Bridging the Boundary: External Activity and … the Boundary: External Activity and Performance in Organizational Teams Deborah G. Ancona Massachusetts Institute of Technology David

Bridging the Boundary:External Activity andPerformance inOrganizational Teams

Deborah G. AnconaMassachusetts Institute ofTechnologyDavid F. CaldwellSanta Clara University

© 1992 by Corned Lrtirversity.0001-8392/92/3704-0634/$1.00.

We thank Lotte Bailyn, Keith Mumighan.Elaine RtMnanelli, the Center forInnovation Management Studies. BobSutton, and tfie anonymous reviewers atAdministrative Science Quarteriy for theirsupport and comments.

This article focuses on the activities teams use tomanage their organizational environment beyond theirteams. We used semistructured interviews with 38new-product team managers in high-technotogycompanies, log data from two of these teams, andquestionnaires completed by members of a different setof AB new-product teams to generate and testhypotheses about teams' external activities. Resultsindicate that teams engage in vertical communicationsaimed at molding the views of top management,horizontal communication aimed at coordinating workand obtaining feedback, and horizontal communicationaimed at general scanning of the technical and marketenvironment. Organizational teams appear to developdistinct strategies toward their environment: somespecialize in particular external activities, some remainisolated from the extemal environment, and othersengage in multiple external activities. The paper showsthat the type of external communication teams engagein, not just the amount, determines performance. Overtime, teams following a comprehensive strategy enterpositive cycles of external activity, internal processes,and performance that enable long-term team success.*

This paper is about groups. Yet the research presented herediffers from that usually found in the dominant socialpsychology paradigm. This research uses on-goingorganizational teams rather than one-time laboratory groups.The tasks of these organizational teams are complex andevolving, not simple and set. The task allocators aremanagers, not academics. The teams' work isinterdependent with other organizational units; teams cannotwork in isolation. The key element that differentiates thisresearch, however, is its focus. Rather than sitting on thegroup boundary and looking inward, we have focusedprimarily on those team behaviors that are directed outward,toward other parts of the organization, using an "external"perspective (Ancona, 1987).

Over the past half century, social psychologists havedevoted substantial attention to the fine-grained analysis ofbehavior within groups. Many frameworks exist for thatanalysis, including models of group decision making(Isenberg, 1986; Nemeth, 1986; Bourgeois and Eisenhardt,1988), task and maintenance activities (Benne and Sheats.1948; Bales, 1983; Schein, 1988), norm development(Bettenhausen and Murnighan, 1985), and evolution (Gersick,1988, 1989), to name a few. Yet it is only recently that theexternal perspective has been studied in depth.

Our research on new-product teams spans the years 1985 to1990 and, along with other studies {e.g., Ancona andCaldwell. 1988; Ancona, 1990). forms the foundation of theexternal perspective. Because much less research hasfocused on external group activity than on behavior within agroup, the first stages of research were necessarilydescription and classification (Kerlinger, 1973; Gladstein andGuinn. 1985). We sought to discover the relatively unknownpattern of groups' external activities with essential others.This discovery phase involved collecting qualitative data.

634/AdministrattV8 Science Quarterly, 37 (1992): 634^665

Organizationai Teams

including interviews with 38 experienced product-teammanagers, daily logs maintained by all members of twoproduct teams, and observation of the activities of those twoteams. These methods allowed us to describe a wide rangeof activities that groups use to carry out their complex tasks.With these data and the extant literature, we generatedhypotheses linking extemal activities, performance, andinternal process. In the classification phase, usingquantitative data from a separate set of 45 new-productteams, we sought to group similar activities intoindependent clusters. At this point, we also classified teamsby the types of external activities they used, to determine ifthere are generic strategies teams use to deal with theirenvironments, and tested the hypotheses about externalactivities and outcomes. This paper describes these severalresearch stages and, thus, the development of the externalperspective.

DESCRIPTION OF EXTERNAL ACTIVITIES

Literature Review

The impetus for describing the external activities of teamscame from a study of 100 sales teams in thetelecommunications industry (Gladstein, 1984), whichinvestigated the relationships between internal groupprocesses and team performance. During preliminaryinterviews, salesmen frequently spoke of the importance oftheir interactions with their firm's installation and repairteams. Thus, several survey questions were added to pursuehow teams interacted with other groups within theirorganization. It was hypothesized that these relationshipswould add another dimension to the task behaviors of thegroup.

The results were surprising. First, group members did notperceive process as separating into the traditional task andmaintenance components (Bales, 1958; Philip and Dunphy,1959; Schein, 1988), Instead, they saw process as dividedinto an internal and an external component, behaviors takingplace among group members versus those with outsiders.Second, while internal group process predicted solelyteam-member satisfaction and team-rated performance,external process was associated only with sales revenue, anobjective, external measure of performance. Thus, an aspectof group process that had been virtually ignored in theliterature affected organizational performance in ways thatinternal processes did not. This study, however, did notspecifically examine those external activities. Research usingthe external perspective began, then, with the simplerealization that organizational group process was not fullyrepresented by internal activities. Group members interactwith one another, but they are also proactive with outsiders,seeking information and resources, interpreting signals, andmolding external opinion.

Although the bulk of small-group research has focused oninternal dynamics, some research has examined externalinitiatives. One set of researchers has studied the amount ofinformation exchanged between teams and theirenvironment. This research indicates that groups must

635/ASa December 1992

match their information-processing capability to theinformation-processing demands of the task environment(AHen, 1984; Nadler and Tushman, 1988; Gresov, 1989).This work has contributed greatly to understanding theimportance of extemal communication to group success.Because the primary focus of the work has been on thefrequency rather than the content or purpose ofcommunication, however, it has not attempted to identifytypologies of extemal interactions.

Other researchers have investigated specific types ofgroups' interactions with others. Those studying innovationhave focused on boundary spanning and the transfer oftechnical information across team boundaries (Aldrich andHerker, 1977; Katz and Tushman, 1979; Allen, 1984), thosestudying interdependence have focused on intergroupcoordination (Malone, 1987), and those studying power andresource allocation have focused on political or persuasiveaction with external constituents (Pfeffer, 1981; Dean,1987). While these researchers have identified keydimensions of external activity, because they were studyingspecific organizational phenomena they did not use thegroup as a focal unit and have not tried to map the full rangeof external activities that groups use to deal with theirenvironments. Although a variety of studies havedemonstrated the importance of external communication togroup or organizational outcomes, there has not been asystematic effort to identify the full set of specific activitiesin which teams engage. Thus, the first task in building theexternal perspective was to describe what externallydependent teams actually do across their boundaries.

Methods

We studied new-product teanns in high-technologyorganizations. These teams enabled us to describe a diverserange of boundary activities because they depend on otherparts of the organization for information, resources, andsupport and must deliver products and services to others.These complex transactions are carried out with a diverseset of functional groups, including marketing, manufacturing,and top management, groups that represent other"thought-worlds" (Dougherty, 1987; 1)—different languages,values, and time frames, as well as different hierarchicallevels.

New-product teams face a highly uncertain and complextask. There may be periods of creativity alternating withtimes when efficiency is the primary outcome of interest.Therefore, most of the interaction with other groups is notclearly programmed in standard operating procedures androutines but evolves to meet task demands. Thecombination of high complexity, high uncertainty, andmultiple forms of dependence with multiple groups of othersseemed well suited for identifying a wide range of boundaryactivities.

Data coHection. To identify the range of external activities,we interviewed the managers of 38 new-product teams,interviews ranged between one and eight hours, with anaverage duration of three hours. Teams were at variousstages of product development: some were just starting,

636/ASQ, December 1992

Organizational Teams

while others had completed projects within the fast month.Using a semistructured interview, we asked each managerto describe the activities that the manager and teammembers carried out with people outside the groupboundaries. Ouestions dealt with the timing, frequency,target, and purpose of interactions across the entire life ofthe product-development process.

In addition, 15 new-product team members from two of the38 teams were asked to keep logs for a week. In these logs,individuals recorded all interactions they had with peopleoutside the team, including meetings, telephone calls,electronic mail messages, or other contact, and the purposeof those interactions. Once the interview and log data werecollected, the two authors and two graduate studentsreviewed transcripts, notes, tapes, and logs to identify allreferences to interactions with outsiders. We also reviewedthe data for reports of decisions not to interact with others;for example, a manager might speak of not wanting to meetwith a member of another group until a specific part of theproduct had been completed. We included these reports inour analysis.

Identification of External Activities

Through a review of the complete set of interactions, weidentified a set of 15 distinct activities, including mapping,gathering information and resources, scanning, feedbackseeking, opening up communication channels, informing,coordinating, negotiating, molding, allowing entry, translating,filtering, classifying, delivering, and protecting, Theseactivities are not unique to product teams, and manyrepresent phenomena that have been extensively studiedfrom many different levels of analysis. For example, whatwe term molding is conceptually simitar to impressionmanagement (Chapman, Bell, and Staw, 1986; Ginzel,Kramer, and Sutton, 1993). Similarly, feedback seeking hasbeen studied at the Individual level of analysis (Ashford andCummings, 1983; Ashford and Tsui, 1991), and mapping hasanalogues In socially constructed realities (Weick, 1979) andsense making (Louis, 1980). Our goal was not to impose acategory scheme from existing literature but, rather, todiscover as wide a range of external activities in groups aspossible. While space does not allow a full description ofeach activity (see Ancona and Caldwell, 1988, for such adescription), four examples illustrate how we moved fromdata to activity.

Mapping. Mapping entails constructing a picture of theexternal environment, including predicting future troublespots or potential allies. Mapping represents the team'sattempt to answer questions such as "Who supports us andwho doesn't?" and "What do people want us to do?"Mapping was often done by combining information thatteam members had from prior experiences with updatedinformation gleaned from conversations with outsiders. Thefollowing quotation illustrates data that helped us to identifythis activity:

The first thing I did was go to talk to lots of people to find out whatthey thought the product was and how to get there.... 1 startedout with the guy who brought me here, he sent me to someone


else, and so it went that 1 came to talk to a lot of high- andmiddle-level people So I gained knowledge about details ofwhat the product ought to be, who the players were, what they didand what they wanted.

