNINGDECISION DEVELOPMENT IN SMALL

GROUPS

The ideas in this section are more thana catalog and less than a theory. Thecontingency factors represent definitebets as to which elements most stronglyinfluence the course of decision-makingactivities. They are based on the mostreliable findings from small groupresearch, and they are unified by a com-mon generative mechanism and descrip-tive model. However, they do not qualifyas a formal theory. For one thing, aworkable theory requires empirical sup-port, and no tests of the hypothesesadvanced here have yet been completed.

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 14: Decision development in small groups, III: A multiple sequence model of group decision development               1

DECISION DEVELOPMENT IN SMALL GROUPS, III 333

They are based on previous results buthave not been established in their ownright. More important, this frameworkby itself tells only half the story. Aworkable theory of decision developmentwould have to incorporate not only anexplanation of group progressions at thelevel of large-scale activity units but alsoan account of how the group's interac-tion accomplished this development. AsPoole, Seibold, and McPhee44 argue,group activities are best understood asstructurational processes whereby thegroup produces and reproduces bothitself and the grounds for members'actions through its own activities andinteraction. From this perspective theanalysis of macro-level patterns (the con-cern of this essay) and micro-level inter-action must be integrated in an adequatetheory.

Several scope conditions apply to thepresent discussion. First, it is limited todecision-making groups; although mostof the propositions could also apply toany task group, they were developedwith decision-making and problem-solv-ing groups in mind and are particularlyadapted to those contexts. Second, allpropositions refer to a single decision (asingle task) rather than to the wholelifespan of the group. Although theremay well be a set of lifespan stages forgroups, I am not concerned with thatissue here. The decision (task) in ques-tion may take a number of meetings andextend over a very long period of time,but I assume that the group has workedprior to and will continue to work afterthe decision. Finally, the propositionsassume the groups satisfy several criteria

44Marshall Scott Poole, David R. Seibold, and RobertD. McPhee, "A Structurational Theory of Group Deci-sion-Making," Conference on Small Group Communi-cation Research, Pennsylvania State University, 1982.For a structurational interpretation of decision develop-ment, see Marshall Scott Poole, "Tasks and InteractionSequences: Multiple Levels of Coherence in GroupDiscussion" in Sequence and Pattern in CommunicativeBehavior, ed. R. L. Street and J . N. Cappella (London:Edward Arnold, in press).

advanced by Bales and Strodtbeck andBormann45 for a "full-fledged" group:(a) The group must not be "zero histo-ry"; (b) The task must not be overlyroutine or an "open-and-shut case"; (c)There must be some pressure or incen-tive for the group to finish its task; (d)There must be some incentive for mem-bers to maintain solidarity and remain inthe group.46

Explanatory Factors

The two explanatory factors I willemphasize are fundamental conditions ofa group's existence—its task and its his-tory. (1) The primary determinant ofdevelopmental progressions is the natureof the group's task. Although this state-ment seems obvious, it is surprising howlittle research has been done regardingthe effects of task on group interaction.4

The evidence we do have is very sugges-tive. In a direct comparison of the effectof different tasks on interaction, Morrisfound that task type significantlyaffected the distribution of almost 60percent of the activities he measured.48

Several other studies report that leaderor executive behavior varies greatlydepending on task type.49 Task also plays

45Bales and Strodtbeck; Ernest Bormann, "The Para-dox and Promise of Small Group Research," Communi-cation Monographs, 37 (1970), 211-16.

46Several other conditions proposed by Bales andStrodtbeck are relaxed in this section.

47Arthur P. Bochner, "Task and InstrumentationVariables as Factors Jeopardizing the Validity of Pub-lished Group Communication Research, 1970-71,"Communication Monographs, 41 (1974), 169-78.

48Charles G. Morris, "Task Effects on Group Inter-action," Journal of Personality and Social Psychology, 4(1966), 545-54.

