Punctuated Equilibrium and Linear Progression Toward a New Understanding of Group Development

13
® Academy of Management Journal 2003. Vol. 46, No. 1. 106-117. PUNCTUATED EQUILIBRIUM AND LINEAR PROGRESSION: TOWARD A NEW UNDERSTANDING OF GROUP DEVELOPMENT ARTEMIS CHANG Queensland University of Technology PRASHANT BORDIA JULIE DUCK University of Queensland This study proposes gaining a new understanding of group development by consider- ing the integrative and the punctuated equilibrium models of group development as complementary rather than competing. We hypothesized that we would observe both punctuated equilibrium and linear progression in content-analyzed data from 25 simulated project teams, albeit on different dimensions. We predicted changes in time awareness and in task and pacing activity in line with the punctuated equilibrium model and changes in structure and process on task and socioemotional dimensions in line with the integrative model. Results partially supported predictions for both models. Group researchers were forced to reconsider their understanding of group processes when Gersick (1988, 1989, 1991) pubUshed the punc- tuated equilibrium model. Gersick (1988, 1989) studied eight field and eight laboratory work groups with definite deadlines, referred to as project teams, and found that instead of develop- ing gradually over time, project teams progressed through an alternation of stasis and sudden change. Drawing on the language of biological evolution, she labeled such a course of develop- ment "punctuated equilibrium." Reviewers (e.g., Bettenhausen, 1991; Guzzo & Shea, 1992) have concluded that this new understanding of change processes challenged the traditional linear mod- els of group development, in which (1) change was conceptualized as a gradual and incremental process, (2) it was assumed that groups progress through a logical sequence of stages over time, and (3) groups were viewed as becoming more effective over time, at least until they move into a final stage of decline and termination (Gersick, We would like to thank Andrew Wollin and Allie Perich for their help at various stages of this research. This research, conducted as part of the first author's doctoral studies at the University of Queensland, was supported by her Australian Postgraduate Award and by a Faculty of Social and Behavioural Sciences Early Ca- reer Research Grant to the second author. An earlier version of this research note received the 2002 Best Dis- sertation Award from the American Psychological Asso- ciation, Division 49. 1991; La Goursiere, 1980; Van de Ven, 1995; Wollin, 1999). We argue that one should not conclude that Gersick's findings discredit stage models of group development without systematically exam- ining the relationship between the punctuated equilibrium model and linear (stage) models of group development. Only three studies (Arrow, 1997; Lim & Murnigham, 1994; Seers & Woodruff, 1997) have directly compared linear and nonse- quential models of group development. However, two of the three studies appear to contain mis- interpretations of some fundamental assump- tions of linear group development models; both Arrow (1997) and Seers and Woodruff (1997) as- sumed that linear models describe a group's pro- gression through "clearly defined" developmen- tal stages. In fact, most linear theorists have defined developmental stages in terms of the pro- portion of time a group spends on issues that are characteristic of a particular stage and do not propose that clearly defined boundaries separate one developmental stage from the other (for ex- ample. Bales, 1953; Wheelan, 1994). This research note presents an empirical study designed to reconcile the punctuated equilibrium model (Gersick, 1988, 1989) and the integrative model (Wheelan, 1994) of group development. The integrative model was chosen to represent stage models because it is a recent integration of previous group development research, including Tuckman's (1965) classic model. We argue that the punctuated equilibrium model and the inte- grative model complement rather than contradict 106

Transcript of Punctuated Equilibrium and Linear Progression Toward a New Understanding of Group Development

Page 1: Punctuated Equilibrium and Linear Progression Toward a New Understanding of Group Development

® Academy of Management Journal2003. Vol. 46, No. 1. 106-117.

PUNCTUATED EQUILIBRIUM AND LINEAR PROGRESSION:TOWARD A NEW UNDERSTANDING OF GROUP DEVELOPMENT

ARTEMIS CHANGQueensland University of Technology

PRASHANT BORDIAJULIE DUCK

University of Queensland

This study proposes gaining a new understanding of group development by consider-ing the integrative and the punctuated equilibrium models of group development ascomplementary rather than competing. We hypothesized that we would observe bothpunctuated equilibrium and linear progression in content-analyzed data from 25simulated project teams, albeit on different dimensions. We predicted changes in timeawareness and in task and pacing activity in line with the punctuated equilibriummodel and changes in structure and process on task and socioemotional dimensions inline with the integrative model. Results partially supported predictions for bothmodels.

Group researchers were forced to reconsidertheir understanding of group processes whenGersick (1988, 1989, 1991) pubUshed the punc-tuated equilibrium model. Gersick (1988, 1989)studied eight field and eight laboratory workgroups with definite deadlines, referred to asproject teams, and found that instead of develop-ing gradually over time, project teams progressedthrough an alternation of stasis and suddenchange. Drawing on the language of biologicalevolution, she labeled such a course of develop-ment "punctuated equilibrium." Reviewers (e.g.,Bettenhausen, 1991; Guzzo & Shea, 1992) haveconcluded that this new understanding of changeprocesses challenged the traditional linear mod-els of group development, in which (1) changewas conceptualized as a gradual and incrementalprocess, (2) it was assumed that groups progressthrough a logical sequence of stages over time,and (3) groups were viewed as becoming moreeffective over time, at least until they move into afinal stage of decline and termination (Gersick,

We would like to thank Andrew Wollin and AlliePerich for their help at various stages of this research.

This research, conducted as part of the first author'sdoctoral studies at the University of Queensland, wassupported by her Australian Postgraduate Award and bya Faculty of Social and Behavioural Sciences Early Ca-reer Research Grant to the second author. An earlierversion of this research note received the 2002 Best Dis-sertation Award from the American Psychological Asso-ciation, Division 49.

1991; La Goursiere, 1980; Van de Ven, 1995;Wollin, 1999).

We argue that one should not conclude thatGersick's findings discredit stage models ofgroup development without systematically exam-ining the relationship between the punctuatedequilibrium model and linear (stage) models ofgroup development. Only three studies (Arrow,1997; Lim & Murnigham, 1994; Seers & Woodruff,1997) have directly compared linear and nonse-quential models of group development. However,two of the three studies appear to contain mis-interpretations of some fundamental assump-tions of linear group development models; bothArrow (1997) and Seers and Woodruff (1997) as-sumed that linear models describe a group's pro-gression through "clearly defined" developmen-tal stages. In fact, most linear theorists havedefined developmental stages in terms of the pro-portion of time a group spends on issues that arecharacteristic of a particular stage and do notpropose that clearly defined boundaries separateone developmental stage from the other (for ex-ample. Bales, 1953; Wheelan, 1994).