Molding. This activity involves a group's attempts toinfluence the external environment to suit its agenda byshaping the beliefs and behaviors of outsiders, in essence,molding involves persuading and influencing; it may meanrepresenting the group in an extremely positive light ifresources are needed or in a less positive light if that iswhat is required. An example follows:

I'm like a cheerleader, trying to get those guys excited about ourpnxJuct.... I went to a meeting and explained that the company wasriding on this prefect and we were going to do it fast and do it right.

Coordinating and negotiating. Although these twoactivities appear to be separate, our interviewees typicallydiscussed them together. Coordinating usually involvesresolving the issues of interdependent schedules. Althoughthe focus of this activity may be integrating work schedules,there is often negotiating going on as wed. This negotiatingis particularly common because of shifting power anddependency relationships between the new-product teamand other groups. One quotation that is indicative of thisactivity follows;

We had to explain (to manufacturing) how certain things worked. 1had lots and lots of meetings about the status of the project. Wewanted some last-minute changes on the machine, butmanufacturing was not able or not willing to put them in all themachines. There were great arguments and the Product Committeewas involved. By April we had worked out a compromiseagreement.

Filtering. Filtering consists of taking information fromoutsiders and delivering a smaller amount to the group.Often filtering would be done to buffer the team or absorbpressure by keeping troubling information or politicalmaneuvering from the team. Buffering also took place whenthe volume of information was considered too great for theteam to absorb. An extreme form of buffering is actually toseparate the group physically from the rest of theorganization. One illustration follows:

Near the end I talked to the top management group a lot. i tried toprotect the group from that kind of pressure, though. It's like TomWest said, we won't pass on the garbage and the politics.''

In all, we found a wide variety of external activities. Somewere taken on by one member of the group while otherswere distributed across many members. In addition, asequence of boundary transactions often combinedelements of several activities. These boundary activitiesplayed a key role m determining how group membersviewed the outside world and how outsiders perceived thegroup. Activities also seemed to relate to one another. Thus,a team that engaged in a high ievel of molding activity mightneed to spend less time protecting the team from undueinfluence later.

CLASSIFICATION OF EXTERNAL ACTIVITfES

IZ^^l^^tZ^N^ After identifying boundary activities, we sought to reduce(Boston: Auantic-Uttte, Brown. 1981). their number of clustenng them into larger categories of



related activities to create a typology of external initiatives.In order to do statistical clustering we needed a largernumber of data points and quantitative measures of thetypes of external activities that team members engaged in.This analysis thus involved a completely different sampleand methods. We used data collected throughquestionnaires completed by members of 45product-development teams, none of which had anymembers who had taken part in the previously describedinterview and log study.

Methods

Description of groups. This phase of the study involvedproduct-development teams in five corporations in thecomputer, analytic instrumentation, and photographicindustries. All of the teams were responsible for developinga prototype new product (not basic research) andtransferring it to their firm's manufacturing and marketinggroups. All the products used new or evolving technologies.One product, for example, automated the sampling processin liquid chromatography. another combined photographicand computer imaging processes. All of the teams weretemporary; they were formed to develop a specificprototype and disbanded once the task was complete. Eachwas formally headed by a project team leader. Teammembers had specific functional or technical skills; thisassignment was typically the individual's primaryresponsibility at work,

Each organization provided access to a set of teams that hadthe following characteristics: (1) ail the teams had to bedeveloping a new product (defined as a major extension toan existing product line or the start of a new product line);(2) to ensure some broad consistency in the complexity ofproducts, all products had a development cycle of one andone-half to three years; (3) for comparability in performanceevaluations, all the teams had to be located within a singledivision; and (4) teams had to range in performance,although company executives did not reveal how teamswere initially classified until all other data had been collected.Team membership was determined from company recordsand verified with team leaders; average size wasapproximately 10 (s.d. = 6.2), with a mode of 12.

The authors met with each team leader to get a history ofthe product and performance evaluations. Team leadersdistributed the questionnaires, including instructions andreturn envelopes addressed to the researchers. Severalfollow-up letters and phone calls were made to teammembers to ensure high response rates.

Data and sample. A total of 450 questionnaires weredistributed to team members and leaders of 47 teams.Because many of the items included in the questionnairerelated to perceptions of the team, the questionnairesdistributed to each team included a list of team members toensure that individuals had a common referent. Completedquestionnaires were returned by 409 individuals, yielding aresponse rate of approximately 89 percent. Response rateswere approximately equal across the five companies; totalresponses per company varied from 39 to 129. Because


Modes of agQregation other thanaveraging were tested to exwninedifferent assumptions about how groupsr^^Bsent ^1 ^^c^^wjual members'amtributions. For exanr^ . we summedindividual scores, under the assumptiontiiat a team's extemal actMty is simplythe ^ m of mdividi^ contritHjtions,

procedures did not ^tnificantiy affectresut^, we ctmtinued to use the meanscores.

much of the analysis was conducted at the group level,teams were included in the final sample only if at leastthree-fourths of their members responded. This reduced thenumber of teams m the final sample to 45.

The average age of the individuals was 38.6 years; 88percent were male; and 77 percent of the respondents wereemployed in the engineering or research and development(R&D) functions of their companies; the remaining 23percent were primarily from the manufacturing or marketingfunctions.

Measures. In this phase of the project we were interestedonly in those items that revealed the types of boundaryactivities undertaken by team members. The fifteenbehaviors that were uncovered in the first phase of thestudy were converted into 24 survey items, including actionssuch as persuading others to support the team, attemptingto acquire resources for the team, and bringing technicalinformation into the group. The number of items is greaterthan the number of behaviors because during pretesting wefound that some constructs, such as coordinating andnegotiating, needed to be separated. We also found that weneeded to specify more clearly the object of a particularactivity. Team members pointed out that scanning forinformation about the competition, for example, wasdifferent from internal technology scanning. We thereforeincluded both. The activities were converted intoquestionnaire items by asking respondents to indicate on a5-point Likert scale the extent to which they felt each of theitems was part of their responsibility in dealing with peopleoutside the team. The complete set of items is shown inTable 1.

Analysis and Results

The 409 individual responses to the 24 boundary-activityitems were factor-analyzed to represent their underlyingstructure. Individuals' ratings of the extent to which theyassumed responsibility for each of the 24 boundary activitieswere analyzed with a principal component analysis and avarimax rotation. Four factors with eigenvalues greater than1.0 explained 60 percent of the total variance, Inspection ofthe Scree plot supported the four-factor solution. Factorscores were then calculated for each individual and averagedto form group scores.^ Table 1 summarizes this analysis andshows the item loadings greater than .40.

Each factor represents a particular type of boundary activity.The first factor contains 12 items (with loadings greater than.50) that reflect both buffering and representational activities.It contains aspects of the filtering and molding activitiesdescribed earlier, as well as some aspects of mapping.Members taking on this set of activities protect the teamfrom outside pressure, persuade others to support the team,and lobby for resources. Because these activities includeboth protective and persuasive goals, we labeled them"ambassador" activities. Members carrying out theseactivities communicate frequently with those above them inthe hierarchy, such as top R&D management, top division,and even top corporate management (Ancona and1990).

640/ASa December

Table 1

Varimax Factor i-oadlngs for Boundary-Management Dimensions {N

Item

= 409)

1

Factor

2

loading

3 4

,785,740

,719.654

.636,602,592

.587,553

,551

,549,519

.416

,417,403

,421,776,660

,657,618,515

Absorb outside pressures for the team so it can work free ofinterference.

Protect the team from outside interference.Prevent outsiders from "overloading" the team with too much

information or too many requests.Persuade other individuals that the team's activities are important.Scan the environment inside your organization for threats to the

product team, .636 ,417"Talk up" the team to outsiders.Persuade others to support the team's decisions.Acquire resources {e,g., money, new members, equipment) for the

team.Report the progress of the team to a higher organizational level.Find out whether others in the company support or oppose your

team's activities, .551 ,449Find out information on your company's strategy or political situation

that may affect the project, .549 ,430Keep other groups in the company informed of your team's activities.Resolve design problems with extemal groups.Coordinate activities with extemal groups.Procure things which the team needs from other groups or

individuals in the company.Negotiate with others for delivery deadlines.Review product design with outsiders, .515 .404Find out what competing firms or groups are doing on similar

projects, .791Scan the environment inside or outside the organization for marketing

ideas/expertise, .719Collect technical information/ideas from individuals outside of the

team. .424 ,645Scan the environment inside or outside the organization for technical

ideas/expertise, 491 .587Keep news about the team secret from others in the company until

the appropriate time. ,823Avoid releasing information to others in the company to protect the

team's image or product it is working on, .817Control the release of information from the team in an effort to

present the profile we want to show, ,592

The second factor was defined by five items that representinteractions aimed at coordinating technical or design issues.Examples of activities in this set include discussing designproblems with others, obtaining feedback on the productdesign, and coordinating and negotiating with outsiders. Welabeled these "task coordinator" activities. Individualscarrying out these activities show high levels ofcommunication laterally through the organization with suchgroups as R&D and manufacturing (Ancona and Caldwell,1990).

The third factor was made up of four items describingbehaviors that involve general scanning for ideas andinformation about the competition, the market, or thetechnology. This factor included many aspects of themapping, information gathering, and scanning activitiesuncovered in the first part of the study. We labeled thisfactor "scout" activity. These items differ from the previousitems in that they relate to general scanning as opposed tohandling specific coordination issues. Individuals carrying out


this activity show high levels of communication withmarketing, sales, and R&D (Ancona and Caidwell, 1990).

The fourth set of activities represent actions that aimed atavoiding releasing information. We labeled the three itemsthat define this factor "guard" activities. Because theseactivities differ from the other three in that they do notrepresent initiatives toward the environment but, rather,intemal activities to keep things from the environment, wedid not include guard activities in subsequent analyses.