49B. M. Bass, M. W. Pryer, E. L. Gaier, and A. W.Flint, "Interacting Effects of Control, Motivation,Group Practice, and Problem Difficulty on AttemptedLeadership," Journal of Abnormal and Social Psycholo-gy, 56 (1958), 352-58; Marvin E. Shaw and J. MichaelBlum, "Effects of Leadership Style Upon Group Perfor-mance as a Function of Task Structure," Journal ofPersonality and Social Psychology, 3 (1966), 238-42; J .Richard Hackman and Charles Morris, "Group Tasks,Group Interaction Process, and Group PerformanceEffectiveness: A Review and Proposed Integration," inAdvances in Experimental Social Psychology, ed. Leon-

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 15: Decision development in small groups, III: A multiple sequence model of group decision development               1

334 COMMUNICATION MONOGRAPHS

an important role in attempts to synthe-size results across contexts. Bochner haspointed out critical problems in compar-ing group communication studies due tothe different tasks used by researchers.50

Hackman and Morris included task as akey variable in their review of input-process-output relationships in smallgroups.51 Poole was able to reconcile anumber of confusing results in researchon group interaction and synergisticeffects merely by including task as anexplanatory variable.52 Finally, as notedabove, the nature of the group sequenceof decision-making phases also variesacross tasks. Mintzberg et al. found thatthe degree of uncertainty associated witha decision was related to the complexityof an organizations decision path. Put-nam and Jones report different phasicsequences for bargaining and problem-solving tasks.54

Task will be defined in terms of twodimensions emphasized in previousresearch:55

(a) Difficulty may be defined as the amount ofeffort required to complete the task. Difficultyvaries directly with the following subcompon-ents: (i) The number of operations and skillsrequired to carry out a task; the morerequired, the more difficult the task; (ii)Cause-effect knowledge about problems and/or solutions; the less clear the knowledge, themore difficult the task; (iii) Goal clarity; theless clear the goal the more difficult thetask.56

ard Berkowitz (New York: Academic Press, 1975), pp.45-99.

50Bochner.51Hackman and Morris.52Marshall Scott Poole, "Group Performance, Mem-

ber Competence, and Synergistic Effects," Unpublishedms., Department of Speech Communication, Universityof Illinois, 1977.

53Mintzberg et al.54Putnam and Jones.55Marvin E. Shaw, "Scaling Group Tasks: A Method

for Dimensional Analysis," JSAS Catalog of SelectedDocuments in Psychology, 3 (1973) MS No. 294; Gor-don O'Brien, "The Measurement of Cooperation,"Organizational Behavior and Human Performance, 3(1968), 427-39.

56Marvin E. Shaw, Croup Dynamics, 3rd ed. (NewYork: McGraw-Hill, 1980); Andrew H. Van de Ven

(b) Coordination Requirements may be definedas the degree to which integrated action ofgroup members is required to complete thetask. Several schemes for measuring or rank-ing coordination requirements have beendeveloped.57

While there are other dimensions anddescriptive systems for tasks, these twohave the strongest support and are themost consistently used in the literature.

(2) History is the second major deter-minant of decision development: Anyperiod or episode of work on a particulartask is suspended in the stream of groupactivity that comes before and will con-tinue afterward. Previous occurrences,as well as members' expectations for thefuture, shape and condition the currentprogression. Perhaps the most tellingevidence of the importance of history ingroup life is the way researchers turn toit as a "cause of last resort." When therelatively small amounts of variance nor-mally explained in group research havebeen discounted, the remainder is con-signed to error and to individual factorsaffecting the groups. These "individualfactors" are precisely the particular his-torical circumstances experienced by thegroup. Bales and Strodtbeck among oth-ers, are careful to list historical idiosyn-cracies which might cause departurefrom their basic sequence.58 However,one reason history is so seldom explicitlyconsidered in group research is its vague-ness; the term covers a multitude ofdifferent events and admits of manyinterpretations. To trace its effect ondecision development we must identifymediating structures that transmit theeffects of history into the group's currentoperations. These mediating structures

and André L. Delbecq, "A Task Contingent Model ofWork Unit Structure," Administrative Science Quarter-ly, 19 (1974), 183-97; Fred E. Feidler, "The Contin-gency Model and the Dynamics of the LeadershipProcess," in Advances in Experimental Social Psycholo-gy, vol. 11, ed. Leonard Berkowtiz (New York: Aca-demic Press, 1978), 59-112.