This research note presents an empirical studydesigned to reconcile the punctuated equilibriummodel (Gersick, 1988, 1989) and the integrativemodel (Wheelan, 1994) of group development.The integrative model was chosen to representstage models because it is a recent integration ofprevious group development research, includingTuckman's (1965) classic model. We argue thatthe punctuated equilibrium model and the inte-grative model complement rather than contradict

106

Page 2: Punctuated Equilibrium and Linear Progression Toward a New Understanding of Group Development

2003 Chang, Bordia, and Duck 107

each other. The following discussion will clarifymisunderstanding in the literature and integratethe two models to offer a comprehensive frame-work for understanding group development andgroup performance. The contribution of thisstudy is in its attempt to clarify the literature andto convey a new understanding of group devel-opment that integrates the punctuated equilib-rium model and the integrative model both the-oretically and empirically. This contribution isvaluable because empirical researcb comparingdifferent theories of group development is largelylacking, given tbe extensive time and resourcesrequired to test just one group developmentmodel empirically (Weingart, 1997).

THEORETICAL BACKGROUND

The Punctuated Equilibrium Model

Presenting her punctuated equilibrium model,Gersick (1988, 1989) argued that, instead of devel-oping gradually over time, work groups experiencelong periods of inertia tbat are punctuated by con-centrated revolutionary periods of quantumchange. According to the punctuated equilibriummodel, groups undergo a two-phase (ratber tban atwo-stage) developmental pattern. Phase 1 is tbefirst period of inertia, tbe direction of whicb is setby tbe end of a group's first meeting. Pbase 1 lastsfor balf of tbe group's allotted time. At the mid-point of the allotted time, the group undergoes atransition tbat sets a revised direction for pbase 2, asecond period of inertia. In addition, Gersick (1989)noted that a group's progress is triggered more bymembers' awareness of time and deadlines than bycompletion of an absolute amount of work in aspecific developmental stage. Moreover, "halfway"emerges as the most likely moment at w^hicb groupswill call attention to time or pacing. Tbe midpointacts as a reminder of tbe approacbing deadline thatinterrupts tbe group's basic pbase 1 strategies andfacilitates tbe midpoint transition and thus tbe on-set of pbase 2.

The Integrative Model

In tbe integrative model, wbicb is based on anintegration of group development researcb over tbelast four decades (e.g.. Bales, 1953; Bion, 1961;Tuckman, 1965), groups are viewed as progressingtbrougb five developmental stages, eacb describedby a unique pattern of bebaviors.

Stage 1 is "dependency and inclusion." Accord-ing to tbe model, members often feel anxious anduncertain wben first entering a group, because tbe

situation is new and not clearly defined. Thus, tbeyare polite and tentative, leader-focused, and some-times defensive.

Stage 2 is "counterdependency and figbt." As tbegroup develops, members start to find tbe leader-focused stage frustrating and confining. Individualmembers seek to clarify tbeir roles, and tbe groupseeks to assert independence from tbe leader (coun-terdependency). Goalitions start to form amongmembers witb similar ideas and values. Figbts andconflicts between coalitions and members with dif-ferent values start to emerge.

Stage 3 is "trust and structure." Resolutions oftbe conflicts of stage 2 clarify goals, increase cohe-sion and member satisfaction, and reduce individ-ual fears of rejection; thus, trust among membersincreases. At stage 3, communication becomesmore open and task-oriented, and more mature ne-gotiations about goals, roles, organization and pro-cedures start to take place. Members wbo baveaccepted a role at stage 1 out of misunderstandingtbe group's goal or mere fear of rejection can nowrenegotiate tbeir roles witb tbe group.

Stage 4 is "work." Work commences immediatelyafter group formation but reaches an optimal stateat tbis stage. Once goals, structures, and norms areestablished, a group can work more effectively.Members sbare the group goals and conform to tbegroup norm of bigb productivity. Furtbermore, asWbeelan (1994) argued, people usually bave someawareness of time wben tbey work. Groups tbat arealways working are probably not working effec-tively, and those tbat start late are definitely notworking effectively.

Stage 5 is "termination." Most work groups bavean ending point; even continuous groups bave tem-porary endings sucb as completion of an assignedproject. At each ending point, the members of afunctional group tend to evaluate tbeir work to-getber, give feedback, and express feelings abouteacb otber and tbe group.

Similarities and Differences between theTwo Models

The two models have tbe following similari-ties. First, in tbe integrative model, stage 1 groupsare cbaracterized by leader dependency and con-cerns about safety and inclusion and, tbus, agroup's members tend to follow tbe dominantmood of the group at tbis stage. Gonsequently,internal and external pressures to perform usu-ally force groups to focus on work very early intbeir development, despite tbeir unreadiness todo so. Tbus, a group's members tend to embracewbatever work processes are proposed at tbis

Page 3: Punctuated Equilibrium and Linear Progression Toward a New Understanding of Group Development

108 Academy of Management Journal February

early stage. Furtbermore, tbese work processestend to remain witb tbe group until members areready to question or challenge prevailing viewswitbin tbe group. Tbis argument accords witbGersick's (1988) assertion tbat tbe approach agroup undertakes at its first meeting will carrythrough the first half of its life span, wbicb ispbase 1 in tbe punctuated equilibrium model.

Second, it is possible tbat tbe integrative modeldescribes cbanges at a more micro level witbineacb pbase of inertia described by Gersick. In theintegrative model, tbere is only a loose boundarybetween stages 1 and 2 and stages 3 and 4. Stages 1and 2 tend to co-occur to form a premature pbase(pbase 1 of tbe punctuated equilibrium model) inwbicb group members struggle witb issues such aspower, structure, and intimacy; and stages 3 and 4tend to co-occur to form a mature pbase (pbase 2 oftbe punctuated equilibrium model) in wbicb issuessucb as power, structure, and intimacy are mostlyresolved, and work is tbe main focus of a group.Tbus, it is possible that a transition marks tbe sbiftof a group's bebavioral pattern from a pbase inwbicb stage 1 and 2 bebaviors dominate to a pbasein wbicb stage 3 and 4 bebaviors dominate. Tbisconceptualization of multilevel change patternsis similar to Wollin's (1999) recent approacb tounderstanding revolutionary and incrementalcbanges in social science. Wollin argued tbat sys-tems bave multilevel deep structures and tbat it isat tbe level of deep structure tbat cbanges deter-mine tbe observed incremental or revolutionarypattern. In other words; changes at more surfacelevels tend to be incremental; revolutionarycbanges occur at a more fundamental level (Wollin,1999).