STRATEGIES FOR APPROACHING THE ENVIRONMENT

While the first two stages of this research described a widerange of external boundary activities and clustered thoseactivities to create a typology, it was clear from ourinterviews that not ai! teams exhibited all of these activities.Teams seemed to develop distinct styles of approachingtheir environment; some seemed to specialize in particularsets of activities, others were generalists, and others did notseem to engage in much activity at alt. Not al! groups havethe willingness or the capacity to exhibit the full range ofactivity. Thus, in the next stage of the research we lookedfor patterns of external activity across teams to see whichcombinations or packages of activities occur naturally andwhether there are pattems that differentiate teams and howthey organize themselves to carry out external activity.

Such an approach has analogues at the individual andorganization levels. At the individual level, literally hundredsof traits have been identified. A person can be introverted orextroverted, have an internal or extemal locus of control, orbe dominant or submissive. But much work has gone intoidentifying personalities, or combinations of traits, thatappear to identify people, Thus, a paranoid personality ismade up of different traits than a compulsive personality,and each represents a very different approach toward theexternal world. At the organization level, strategyresearchers have long been interested in classificationschemes, and several typologies exist, including those byMiles and Snow (1978)—defenders, analyzers,prospectors—and Porter (1980)—cost leadership,differentiation, focus.

We followed the same logic at the group level as we soughtto determine the external strategies that groups withinorganizations use. A typology of strategies would allow us tocategorize groups by the distinctive styles they seem toexhibit in interacting with their environments, differentiatetheir form, and examine the implications of those forms. Weuse the term strategy to label the patterns of externalactivity that are found. This is not to suggest that suchpatterns are necessarily intentional. Rather, they representthe subset of activities a team has demonstrated for a givenperiod of time.

Methods

We used Q-factor analysis of the questionnaire data fromthe 45 teams described earlier to produce a taxonomy ofexternal strategies indicating how external activities act incombinations. While regression analysis statistically isolates



the independent effects of each activity set. such atechnique does not tell us which combinations or gestaltsnaturally occur (Hambrick. 1983). 0-factor analysis groupstogether those teams that share common approaches to theenvironment, and these groupings can then be comparedalong other dimensions such as internal process andperformance. In contrast to the more conventional R-factoranalysis, 0-factor analysis is based on the respondentsrather than the variables and condenses respondents intodistinctly different groups. It differs from cluster analysis inthat the groupings are based on intercorretations betweenmeans and standard deviations of the respondents ratherthan on the absolute distances between the respondents'scores. Thus. 0-factor analysis may be more sensitive topatterns among the variables than absolute differences inmagnitude.

Results

A 0-factor analysis with a varimax rotation was performedon the 45 groups, using scores on the ambassador, taskcoordinator, and scout variables. This analysis identified fourdistinct sets of groups. The first part of Table 4. below,shows the results of three one-way analyses of variance(with group size as a covariate), using groups' scores on theambassador, task coordinator, and scout variables asdependent variables.

Our analysis suggested that groups' activities clustered intofour distinct strategies that showed no significantrelationship to a number of the characteristics of the productunder development including the extent to which theproduct was revolutionary, the level of competition in themarket, the organization's experience with the technology,and the availability of resources to the team. The firststrategy concentrated on ambassador activity and very littleelse; we labeled it "ambassadorial." The second combinedscout activities with some task coordination; we called it"technical scouting." The third strategy was low on alldimensions, with some minima! scout activity. We labeledthis pattern "isolationist." Finally, groups using the fourthstrategy had members who felt responsible for bothambassadorial and task-coordinator activities but did littlescouting. This strategy avoided general scanning; it focusedon external interaction both to persuade others that theteam's work was important and to coordinate, negotiate, andobtain feedback from outside groups. We called it"comprehensive." The number of teams displaying thesefour strategies was relatively equal, with 13 showing anambassadorial strategy, 12 a technical-scouting strategy. 10an isolationist strategy, and 10 a comprehensive strategy. Inaddition, these strategies were distributed roughly equallyacross organizations.

FORMULATING AND TESTING HYPOTHESES

While description and classification can be seen as ends inthemselves at the start of a scientific inquiry, later progressis made by proposing hypotheses and testing them. By thetime we were collecting the questionnaire data, the researchhad advanced to a point where we could generate


hypotheses. In addition, our own qualitative data wereconverging with the theory to suggest that external activitiesare related to team performance and to intemal groupprocesses in specific ways. Group theory has tended toconcentrate on the link between intemal process andperformance. If extemal activities become relevant variables,then group theory will need to be extended to include them.In addition, this hypothesis testing may suggest ways thatorganizational teams can improve their performance. Thus,we moved from trying to map how highly interdependentteams approached their environment to trying to understandhow these external activities related to team performanceand internal processes. This section reports on thequalitative data and literature that led to a set of hypotheseslinking these variables, describes the methods used to testthose hypotheses, and reports the results.

Oualitative Data

The initial set of 38 interviews provided some very rich anddetailed accounts of the histories of numerous new-productteams in the high-technology environment. The accountswere quite varied; some told of marvelous successes,others of failure, and others of troubled teams that hadsomehow dealt with issues as they came up and turnedadversity into surprising performance. Some accounts wereretrospective, and some were about current activities. Someteams were quite cohesive, and others had members thathardly knew each other. Some had leaders who spokemainly of technical issues, while others were completelywrapped up in politics. These rich data, combined with ourknowledge of previous literature, helped us to develophypotheses connecting external activities to performanceand intemal activities that could then be tested with thesample of 45 teams. We describe four cases here that,together, represent disparate points along a range ofperformance and of strategies for dealing with others. Datalike these, from ail 38 interviews, were used to formulatethe hypotheses.

Swallow. The Swallow story is one of great success.Swallow is a pseudonym for an internal project name givento a computer being designed to serve the computer-aideddesigrv-computer-aided engineering market. The project wasof major import to the organization and had the difficult goalof coming in at three times the performance of theorganization's current offering at the same price. Ourinterview with the Swallow leader came just as the projectwas being completed. He reported that they had done theimpossible in terms of meeting schedule and productspecifications. There was already great market demand.Elsewhere in the organization the Swallow was viewed as amodel project.

The Swallow story was told over seven hours and wasbased on a daily journal the product-team leader had keptthroughout the project's history. Key to his account was agreat emphasis on extemal activity. This activity was presentfrom the start of the project and v^s maintained throughoutits life.

644/ASa December


The project leader had engaged top management from thevery beginning. As the project was being discussed by theOperating Committee, a top management committee thatmanaged new-product development projects, he met oftenwith committee members. He worked to convincecommittee members that they should go with arevolutionary design rather than a simple upgrade. Heinsisted they had the talent and motivation to make a greatproduct and make it quickly. When the decision was madeto go ahead with the project, the leader asked the presidentof the company and the vice president of R&D to come tothe first meeting to explain the importance of the product tothe company and to communicate their support of the team.The leader remained in close contact with the presidentthroughout the project.

Lateral communication was also quite high throughout theproject. When the team had put together some preliminarydesign ideas, the leader set up a design review with multiplerepresentatives from R&D. The review was set up to "makesure we weren't going off in crazy directions." The reviewgroup met with the team for two weeks to make sure allquestions were asked and numerous points could bedebated.

Three people from the manufacturing function were broughtin at the very beginning to serve as experts to the team.These people were responsible for reacting to thecomponents the engineers were considering to determinewhich could be obtained easily and which would be difficultto find. The three men also continuously evaluated the easeof manufacturing the computer and acted as liaisons toothers in their area, providing information on productprogress and details of what would be expected ofmanufacturing and when it would be expected. As theleader explained, "Coordinating was a major task throughoutthis project. We could never have gotten it out the door ifteam members hadn't been checking on other parts of thecompany and pushing to make sure things happened." Heexplained that things had to be checked and double checkedconstantly. People would say that components had beenordered, but then no one would follow up on why they hadnot shown up when promised.

Each week a meeting of the team was held to which manyoutside groups were invited, including purchasing,manufacturing, production planning, diagnostics, andmarketing. People were informed of progress and changes.Notes for each meeting were taken and typed up on thecomputer, where they could be accessed by these othergroups. All members of the Operating Committee wereautomatically sent copies of tfie minutes.

Not all members of the team were engaged in extemalactivity. The team was housed in a somewhat remotelocation, making it difficult for outsiders to disturb theengineers at work. This meant extra work for those teammembers who were engaged in external activity. Many ofthose people spent days away from the team site, returningregularly to report on "letters from home."


Toward the end of the project, external activity increasedeven more as the product moved to manufacturing, andsates and marketing got more active. There were many,many meetings to train manufacturing people, to makestatus reports as schedules slipped and tensions rose, andto settle big fights about when the product was "reallyready" to be shipped. When the product was finally "let go"the team was credited with a big win for the company,though at the time many team members were tooexhausted to take in what they had been able to achieve.

The Devices team. We heard a very similar story of extemalactivity at another company from a team that was making avep/ different type of product. The team leader had justbeen promoted to manager, and he was explaining thehistory of his last project and what he had learned from it.He had been with the company for many years, starting outas a design engineer. Apparently, in the "olden days"products were not that complex and time pressures werealmost nonexistent. Thus, one person could do most of thedesign work, and the product would be easily sold to atechnically sophisticated customer. Then things began tochange. The products became more complex, andcompetitive pressures meant that the company had toreduce the time to market from five years to one year. Themarket for the products had grown to include anonengineering market that required more help from sales,service, and marketing. The leader decided that his nextproject would have to involve a team of engineers thatsomehow split the work into smaller pieces and that workedmore closely with other parts of the company.

The leader then described how he and five other engineerswho thought they "had the necessary technical experienceand could work well together" started meeting to figure outhow to design a device in parallel rather than alone. At thetime, many people in the company did not support this typeof procedure. The leader found a vice president who wasintrigued with the idea and who provided some initialfunding.

The leader and one member described their roles as "verymuch the lobbyists for the product." They spoke of "selling"both the product and their team approach throughout theorganization. They reported that there was no formalprocedure for gaining support as they needed it. You need"personality, good connections, and the perseverance tokeep working to keep everyone informed and keep everyonecontributing what they've committed to," said the leader.The leader had biweekly discussions with the vice president,mostly concerning the major issues around resources.