57O'Brien; Shaw, Group Dynamics.58Bales and Strodtbeck.

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 16: Decision development in small groups, III: A multiple sequence model of group decision development               1

DECISION DEVELOPMENT IN SMALL GROUPS, III 335

should be directly relevant to the work-ing group and should generalize acrossgroups to at least some extent.

Three such structures will be consid-ered here, generically called "workingrelationships:"

( a) Members' level of involvement in the group.Following Gustafson et al. we will distin-guish three types of involvement:59 (i) Pseu-domutuality; where members have a lowlevel of involvement and dependence on eachother and there is little trust or commitment.The group's solidarity is maintained bynever pushing or testing the relatively weakbonds of trust or reliability.

(ii) Dominance, where one or more mem-bers dominate others, and the others dependon them for leadership, evaluation, and mor-ale. The group's solidarity is maintained byrelations of precedence and hierarchy amongmembers and the group's direction is deter-mined by those in control.

(iii) Mutuality, where members are inmore or less equal relations and high statusdoes not imply dominance. The direction ofthe group is determined by shifting groups ofmembers and control of the group is rela-tively open to all. Members are highly inter-dependent and there is a high level of trustamong members, because all feel some degreeof "ownership" of the group. Group activi-ties are directed toward achieving the goalsand growth of all members, rather than thoseof a dominant clique.

These types are clearly related to theindividualistic, competitive, and cooperativeclimates discussed by Deutsch,60 Rubin andBrown,61 and others. They incorporate thehistory of the group as it affects the coactingand cooperative behavior of group members.They also reflect the outcome of power strug-gles, the terms in which members have cometo think of one another, and the assumptionsabout group work which members haveaccepted.

(b) Members' degree of consensus about leader-ship and status relationships in the group.

59Gustafson and Cooper; James P. Gustafson, LowellCooper, Nancy C. Lathrop, Karin Ringler, Fredric A.Seldin, and Marcia K. Wright, "Cooperative andClashing Interests in Small Groups," Human Relations,34 (1981), 339-78.

60Morton Deutsch, The Resolution of Conflict (NewHaven: Yale University Press, 1973), chs. 2 and 13.

61Jeffrey Z. Rubin and Bert R. Brown, The SocialPsychology of Bargaining and Negotiation (New York:Academic Press, 1975), ch. 8.

This reflects the degree to which the grouphas previously worked out executive relation-ships. It also reflects previous levels of conflictin the group and the strategies used to managethose conflicts. This factor indexes the stabil-ity of the involvement type that holds in thegroup at a given time.

(c) Procedural norms are sets of roles the groupuses to guide its work, such as rules onagendas and the reflective thinking format.Bales and Strodtbeck recognized the power of

. such procedures to retrack or create develop-mental sequences. These norms result fromprevious successful (or unsuccessful) experi-ences of the group. As Putnam has argued,they may also reflect composition effects (or alearned preference for norms).62

I have chosen to emphasize factorsdirectly related to the group's ability tocoordinate its members. This choice isbased on the assumption that history hasits most immediate effects through thepreconceptions members have about howthe group should work, and about howand to what extent they should partici-pate in this work, as well as through theenmities or friendships it creates amonggroup members. There may also be othermedia for channeling the effects of his-tory into the present. 3

Propositions

These two factors can affect the courseof decision development in at least fourways: (1) They can influence the types ofactivities groups engage in and the orderin which these occur; (2) They caninfluence the number of cycles of anactivity (or sequence of activities) thegroup goes through (e.g., comprehensioncycles); (3) They can influence the typeand frequency of breakpoints that occur;and (4) They can influence the degree towhich development in the three strands

62Linda L. Putnam, "Preference for ProceduralOrder in Task-Oriented Small Groups," Communica-tion Monographs, 44 (1979), 193-218.

63For example, the group's level of aspirations asdiscussed by A. Zander, "Group Aspirations," in GroupDynamics: Research and Theory, 3rd ed.: ed. D.Cartwright and A. Zander (New York: Harper andRow, 1968) pp. 418-29.