Besides tbe similarities, tbere are also some dif-ferences between tbe punctuated equilibriummodel and tbe integrative model. First, in terms ofspecificity, tbe punctuated equilibrium model de-scribes cbanges in tbe way a group works on itstasks over time, wbereas the integrative model de-scribes the overall developmental pattern of agroup over time. Tbis difference in specificity isreflected in the difference in tbe two coding sys-tems. Gersick's observational system focused on"ideas and decisions tbat gave tbe product its basicsbape or tbat would be the fundamental cboices ina decision tree if tbe finisbed product were to bediagrammed . . . and points wbere milestone ideaswere first proposed, whether or not tbey were ac-cepted at tbat time" (1988: 14). Tbe integrativemodel, on tbe otber band, was developed tbrougbuse of tbe Group Development Observation System(GDOS; Wbeelan, Verdi, & McKeage, 1993), wbicbcaptures temporal cbanges in groups' structures

and processes along both socioemotional and task-related dimensions. For example, tbe coded tran-script on tbe rigbt in Table 1 demonstrates tbeintegrative model's focus on group processes sucbas work (coded witb "W"), fligbt ("FL"; avoidanceof intimacy or work), and pairing ("P"), wbicb re-fers to relationsbip building. In contrast, tbe leftside of Table 1 demonstrates bow tbe same groupinteraction would be coded under tbe punctuatedequilibrium model. Applying tbe latter, we ab-stracted tbemes from the group's discussion, cod-ing tbe same group interaction as tbe group exam-ining its resources. Because tbe punctuatedequilibrium model focuses a group's approacb toits work, statements tbat represent relationsbipmaintenance or avoidance of intimacy are lessrelevant.

Second, in terms of generalizability, tbe punctu-ated equilibrium model describes developmentalpatterns tbat apply only to "groups tbat bave someleeway to modify tbeir work processes and mustorient tbemselves to a time limit" (Gersick, 1988:36). Tbe integrative model was designed to de-scribe tbe developmental patterns of all types ofgroups—for example, IGU nurse teams (Wbeelan &Burcbill, 1999), executive teams (Buzaglo &Wbeelan, 1999), faculty member groups (Wbeelan& Tilin, 1999), and financial services and botelindustry teams (Wbeelan, Murpby, Tsumura, & Ky-liene, 1998).

HYPOTHESES

Gomparison of tbe two models described in tbepreceding section led to our first bypotbesis:

Hypothesis 1. Simulated project teams un-dergo both punctuated equilibrium and linearprogressive developmental patterns, albeit ondifferent dimensions. Over time, changes in agroup's time awareness, pacing activities, andtask activities will occur, consistent with thepunctuated equilibrium model, and changes intask and socioemotional activities will occur,consistent with the integrative model.

Furtbermore, we argue tbat tbe presence of punc-tuated equilibrium does not preclude tbe presenceof linear progression (Wollin, 1999).

Hypothesis 2. The integrative model describeschanges at a more micro level within eachphase of inertia described by Gersick, with themidpoint transition marking a group's movinginto the more productive developmentalstages.

Page 4: Punctuated Equilibrium and Linear Progression Toward a New Understanding of Group Development

2003 Chang, Bordia, and Duck 109

TABLE 1Meeting Transcript Sample Coded by Two Methods

Gersick-Style Meeting Map" Group Development Observational System Coding**

0:00-3:00: examined resources, looking atthe tapes available.

0:30-1:30: proposed content. One persontalked about the movie "Crazy People."

Wanted to use the idea shown in themovie: "Millions of people get killedevery year but we have the fewestnumber." The group then evaluated theproposed content—"not original."

1:30-4:20: listened to all the music tapes,commented on the music, andproposed ideas that go with the music."This is captain someone," "This is likean Asian music," "But that doesn'tmake Asian sound exciting, it justmakes it sound relaxing." "We've alsogot to mention the country I guess."

B These are all the music tapes (W)A tapes (W)C I wonder what the music is? (W)B while you are sleeping (W), silk road of theme (W), pearl shells (W), which I

never hear of (FL), while you are sleeping (W), is that the movie? (FL)A hm (FL)C yes (W)B we should use that one (W)D . . . (U)A I am not going to play it (W), I am just going to . . . (W)D ok (W)E has any one seen crazy people (FL)D pardon (U)E crazy people (FL), the movie (FL)B no (answering B) (FL), that's good (referring to the tape recorder) (W),C ok so (W), so we got to basically advertise (W)E they did this thing on the movie (W) where there like was this advertising

guy (W), and you know like how they usually say those safety stuff (W)1:00 E and he goes millions people got killed every year but we have the fewest

number (W), like people get killed (W)D oh, really (P)E they just do all this crazy thing (FL) and people really like it (FL) cause he

just like do all these crazy commercials (W)B that's good though (W)E but all these crazy people helps him to make up these ideas (FL), but it's not

original (W), but its creative (W)C but they wouldn't know whether it's not original (W) unless they have seen

it(W)D yeah (W), but they have the tape (W)E but . . . (U)oh yeah (W)E elevating music (playing music tape 1) (W)B it's like airplane music when you land (W)A oh. Yeah (W)D or when you are taking off (W)A yeah (P), it's true (W)E what's that (the music) (W)

2:00 B so we would have to like do the other ones . . .

" The left-hand column is a five minute-sample segment of the meeting map describing changes in the central theme of a group'sdiscussion over time. This map was constructed following guidelines provided by Gersick (1989). The coding of the segments is in boldtype. The numbers in the left column represent minutes elapsed since project inception.

^ The right-hand column shows a GDOS transcript for only two minutes of the group's interaction. The letters at the start of each lineidentify group members. The coding of each statement is in parentheses, in bold type. The coding reflects the purpose of the communi-cations and the nonverbal behaviors accompanying them.

In tbis study, we aimed to test tbe bypotbeses byreplicating Gersick's (1989) laboratory study onsimulated work groups. Laboratory groups werestudied because tbeir development could be ob-served from project inception to termination. Moreimportantly, a laboratory study offered a controlledenvironment in wbicb similarities and differencesbetween tbe two models could be tbe most effec-tively investigated. Furtbermore, Gersick demon-strated tbat laboratory groups display developmen-tal patterns tbat are similar to tbose of field groups

and tbus tbat data derived from laboratory groupsbave external validity.

Tbe integrative model (Wbeelan, 1994), in con-trast to tbe punctuated equilibrium model, was de-veloped in tbe field and bas not been tested in alaboratory setting. However, given tbat tbe model isgeneralizable to a wide range of groups, we ex-pected laboratory groups to display similar devel-opmental patterns, witb tbe exception tbat tbe lab-oratory setting migbt lead to less socioemotionalinteraction and/or conflict as participants would

Page 5: Punctuated Equilibrium and Linear Progression Toward a New Understanding of Group Development

110 Academy of Management Journal February

bave limited vested interest in tbe task and wouldonly be interacting for 40 minutes.