The leader reported that although each member of the groupcould go out and coordinate for his or her part of the project,this was not thought to be a good idea. He stated that it isdifficult to communicate with all the other groups. You"need to know where to go, who to see, how to talk tothem," and not everyone can do that. Besides, the leaderreported, some design engineers work better withuninterrupted time, so specialized liaison activity wasorganized.



This project was rated a big success, capturing a largemarket share. The leader attributed much of the success tothe team's real belief in the product. This belief, he thought,got translated into an ability to convince others that theproduct was a real priority, in the willingness to push othersto follow through, and in sustaining the team and its alliesthrough the difficult periods.

The Beaver project. This project was done in the sameorganization as the Swallow project, but it was not such agreat success. Although the computer ended up beingshipped within a reasonable amount of time, it experiencedmany delays that the company considered costly. Also, themanager of the project was demoted during the computer'sdevelopment because management felt that someone elsecould manage the project more effectively. What is sofascinating about this project is that the changes made toimprove this team mirror activities that were "naturally"undertaken in the high-performing teams.

Using numerous interviews with the team leader and severalkey informants over the life of this project, we put togethera history of it. This was the only project we followed as itevolved rather than examining retrospectively. The projectleader was from R&D and preferred working on technicalissues. Throughout the first several interviews with him, hisdescriptions of work included interaction with topmanagement, his team, and other members of R&D aroundproduct specifications. When asked about how the teamwould coordinate with manufacturing and marketing, theleader replied, "Have common goa! statements. !n otherwords, we have goals around time to market, we have goa!saround product cost, we have goa!s around productfunctionality, we have goals around product quality,"

Unfortunately, the reliance on goal statements to handleinterdependencies among functions did not work very well.About six months into the project another team leader wasput on the team to "help, as an additional resource." Whenasked why he was put on the team, the new memberreplied, "I've been a catalyst. . . . to sell the product to otherpeople, to be the champion of the product outside." Inaddition, the new person was brought in to implement achange in the product's components and to improve theschedule.

When asked why he could do what the team leader couldnot, the new member replied, "He just didn't know how tomake it happen. That was the part he had the biggeststruggle with. Because he kept sort of saying we ought todo it this way and nothing would happen. . . . I helped worksome of the resource issues and 1 try to help out when hehas a problem." In contrast, the team leader would talkabout his frustration with getting things done with othergroups, While the new "resource" worked his vertical andhorizontal interactions informally, pushing team membersinto their own direct negotiations with outsiders, the teamleader often escalated problems up the hierarchy. As theleader stated, "Every time you elevate something itbecomes a great deal more painful.... As things getelevated it causes more work because you're dealing with


people who don't ur>derstand the fundamental issues andyou have to educate them, you're dealing with more players. . . you've got to deal with engineering and marketing VPs,so you end up with the pyramid growing and growing."

But the issues with marketing and manufacturing keptgrowing. There was a stalemate with marketing over howmany configurations of the product had to be made tosatisfy a broad range of customers, and this issue gotescalated up the hierarchy. Then, when the product wasfurther along, major difficulties with manufacturingdeveloped. At that time, the team leader was demoted anda new leader from manufacturing was put in place. As theold leader tried to make sense of what had happened, heremarked, "Typically What happens, and I'm just finding out. . . there is no justice and you're never right—justunderstand that and you'll be all set. ! try to deal with thingsin a rational way, manufacturing at this company is notnecessarily that nature of beast. Many things that wereplanned for Beaver six, eight, ten months ago weren't everreally committed by manufacturing."

Thus, Beaver got back on track, thanks to the additional"resource," who worked on many of the management andmarketing issues, and to the new team leader frommanufacturing, who was able to get real commitments frommanufacturing. Both additions pushed team members tohandle extemai negotiations rather than relying onhierarchical intervention. In the end. Beaver filled a big gap inthe company's product line and sold very well.

The ID printer project. This project was set up to 'take thecompany into a new sphere of activity by combining twoexisting technologies into a revolutionary new product. Theteam was put together with very talented engineers and anexperienced team leader. It was a complete failure. Thestory was related by a division manager who had supportedthe project at its inception but gradually became verydisenchanted with its progression. He reported onconversations he had had with numerous people throughoutthe organization about the project.

The team effort had started with a big send-off, with topmanagers expressing their hope that the team could takethe company in a new direction. The team set out to workwith ample financial and personnel resources and six monthsto "play in the sandbox," meaning to experiment withdifferent technologies and product designs. The work wasdivided up so that different team members worked onseparate aspects of the new design and got input fromdifferent parts of the organization and outside world. Theteam had a great deal of information about "the marketpotential, all the amazing applications that we could use thistechnology for," and "the way we could put this thingtogether." The team v as very excited and worked longhours, getting as many new ideas as possible out on thetable. After nine months, however, team members could stillnot agree on what exactly the first product was going to do,nor what components t h ^ were going to use in its design.Their response to being behind schedule was to continue toseek additional ideas and change the product design. At first

648/ASQ, Deoember 1992


top management was quite lenient, but then "the pressurestarted." While management tried to get commitments onschedules, the team was always late. A design would comein. and then it would be followed by another design thatostensibly had some "new and exciting feature that wouldmake the product even better." Design and redesigncontinued for several more months before the project wasfinally killed. During this time, the team leader was oftenhard to find. He often avoided meetings with top managers,always promising that the solution was "coming soon,"when he did meet them. The division manager wasextremely disappointed in the outcome and later puttogether another team to continue the work. He never didunderstand why the team could not get past the continuoussearch for a better product design.

Members of the ID printer project seem to have beenunable to move beyond exploration to commitment to agiven design. They seemed to get stuck in this explorationmode and continued to modify concepts rather than thinkingthrough their implications for a final product. While theleader tried to move the group, the engineers kept comingup with more ideas and changes. As the volume andcomplexity of ideas mounted, the group becameoverwhelmed with alternatives, and closure on a final designbecame even harder.

These four cases are similar to many others. Collectively, thecases combine to illustrate several patterns, suggestinghypotheses that might be tested with the larger data set.

Hypotheses

One clear pattern was that all successful teams had highlevels of ambassadorial activity. The Swallow team and theDevices team both had leaders that started to exhibitambassadorial activity even before the project teams wereput together, and they continued that activity with otherteam members throughout the life of the team. Theless-successful Beaver team imported a high-levelambassador to help it progress when the leader seemedunable to implement many of his ideas. Top managersconsidered this activity essential, and when it was notforthcoming, they imposed it on the team. The one failedteam, ID, started out with top management support,.but itwas not garnered through ambassadorial activity. Also, asthe team continued to fall behind, the leader avoided topmanagement. He did not engage in impression managementbut, rather, let others form their own perceptions of why theteam was not getting the design out on time.

This positive relationship between performance andambassadorial activity is also posited in the literature atmultiple levels of analysis. Dutton and Ashford (1992) arguedthat the "issue selling" of middle managers helps to set theagenda of top management and is related to organizationalperformance. Issue selling is also linked to enhancingindividual visibility, perceptions of personal competence, andindividual power (Burgelman, 1983; Dean, 1987). Our datashow these same effects at the group level. Ambassadorialactivity is also similar to impression management, wherebyan individual, group, or organization tries to control extemal


images of itself (Schenkler, 1980; Tedeschi, 1981). The dataand literature combine to form the basis for our firsthypothesis:

Hypothesis 1: Team performance is positively associated withambassadorial activity.

The cases also suggest that task-coordinator activity plays apivotal role in team success. The Swallow team engaged inhigh levels of interaction with R&D, getting feedback on theproject design, and with manufacturing, getting information,coordinating, and obtaining feedback. Coordination andinformation sharing were considered major tasks throughoutthe project's history. The Devices team actually organizedspecialized liaisons who would carry out task-coordinatoractivity for the whole team. Again, this activity wasconsidered essential for team success. The Beaver team didsuch a poor job of coordinating with manufacturing that theleader had to be demoted in order to increase the linkagesbetween the team and this area. The activity was consideredso important that when it did not naturally occur, it wasmandated. Finally, the ID team never got to the point ofcoordinating. Members spent so much time designing theproduct that they did not get to the specialized tasks ofobtaining feedback and negotiating around innplementation ofthe design.

Existing research from numerous sources provides separatesupport for a positive relationship between task-coordinatoractivity and team success. The boundary-spanning literatureshows positive relationships between the frequency ofexternal lateral communication and performance (Tushman,1977, 1979; Allen, 1984). While studies of feedback seekingat the individual level show more equivocal results (Ashfordand Cummings, 1983), there is still enough empirical andtheoretical evidence to suggest the following hypothesis:

Hypothesis 2: Team performance is positively associated withtask-coordinator activity.

The cases seem to indicate that scouting activity isnegatively related to team performance. Although these dataseem to contradict the literature postulating the need forunderstanding current market and technology trends (Zurgerand Maidique, 1990). they do not. Rather, the data coincidewith the notion that teams need, at some point, to movebeyond exploring the possibilities of a product toimplementing product design and exploiting a chosentechnology and market niche (Roberts, 1977; Gersick, 1988).Scouting cannot become an enduring pattern. The ID projectfailed because the team did not stop scouting activity andcommit to a plan. Rather, the team always looked for abetter technology or a better design to win in the market.Whiie other teams did engage in scouting activity, thisactivity apparently took place very early in the life of theproject and often involved a subset of the team, either in theearly phases of the project or before the team was even puttogether. Teams that were, unable to move from this generalscanning to a defined product design early on seem destinedto fail; We thus predict the following:


Ctaânizational Teams

Hypothesis 3: Team performance is negatively associated withprolonged scouting activity.

In all the cases we reviewed, successful teams were deeplyengaged in communications with outsiders. This observationis consistent with research by information-processingtheorists, who focus on the amount of information that theteam acquires from its environment (Lawrence and Lorsch,1967; Thompson, 1967; Galbraith, 1977). Theinformation-processing approach is normative, positing thatgroups must match their information-processing capability tothe information-processing demands of the externalenvironment (Nadier and Tushman, 1988). Support for thisapproach comes from studies showing that teams carryingout complex tasks in uncertain environments need highlevels of external interaction to be high performing(Tushman, 1977, 1979; Gresov, 1989; Ancona and Caldwell,1992). In research and development teams, for example,frequency of communication within the teams has shown norelation to performance, while increased communicationbetween the teams and other parts of the laboratory wasstrongly related to project performance (Allen, 1984).High-performing teams also showed higher frequencies ofcommunication with organizational colleagues outside ofR&D than their low-performing counterparts. Given that thenew-product development task in high-technology firms iscomplex, and the environment uncertain, this suggests thefollowing:

Hypothesis 4: Team performance is negatively associated withisolationism, or very low levels of external activity.