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 17: Decision development in small groups, III: A multiple sequence model of group decision development               1

336 COMMUNICATION MONOGRAPHS

is coordinated, whether the strandsdevelop unevenly or in concert. Follow-ing is a list of propositions about decisiondevelopment based on the multiplesequence model.

Effects of Structural Requirements. Inthe previous section I stressed the role ofthe accumulation of components for taskaccomplishment in decision develop-ment. The particular ensemble of com-ponents the group requires to fill out itsstructure will exert a profound influenceon the occurrence and sequence of taskprocess activities. Furthermore, for anyparticular ensemble it should be possibleto arrange the missing components inorder of logical priority (for example,problem definition logically comes beforesolution adaptation). This orderingshould influence the developmentalsequence of task process activities.64

Proposition 1. The group will engage in thosetask process components that cor-respond to the missing compo-nents of the task completionstructure.

Proposition la. The order of the task processcomponents will correspond tothe logical priority ordering ofthe missing task components.

At least two distinct definitions of thedecision task may occur, each of whichimplies different ensembles of task com-ponents. In the first instance, a groupmay look on a decision as a "full-fledged" problem which requires allcomponents—problem diagnosis, solu-tion search, testing, and a decision. Thisinterpretation is assumed in most devel-opmental theories, which take problemanalysis as a necessary condition of deci-sion-making. In the second instance, agroup may look on a decision simply as astimulus for action on grounds that arealready well-understood. This definitionimplies omission of problem analysis andemphasis on solution development. The

64Poole, "Decision Development, II."

second instance is clearly truncated com-pared to the first, but it does represent afairly large proportion of the groupsencountered in actual situations.65 Otherensembles of components may occur aswell, but these two seem the most likely.

A number of researchers have empha-sized the importance of procedures toeffective decision-making and problem-solving.66 However, procedural sugges-tions may conflict with the accumulationof components of the completion struc-ture. For example, emphasis on brain-storming solutions and voting proceduresmay distract the group from neededwork on problem definition and diag-nosis. Premature work on solutions willvery likely be undermined by lack ofagreement about the problem, requiringthe group to retrace its steps and renego-tiate solutions.67

Proposition 2. Procedural norms about theorder of decision activities willinfluence the occurrence andorder of task process and rela-tional activities.

Proposition 2a. If procedural norms do not fulfillmissing task components, thegroup will recycle its task processactivities to fulfill the componentand then retrace its steps throughthe former sequence of task pro-cess activities. It will recycle tofulfill the most basic missingcomponent in terms of logicalpriority. There will also be anincrease in procedural task pro-cess activities over cases wherenorms match requirements.

65Norman R. F. Maier, Problem-solving Discussionsand Conferences: Leadership Methods and Skills (NewYork: McGraw-Hill, 1963), chs. 3 and 4.

66Carl E. Larson, "Forms of Analysis and SmallGroup Problem Solving." Communication Mono-graphs, 36 (1969), 452-55; J . Hall and W. H. Watson,"The Effects of a Normative Intervention of GroupDecision-Making Performance," Human Relations, 23(1970), 299-317; Charlan Nemeth, "InteractionsBetween Jurors as a Function of Majority versus Un-animity Decision Rules," Journal of Applied SocialPsychology, 7 (1977), 38-56.

67Mintzberg et al.

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 18: Decision development in small groups, III: A multiple sequence model of group decision development               1

DECISION DEVELOPMENT IN SMALL GROUPS, III 337

Effects of Task Properties. Severalpredictions can be ventured about theeffects of the group's task on develop-mental processes. First, tasks high indifficulty will result in more and morevaried kinds of all three types of activi-ties, more recycling, and more break-points than low difficulty tasks. Morespecifically, if the group's goal is unclearthere will be more problem-related taskprocess activity and more recyclingthrough such activity than if the goal isclear. If the cause/effect knowledgeabout the solution is unclear, there willbe more solution-related task processactivity and more recycling through suchactivity than if the cause/effect knowl-edge is definite. If the coordinationrequirements are high, there will bemore conflict, bargaining, and integra-tion relational activities than if they arelow. Other, more specific predictions canbe made for particular tasks.