METHODS

Participants and Procedures

Twenty-five groups of first-year university psy-cbology students (8 groups of five and 17 groups offour) participated in tbe experiment for partial ful-fillment of psycbology course requirements. Tbesample consisted of 69 female and 38 male studentstbat formed 2 all-male, 7 all-female, and 16 mixedgroups. Groups' gender composition was not con-trolled in tbis experiment because previous re-searcb (Verdi & Wbeelan, 1992) bas suggested tbatit bas no influence on patterns of group develop-ment. However, observation of tbe groups' interac-tion sbowed tbat tbe all-male groups were lesscommitted to tbe task tban were tbe otber groups.

Tbe task in our study was modeled closely ontbat in Gersick's (1989) laboratory study. Partici-pants were told to assume tbe role of professionaladvertising writers at a major urban radio stationwbo were to design a pilot commercial for a well-known airline. Eacb team received a folder tbatinformed them about tbe client's special requestand tbe costs of producing an advertisement. Eacbgroup also received an audio tape player, a taperecorder, and tbree different music tapes. Tbe bud-get allowed eacb group to use only one of tbe musictapes. Participants were asked to note tbe time tbeirgroup began on tbeir watcbes and to bave tbe radiocommercial ready for collection wben tbe experi-menter (one of tbe autbors) returned in exactly 40minutes. Participants were told tbat several otberadvertising agencies were interested in tbis projectas well and tbat if tbeir team did not produce apilot commercial tbat satisfied tbe client, tbeiragency would lose tbis contract. Tbe competitionwas intended to motivate tbe students to finisb ontime and to attend to tbe requirements and evalu-ation criteria. At tbe end of tbe 40 minutes workingtime or upon completion of a group's first present-able product (wbicbever occurred first), tbe exper-imenter returned to tbe room and debriefed tbeparticipants. Meetings were videotaped for furtberanalysis. Eacb group's interaction was transcribedword-for-word into a written script for coding andanalysis; tbe average script contained 26.8 (s.d. =5.48) pages of double-spaced, 12-point type and1,113 (s.d. = 269.32) complete sentences.

The Coding Systems

Coding for the punctuated equilibrium model.Responses were coded as far as possible according

to tbe coding scbeme described by Gersick (1989),altbougb sbe did not fully describe a unit of anal-ysis for testing tbe punctuated equilibrium model.Her coding system contained two broad classes.Tbe first categorized tbe actions group memberstook to manage a work process. Tbere were tbreetypes of action statements: Process statements ("P")were members' suggestions about bow tbeir groupsbould proceed witb tbe work (for instance, "Wbydon't we just toss out some ideas tbat we could getinto tbe commercial?"). Time-pacing statements("T") were group members' direct references totime—noting wbat time it was or bow mucb timebad elapsed or was left—and members' attempts topace tbeir group by saying wben, in terms of tbeallotted time, sometbing sbould be done (for exam-ple, "toward tbe end" and "before too long") or bymentioning bow long an action would take, or fin-isbing on time (for instance, "We bave got 20 min-utes left!"). Resources-requirements ("R") state-ments were members' references to tbeir group'sresources, requirements, or criteria for tbe task andexplicit attempts to sbape tbe product in accor-dance witb tbe group's resources or requirements(for instance, "Tbat's $200 per thing, so we basi-cally bave tbe cboice of one.").

Tbe second class of categories included state-ments about tbe product: Content statements ("#c")were group members' mentions of selling points tobe pusbed in tbe ad, ideas for content themes orstory lines, tbe content of dialogue, or informationto be presented. Detail statements ("#d") were ideasabout small modifications or fine points of ad con-tent (for instance, "Should the brakes slam ornot?"). Format statements ("#f") were ideas for tbebasic format of tbe ad, the vehicle through whicbtbe information would be conveyed (for instance,"Wbat if we bad a conversation between two peo-ple?"). Procedure statements ("#p") were ideasabout tbe process of acting out the ad and aboutwbo would do wbat for tbe recording session (sucbas "I'll do tbe second person."). Following proce-dures similar to tbose reported by Gersick (1989),we constructed a qualitative map for eacb group(see Table 1). Tbe map described cbanges in tbecentral tbeme of a group's discussion over time.

Group Development Observation System(GDOS) coding for the integrative model. Workstatements ("W") were tbose that represented pur-poseful, goal-directed activity and task-oriented ef-forts (for example, "Why don't we start writing thisdown?"). All categories described by tbe punctu-ated equilibrium model were coded as work state-ments in tbe GDOS coding. Fight statements ("Fl")were tbose that implied argumentativeness, criti-cism, or aggression (such as "That is a stupid

Page 6: Punctuated Equilibrium and Linear Progression Toward a New Understanding of Group Development

2003 Chang, Bordia, and Duck 111

idea."). Flight statements ("FL") were those thatindicated avoidance of the task and confrontation(such as "Did anyone watch the movie on SBS lastnight?"). Pairing statements ("P") were those thatincluded expressions of warmth, friendship, sup-port, or intimacy with others (for example, "Goodwork, John!"). Counterpairingstatements ("CP") in-dicated avoidance of intimacy and connection anda desire to keep the discussion distant and intellec-tual (for example, "Can we talk about the com-mercial instead?"). Dependency statements ("D")showed an inclination to conform with the domi-nant mood of the group or to follow suggestionsmade by the leader and, generally, demonstrated adesire for direction from others (for example,"What do you think we should do?"). Counterde-pendency statements ("CD") asserted indepen-dence from and rejection of leadership, authority,or other members' attempts to lead (such as "Whydon't we try my idea first?"). It should be noted thatall statements needed to be coded on the basis oftheir purpose, the group's history, the nonverbalcues given by the speaker, and the reaction of therecipient. For example, "Can we talk about thecommercial instead?" was coded as a counterpair-ing statement when it was a reaction to a group'srelationship-building conversation. In contrast,"Did any one watch the movie on SBS last night?"was coded as flight when it was stated to avoidworking on the commercial, when the group foundthe task too difficult to proceed with.