Ail four hypotheses are consistent with a recent qualitativestudy of five consulting teams in a state-run service thatexamined team strategies toward the environment and theirlink to performance (Ancona, 1990). Three strategies wereidentified there: informing, parading, and probing. Informingteams remain relatively isolated from their environment,although they intend to inform others of their planssometime in the future; parading teams are primarilyconcerned with making their own activities visible tooutsiders; and probing teams actively engage outsiders.

Although Ancona's (1990) study uses teams with a differenttask and a different organizational environment thannew-product teams, its findings are consistent with theproposed hypotheses. For example, the highest performers,as rated by top management one year after team formation,were the probing teams, who combined upward persuasionwith lateral feedback seeking, coordination, and testing ofsolutions. This suggests that performance is enhancedthrough ambassadorial and task-coordinator activity(hypotheses 1 and 2). In addition, parading teams, whichwere very visible in the environment as they continuouslymonitored external trends and conditions, were not viewedas high performers because they did not achieve highresults. This suggests that performance is hindered bycontinuous scouting activity (hypothesis 3). The lowestperformer was the informing team, which remained isolatedfrom both its external task environment (the customer) andtop management. Its members were incorrect either inthinking that they had sufficient information to complete the


task by themselves or in assuming that their performancewould be evaluated independent of the process and visibilitythey displayed during the task's lifetime. This findingsuggests that performance is negatively related toisolationism, supporting hypothesis 4.

Internal and External Activity

As group theory evolves, it will also be important to take intoaccount how internal and external activities affect oneanother, as well as how they affect performance. While atarge amount of previous work has established linksbetween intemal group process and performance, the causa!direction between these variables is open to question (Lottand Lott, 1965; Chapman, Bell, and Staw. 1986), and therelationship between external activities and what goes onwithin the group has received less direct study.

Although our data from the 38 interviews were equivocalabout the relationship between these variables, mostprevious research suggests that extemai activities interferewith the development of effective internal operations. Theinternal cohesion that exists under conditions of groupthink(Janis. 1982, 1985) can promote extemai stereotyping andeliminate the import of external information that mightdamage group consensus. The intergroup literature (Smith,1983, 1989) also suggests that a negative relationshipbetween internal and external activities can develop. Groupscan be underbounded—having many external ties but aninability to coalesce and motivate members to pull togethertheir external knowledge—or overbounded—having highinternal loyatty and a complex set of internal dynamics butan inability to reach out to the external world (Sherif, 1966;Alderfer, 1976). Finally, the conflict literature predictsintensified intragroup conflict when group members collectinformation from outsiders with different goals, cognitivestyles, and attitudes (Shaw, 1971; Schmidt and Kochan. 1972).

Yet not alJ studies indicate a negative relationship. In a studyof eight task forces. Gersick (1988) found that groups shifttheir basic assumptions and operating procedures over time.Her study suggests that teams may deal with internal andexternal demands sequentially, first acting on initialinformation from the environment in isolation and thenemerging to get further feedback and information fromoutsiders. Ancona's (1990) study of five consulting teamsalso suggests that the point at which the group is observedmay be important. Her results suggest that, in the shortterm, those teams that concentrated on ambassadorial andtask-coordinator activity suffered from low levels of cohesionand high (evels of disorganization; over time, as theysatisfactorily interacted with their environment, internalprocesses improved. For scouting teams, short-termcohesion and clarity of roles and goals eventually gave wayto intema! dissension as interaction with the extemaienvironment proved unproductive. This is consistent withHomans' (1950) observation that complexity and conflict inthe extemai environment will be replicated within the groupwhen Infomnation from the environment is imported into thegroup. We propose a final hypothesis based primarily on this

work:

652/ASQ, DecOTiber 1992

Hypothesis 5: External activity is related to internal cohesion andinternal task processes.More specifically, we posit that ambassadorial andtask-coordinator activities are positively associated withinternal cohesion and smooth intemal processes, while theassociation is negative for scouting activity, but thishypothesis must be qualified. As we argued earlier, externalprocess is likely to influence team performance. Asubstantial amount of literature has also established linksbetween performance and these internal variables. We arepostulating an additional direct link between external activityand internal process. Because external activity also affectsinternal process through its impact on performance,however, this direct link will be hard to isolate, observe, andmeasure. Hypothesis 5 is thus somewhat tentative.

RELATIONSHIPS BETWEEN EXTERNAL ACTIVITIES ANDGROUP PROCESS AND PERFORMANCE

This part of the study tests the impact of external teamactivities and strategies on team performance and internalprocesses, using the same sample of 45 new-product teamsdescribed earlier. It relies primarily on the questionnaire data,although it uses a broader set of questions than describedearlier. Thus, this section reports on measures, analysis, andresults of hypothesis testing.

Measures

In addition to the measures of boundary activities, threeother sets of variables were assessed: frequency ofcommunication, team performance, and internal processes.

Frequency of communication. Team members were askedhow often they communicated with nonteam individuals inthe marketing, manufacturing, engineering, andproduct-management functions during the previoustwo-week period. They responded on 6-point scales,anchored by 1 ~ "Not at all" and 6 = "Several times aday." Because these functional groups had different namesin the different companies, the questionnaires were modifiedto conform to company-specific terminology. Because thesefour groups represented everyone with whom teammembers would normally communicate in their work, theseresponses were averaged. Team scores were computed byaveraging the individual scores (X = 2.54, s.d. = .78).

There has been a debate in the literature as to whetherorganization members can accurately assess communicationpatterns. Bernard and his colleagues (Bernard, Killworth, andSailer, 1980; Bernard et al., 1984) claimed that asking peoplehow much they talk to others produces inaccurate results.Individuals forget some communications and overcountothers. Other researchers have countered this criticism byshowing that organization members may not reproduceexactly the communications that have just occurred, buttheir bias is in the direction of long-term patterns ofcommunication (Freeman, Romney, and Freeman, 1987).Respondents are thus not actually answering the question,"Who did I speak to in the last two weeks" but "In a typicaltwo-week period, with whom am I likely to have spoken."Because our focus was on this more general pattern of


communication, the broad measure of communicationfrequency we used is appropriate.

Team performance. Following the stakeholder view oforganizations, team performance cannot be seen as asimple, unidimensional, construct. First, as Goodman, Ravlin,and Schminke (1987) argued, group measures ofperformance must be both fine-grained and related to thetask. If a group is responsible for completing an innovativenew product, for example, then performance measuresshould include the group's innovativeness not just generalmember satisfaction. Second, Giadstein (1984) found thatevaluations of group performance differ depending onwhether group members or managers are doing the rating.This supports Tsui's (1984) contention that differentconstituencies often have different definitions ofperformance and suggests that ratings from these variousconstituencies be included in a study of group performance.The multidimensionality of performance may be particularlyrelevant when the outputs of a group are being assessed(Hackman and Walton, 1986). Finally, group researchers havefound a lag effect between group process and performance(Giadstein. 1984; Bettenhausen and Murnighan. 1985;Ancona, 1990). This suggests that processes exhibited attime 1 may affect performance at time 1 or time 2, Goingone step further, certain processes may have a positiveeffect in the short term but turn out to be negative overtime. Thus, this research examines the impact of externalprocesses on several measures of performance, as rated byboth group members and top management, in the shortterm-and at project completion.

We used subjective ratings of performance, albeit frommultiple sources. While more objective ratings such aspercent over budget or actual sales have been suggested(Clark and Fujimoto, 1987). it was our experience that thesenumbers were often interpreted through subjective lenses,were influenced by numerous external factors not under thecontrol of the team (e.g., an economic recession), and wereless important than managerial ratings in determiningpromotions, future job assignments, and performanceevaluations.

Performance data were collected at two points. The initialperformance ratings and measures of boundary activitieswere collected when the teams had been together anaverage of 10 months. Thus, the teams were approximatelymidway through their originally scheduled process (time 1).A second wave of performance data was collectedapproximately two years later, when teams had completedtheir projects or were in the final stages (time 2). Topdivision managers were asked at both time 1 and time 2 toassess the teams in their company on dimensionssuggested by Hauptman (1986). Using 5-point Likert scales,they rated each team's efficiency, quality of technicalinnovations, adherence to schedules, adherence to budgets,and ability to resolve conflicts. Although the sample size wassmalt, the performance items at each time were subjectedto a principal components analysis to identify underlyingpatterns. Using the data collected at time 1, two factorsemerged. One factor was defined by the questions about


Orgamzationri Teams

adherence to budgets and adherence to schedule. Weaveraged those two items to form a single variable welabeled budgets and schedules. The remaining three itemsloaded on the second factor; we averaged them to create avariable we called efficiency of innovation. A different factorstructure emerged when we analyzed the performancemeasures collected at time 2. One factor, which we labeledinnovation, was defined solely by the single item about thequality of technical innovations produced. The second, whichwe labeled team operations, was defined by the remainingfour items, which were averaged to form a scale score. Toassure comparability of the performance ratings acrosscompanies, we adjusted individual scores for each team bysubtracting the mean of the scores assigned to teams withinthat company. Thus, the performance scores were adjustedfor company and the overall means set to zero.

One additional performance measure was collected at tinne1. Team members were asked in the questionnaire to ratethe performance of their teams on six dimensions, includingefficiency, quality, technical innovation, adherence toschedules, adherence to budgets, and work excellence.These items were completed by all individuals, allowing usto do a principal components analysis of the items. Theanalysis yielded a single factor. A score, which we calledteam rating, was assigned to each team by averaging theindividual members' scores (alpha = .83) (X = 3.63, s.d. =.38),

Internal processes. As Goodman, Ravlin, and Schminke(1987) have noted, task-oriented group processes may bemore directly related to performance than more traditionalaffect-based measures of group process. Members'perceptions of the teams' work-related group process wereassessed with three items. Individuals used 5-point Likertscales to indicate the team's ability to develop workableplans, define goals, and prioritize work; high scores definedbetter perceived processes (see Hackman, 1983). Since aprincipal components analysis yielded a single underlyingfactor, these three items were averaged to form a singlescale (alpha = .86), A score was then computed for each ofthe 45 teams by averaging the individual scores of themembers of the team (X = 3.69, s.d. = .43).