Proposition 3. High difficulty tasks will resultin more and a greater variety ofall three types of activity thanlow difficulty tasks.

Proposition 3a. Unclear goals will result in moreproblem definition and diagnosisand more recycling through thesethan will clear goals. They willalso result in more orientationand process comments.

Proposition 3b. Unclear cause/effect knowledgeabout solutions will result inmore solution-related task pro-cess behaviors than will definitecause/effect knowledge in thisarea.

Proposition 3c. If coordination requirements arehigh, there will be more conflict,bargaining, and integration ac-tivity than if coordination re-quirements are low.

Effects of Involvement. The degree ofmember involvement should stronglyinfluence decision development. Mem-bers' expectations about how they will beincluded effect their contributions and,as a result, the types of activities and

quality of work in the group.68 Motiva-tional climates, closely akin to theinvolvement types, have demonstrableeffects on interaction.69 These effectsshould manifest themselves primarily inthe relational activity dimension becausethis strand is most sensitive to members'working relationships. Predictions flowdirectly from the descriptions of theinvolvement types above.

Proposition 4. Member involvement will in-fluence relational activity.

Proposition 4a. Pseudo-mutual involvement willresult in fewer occurrences ofconflict, critical work, and inte-gration, as well as less recyclingthrough these activity periodsthan will the other two involve-ment types.

Proposition 4b. Dominant involvement will re-sult in less integration thanmutuality and in less bargainingthan either mutuality or pseudo-mutuality. It will also create lessrecycling through relational ac-tivities than mutuality.

Proposition 4c. Mutual involvement will resultin more bargaining, criticalwork, and integration than willthe other two involvement types.

Effects of Consensus. Level of consen-sus has been shown by Heinecke andBales to effect the levels of agreementand conflict in group interaction, as wellas member satisfaction and group effec-tiveness.70 As would be expected, highconsensus groups enacted more agree-ment and less conflict and had greatersatisfaction and effectiveness ratingsthan low consensus groups. Consensusalso seems to effect "executive" behav-iors, as Stogdill's summary of leadershipliterature shows.71 In general, leadership

68Gustafson and Cooper: Shaw, Group Dynamics, pp.276-77; Ralph Stogdill, Handbook of Leadership (NewYork: Free Press, 1975), chs. 28 and 31.

69Deutsch; Rubin and Brown.70Christoph Heinecke and Robert F. Bales, "Devel-

opmental Trends in the Structure of Small Groups,"Sociometry, 16 (1953), 7-38

71Stogdill.

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 19: Decision development in small groups, III: A multiple sequence model of group decision development               1

338 COMMUNICATION MONOGRAPHS

behavior seems to be more definite andeffective under conditions of high con-sensus. With low consensus a largernumber of procedural activities function-ing as leadership bids should occur andthe group may be less organized andeffective (Hirokawa reported much moreprocedural activity in ineffective thaneffective groups). The greater amountof conflict in groups with low consensusalso implies more topic shifts than ingroups with high consensus, as membersattempt to control the discussion. Final-ly, consensus serves as a mediating vari-able governing the stability of dominantgroups with dominant involvement pat-terns. With low consensus such groupsshould be much less stable and thereforeexhibit more open conflict.

Proposition 5a. Low levels of consensus willresult in relatively fewer occur-rences of focused and criticalwork and relatively more con-flict, bargaining, and ambiguitythan will high levels of consen-sus. There will be more recyclingthrough periods of conflict, bar-gaining, and ambiguity with lowconsensus than with high consen-sus.

Proposition 5b. Low levels of consensus willresult in more activity related toestablishing criteria and moreorientation and process commentactivity than will high levels ofconsensus. Low consensus willalso result in more cycling backthrough these activities than highconsensus.

Proposition 5c. Low levels of consensus willresult in more topic shifts thanhigh levels of consensus.

Proposition 5d. In groups with dominant in-volvement, low levels of consen-sus will result in more conflictthan high levels of consensus.