The first author coded all the transcripts twice,once with Gersick's coding system, and once withthe GDOS. A research assistant coded a randomlyselected 10 percent of the scripts. Coefficients forinterrater reliability between the first author andthe research assistant were calculated both for unit-izing and for coding. For all of the integrativemodel codings and for the pacing and time aware-ness statements of the punctuated equilibriummodel, a unit was defined as a simple sentence thatpresents a complete thought to another person orpersons. A unit did not have to be a grammaticalsentence as long as a communication providedenough information that it could be interpreted byothers and could stimulate a reaction in them.Thus, "I agree" is a unit, and "Absolutely" is a unitif said in response to another statement (Wheelan etal., 1993: 44). Unitizing reliability was assessedusing Guetzkow's U, and the global and category-by-category reliability measures of the GDOS werecalculated using Cohen's kappa (Folger, Hewes, &Poole, 1984). The unitizing reliability was calcu-lated to be 0.99 for the integrative model. Theglobal level of interrater reliability was 0.74 for theintegrative model, and the category-by-category

reliability coefficients for work, flight, pairing,counterpairing, dependency, counterdependency,uncodable, and pacing and time statements were0.99, 1.00, 0.94, 0.97, 0.81, 0.75, 1.00, and 1.00respectively. Cohen's kappa was not calculated forthe fight category because no fight statements wereobserved in this study (see below). For the punctu-ated equilibrium model, interrater reliability wasonly calculated on the percentage of agreement onthe linear or honsequential pattern of developmentand for the time and pacing statements (as de-scribed above). We relied on these two measuresbecause of the lack of detail in Gersick's (1989)paper on how to analyze the data from other codingcategories. Another research assistant who had notbeen involved in the initial coding of the tran-scripts used GDOS to code 25 percent of the tran-scripts; the percentage of agreement between thisresearch assistant and the first author on the linear/nonsequential pattern of development was 100 per-cent. For all observed transitions, the two ratersalso agreed on the timing of the transitions and theevents that constituted them.

RESULTS

The Punctuated Equilibrium Model

Support for the punctuated equilibrium modelwas assessed in two ways: First, via examination ofthe overall pattern of attention to time and pacingthroughout a group's allotted time and second,through examination of qualitative shifts in theway the group performed its task.

Figure 1 demonstrates the distribution of pacingand time statements across all 25 groups. Thegroups showed more awareness of pacing as timeprogressed. To examine whether this pattern of in-crease in time awareness might be interpreted asboth linear and nonsequential, we performed twoseparate Friedman's chi-square analyses (Howell,1990). In the first analysis, we compared timeawareness in the two phases (20 minutes each) toreplicate Gersick's finding. In the second, we ex-amined time awareness across the four 10-minutetime intervals to determine if there was evidence ofa gradual increase in time awareness over time—that is, a linear trend. A Bonferroni adjustment wasmade so that alpha was set to .025 for the two"main effect" analyses and to .004 for the four"pairwise" comparisons.

Results of the two chi-square tests supportedboth a nonsequential and a linear increase in timeawareness. Time awareness significantly increasedbetween the phases: phase 2 contained 65 timereferences, as compared to 34 in phase 1 ix^^ =

Page 7: Punctuated Equilibrium and Linear Progression Toward a New Understanding of Group Development

112 Academy of Management Journal February

# #

FIGURE 1Patterns of Time Statements Collapsed across the 15 Groups^

# ## # # # #

# # , # # # # # # # # ## # # # # # # # # # # # # # # ## # # # # # # # # # # # # # # # # # # # # # # # # # # #

# # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # #

Time

" Each "#" represents the time statements of a different group.

9.71, p < .01). The four-stage chi-square testshowed a significant difference in groups' timeawareness over the four 10-minute intervals; therewere 14, 20, 26 and 39 time references over theseperiods ix^^ = 13.85, p < .01). Furthermore, pair-wise comparisons of the time awareness statementsacross the four 10-minute intervals showed that theonly significant difference in time awareness wasbetween intervals 1 and 4 [x^^ = 11.79, p < .001;X^i = 1.06, 0.78, and 2.60 for the comparisonsbetween intervals 1 and 2, 2 and 3, and 3 and 4,respectively). Thus, both linear and nonsequentialchanges of time awareness were observed in thisstudy. To understand if the published data fromGersick's (1989) study could also be interpreted assupportive of both linear and nonsequential pat-terns, we reanalyzed those data and found a similarpattern. When the group's allotted 60 minutes wasdivided into four 15-minute intervals, a significantdifference in time awareness was found [x^^ = 9.57,p < .0025; 7, 12, 14, and 23 time references), andnone of the pairwise comparisons were significantexcept for the comparison between intervals 1 and4 (A'% = 8.53, p < .001; x^i = 1-32, 0.15, and 2.19for the comparisons between intervals 1 and 2, 2and 3, and 3 and 4, respectively). Thus, reanalysisof Gersick's data also indicated that the resultscould be interpreted as both linear and nonsequen-tial, depending on the unit of analysis that is used(see Wollin, 1999).

Midpoint Transition

Using Gersick's coding system as a guideline, weconstructed a time map describing the changes inthe central theme of discussion over time for eachgroup; Table 1 presents an example of a map. Nineof the 25 groups studied here showed some form oftransition around the midpoint of their allottedtime. A transition point was defined by Gersick as

"the moment when group members made funda-mental changes in their conceptualization of theirown work" (1989: 277). Gersick argued that thisshift could occur in two ways: "One way consistedof summarizing previous work, declaring it com-plete, and picking up a next subtask. A second waywas observed in groups whose phase 1 agendasappeared to be floundering. These groups justdropped stalled phase 1 approaches and reachedout for a fresh source of inspiration, somethingaround which to crystallize further efforts" (1989:303). Like Gersick, we observed that transitionsthat occurred around the midpoint were mostlikely to focus on new content or a new format thatwould help a group to integrate the materials gen-erated up to that point to create the basic structureof the commercial. For example, one group spentthe first half of its time talking about possible con-tent ideas for the commercial. After 19 minutes ofdiscussion, one member said, "All I can think of is'I Like Aeroplane Jelly'" (a reference to a jingle inan existing TV ad). Another group member re-sponded, "We can change the words of 'I Like Aero-plane Jelly' to 'I Love Air Australia.'" This sugges-tion was not immediately taken up by the group,but 2 minutes later the group used this idea andstarted making up words for this song, which be-came a major part of the commercial.

Gersick (1989) studied 8 groups in the field and 8groups in the laboratory. In both cases, she foundmidpoint transitions in most groups. She con-cluded that the midpoint is the most likely, but notthe only, transition point in project teams. How-ever, in our study, with a larger sample of 25groups, we found that only 9 groups underwentmidpoint transitions. The other 16 groups fitted thedescription of linear progression more closely thanthat of a progression marked by a midpoint transi-tion. Small and incremental changes occurredthroughout these groups' life spans, without clear.

