product development tensions: exploring contrasting styles...

* Acadsmy of Management Jaurnal2002, Vol. 45, No, 3, 546-564,

PRODUCT DEVELOPMENT TENSIONS:EXPLORING CONTRASTING STYLES OF

PROJECT MANAGEMENT

MARIANNE W. LEWISM. ANN WELSH

University of Cincinnati

GORDON E. DEHLERThe George Washington University

STEPHEN G. GREENPurdue University

Successful product development requires managing tensions—coping with fluctuatingcontingencies to foster innovation and efficiency. To investigate this challenge, weexplored the nature, dynamics, and impacts of contrasting project management styles.Our conceptual framework details emergent and planned styles. Following 80 projectsover two-year periods, we find that these styles offer disparate but interwoven ap-proaches to monitoring, evaluation, and control activities; use of these activitiesfluctuates over time; a paradoxical blend of styles enhances performance; and uncer-tainty moderates project management-performance relationships.

Tensions cannot be eliminated because they are in-herent in [product development] activities and helppower the innovation process. (Dougherty. 1996:430)

Product development is a potentially vital sourceof competitive advantage and organizational re-newal, but success is often elusive. Product devel-opment projects spark conflicts between externaldemands and internal competencies, needs forspontaneity and for structure, desires for changeand for stability (Dougherty, 1996; Jelinek &Schoonhoven, 1990). Project teams strive to de-velop technical knowledge and achieve commer-cial objectives hy building innovative capacity. Yetsuccess also requires efficient execution to keepprojects on schedule and within budget.

For over 30 years, researchers have sought todiscover the contingencies that impact product de-velopment performance. Recent reviews (e.g..

Many thanks to Editor Tom Lee and three anonymousreviewers, as well as to Suzanne Masterson and SteveWheelwright, for their insightful comments on earlierversions of this article.

This research is based on work supported by the Na-tional Science Foundation under Grant No. SES-8519415and by the Center for Innovation Management Stndies,Lehigh University, under Grant No. 429551-802, The au-thors thank the University ot'Gincinnati and Purdue Uni-versity for their support.

Brown & Eisenhardt, 1995; Montoya-Weiss & Cal-ontone, 1994) have catalogued studies of projectfactors (such as novelty and scope), organizationalfactors (like structure and team composition), andenvironmental factors (such as market potential),stressing their mixed results. This literaturehighlights the tensions surrounding product devel-opment: project managers must cope with multi-ple—and often conflicting and fluctuating—con-tingencies as they seek to foster innovation andefficiency. Yet critics also note the limitations of acontingency approach (e.g.. Dougherty, 1996; Pen-nings, 1992). To make sense of seemingly conflict-ing findings, researchers need more encompassingtheoretical frameworks. Such frameworks mightenahle insights into how managers adjust their be-haviors and organizations to changing contingen-cies. From a practical standpoint, however, contin-gency studies need greater precision. By virtue ofoperating at high levels of analysis, research oftenlacks an action orientation and offers limited man-agerial guidance.

Quinn (1988) proposed an alternative approachthat may foster simultaneously broad and detailedunderstandings of management. He claimed thatresearchers might extend contingency studies byexamining more general managerial capabilities orstyles. Management style denotes an underlyingmode of thinking and behaving that in turn pro-motes a specific repertoire of actions that managers

546

2002 is, Welsh, Dehier, and Green 547

draw upon in contexts of varying complexity anduncertainty. Historically, the research literature haspolarized management styles, framing them in termssuch as convergent/divergent thinking, transactional/transformational leadership, theory X/theory Y. Like-wise, product development researchers often stressdivergence between an emergent and a planned style.Some advocate an emergent, fluid style of manage-ment as a means to foster creativity and improvisa-tion (e.g., Doughtery, 1992; Moorman & Miner, 1998).Others prescribe disciplined planning as a way tofocus and speed project efforts (e.g.. Wheelwright &Clark, 1992; Zirger & Maidique, 1990). Yet propo-nents of both styles also recognize the need for bal-ance. Case studies richly depict the interplay of emer-gent and planned styles [e.g., Jelinek & Schoonhoven,1990). Laufer's (1997) case collection, for example,illustrates how exceptional project managers are ca-pable of moving iteratively between contrasting stylesin response to contending demands.