Effects of Procedural Norms. Proce-dural norms serve as a "guidance mecha-

72Fishcr, Small Group Decision-Making, pp. 207-20;see also Hirokawa.

nism" for the group's work as we notedin the previous section. Previous unitarysequence phasic models assumed allgroups followed the same proceduralnorms. Although this is clearly notalways the case, unitary sequence normswill predict activity orders under certaincircumstances. These include high con-sensus, a low to moderately difficult task,and a preference for procedural orderamong group members. As Bales andStrodtbeck note, departures from theseconditions will diversify the sequencebeyond the unitary norm.73 Previousresearch on unitary norms allows us tospecify a priority ordering of both taskprocess and relational activities:

Proposition 6a. Given high consensus, low tomoderate task difficulty, mediumto high coordination require-ments, and group preference forprocedural order, the groupshould follow the following se-quence of task process activities:orientation::problem analysisand definition of solution guide-lines::design and evaluation ofsolutions::selection of solu-tions::integration/reinforcement.

Proposition 6b. Given high consensus, low tomoderate task difficulty, mediumto high coordination require-ments, and group preference forprocedural order, the followingsequence of relational activitiesshould emerge for each phase oftask process activities: ambigui-ty::conflict-bargaining::work::in-tegration. This four-step cycleshould hold for each period out-lined above, e.g., for the problem-analysis and definition of solu-tion criteria period.

These propositions assume that our"logical" priorities of the task processand relational activities are most likely tobe used by groups when they successfullycomplete a task. Some groups maydevelop considerably more complicated

73Bales and Strodtbeck.

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 20: Decision development in small groups, III: A multiple sequence model of group decision development               1

DECISION DEVELOPMENT IN SMALL GROUPS, III 339

procedures, and this simple orderingwould not be expected to hold for thiscase.

Breakpoints. The occurrence and typeof breakpoints will also be effected by thetwo factors. We have already mentionedseveral instances where recycling isexpected, which introduces delay break-points. Generally, the more difficult thetask the more breakpoints there will bebecause members will have to retracetheir steps as they discover errors or newideas and as they recognize failures intheir work. Further, low consensus willalso generate breakpoints when mem-bers get into conflicts and try to retracesteps to establish consensus.

Proposition 7a. There will be more delays andfailure breakpoints with a highlydifficult task than with low diffi-culty.

Proposition 7b. There will be more delays andconflict breakpoints with a lowconsensus group than with a highconsensus group.

Synchrony Among Strands. Finally,the two factors will influence the degreeto which there is even, coordinated devel-opment among the three tracks, i.e., theextent to which activities on differenttracks co-occur and have simultaneousbreakpoints. Even development amongthe tracks implies a stable internal envi-ronment in the group—high consensuson leadership and operating proce-dures—and a high degree of memberinvolvement—either dominant or mu-tual involvement types. These are neces-sary prerequisities for the coordinationof activities and for smooth, nonprob-lematic transitions from facet to facet ofgroup activity. Moreover, it also seemslikely that a coordinated developmentrequires a task of low to intermediatedifficulty; a highly difficult task wouldthrow the group out of kilter and preventsmooth progressions.

development among the threeactivity types for groups with (i)high consensus, (ii) mutual ordominant involvement types, and(iii) low to intermediate difficultytasks.

Additional Considerations

The propositions have been developedas though they are independent of oneanother, but they can be combined togenerate more specific predictions aboutdevelopment. For example, in the case ofgroups with high consensus and strongpreference for procedures, but a taskwith unclear goals, propositions 6a and3a can be combined to predict extracycles of problem definition and diag-nosis in the basic task process sequence,as well as possible cycling back to prob-lem definition from the solution search-design-evaluation step. There are un-doubtedly some gaps in this explanatoryframework, but it should allow precise,testable predictions for quite a diverserange of situations. What it will notallow is exact prediction of sequences.The predictions offered here are onlygeneric; the exigencies of particulargroups will always introduce idiosyncra-cies which can be understood andexplained but are not predictable.

An important caveat is necessary: Thepropositions assume that the explanatoryfactors are stable input conditions. How-ever, working relations and even tasksonly exist and effect the group throughan ongoing process of structuration thatproduces and reproduces them as struc-tural features of decision-makinggroups.74 This process not only accountsfor their effect on interaction; it may alsochange these factors over time. Homans,among others, has shown the dependenceof status structure on member behavior:members produce and reproduce thegroup's hierarchy by consulting, obey-

Proposition 8. There will be even, coordinated 74Poole, Seibold, and McPhee.