Page 8: Punctuated Equilibrium and Linear Progression Toward a New Understanding of Group Development

2003 Chang, Bordia, and Duck 113

sudden changes in direction. Nevertheless, of the16 groups that did not display midpoint transi-tions, 12 did undergo some transitions, most ofwhich occurred within the first 10 minutes of theirlife spans.

The presence of both linear progressive andpunctuated developmental patterns illustratedthat groups can follow various developmentalpatterns—punctuated equilibrium, linear pro-gression, or a combination. This finding is notsurprising, given that, even with a sample size of8 groups, Gersick observed two ways of makingthe midpoint transition. More importantly, thefirst way of making a transition she observed(summarizing previous work, declaring it com-plete, and picking up a next subtask) describeselements of linear progression as well as ele-ments of nonsequential transition. Variations inthe "sizes" of transitions and the number of tran-sitional points resulted in our observation ofmultiple developmental patterns.

The results of this study accorded with those ofLim and Murnighan (1994), who found thatchanges in negotiation activities over time couldbe interpreted as forming either a nonsequentialpattern in which a transition point occurred im-mediately before a deadline, or as forming a lin-ear pattern in which gradual and incrementalchanges occurred throughout a group's life span;Lim and Murnighan suggested that their studydid not necessarily discredit the punctuatedequilibrium model; rather, they saw it as demon-strating that if a group does display nonsequen-tial developmental patterns, the nature of the taskhas a dramatic impact on when in the group's lifespan a transition will take place. The fact thatmost of the groups in the present study did showsome form of transition during their life spanssupported the validity of the punctuated equilib-rium model but, like Lim and Murnighan (1994),we found that transitions do not always occur atthe midpoint.

The Integrative Model

To examine group developmental patterns fromthe perspective of the integrative model, we di-vided the groups' 40 minutes of allotted time intofour 10-minute intervals. For each interval, the pro-portion of time allocated to each category was cal-culated by dividing the number of statements madein that particular category by the total number ofstatements made. Figures 2a-2e illustrate the de-velopmental patterns of the groups over time.

Visual inspections of the data were conducted asa starting point to the interpretation of the results.

Inferences based on the visual inspections are cru-cial to this study; given the low statistical power ofthe study, we did not expect many of the statisticaltests to be significant. The figure does not containcounterpairing and fight graphs. With only onegroup yielding only two counterpairing statements,these were very rare, possibly because group mem-bers' personal involvement in the task was low.The standard deviations for all five categories werelarge, although this was more a result of the groups'displaying different baselines (for example, somegroups spent 40 percent of their time on flight andothers spent less than 10 percent of their time onflight) than a consequence of their displaying dif-ferent developmental patterns. We inferred generaldevelopmental trends from the data aggregated forthe 25 groups.

Friedman's rank tests were performed on thisaggregated data for each category as a means todetermine if there was any significant difference inthe amount of time allocated to the particular cat-egory over the four time intervals. No statisticaladjustment was made because the categories aremutually exclusive. If a significant chi-square wasobtained for the particular category, we conducteda series of pairwise comparisons using Friedman'srank test (with an alpha set to .008, after the Bon-ferroni adjustment) to determine whether thechanges over time were linear or nonsequential.

Work. As expected, work statements dominatedthe groups' interactions at all times. Roughly 70 to80 percent of the statements were work statements.Visual inspection of Figure 2a and the Friedman'srank tests support the integrative model hypothesisthat attention to work will increase gradually overtime. There was a significant difference in the pro-portion of time allocated to work statements overthe four time intervals (A-^ = 26.67, p < .0001).Follow-up pairwise comparisons showed that dif-ferences in the amount of work in adjacent timesegments were nonsignificant, whereas compari-sons between nonconsecutive time segments weresignificant {x^j = 14.44, 11.56, and 10.67, p < .001for times 1 and 4, 2 and 4, and 1 and 3; x^i ~ 1-5;1.5 and 6.0, n.s., for times 1 and 2, 2 and 3, and 3and 4, respectively).

Flight In contrast, flight statements only occu-pied 6 to 7 percent of groups' interaction time. Mostflight statements were made to avoid the task ratherthan to avoid confrontation in the group. Visualinspection of Figure 2b supports the integrativemodel prediction that flight statements will in-crease at time 2 and then decrease over time 3 andtime 4. However, the difference in flight over timewas not statistically significant (x^^ = 2.79, n.s.).

Page 9: Punctuated Equilibrium and Linear Progression Toward a New Understanding of Group Development

114 Academy of Management Journal February

FIGURE 2Developmental Patterns Averaged across the 15 Groups^

(2a) Work Statements100%

(2b) Flight Statements30%

-5%

Time 1 Time 2 Time 3 Time 4 Time 1 Time 2 Time 3 Time 4

(2c) Pairing Statements

25% -

20% -

15% -

10% -

5% -

0% -

• T 1

i

1 1 1

(2d) Dependency Statements30% •

20% - -

10% - -

Time 1 Time 2 Time 3 Time 4 Time 1 Time 2 Time 3 Time 4

(2e) Counterdependency Statements5%

Time 1 Time 2 Time 3 Time 4

* The five graphs describe the means and standard deviations of the work, pairing, dependency, counterdependency, and flightstatements made over the four 10-minute intervals.

Pairing. Pairing statements occupied 10-16 per-cent of the groups' interaction. Most pairing state-ments were made to show reflective listening toother group members. Visual inspection of Figure2c and the Friedman's tests led to the conclusionthat the proportion of time allocated to pairingstatements gradually decreased over time. Therewas a significant difference in the proportion oftime allocated to pairing statements over the fourtime intervals (;̂ 3̂ = 20.75, p < .0001). Pairwisecomparisons showed that pairing statements stayedrelatively constant from time 1 to time 3 and de-creased significantly at time 4 {x^.^ = 8.9, 13.5, and8.17, p < .004 for times 3 and 4, 1 and 4, and 1 and

3; x^^i = 0.0,1.63, and 6.0 for times 1 and 2, 2 and 3,and 2 and 4, respectively).

According to the integrative model, the propor-tion of time allocated to pairing should increasegradually over time until groups reach the truststage (stage 3), and then decrease slightly at thework stage (stage 4). Neither visual inspection ofthe graph nor the Friedman's rank tests supportedthis pattern of development. Instead, both sug-gested that pairing decreased slowly over time, es-pecially at time 4.

The lack of increase in pairing over time could beexplained by the artificial group setting. Given thatthese simulated work groups were highly task-

Page 10: Punctuated Equilibrium and Linear Progression Toward a New Understanding of Group Development

2003 Chang, Bordia, and Duck 115

oriented and that group members had no expecta-tions of future interaction, they might not have feltthe need to engage in the maintenance activitieslikely to occur in naturally occurring work groups.A lack of socioemotional activity is expected inlaboratory groups; however, the proportion of timeallocated to pairing statements (ranging from 16 to10 percent) was much lower than that observed inother studies using similar groups (e.g., Bales,1953). On the other hand, the decrease in pairingstatements at time 4 is supportive of the model.