To date, however, product development litera-ture lacks clear conceptualizations and correspond-ing measures of project management styles. Indeed.Dougherty (1996) called for constructs that capturetbe tensions of product development. Prescriptionsfor emergent and planned styles may reflect thesetensions. Rather than another normative model, re-searchers need measures that help assess disparate,but potentially complementary, ways of managing.She also argued that more effective constructswould build from tbe activities of project manage-ment. Identifying varied approaches to specificmanagerial activities would enhance research pre-cision and relevance. Without such constructs, re-searchers and practitioners are left questioning thenature, dynamics, and impacts of project manage-ment styles: What are the characteristics of con-trasting styles? Do managerial styles vary as aproject progresses? And, does style matter, signifi-cantly affecting project performance? This studyaddresses these issues to contribute a fine-grainedtreatment of project management styles and theirroles in product development.

VARIED PRESCRIPTIONS FOR PROJECTMANAGEMENT

In our tough, dynamic, and demanding world, "ei-ther/or" approaches are no longer viable . . . today'schallenges of fast change and uncertainty require"hoth/and" approaches to thinking and working.Success demands that project leaders adopt both aninward and outward orientation, develop both for-mal and informal procedures. (Ian Mitroff [Laufer,1997: xi])

Increasingly, researchers claim that managingtensions is crucial to product development success.Jelinek and Schoonhoven's (1990) case studies de-pict how conflicts between stability and changerequire continuous interplay between preexistingplans and evolving understandings. Dougherty(1996) concurred, expanding the set of tensions toinclude freedom/responsibility, emergent/deter-mined, outside/inside. Revisiting Burns andStalker (1961). she theorized that organic ap-proaches (for instance, fluid roles, team autonomy)foster employee commitment and creativity, andthat mechanistic approaches (for instance, formalmethods, centralized authority) keep projects ontarget and linked to wider organizational goals. Yetproduct development studies often appear polar-ized within divergent streams, whicb offer seem-ingly contradictory prescriptions (Benghozi. 1990;Brown & Eisenhardt, 1995). Broadly speaking, workin these two streams advocates either an emergentor a planned style of project management. An emer-gent style involves facilitating team members' cre-ativity, flexibility, and improvisation, and aplanned style provides managerial discipline anddirection.

The emergent stream is attracting increased at-tention, tapping insights from eclectic studies ofjazz, sports, firefighting, and theatre. Social con-struction and cognition theories help researchersdepict processes of "sense making" and improvisa-tion (e.g., Dougherty, 1992; Weick, 1998). Moormanand Miner (1998) explained that improvisation oc-curs as teams learn experientially. building knowl-edge as they work through uncertainty and experi-ment with diverse product designs. Yet researchersalso call for balance, as planning and structure offera frame within which creativity occurs. Comparingthe need for routines within product developmentteams to the needs of jazz ensembles. Pasmorenoted this: "The lesson for us . . . is that flexibilityis always possible without control, but that randomactivity does not produce jazz; it produces noise"(1998: 563).

Proponents of a planned style, in contrast, stressdisciplined problem solving. Illustrative studieshighlight the value of extensive planning and cal-culated implementation (e.g., Zirger & Maidique,1990) and the need to methodically expand or-ganizational knowledge (e.g., Clark & Fujimoto,1991). A practitioner orientation aids discovery ofthe "best practices" to guide the developmentprocess; formal reviews are an example. Many,however, also endorse balance or "subtle control,"calling for managers to set explicit goals for auton-omous teams (e.g., Wheelwright & Clark, 1992).Eisenhardt and Tabrizi (1995), for instance, used

548 Academv of Management Journnl June

notions from the emergent stream to theorize howplanning methods (like setting frequent milestonesand having powerful leaders) foster improvisation.

Rather than sharpening the focus of researcii ef-forts, work accumulating in these streams has led toincreasingly diffuse conceptualizations of projectmanagement styles. Dougherty (1996) highlightedtwo particular concerns. First, she criticized exist-ing constructs for failing to capture the tensions ofproduct development. Case studies depict contrast-ing styles in rich but ambiguous terms, and surveyresearcb offers explicit but oversimplified either/ormeasures. For instance, cases often describe howmanagers who are performing well learn to accom-modate tensions, moving back and forth betweendisparate styles to meet fluctuating demands (e.g.,Jelinek & Schoonhoven, 1990; Laufer. 1997).