Many of the arguments suggesting a link between externalactivities and internal process use cohesiveness as anindicator of process. This more traditional affect-basedmeasure was assessed using Seashore's (1954) four items,which ask members (1) how willing they are to defend oneanother from criticism; (2) how well they help each other;(3) how well they get along; and (4) the extent to which theystick together. These four items were averaged to form asingle scale (alpha = .91). A score was then computed foreach of the 45 teams by averaging the individual scores ofthe members of the teann {X = 3.7, s.d. = .81).

Analysis

If the composite variables created by combining individuals'assessments of the teams' performance, intemal process,and group cohesiveness represent distinct characteristics ofa group, small within-group variances and large


between-group variances should be observed. In order totest the extent to which these variables were distinct fromone team to the next, an analysis of variance was conductedusing teams as the independent variable and the threecomposite scores as dependent variables. Results indicatethat between-group variance was significantly greater thanwithin-group variance for all three variables {team-ratedperformance: p < .001; task process: p < .01;cohesiveness: p < .001), suggesting that the compositescores reliably represented and distinguished teams.

Regression analysis was used to determine the relationshipsbetween both frequency of communication and type ofactivity and performance. This analysis allowed us toevaluate the usefulness of activity types, independent of theamount of communication, in predicting performance.Correlation analysis was used to examine the relationshipsbetween internal and external activity. Finally, additionalinsight into the relationships among strategies, performance,and internal process was assessed via analysis of varianceacross the four strategies.

Results

The correlational analysis indicates some significantrelationships among external activity sets and performancemeasures. Table 2 shows means, standard deviations, andcorrelations among all variables. Ambassadorial activities arepositively correlated with task-coordinator activities andnegatively with scouting activities. Some performancemeasures are also interrelated, perhaps because themanagements' ratings of performance were adjusted by asubtraction of the company mean to ensure comparabilityacross companies. The two time-1 management ratings ofperformance were positively related and were both relatedto the time-2 measures of innovation. The teams' ownratings of performance, however, were unrelated tomanagement's ratings of performance.

Of central interest in this part of the study are therelationships between the boundary activities andperformance and between the boundary activities andinternal processes. There is clear support for hypothesis 1,which posited a positive relationship between ambassadorialactivity and performance. Correlational analysis showsambassadorial activity to be positively associated withmanagements' ratings of teams' ability to meet budgets andschedules Uime 1) and of team members' ratings of theirown performance (time 1). There is also some relationshipwith managements' ratings of innovation (time 2). Table 3reports regression equations for each of the fiveperformance measures. Here, ambassadorial activity is onlypositively related to adherence to budgets and schedules(time 1), thus partially supporting hypothesis 1.

Hypothesis 2 posited a positive relationship betweentask-coordinator activity and performance. Correlationalanalysis shows that higher levels of task-coordinator activitywere associated with higher ratings (two significant at p <.05; two marginal at p < .10) on all four management-provided performance measures (time 1 and time 2). The


OrgairizathHial Twims

Table 2

Corrdations among AH Varirirfes

1. Frequency ofcommunication

2. Ambassadcmalactivities

3. Task-coordinatoractivities

4. Scoutingactivities

5. Guard acti>nties6. Intemal process7. Cohesiveness8. Budgets &

schedules(time!)

9. Efficiency ofinnovation(time 1)

10. Team rating(time 1)

11. Innovation(time 2)

12, Teamoperations(time 2)

• p < .10; " p < ,05;'

Mean

2.61

.14

.02

.02-.153.693.30

.01

.03

3.63

.00

.00

•~p < .01.

S.D.

.86

.47

,43

.38

.38

.44

.46

.95

-79

.38

.64

,63

1

-

--13

.14

.00- .05--16--54«~

.24"

.2T

- . 4 8 ™

,13

.14

2

-

.26^

- . 3 8 " "-,01

,47"*,24*

-45""

.14

, 3 1 "

-23*

.02

3

_

-02.07

-.01- .10

.30"

.22»

-07

.43"*

.21"

4

-,18

- .22*- .23"

- . 4 3 * "

- . 3 6 " "

- .19"

- . 4 1 * "

-02

5

-- ,12

,25"

-,08

,04

.18

-.11

-.08

6

-,39""

,12

,17

.58""

,09

.08

7

.02

-.04

.89""

-.03

.09

8

-

42" *

.03

.53""

.17

9

-

.15

.46""

.27

10

.03

.00

11

-

.20

regression analysis shows that task-coordinator activity is nolonger related to time-1 management-rated performance,perhaps due to multicolinearity. The relationship withinnovation (time 2) remains very strong, showing partialsupport for hypothesis 2.

Hypothesis 3 posited a negative relationship betweenprolonged scouting activity and performance. Because mostof our sample is measured welt into the design period, weconsider any reports of scouting to reflect prolongedscouting. Correlational analysis shows that scouting activitieswere negatively associated with ratings on the time-1measures of budgets and schedules and innovationefficiency and the time-2 measure of innovation and teamratings of performance. The regression results also supportthe hypothesis, with scouting activities being negativelyrelated to budgets and schedules (time 1). efficiency ofinnovation (time 1). and innovation (time 2).

Hypothesis 4 posited a positive relationship between theamount of external activity and performance. Correlationalanalysis shows that frequency of communication wasmarginally associated with time-1 meeting budgets andschedules and efficiency of innovation and highly negativelyrelated to team ratings of performance. The regressionanalysis shows only a marginal positive relationship betweenfrequency of communication and adherence to budget andschedule (time 1) and a significant and negative relationshipwith the teams' ratings of their own performance. Thusthere is no support for hypothesis 4.

Hypothesis 5 posited relationships between internalprocesses and external activities. Correlational analysisshows groups with high levels of ambassadorial activitiesreported higher ratings of intemal process and marginallyhigher ratings of cohesiveness than groups with tow levels


TabJe3

Regresshm BesuHs*

Variable

Frequency ofcommunication

Ambassadorialactivities

Task-coordinatoractivities

Scouting activities

Adjusted rF

• p < .10; —p < .05; ***p* Entries are standardized

Budgets &schedules

(time 1)

.25*

. 3 1 -

.19- . 3 2 -

.315.24—

< .01.

Efficiency ofinnovation

(time 1)

.17

-.04

.21- . 3 8 -

.122.26*

regression coefficients.

Teamrating

(time 1}

- . 4 6 —

.18

.09-.12

.244.61 —

Innovation(time 2)

,06

- .05

.45—- . 4 4 —

.294.87—

Teamoperations

(time 2)

.11

-.01

.20

.01

.000.51

of this activity. No significant relationships betweentask-coordinator activity and internal process were found.Significant and negative relationships between scoutingactivity and cohesiveness and between scouting activity andinternal task processes were observed, again showingsupport for hypothesis 5. This pattern of results generallysupports hypothesis 5 and illustrates the complexity ofthe relationships between external activity and groupprocess.

An analysis of variance using group strategy as theindependent variable revealed significant differences incommunication patterns across strategies, as shown in Table4. Those teams following the ambassadorial strategy and theisolationist strategy have the lowest frequency ofcommunications with outsiders and. although not shown,members of these teams spend the lowest percentage oftheir time with outsiders (12 percent and 10 percent,respectively). In contrast, technical-scouting andcomprehensive teams have the highest frequency ofexternal interaction and spend the highest percentage oftheir time with outsiders (18 percent and 16 percent,respectively). More in-depth analysis (Ancona and Caldwell,1990) shows that ambassadorial activity (found in teamsusing ambassadorial and comprehensive strategies) mayshow low levels of external communication becauseindividuals have high levels of communication withtop-division and top-corporate management. Thisconcentrated communication requires a lower frequency ofinteraction than the more diffuse communication patternsfound with strategies involving scouting and task-coordinatoractivity. The latter reported high levels of interaction acrossmanufacturing, marketing, and R&D.

The external strategies show different relationships toperformance. While both ambassadorial and comprehensivestrategies are related to achieving budgets and schedules inthe short term (time 1), only the comprehensive strategy ispositively related to performance over time (innovation, time2). Both the technical-scouting teams and the isolationistteams have poor performance over time.


Organiza^nal Teams

Table 4

Means off Dependent Measures aiôss

Strategies

CharacteristicsAmbassadorialTask coordinatorScouting

Frequencyof communication

Internal processesTask processCohesiveness

PerformanceBudgets & schedule

(time 1)Efficiency of innovation

(time 1)Team rating (time 1)Innovation (time 2)Team operations

(time 2)

• Common superscriptst Significance refers to

Four Strategtos*

TechnicalAmbassadorial scouting

.52^-,17^- .20^

2.17^

3.89^3,14^

.25^'

- .033.61- .08 '

- .22

W]Xh\n the tablemain effects fron

- , 2 1 ^,24''.38''

2 69(ib

3.49''2.59^

^ - .38^

- .173.40

3 Q-]ab

- , 04

Isolationist

-,07"^-.32^

,18 ^

2.47^

3.70 *2.80^

-.37^

- .013.45-.42^

.13

reflect no significant differences in post-hoc1 one-way analyses of variance using size as

Comprehensive

.33^

.36^-,2r

3.24^*

3.78^"2.64''

.83^^

.393.55

37''

.24

mean comparisons.a covariate.

p t

<.O1<.O1<.O1

<,05

<.1O<,05

<.O1

N.s.N.S.<.1O

N.S.