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 21: Decision development in small groups, III: A multiple sequence model of group decision development               1

340 COMMUNICATION MONOGRAPHS

ing, and rewarding informal leaders.75 Inthe same vein, members can change thenature of the system by altering theiractivity and interaction patterns: a mem-ber may quietly emerge as a leadershipcontender merely by showing her compe-tence; accepted orders may becomeunbearable and promote rejection andrevolt. Poole, McPhee, and Seibold havesupported a similar view of proceduralnorms: norms only affect groups insofaras they are used in interaction, and thisleaves them open to alteration or rejec-tion.76 Tasks, too, are constituted bygroup interpretations of work require-ment and hence are liable to change. Ineach case we must recognize the possibil-ity that our explanatory factors maychange during the course of the discus-sion, thereby making a more dynamic,dialectical explanation necessary. Al-though the two sets of factors mayremain stable in some cases, we mayhave to explain the coevolution andmutual effects of decision behavior andthese factors.

Finally, we have made no reference toeffectiveness or any other output mea-sure during this discussion. This over-sight has been purposeful because intro-ducing considerations of outcomes wouldhave greatly complicated our task. Not-withstanding, several authors haveshown effectiveness to be related todevelopmental sequence.77 Eventuallythis framework must be extended tocover the entire input-process-outputsequence.

CONCLUSION

The model of decision developmentoffered here is built upon a descriptivesystem more comprehensive and flexiblethan the traditional phasic model. It

75George C. Homans, The Human Croup, (NewYork: Harcourt, Brace, 1950).

76Poole, McPhee, Seibold.77Hirokawa; Hackman and Morris.

consists of three interlocking tracks ofgroup activity, interrupted at irregularintervals by breakpoints, and serving toaccumulate a structure of componentsfor task accomplishment. This descrip-tion posits task (in this case, decision) asthe generative mechanism for groupactivity; the three activity types, thebreakpoints, and the components allevolve from an analysis of task-directedactivity. The model also proposes twosets of explanatory factors representingfundamental parameters of group activi-ty: the group's task and its historicalcontext. A number of propositions weredeveloped and advanced as an initialframework for a contingency theory ofdecision development.

If it can be realized, the power of thiscontingency theory will be in its capacityto predict and explain the general config-urations of development for a diverse setof groups and conditions. It should pro-vide greater resolution and better fit thanphasic models. However, it is unlikelythat this or any developmental theorywill be able to predict specific sequencesof activities. For one thing, even if thesame forces are acting on two groups andthe groups are subject to the same devel-opmental "laws," the activity tracks willdevelop unevenly, resulting in differentdevelopmental sequences. Second, criti-cal events unique to each individualgroup may cause it to take a differentpath from other similar groups. Suchevents lie in the realm of idiographicexplanations; they are beyond the pur-view of general theory.

The greater flexibility of the multiplesequence model also promises practicalbenefits. The ability to track unevendevelopment may facilitate the evolutionof intervention strategies consistent withthe complexities of group process. Toooften, a task process strategy such asNominal Group Technique is used with-out regard for its positive or negativerelational effects on the group. The mul-

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014

Page 22: Decision development in small groups, III: A multiple sequence model of group decision development               1

DECISION DEVELOPMENT IN SMALL GROUPS, HI 341

tiple sequence model may make it easierto assess empirically the effects of suchinterventions on group process and adaptthese techniques in an appropriate man-ner.

Like the phasic model, the multiplesequence formulation will one day besuperseded. At present, it can serve avaluable function by jolting develop-

mental research out of its well-worntracks. For thirty years, researchers ongroup development have been conduct-ing the same study with minor altera-tions. Freud observed that the "narcis-sism of small differences" has a valuableplace in everyday social relations. It is,however, only a detriment to scientificprogress.

Dow

nloa

ded

by [

Cen

tral

Mic

higa

n U

nive

rsity

] at

05:

37 0

5 D

ecem

ber

2014