Most survey measures, however, position con-trasting styles at the opposite ends of a single con-tinuum, portraying them as mutually exclusive.McDonough and Barczak (1991), for example, in-vestigated two styles of control. They asked. Isproduct development accelerated by an emergent,participative style, which offers teams considerablediscretion, or by a more directive style, where man-agers oversee project details? Although results sup-ported a more emergent style, it should be notedthat they used a single measure of control to assessboth styles. Shenhar and Dvir's (1996) survey studyoffered very different results. They conceptualizedvaried styles of control, communication, and eval-uation and obtained results suggesting that thegreater a project's scope and uncertainty, the morea planned style (that is, firm control, formal docu-mentation, external reviews) enhanced perfor-mance. Yet their measures also placed the stylesalong simple continua. According to Quinn, moreinsightful conceptual frameworks would detail sty-listic differences, but "move from a traditional,schismogenic, either/or approach to a both/and ap-proach, thus making it possible for us to see man-agement behavior in genuinely new ways" (1988:85). Corresponding measures would assess eachstyle separately, rendering it empirically possiblefor managers to go back and forth between contrast-ing styles of a behavior.

Dougherty's (1996) second concern is reflected incalls for more activity-based constructs (e.g., Pen-nings, 1992; Shenhar & Dvir, 1996). Jelinek andSchoonhoven (1990) criticized the lack of concep-tual development around key managerial activities.Most studies analyze a narrow range of activities ordepict broad product development strategies. Theresult is a limited understanding of what firms ac-tually do to manage projects and how differentstyles affect performance. Building constructs from

the activities of project management would enablea more thorough treatment of the "processes, dy-namics, and events underlying this kind of innova-tion" (Dougherty, 1996: 424).

Practitioner-oriented texts, in particular, oftenstress the importance of three activities: projectmonitoring, evaluation, and control. Unlike envi-ronmental contingencies (like market fluctuationsand competitive intensity), these activities denoteelements of product development over which man-agers have some influence—and pose a largely un-tapped area of study (Rosenau & Moran, 1993). Inaddition, these managerial activities are ubiqui-tous, impacting project execution from idea gener-ation and product design through production andproduct launch (Wheelwright & Clark, 1992). Yetmanagers' approaches to each activity may varydepending on whether they apply an emergent or aplanned style exclusively or move between thesestyles.

Styles of Approaches to ProjectManagement Activities

At the same time the firms we studied must stressinnovation, they also require great precision andpredictahility to enable them to master complex de-sign and production details. The implications of thisdilemma must be worked out in management style.(Jelinek & Schoonhoven, 1990: 56)

To elaborate the natures of emergent and plannedstyles, we propose a conceptual framework thatjuxtaposes the respective approaches they entail toproject monitoring, evaluation, and control. Figure1 summarizes each activity set. As noted previ-ously, proponents of both styles have called forbalance but emphasized contrasting approaches.Building from the literature, we purposefully po-larize these styles to accentuate their differencesand similarities and to facilitate subsequent opera-tional definition.

Emergent and planned styles reflect varied man-agerial assumptions and goals. Advocates of anemergent style view product development as inher-ently ambiguous (Dougherty, 1996). In this light,project managers seek to facilitate team members'improvisation and experiential learning, helpinginnovation emerge from the bottom up. Barrett(1998) explained that, whether a team is engagedin theatre or in product development, a fluid andspontaneous approach to management sparkscreativity. As teams discover new opportunitiesand challenges, managers may encourage mem-bers to monitor their evolving understandings,gather information from diverse sources, and/orexperiment with new designs as needed. Hence,

2002 Lewis, Welsh, Debter. and Green 549

FIGURE 1Contrasting Styles of Project Management

Emergent Style Planned Style

Monitoring Understandings

Tracking emerging project skills,

knowledge, and focus (e.g.. Barrett,

1998; Mirvis, 1998; Van de Ven &

Polley, 1992)

Monitoring

Monitoring Milestones

Comparing progress to

predetermiQed standards and goals

(e.g., Eisenbardt & Tabrizi, 1995;

Schoonhoven, 1990)

Information Gathering

Boundary spanning and

environmental scanning by team ^ ^ H Evaluation •

members (e.g., Ancona & Caldwell,

1992; Dougherty, 1992)

Formal Review

Systematic assessment of project

^ by top managers and/or review

board (e.g.. Rosenau & Moran, 1993;

Wheelvifright & Clark, 1992]

Participative Gontrol

Team autonomy, discretion, and

decision-making accountability ^ |

(e.g., Dougherty, 1996; McDonough &

Barczak, 1991; Weick, 1998)

• Control •

Directive Control

Close managerial involvement in

^ ^ project details, feedback, and

adjustments (e.g., Shenhai & Dvir,

1999; Eisenhardt & Tabrizi, 1995)

monitoring, evaluation, and control are looselycoupled activities.