DISCUSSION

External Activities

We began this study with relatively few models of externalactivity. Moving from qualitative information gathering toquantitative clustering techniques, we were able to form atypology of four main activities and team strategies directedtoward the environment. Ambassadorial activities provideaccess to the power structure of the organization asmembers promote the team, secure resources, and protectthe team from excessive interference. Task-coordinatoractivities provide access to the workflow structure; they areaimed at managing horizontal dependence. Throughcoordination, negotiation, and feedback, these activitiesallow for a tighter coupling with other organizational units,often filling many of the gaps left by formal integratingsystems. Scouting activities provide access to theinformation structure; they are aimed at adding to theexpertise of the group. These activities allow the group toupdate its information base, providing new ideas abouttechnologies and markets.

Interestingly, not all teams engaged in all forms of externalactivity but. instead, developed several distinct strategiestoward outsiders. Some teams specialized in one particularactivity, e.g.. ambassadorial and technical-scouting teams;others showed relatively low scores on all activity sets, e.g.,isolationist teams; and still others were more likegeneraiists, combining ambassadorial and task-coordinatoractivity, e.g., comprehensives. These four group strategiesare very similar to those found by Ancona (1990). and,together, these two studies provide support for the validityof these strategies as representing real pattems found inorganizational teams today. In addition, this taxonomy

^9/ASQ, December 1992

provides a basis for categorizing groups and differentiatingtheir forms and the implications of those forms.

These extemal activities and strategies need to be added toour conceptualization of group process to represent morefully the wide range of what group members actually do.Perhaps by adding extemal initiatives to the already-presenttask and maintenance activities we can then go on to predicthow extemal activities will influence the relationshipsbetween group process and other constructs. It also remainsfor future research to delineate more clearly the antecedentsof external strategies. Hackman and Walton (1986)suggested examining the organization's treatment andsupport of groups. Top managers may approach certainteams and not others. Some organizational environmentsmay reinforce some strategies and not others. Ancona(1990) has suggested that the team leader also plays a largerole in determining a team's extemal strategy. The questionis how do leaders choose how to direct their team?

External Activities, Performance, and Internal Process

The results of our hypothesis generating and testing point totwo major findings. First, the pattem of external activities isa better predictor of performance than simply the frequencyof communication. Second, and on a more speculative note,cycles may play a role in team behavior, so that a strategythat works in the early life of a group may not supportpositive performance over time.

Whiie the information-processing theorists have long pointedto frequency of communication as a key representation ofextemal activity, this study points to the advantages ofexamining the type of external activity. Our results indicatethat frequent communication is marginally related tomanagements' ratings of performance at time 1 but not attime 2 and negatively related to members' ratings ofperformance. The activity sets are more strongly related tomanagements' ratings of performance than to the frequencyof communication.

The comparisons of group strategies also illustrate thecontribution of a content-based rather than afrequency-based approach to external interaction. If we wereto look at external frequency alone, teams following anisolationist and ambassadorial strategy would be groupedtogether as low-frequency communicators, and teamsfollowing a technical scouting and comprehensive strategywould be grouped together as high-frequencycommunicators. Yet such a classification would mask greatdifferences between teams. The technical scouting teams,for example, show some of the lowest performance andintemal process scores, while those following thecomprehensive strategy show some of the highest scores.Thus, while information-processing theorists would postulatethat teams with a high frequency of communication wouldbe better performers given the uncertainty and complexity ofthe high-technology, new-product team task andenvironment, this is not always the case, as shown.

This study also tells us something about the types ofextemal activity needed for team effectiveness. The activity



sets related to the different performance measures insingular ways. Ambassadorial activities were related tomanagements' ratings of the teams' adherence to budgetsand schedules at time 1. Task-coordinator activities,however, were positively related to managements' ratings ofinnovation at time 2. In contrast to this pattern, very generalintelligence gathering—defined by a high level of scoutingactivity—was associated with low managerial ratings ofperformance at both time 1 and time 2.

Analyzing the team strategies provides a little more insightinto the complex pattern of relationships between externalactivity and performance. While ambassadorial activitiesseem to be key to performance, their effect over the longterm seems to hold only in combination withtask-coordinator activities. Pure ambassadorial teams andcomprehensive teams move along on budget and scheduleat time 1. At time 2, however, the ambassadorial teams arepoor at innovation and team operations, while thecomprehensive teams continue to be the highestperformers. Too much scouting activity, as shown bytechnical-scouting teams, is related to low performanceratings.

A very different pattern emerges when the team rates itsown performance. Teams feet that they perform well whenthey concentrate their efforts internally; they revealperceptions of performance that are negatively related to thefrequency of communication and positively related to cleargoals and priorities and high cohesiveness. Thus, predictorsof management and team-rated performance are verydifferent.

Ratings of internal task processes and cohesiveness showyet a different pattern. White frequency of communicationalone and scouting activities are negatively related tocohesiveness and task processes, ambassadorial activitiesare positively related to internal measures. Comprehensiveteams that combine ambassadorial and task-coordinatoractivities show positive internal task processes but lowcohesiveness compared with the pure ambassadorial groups.Finally, isolationist teams show relatively high internalscores.

This set of results suggests that managing the powerstructure alone produces many positive outcomes in theshort term. Ambassadorial teams can move along quickly onbudget and schedule and develop effective task processesand cohesiveness. In terms of managerial ratings ofperformance, however, it is only teams that manage boththe power structure and the workflow structure that are ableto maintain performance over time. The comprehensiveteams move ahead quickly on budgets and schedule in theshort term, but they also manage to produce the mostinnovative products over the course of the developmentprocess. This finding is similar to that reported by Zurger andMaidique (1990). These teams pay a price, however, in thatthey are less cohesive than pure ambassadorial teams in theshort term.

The technical scouting teams fare the worst on alldimensions. These teams have low performance ratings


from top management at time 1 and time 2 and low scoreson intemal task processes and cohesiveness. It may be thatsuch teams constantly react to genera! environmental dataand become unable to commit to producing a specific endproduct at a specific time. Or it may be that high levels ofscouting activity somehow reduce the efforts teammembers put into the more performance-relevant externalactivities or into building effective intemal processes.Another possibility is that the organization has given theteam too general an assignment or expects too much, thuspushing the team to keep searching for the answer to animpossible question. This search can eventually undermineinternal cohesion.

The isolationist teams, which most neglect external activity,also do quite poorly in terms of their performance at time 1and time 2, yet their scores on internal task processes andcohesiveness are quite high. These teams may simplyconcentrate on internal activities and be oblivious to thenegative feedback from other parts of the organization.

These different patterns of external activity, performance,and internal behavior suggest that some teams may entercycles of activity early on that reinforce themselves overtime and determine team outcomes. In this interpretation,external activity, intemal processes, and performance allinteract and influence one another. For ambassadorial teamsthis cycle might start with favorable managerial evaluationsthat give team members confidence and facilitate thecreation of effective task processes and cohesiveness. butthis cohesiveness, in turn, may cause teams to lessenextemai activity (Janis, 1982; Katz, 1982) and thus to fail toget the necessary feedback and do the coordination neededto produce an innovative product, This would have been thecase with the Beaver project if people outside the team hadnot forced it to increase task-coordinator activity with otherparts of the organization. Comprehensive teams, by contrast,begin both ambassadorial activity and task coordination early,thus providing feedback to the team from other groups fromthe very start. These networks pave the way for coordinationand negotiation throughout the project, as seen in theSwallow and Devices teams. The constant interaction withoutsiders and their divergent views and values may,however, account for these teams' inability to achieve theinternal cohesion and efficiency achieved by theambassadorial teams.

The technical-scouting teams seem to enter a cycle ofcomplexity and negative performance. Our observations ofthese teams suggest that they enter a pattern of explorationthat they cannot escape. Whether due to external conditionsor their own beliefs, members of these teams continue tosearch for new approaches to their product. In turn, thisexploration brings large amounts of conflicting informationinto the team, thus requiring complex internal interaction. Asthe complexity grows, both extemai and intemal frustrationdevelops, leading to negative managerial ratings ofperformance and internal conflict. Again, this cycle may feedon itself so that these teams cannot move on to reduce thecomplexity, define a product, and move it through theorganization.


Cênizational Teams

Finally, the isolationist teams create impermeable boundariesthat allow them a cocoon-like existence. Internally, theywork efficiently and cohesively, a cycle that reinforces thebenefits of ignoring the outside world. The question here iswhat would happen to this cycle if these teams eventuallywere to realize how others view their work. Thecohesiveness they've cultivated could dissipate if the outsideworld views their product unfavorably.

Organizations' increasing reliance on teams to developproducts and processes requires that teams span traditionalorganizational boundaries. Furthermore, teams are beinggiven increasing responsibility to define, market, carry out.and transfer their work. These new responsibilities requireextensive external interaction with organization membersoutside the group's boundaries. The study of such groupsthus must reach beyond the traditional research boundariesof groups and their internal processes to the widerorganizational arena in which the group does its work.

REFERENCES

Afderfer, Clayton P.1976 "Boundary relations and

organizational diagnosis." InM. Meltzer and F. Wickert(eds.), HumanizingOrganizational Behavior:142-175. Springfield, IL:Charles C. Thomas.

Aldrich, Howard E., and DianeHerker1977 "Boundary spanning roles and

organization structure."Academy of ManagementReview. 2: 217-230.

Allen, Thomas J.1984 Managing the Flow of

Technology: TechnologyTransfer and theDissemination ofTechnological Informationwithin the R&D Organization.Cambridge. MA: M I T . Press.

Ancona, Deborah G.1987 "Groups in organization:

Extending laboratory models."In C. Hendrick (ed.), AnnualReview of Personality andSocial Psychology: GroupProcesses and IntergroupProcesses: 207-231. BeverlyHills, CA: Sage.

1990 "Outward bound: Strategiesfor team survival in theorganization." Academy ofManagement Journal. ^ :334-365.

Ancona, Deborah G., and David F.Caidwell1988 "Beyond task and

maintenance: Definingextemal functions in groups."Group and OrganizationStudies. 13: 468-494.

1990 "Beyond boundary spanning:Managing externaldependence in productdevelopment teams." Journalof High TechnologyManagement Research, 1:119-135.

1992 "Demography and design:Predictors of new productteam performance."Organization Science, 3:321-341.

Ashford, Susan J., and Larry L.Cummings1983 "Feedback as an individual

resource: Personal strategiesof creating information."Organizational Behavior andHuman Performance, 32:370-398.