Dependency. Dependency statements occupiedonly 3-5 percent of the interaction time. However,this category was expected to have low frequenciesbecause the groups did not have designated lead-ers. Visual inspection of Figure 2d and the Fried-man's tests support the integrative model argumentthat dependency will decrease gradually over time.There was a significant difference in the proportionof time allocated to dependency statements overthe four time intervals (x^^ = 13.85, p < .003).Pairwise comparisons showed that only the differ-ence between time 1 and time 4 was significant atthe adjusted alpha level (A'̂ = 9.78, p < .002; x^i =0.18, 0, 0.65, 5.0, 3.85, n.s., for times 1 and 2, times2 and 3, times 3 and 4, times 1 and 3, and times 2and 4, respectively).

Counterdependency. Counterdependency state-ments occupied only 1 percent of the groups' inter-action time. This category was expected to be ofvery low frequency because of the groups' artificialsettings and the group members' lack of vestedinterest in the task. Visual inspection of Figure 2esuggests that counterdependency statements werehighest at time 2, and slightly lower at time 1, butmuch lower at time 3 and time 4, supporting theintegrative model argument that counterdepen-dency will be higher in the early stages of groupdevelopment (stages 1 and 2) and lower in the laterstages of group development (stage 3 and 4), with apeak occurring at the counterdependency/fightstage. However, this result needs to be interpretedwith caution because, although there was a signif-icant difference in the proportion of time allocatedto counterdependency statements over the fourtime intervals ix^^ = 9.83, p < .02), no pairwisecomparisons were significant at the adjusted alphalevel of .008 ( / i = 2.0, 4.5,1.8,1.8,1.8, and 5.4 fortimes 1 and 2, 2 and 3, 3 and 4,1 and 4,1 and 3, and2 and 4, respectively).

DISGUSSION AND CONGLUSION

In summary, results of this study partially supportdevelopmental trends that are in line v̂ rith the inte-grative model. According to this model, in a group.

statements showing attention to work and pairingshould increase over time, whereas those indicatingdependency should decrease over time. Furthermore,the integrative model prediction is a peak in counter-dependency, fight, and flight statements in phase 2 ofa group's development and a decrease from thereonward. Results of this study support the predictedpattern of change for work and dependency, but notfor pairing. Visual inspections of changes in flightand counterdependency also support the predictedpatterns, although corresponding statistical tests con-ducted at a conservative alpha level do not. More-over, no fight or counterpairing statements were ob-served in the groups' interaction. On the other hand,results of this study also suggested some reservationsconcerning the use of the Group Development Ob-servation System (Wheelan et al., 1993) vwth lab-oratory work groups. As expected, the ad hoc labora-tory groups in this study were very task-focused, andit was difficult to study group development on the 'socioemotional level.

Results of this study supported Hypothesis 1.Both punctuated equilibrium and linear progres-sion were observed simultaneously in simulatedproject teams, with the dimension of observationand the unit of analysis accounting for the type ofdevelopment that was observed. The punctuatedequilibrium model describes changes in a group's,,time awareness and pacing activities over time aswell as changes in its task activities over time. Theintegrative model, on the other hand, describeschanges in a group's structure and process alongboth task and socioemotional dimensions. The dif-ferent scales of changes described by the two mod-els can be seen by comparing the coding of a sam-ple transcript made using the GDOS with a meetingmap of the same interaction constructed using Ger-sick's coding system (see Table 1). For example, inthe first 30 seconds of the group's interaction,group members showed avoidance of the task byengaging in off-task discussion of whether the mu-sic provided by the experimenter was taken from amovie. This information would be coded by theGDOS as an indication of the group's flight awayfrom the task. However, when one examines the cen-tral theme of the group's task-oriented discussion,this information would simply be coded as part of thegroup's listening to and commenting on the music

• tapes provided. Thus, the GDOS aids understandingof subtle changes in a group's processes and struc-ture, whereas Gersick's coding system guides under-standing of the overall changes in a group's approachto its task. The two models complement each other toprovide rich information on the developmental pat-terns of project teams over time.

Results of this study also demonstrate that

Page 11: Punctuated Equilibrium and Linear Progression Toward a New Understanding of Group Development

116 Academy of Management Journal February

changes in pacing and time statements can be in-terpreted as both linear and nonsequential, de-pending on the unit of analysis used. This findinghighlights the significance of the unit of analysis indetermining the developmental patterns that areobserved in groups. The work conducted followingthe punctuated equilibrium model used a largertime fi:ame (20 minutes) and demonstrated discon-tinuous change from phase 1 to phase 2, whereasthe integrative model work used a smaller timefi-ame (10 minutes) and demonstrated incrementalchanges over times 1, 2, 3, and 4. With the excep-tions of the absence of an increase in pairing state-ments and the nonsignificance of the change inflight statements over time, the overall pattern ofchanges in GDOS categories closely matched theintegrative model's predicted pattern, thus sup-porting the developmental sequence proposed inthe linear model. However, post hoc pairwise com-parisons for the work and pairing statements alsosuggested that the patterns of change could be de-scribed as an incremental step occurring betweentimes 1 and 2 combined and times 3 and 4 com-bined, rather than as linear increase over times 1, 2,3, and 4. This pattern of results suggested that themidpoint transition also marked the group's reso-lution of early developmental issues such as lead-

o ership and work structure and a move forward inthe production of the pilot commercial. This pat-tern—groups' moving toward more productive de-velopmental stages (3 and 4) after a midpoint tran-sition—was further supported by thematic changesin the discussions: In most cases, midpoint transi-tions moved the groups from an earlier conversa-tion on the content and process of the commercialto later dialogue on procedure and on details of thecommercial. Furthermore, most groups displayedmore effective structuring (such as forming sub-groups to work on different aspects of the task)toward the ends of the their life spans. Thus, weconclude that Hypothesis 2 is also supported. Theintegrative model appears to describe changes at amore micro level within two phases of inertia. Thisfinding accords with Wollin's proposition thatpunctuated equilibrium should be viewed as "astepped continuum of change in an organizationalsystem, from revolutionary discontinuous changeto more incremental change, reflecting the differentlevels of its deep structure" (1999: 365).