In comparison, proponents of a planned stylepresume that product development is more predict-able and rational, best managed as a top-down pro-cess. Successful senior executives and project man-agers provide discipline and structure, striving todirect team efforts and link project and organiza-tional goals (Wheelwright & Clark, 1992). From thisperspective, monitoring, evaluation, and controlactivities appear tightly coupled, interwovenwithin a systematic cycle. Milestones help a teammethodically track a project; formal reviews enablecritical assessments that inform major decisions(for instance, continue/terminate project; resourceallocation); and directive control allows managersto adjust project resources and objectives as neces-sary (Rosenau & Moran, 1993).

Juxtaposing emergent and planned styles ofmonitoring, evaluation, and control accentuatesthe disparate, but potentially complementary,approaches to managing they imply. Monitoringactivities track projects in progress. In applica-tions of an emergent style, monitoring under-standings gauges progress in terms of team mem-bers' experiential learning—that is, how theirproject focus, knowledge, and skills are developingas they interact with others and with the productitself (Van de Ven & Polley. 1992). Barrett (1998)theorized that, in innovation processes, sense mak-ing is often retrospective, occurring through andafter action. Monitoring understandings aids retro-spective sense making by helping team memherscomprehend how work is unfolding across variedactors and by fostering a greater mindfulness ofothers' expertise. Teams "are able to innovate and

550 Academy of Management Journal June

elaborate on ideas with the assurance that they areoriented to a commonplace" (Barrett, 1998: 612).Monitoring understandings also helps team mem-bers translate their experiences into "superordinategoals, or a shared vision, and in interpersonal rela-tions, into efforts to build empathy and mutualunderstanding" (Mirvis, 1998: 59).

In contrast, a planned style promotes monitoringexplicit milestones. According to Jelinek andSchoonhoven (1990), milestones ideally convert aproject strategy into analyzable technical, budget-ary, and time-related objectives. Interim goals,schedules, and tests offer managers clear indica-tions of a project's progress and a common frame ofreference, but also serve as guides for team mem-bers {Eisenhardt & Tabrizi, 1995; Wheelwright &Clark, 1992). Such predetermined standards mayaid team coordination and ensure that projects donot absorb unnecessary resources.

Dougherty (1996) claimed that varied approachesto monitoring reflect conflicts between emergenceand determination. If firms rely strictly on devel-oping knowledge from the bottom up, projects maynot build on one another. Likewise, if teams areforced to follow top-down plans, they might notexplore novel scientific or commercial opportuni-ties. Emergent and planned styles may help teamscope with these tensions, offering complementarymeans of tracking and spurring progress. Monitor-ing understandings gauges whether team membersare developing new skills and a shared vision,whereas milestones guide these developments to-ward the firm's strategic objectives.

Hypotbesis la. Monitoring understandings (acomponent of an emergent style) and mile-stones (as in a planned style) are distinct yetinterrelated approacbes to project monitoring.

Evaluation denotes the appraisal of a project'svalue (for instance, its fit with market demands andorganizational competencies). Applying an emer-gent style, team members evaluate their project bygatbering information from external sources. Thisis a potentially vital boundary-spanning activity.By interacting with those outside the project, teamsmay gauge organizational support and manage oth-ers' impressions of the project (Ancona & Caldwell,1992). Depicting functional groups as "thoughtworlds," Dougherty (1992) claimed that insightsfrom different functions might clarify how otherorganizational actors envision a project, often re-vealing unexpected interpretations. Informationgathering also may spur improvisation. As Crossanexplained, "The environment will teach you if youlet it" (1998: 595). Scanning the environment fordivergent views fosters a sense of more compli-

cated alternatives, helping teams break out of exist-ing mind-sets to see a project in new light.

A planned approach entails more systematicevaluation, or formal review. Wheelwright andClark (1992) stressed the value of reviews con-ducted by senior executives, divisional reviewboards, or cross-functional groups. As gatekeepers,reviewers determine whether a project will contrib-ute significantly to its firm's product portfolio, isready for transfer fTom R&D to manufacturing, orshould be terminated. Within the cycle of tightlycoupled planning activities, monitoring milestonesmay inform such appraisals by triggering alarmsthat a project is off course or exposing the need toconnect the project to wider organizational goals(Rosenau & Moran, 1993: 166). Formal reviews alsooffer teams feedback that may delineate theirproject's scope (Shenhar & Dvir, 1996),

These contrasting activities correspond to diver-gent/convergent tensions of product development.Teams must look outward to cull fresh insights andalternative interpretations. Yet projects must alsoremain focused and build from existing organiza-tional competencies (Dougherty, 1996). In re-sponse, emergent and planned styles of evaluationmay spur the iterations between divergent andconvergent thinking that fuel innovation (Mirvis,1998). Information gathering encourages teammembers to explore opposing views, whereas for-mal reviews contribute assessments that help man-agers meld a project with their firm's vision ofproduct development.