Ashford, Susan J., and Anne S.Tsui1991 "Self regulation for managerial

effectiveness: The role ofactive feedback seeking."Academy of ManagementJournal, 34: 251-280.

Bales, Robert F.1958 "Task roles and social roles in

problem solving groups." In E.Maccoby, T. M. Newcomb.and E. L. Hartley (eds.),Readings in SocialPsychology. 3d ed.: 437-447.New York: Holt, Rinehart andWinston.

1983 "SYMLOG: A practicalapproach to the study ofgroups." in H. H. Blumberg,A. P. Hare. V. Kent, and M. F.Davies (eds.). Smalt Groupsand Social Interaction. 2:499-523. New York: Wiley.

Benne, Kenneth D., and PauiSheats1948 "Functional roles of group

members." Journal of SocialIssues, 2: 42-47,

Bernard, H. Russell, Peter D.Killworth, David Kronenfeld, andLee Sailer1984 "On the validity of

retrospective data: Theproblem of informantaccuracy." In B. S. Siegel,A. R. Beais, and S. A. Tyler(eds.). Annual Review inAnthropology: 495-518. PaloAlto, CA: Stanford UniversityPress.

Bernard, H. Russetl, PeterKiltworth, and Lee Sailer1980 "Informant accuracy in social

network data, IV: Acomparison of clique-levelstructure in behavioral andcognitive data." SocialNetworks, 2: 191-218.

Bettenhausen. Kenneth, and J.Keith Murnighan1985 "The emergence of norms in

competitive decision-makinggroups." AdministrativeScience Quarterly. 30:350-372.

Bourgeois, L. Jay, III, andKathleen M. Eisenhardt1988 "Strategic decision processes

in high velocity environments:Four cases in themicrocomputer industry."Management Science. 34:816-835.


Burgelman,1^3 "A process nDodel of internal

oirporate venturing in tinedivereified major firm."Administrative ScienceQuarterly. 28: 223-244.

Chapman, Jenntter A., Nancy E.Bell, and Barry M. Staw1986 "The managed thought: The

role of self justification andimpression management inorganization settings," in H.Sims and D. Gioia (eds.). TheThinking Organization:191-214. San Francisco:Jossey-Bass.

Clark. Kim B., and Takahirofujimoto1967 "Overlapping problem solving

in product development."Working Paper, HarvardUniversity, Graduate School ofBusiness Administration,

No. 87-048.

Dean, James W., Jr.1987 Deciding to Innovate:

Decision Processes in theAdoption of AdvancedTechnology. Cambridge. MA.:Ballinger.

Dougherty,. Deborah1987 "Nevtf products in old

organizations: The myth ofthe better mousetrap insearch of the beaten path,"Unpublished Ph.D,Dissertation. MassachusettsInstitute of Technology, SloanSchool of Management.

Dutton, Jane E., and Susan J.Ashford1992 "Selling Issues to top

management." WorkingPaper, University of Michigan,School of BusinessAdministration,

Freeman, Umon C , A. KimballRomney. and Sue C. Freeman1987 "Cognitive structure and

infonnsnt accuracy."American Anthropologist, 89:310-325.

Galbraith, Jay R.1977 Organization Design. Reading

MA:

Gersick, Connie J. G.1988 "Time and transition in v>/ork

teams: Toward a new modelof group development."Academy o1 ManagementJoumal 31: 9-41.

1989 "Maricing time: Predictabietransitions in task groups."Academy of ManagemerrtJoumal, 32: 274-309.

Ginzel. Unda E., Roderick M.Kramer, and Robert 1. Sutton1393 "Organizaitional images as

negotiated accounts: Adynamic conception ofimpression managenrient." InL. L. Cummings and B. M.Staw (eds.). Research inOrganizational Behavior, vol.15. Greenv^rich. CT: JAI Press(forthcoming!.

Giad&tein, Deborah1984 "Groups in context: A model

of task group effectiveness,"Administrative ScienceQuarterly, 29: 499-517.

Giadstein, C^borah, and James B.Quinn1985 "Making decisions and

producir>g action: The twofaces of strategy." InJohannes M. Pennings andAssociates (eds.).Organizationa) Strategy andChange: 198-216. SanFrancisco: Jossey-Bass.

Goodman, Paul S., ElizabethRavlin, and Marshall Schnrtinke1987 "UnderstarKiing groups in

organizations." In L, L.Cummings and B, M, Staw(eds), Researdi inOrganizational Behavior. 9:121-175. Greenwich. CT: JAIPress.

Gresov, Christopher1989 "Exfrforing fit and misfit with

multiple contingencies."Administrative ScienceQuarteriy. 34: 431-453,

Hackman, J. Richard1963 'The design of work teams,"

In J. W. Lorsch (ed.).Handbook of OfganizationalBehavior: 315-342,Englewood Cliffs. NJ:Prentice-Hall,

Hackman, J. Richard, and Richarde. Watton1986 "LeacSng groups in

organizations." tn P, Goodman(«!,). Designing EffectiveWork Groups: 72-119. SanFrancisco: Jossey-Bass.

Hambrick, DonaW C.1983 "Some tests of the

effectiveness and functionalattributes of Mites andSnow's strategic types."Academy of ManagementJoumal, 26: 5-26,

Homans, ^1950 The Human Group. New York:

Hareourt Brace Jcwanovich,

Isenberg, Daniel J.1986 "Group polarization: A critical

review and meta-analysis."Journal of Personality andSocial Psychdogy, 50:1141-1151.

Janis, Irving L.1982 Groupthink. Boston: Houghton

Mifflin.1985 "Sources of error in strategic

decision making." inJohannes M. Pennings andAssociates teds,).Organizational Strategy andChange: 157-197. SanFrancisco: Jossey-Bass.

Katz, Ralph1382 "The effects of group

longevity on projectcommunication andperformance," AdministrativeScience Quarterly, 27:81-104.

Katz, Ralph, and Michael LTushman1979 "Communication pattems.

project performance, and taskcharacteristics: An empiricalevaluation and integration inan R&D setting."Organizationai Behavior andHuman Performance, 23:133-t62.

Kerfinger, Fred N.1973 Foundations of Behavioral

Research, 2d ed. New York:Holt, Rinehart and Winston.

Lawrence, Paul R., and Jay W.Lorsch1967 Organization and

Environment: ManagingDifferentiation and Integration.Homewood, IL;

Hauptman,1986 "The influence of task type on

the relation betweencommunication andperformance: The case ofsoftware development." R&DManagement, 6: 218-244,

Lott, Albert J., and Bernice E. Lott1965 "Group cohesiveness as

interpersonal attraction: Areview of the relationshipswith antecedent andconsequent variables,"Psychoiogica! Bulletin, 64:259-309.

Louts, Meryl R.1980 "Surprise and sense making:

What newcomers experiencein entering unfamiliarorganizational environments."Administrative SdenceQuarterly, 25: 226-251.

Malone, Thomas W.1987 "Modeling cooixlination in

•organizations and markets."Management Science. 23:1317-1332.

664/ASQ, December 19S2


Mites, Robert E., and Charles C.Snow1978 Organizational Strategy,

Structure and Processes. NewYork: McGraw-Hill.

Nadler, David A., and Michael LTushman1988 Strategic Organization Design:

Concepts, TooSs, andProcesses, Glenview, IL:Scott. Foresman.

Nemetfi, Christine J.1986 "Differential contributions of

majority and minorityinfluence." PsychologicalReview, 93: 23-.32.

Pfeffer, Jeffrey1981 Power in Organizations,

Marshfield, MA: Pitman,

Philip, Hugh, and Dexter Dunphy1959 "Developmental trends in

small groups." Sociometry,22: 162-174.

Porter, Michael E.1980 Competitive Strategy, New

York: Free Press.

Roberts, Edward B.1977 "Generating effective

corporate innovation."Technology Review,October/November: 21-40.

Schein, Edgar H.1988 Process Consultation: Its Role

in Organization Development,vol, 1, Reading, MA:Addison-Wesley,

Schenkler, Barry R.1980 Impression Management.

Monterey, CA: Brooks/Cole.

Schmidt, Stuart M., and ThomasA. Kochan1972 "The concept of conflict:

Toward conceptual clarity."Administrative ScienceQuarteriy, 17: 359-370.

Seashore, Stanley1954 Group Cohesiveness in the

Industrial Work Group. AnnArbor, Ml: Institute for SocialResearch, University ofMichigan,

Shaw, Marvin1971 Group Dynamics: The

Psychology of Small GroupBehavior. New York:McGraw-Hill.

Sherif, Muzafer1966 in Common Predicament:

Social Psychology ofIntergroup Conflict andCooperation, Boston:Addison-Wesley,

Smith, Kenwyn K.1983 "An intergroup perspective on

individuai behavior." In J. R,Hackman, E, E, Lawler, andL W. Porter (eds.).Perspectives on Behavior inOrganizations: 397-407, NewYork: McGraw-Hill.

1989 "The movement of conflict inorganizations: The jointdynamics of splitting andtrlangulation," AdministrativeScience Quarterly. 31: 1-20,

Tedeschi, James T.1981 Impression Management

Theory and SocialPsychological Research, NewYork: Academic Press,

Thompson, James D.1967 Organizations in Action: Social

Science Bases of^ministration Theory. NewYork: McGraw-Hill,

Tsui, Anne S.1984 "A role set analysis of

managerial reputation."Organizational Behavior andHuman Performance, 34:64-36,

Tushman, Michael L1977 "Special boundary roles in the

innovation process,"Administrative ScienceOuarterly, 22: 587-606.

1979 "Work characteristics andsubunit communicationstructure: A contingencyanalysis," AdministrativeScience Ouarterly, 24: 82-98,

Weick, Karl E.1979 The Social Psychology of

Organizing. 2d ed, Reading.MA: Addison-Wesley.

Zurger, Billie Jo, and Modesto A.Maidique1990 "A model of new product

development: An empiricaltest." Management Science,36:


Documents

Bridging the Boundary: External Activity and … the Boundary: External Activity and Performance in Organizational Teams Deborah G. Ancona Massachusetts Institute of Technology David