The results of our study clarify an apparent mis-understanding in the current group literaturewhereby Gersick's (1988, 1989) work is typicallyperceived as contradicting the linear developmen-tal patterns proposed in traditional stage models(e.g., Bettenhausen, 1991; Guzzo & Shea, 1992). Ourstudy demonstrates that both the integrative model

and the punctuated equilibrium model describevalid developmental patterns of project teams. Fur-thermore, the two models complement each otherto better inform researchers and practitioners onthe development of different aspects of a group'sfunctioning.

Depending on the temporal dimension they areinterested in, researchers and practitioners work-ing with organizational project teams can chooseeither the integrative model or the punctuatedequilibrium model. For example, when planningto implement significant change in a work group,a change agent can use insights from both modelsto prepare the group for the upcoming change.First, the agent can work with the group on earlydevelopmental issues to facilitate trust amonggroup members and effective work processes(using the integrative model), and can thus en-hance the group's ability to cope with interrup-tions to work flow. Second, the change agent canthen introduce small internal or external changesto interrupt the group's current state of inertiaand create an environment of instability, whichwill in turn increase the group's propensity forlarger changes (using the punctuated equilibriummodel). For example, replacing a group membercan facilitate the group's examination of currentstructure and processes and thus provide oppor-tunities for introducing changes to one or bothaspects (Arrow & McGrath, 1993). Once changeshave been introduced, early developmentalissues might need to be revisited (the integrativemodel) to facilitate effective work under the newworking conditions. By integrating the two mod-els, change agents can help groups to make tran-sitions more successfully and in a preferreddirection.

The artificial laboratory setting and the smallsample here are the study's most important limita-tions. It was clear from the observational data thatthese simulated work groups in a laboratory settingdid not display the same socioemotional develop-ment that naturally occurring work groups do, sowe could not test some specific predictions basedon the integrative model. However, the resourceintensity of group development observational re-search makes it difficult to study field groups usingsimilar coding schemes. Thus, in the future, re-searchers should aim to both streamline the presentcoding system so it can be used for field studies andto conduct more laboratory studies in controlledenvironments with creative interventions that pro-mote the socioemotional development of the simu-lated work groups.

Page 12: Punctuated Equilibrium and Linear Progression Toward a New Understanding of Group Development

2003 Chang, Bordia, and Duck 117

REFERENCES

Arrow, H. 1997. Stability, bistability, and instability insmall group influence patterns. Journal of Person-ality and Social Psychology, 72: 75-85.

Arrow, H., & McGrath, J. E. 1993. Membership matters:How member change and continuity affects smallgroup structure, process, and performance. SmallGroup Research, 24: 334-361.

Bales, R. F. 1953. The equilibrium problem in smallgroups. In T. Parsons, R. F. Bales, & E. A. Shils (Eds.),Working papers in the theory of action: 111-161.New York: Free Press.

Bettenhausen, K. L. 1991. Five years of groups research:What we have learned and what needs to be ad-dressed. Journal of Management, 17: 345-381.

Bion, W. R. 1961. Experiences in groups. New York:Basic Books.

Buzaglo, G., & Wheelan, S. A. 1999, Facilitating workteam effectiveness: Case studies from central Amer-ica. Small Group Research: 30: 108-129.

Folger, J. P., Hewes, D. E., & Poole, M. S. 1984. Codingsocial interaction. In B. Dervin & M. Voight (Eds.),Progress in communication sciences, vol. 4: 115-161. Norwood, NJ: Ablex.

Gersick, C. J. 1988. Time and transition in work teams:Toward a new model of group development. Acad-emy of Management Journal, 31: 1-41.

Gersick, C. }. 1989. Marking time: Predictable transitionsin task groups. Academy of Management Journal,32: 274-309.

Gersick, C. }. 1991. Revolutionary change theories: Amultilevel exploration of the punctuated equilib-rium paradigm. Academy of Management Review,16: 10-36.

Guzzo, R. A., & Shea, G. P. 1992. Group performance andintergroup relations in organizations. In M. D. Dun-nette & L. M. Hough (Eds.), Handbook of industrialand organizational psychology (2nd ed.), vol. 3:269-313. Palo Alto, CA: Consulting PsychologistsPress.

Howell, D. C. 1990. Statistical methods for psychology.Belmont, CA: Duxbury Press.

LaCoursiere, R. B. (Ed.). 1980. The life cycle of groups:Group development stage theory. New York: Hu-man Science Press.

Lim, S. G. S., & Murnighan, J. K. 1994. Phases, deadlines,and the bargaining process. Organizational Behav-ior and Human Decision Processes, 58: 153-171.

Seers, A., & Woodruff, S. 1997. Temporal pacing in task

forces: Group development or deadline pressure.Journal of Management, 23(2): 169-187.

Tuckman, B. W. 1965. Developmental sequence in smallgroups. Psychological Bulletin, 63: 384-399.

Van de Ven, A. 1995. Explaining development andchange in organizations. Academy of ManagementReview, 20: 510-540.

Verdi, A. F., & Wheelan, S. A. 1992. Developmental pat-terns in same-sex and mixed-sex groups. SmallGroup Research, 23: 356-378.

Wheelan, S. A. 1994. Group processes: A developmen-tal perspective. Sydney: Allyn & Bacon.

Wheelan, S. A., & Burchill, C. 1999. Take teamwork tonew heights. Nursing Management, April: 28-31.

Wheelan, S. A., Murphy, D., Tsumura, E., & Kline, S. F.1998. Member perceptions of internal group dynam-ics and productivity. Small Group Research, 29:371-393.

Wheelan, S. A. & Tilin, F. 1999. The relationship be-tween faculty group development and school pro-ductivity. Small Group Research, 30:59-81.

Wheelan, S., Verdi, R., & McKeage, R. 1993. The GroupDevelopment Ohservation System: Origins andapplications. Philadelphia: PEP Press.

Wollin, A. 1999. Punctuated equilibrium: Reconciling the-ory of revolutionary and incremental change. SystemsResearch and Behavioral Science, 16: 359-367.

Artemis Chang ([email protected]) is a lecturer in theSchool of Management at Queensland University ofTechnology. She received her doctoral degree from theUniversity of Queensland. Her research interests includegroup processes and performance, time, change, humanresource information systems, and employee turnover inthe IT industry.

Prashant Bordia (Ph.D., Temple University) is a seniorlecturer in the School of Psychology, University ofQueensland. His research interests include group devel-opment, computer-mediated communication, and ru-mors in organizations.

Julie Duck (Ph.D., University of New England) is a seniorlecturer in the School of Psychology at the University ofQueensland. Her ciurrent research focuses on group andintergroup processes, especially as they apply to organi-sational contexts and to the impact of mass communica-tion on perceptions, attitudes, and behavior.

Page 13: Punctuated Equilibrium and Linear Progression Toward a New Understanding of Group Development