Hypotbesis ib. Information gatbering (reflect-ing an emergent style) and formal review (aplanned style) are distinct yet interrelated ap-proacbes to project evaluation.

Control practices influence the locus of projectdecision making. An emergent style fosters a par-ticipative approach. Participative control offersteams the freedom to challenge existing ideas andsolve problems regarding product design and theirown tasks (McDonough & Barczak, 1991). Accord-ing to Dougherty (1996), product development of-ten requires greater commitment and task varietythan other types of projects. Participative controlmay build trust and enable team members' roles toevolve over time. In addition, improvisation thrivesas members "are authorized to paraphrase, embel-lish, and reassemble their prevailing routines ex-temporaneously . . . [when] they are encouraged tothink while doing rather than being guided solelyby plans" (Weick, 1998: 549).

A planned style promotes directive control.Project and senior managers seek to provide teamsguidance and support. In their research, Shenhar

2002 Lewis, Welsh, Dehler, and Green 551

and Dvir observed this: "Managers were concernedwith finishing the project on time, within budget,and according to the initial plan and design" (1996;616). To meet these goals, they remained closelyinvolved in project details. Others have proposedthat directive control spurs action, allowing man-agers to offer feedback and adjust resources asneeded (e.g., Eisenhardt & Tabrizi, 1995). This ac-tivity builds from the other planned approaches, asmonitoring milestones and formal reviews helpmanagers recognize when and where to intervene.

These control activities mirror tensions betweenfreedom and structure—the need to encouragecommitment and creativity and to ensure timelyand appropriate action (Benghozi, 1990). Research-ers suggest that "subtle control" is the blend ofparticipative and directive control apparent in"heavyweight" teams, which provide members dis-cretion and the support of a powerful leader (e.g.,Wheelwright & Clark, 1992). As Dougherty furtherexplained, "It is desirable that people feel free togenerate ideas, create possible solutions to prob-lems, and experiment with various courses of ac-tion. It is also desirable that people feel responsibleto work toward common goals" (1996; 430).

Hypothesis lc. Participative control (reflectingan emergent style) and directive control (aplanned style) are distinct yet interrelated ap-proaches to project control.

The Dynamics of Project Management Styles

Researchers have proposed that project manage-ment styles are dynamic, evolving to meet fluctu-ating demands (e.g., Ancona & Caldwell, 1992).Managers, for instance, may alter their approachesin response to new resource allocations, changes inmarket demand, progress by competitors on similarprojects, or novel scientific discoveries (Van deVen & PoUey, 1992). Yet, if styles fluctuate, re-searchers must pay close attention to when theyanalyze project management. Studies using cross-sectional or retrospective data may provide onlypartial or even inaccurate views of relationshipsbetween project management and performance(McDonough & Barczak. 1991). In contrast, longitu-dinal data might reveal whether—or how—manag-ers adjust their actions to contextual changes (Pen-nings, 1992).

Existing literature suggests that activity stylesfluctuate in different ways. Emergent activities maybe crucial when a team is entering the unknown(Barrett, 1998). Van de Ven and Polley (1992) ar-gued that an emergent style is most prominent atthe start of a project, when teams seek to cultivate

new ideas. Then, as the product development pro-cess moves toward transforming ideas into an im-plemented reality, emphasis on experiential learn-ing declines. Likewise, Katz (1982) found thatinformation-gathering activities decreased duringlengthy projects, potentially isolating team mem-bers and impeding their performance. In sum, ef-forts to foster improvisation may wane as a projectshifts from idea generation to manufacturing andeventual product launch.

Hypothesis 2a. The use of emergent activitieswill decrease as projects progress.

A planned style is likely to be more stable, but itsrole may change over time—helping reduce uncer-tainty in the early stages, then pushing a project tocompletion. Initially, planned activities serve toclarify project details. Close managerial control andfrequent monitoring are vital, because the earlier aproduct is conceptualized, the faster its develop-ment (Clark & Fujimoto, 1991). Planned approachesare also more prone to inertia (Van de Ven & Polley,1992). Over time, these managerial activities evolveinto reliable routines, such as the use of systematicmilestones and formal reviews. According toRosenau and Moran, '"imitation, standards, androutine" lie at the core of planned monitoring andevaluation (1993: 66). Coupled with directive con-trol, these activities provide a consistent course ofaction for project teams.

Hypothesis 2b. The use of planned activitieswill remain stable as projects progress.

Project Management-Performance Relationships

This section addresses our concluding question:Does style matter? Project performance is multifac-eted, encompassing the need for innovation andefficiency. As Bowen and colleagues (1994) ex-plained, a successful product development projectreflow contributes technical knowledge that en-hances both the product and future R&D efforts andproduces a final product capable of attaining itsprojected market share. Simultaneously, projectsmust remain on time and within budget. Yet fewstudies have investigated the effects of project man-agement styles on multiple facets of performance(Brown & Eisenhardt, 1995; Jelinek & Schoon-hoven, 1990). Moreover, Doughtery (1996) claimedthat new, activity-based constructs are needed toidentify specific levers of performance. In response,we employed our conceptual framework to examinehow contrasting managerial activities impact innova-tion and efficiency.

Relationships between project management and

552 Academy of Management journal

performance may vary with managerial style. Inparticular, researchers have proposed a positivelink between an emergent style and innovation. AsMoorman and Miner (1998) noted, the primary goalof this approach is to facilitate improvisation.Emergent activities help teams build technicalknowledge by exploring novel scientific conceptsand help them to achieve their commercial objec-tives by experimenting with alternative product de-signs. Information gathering potentially fuels cre-ativity as teams grapple with divergent insightsprovided by external departments (Dougherty,1992), whereas participative control and monitor-ing understandings may foster the experientiallearning of team members (Van de Ven & Polley,1992).

Hypothesis 3a. An emergent style of manage-rial activities will be positively related toproject innovation (building technical knowl-edge and achieving commercial objectives).

In contrast, a planned style may enhance projectefficiency—its principle objective (Shenhar & Dvir,1996). Managers seek to ensure that teams havesufficient support and remain targeted on projectgoals. According to Wheelwright and Clark (1992),schedule- and budget-based milestones keep teamsaware of their scarce resources. Likewise, formalreviews guide decision making regarding resourceallocation. Directive control also may defineproject boundaries, reducing the likelihood ofwasteful explorations and costly errors (McDon-ough & Barczak, 1991). In sum, planning activitiesprovide discipline that may motivate action andbolster efficiency.

Hypothesis 3b. A planned style of managerialactivities will be positively related to projectefficiency (keeping projects on time and withinbudget).

Studies suggest that project uncertainty moder-ates project management-performance relation-ships (e.g., Eisenhardt & Tabrizi, 1995; Moorman &Miner, 1998; Shenhar & Dvir, 1996). Examiningchange in project uncertainty may extend this workand offer further insights into the dynamics of man-agement styles. As Dougherty (1996) explained,fluctuations in uncertainty intensify product devel-opment tensions, straining managers' capacity togauge critical contingencies and meet conflictingdemands. Hence, varying uncertainty surroundingthe technology and/or the commercialization of adeveloping product pose an exceptional test ofmanagerial capabilities.

According to Pennings (1992). changes in uncer-tainty interact with managers' perceptions and ac-

tions. Exploring these interactions may enable agreater understanding of how managers cope withnew opportunities and problems. For instance,technical uncertainty may increase when teamsmake dramatic scientific breakthroughs, and com-mercial uncertainty rises if teams recognize thattheir firm lacks the ability to take their design tomarket (for instance, it is inexperienced in manu-facturing or selling that type of product). Copingwith such variations demands greater managerialflexibility. The improvisation fostered by an emer-gent style may help teams work through heightenedambiguity. On the other hand, more structured,planned activities may offer a sense of order thatreduces team members' anxieties.

Hypothesis 3c. Change in project uncertaintymoderates project management-performancerelationships.

METHODS

For this exploratory study, we sought an organi-zation that would provide access to numerous anddiverse projects. The chosen site was a U.S. firm inthe chemical industry (over 26,000 employees, $4billion in net sales, and $590 million in annualR&D investment). This firm competed in four linesof business, each with its own R&D division dedi-cated to developing a different type of product:agricultural chemicals (for instance, ear tags forfarm animals), consumer products (such as a newelevator mechanism for a stick deodorant), indus-trial chemical intermediates (such as wet strengthresins for paper), and pharmaceuticals (such asnew formulations of over-the-counter vitamins andprescription antiarthritics). This site also enabledus to focus on R&D, a particularly tenuous phase ofproduct development that requires considerablecreativity and experimentation within clear timeand budgetary constraints. The firm had strongcompetencies in manufacturing, but it was deeplyconcerned with improving the management of itsR&D. Divisional managers, project managers, andteam members generously contributed their timeand energy to the study.

Within the research site, data were collected on asample of 80 product development projects. Threecriteria were used to select projects: (1) the projecthad to be judged by R&D personnel as promising,but presenting significant technical uncertainty;(2) the project had to have been authorized andassigned personnel and budgetary resources; and(3) the research had to have progressed beyond thebasic level and be directed at a specific usage, butnot necessarily associated with a defined product.


These criteria helped ensure that projects in thestudy had a reasonahle probability of continuingfor a significant period of time hefore completion ortermination.

Because a limited numher of projects were initi-ated across the four divisions in any single year, weused a staggered data collection strategy to buildthe sample. We examined 11 projects started inyear 1 of the study, 39 projects initiated in year 2,and so on until a complete sample of 80 projectswas ohtained. This sample size provided sufficient,albeit limited, degrees of freedom for analysis andwas feasihle given the demands of our researchdesign. Employing a multiple panel approach, wecollected data in real time, following each projectuntil it was completed or for a maximum of twoyears. The survey was first administered sixmonths after a project had begun and was thengiven approximately every six months. At the endof the study (which spanned more than six years),23 projects had heen completed, 30 projects hadheen terminated, and 27 projects were still ongoing.

Project informants were selected through dis-cussions with an R&D manager in each division,who served as a liaison between our researchteam and that division's project teams. Four cri-teria helped identify appropriate informants: (1)they needed to have heen highly involved inproject efforts at the time data were being col-lected; (2) they needed to hold a position thatwould allow them to knowledgeahly answer thetypes of questions being posed; (3) when possi-hle, independent informants were selected to re-duce common method variance bias; and (4)where feasihle, multiple informants were chosento respond to questions about the project. Ques-tionnaire packets for each respondent were dis-tributed hy the liaison and returned in sealedenvelopes. Both phone calls and direct follow-upby the liaison were used to correct any problems{such as ohviously missing data) and to requestthe return of questionnaires. This procedureyielded a 91 percent response rate.

Prior to each panel of data collection, the divi-sion liaisons reviewed the respondents to ensurethat they continued to meet the criteria. If a respon-dent was no longer actively involved in a project,she or he was replaced. Memhers of the projectteam provided information about project manage-ment and uncertainty. The project leader and oneindependent source—typically, an upper-levelR&D manager—provided information regardingperformance. For the purposes of this article, weanalyzed data collected at three points in time, in-cluding the first panel (time 1) and tbe last panel(time L, which was approximately 24 months after

time 1), as well as the panel immediately precedingthe last panel (time L - 1, approximately 18 monthsafter start-up.

Measures

Questionnaires for this study were developedthrough a year-long joint effort of academic andfirm R&D personnel. Factor analyses conducted onthe resulting data aided development of the projectmanagement, performance, and uncertainty mea-sures (Tahle 1. below, gives all items and factorloadings). To examine project management styles,we operationally defined six constructs that distin-guish between emergent and planned approachesto monitoring, evaluation, and control. Confirma-tory factor analyses (discussed in the Results sec-tion) served to test our proposed conceptual frame-work. We also sought to identify varied facets ofproject performance (building technical knowl-edge, achieving commercial ohjectives, and stayingon time and within budget) and of project uncer-tainty (technical and commercial). As Hurley andcolleagues (1997) suggested, exploratory methodswere used in lieu of hypothesized models. Princi-pal factor analyses with varimax rotation aidedscale development. The four performance factorsand two uncertainty factors were determined usingseveral decision rules: theoretical rationales, eigen-values greater than one, scree plots, and high factorloadings.

Project management styles. We sought to mea-sure emergent and planned approaches to eachactivity. For project monitoring, an emergentstyle stresses monitoring understandings. Fouritems assessed the importance, in a given project,of team members' demonstrating an understand-ing of project objectives, technical knowledge,awareness of project details, and devoting suffi-cient time to the project. In contrast, a plannedstyle calls for monitoring milestones. Four itemsgauged the importance, in a given project, of theproject schedule, interim goals, and test resultsfor tracking progress. Monitoring items wererated on a scale ranging from 1, "not very impor-tant," to 5, "very important."

The emergent style of project evaluation, or in-formation gathering, encourages team memhers toseek ideas and feedback from people outside of theproject. Our three-item measure assessed the de-gree to which teams received useful informationfrom regulatory and marketing personnel on tech-nical and commercial issues (1, "little useful infor-mation," 5, "much useful information"). For theplanned style, formal review signifies systematicappraisals by senior managers. Three items mea-


TABLE 1Results of Factor Analyses for Project Management, Performance, and Uncertainty Variables*

Analysis 1 2 3 4 5

Confirmatory factor analysis: Project management

Monitoring understandingsClear understanding of technical objectives .52Show current technical knowledge ,63Aware of all project details and work .71Devote sufficient time to working on the project .72

Monitoring milestonesBeing on schedule .83Being ahead of schedule .61Producing "hard data" (for instance, test results) .38Accomplishing specific interim project goals ,51

Information gatheringTechnical/scientific: information from regulatory personnel • .54Technical/scientific information from marketing personnel ,00

Commercial information from marketing personnel .78

Formal reviewBy corporate executives .80By the MRPC (the firm's expert review board} .79By development research personnel .49

Participative controlSufficient freedom in running your part of the project .74Sufficient decision authority regarding resource allocation .86Accountability for the success of your part of the project .60

Directive controlInvolvement by project/upper management in details, tightening schedulesFeedback from project manager ahout how the project is progressingInvolvement of project manager in altering resources, objectives, and

schedules

Exploratory factor analysis: Project performance

Technical knowledgeProbability of yielding a major breakthroughExtent yielded information suggesting new market areasExtent yielded information suggesting new product areasExtent yielded valuable technical knowledgeProbability of yielding proprietary technical knowledgeProbability of yielding an incremental advanceExtent yielded information helpful to other ongoing projects

Commercial objectivesProbability of getting product to market in timely fashionProbability of introduction in time to capture desired market shareProbability of meeting commercial objectivesProbability of developing a product with reasonable manufacturing costs

On limeIs project on time in terms of its projected schedule?How does progress compare to its projected schedule?

Within budgetHas this project been more costly than expected or overspent?How do actual costs compare with estimated costs?Is this project within budget in terms of its projected costs?

EigenvaluePercentage of variance explainedCumulative percentage of variance explained

.83

.78

.78

.72

.67

.69

.66

.10

.15

.10

,03

.09

.10

-.14-.06-,03

4,8330.230.2

.07

.03

-.09.17

.17

.35

-.03

.82

.86

.88

.78

.29

.33

.02

-.08.01

3.1319.649.8

-.13.04

.07

-.20-.18-,11

.06

,05

,01

-.04-.06

,08.09

.88

.84

.79

2.1313.363.1

- . 12.17

.20

.10

-.08

.18

.41

.31

.10

.05

.84

.84

.05

.07

,05

1,086.74

69.8

.68

.61

.54


TABLE 1continued

Analysis

Exploratory factor analysis: Project uncertainty

Technical uncertaintyAdequacy of science and technology (S&T) used in this projectDivision's state of the art in this project's S&TRecognition of technical objectives for this projectTeam members' familiarity with the S&T used in the projectTeam memhers" actual experience with this project's S&TDefinition of technical objectives for this projectPredictability of progress of S&T knowledge development in divisionWhether a particular S&T strategy is being followed in this projectRate of progress of S&T knowledge development in divisionDivision's rate of entry into this project's S&T area

Commercial uncertaintyFirm experience withFirm experience withFirm experience withFirm experience withFirm experience with

manufacturing the productpackaging the productadvertising the productselling the productdistributing the product

EigenvaluePercentage of variance explainedCumulative percentage of variance explained

.32-.08-.07

.15

.19

.25

.40

.29-.17

.27

.75

.86

.89

.88

.87

5.2635.135.1

.62

.59

.60

.62

.65

.66

.53

.57

.61

.41

.05

.15

.11

.11

.11

2.4916.651.6

Bold type indicates the factor loadings are significant at p < .01.

sured the extent to which corporate executives, anexpert review board, and divisional developmentresearch personnel reviewed the project (1, "neverevaluate," 5, "often evaluate").

Participative control denotes the emergent styleof control, whereby managers offer teams consider-able autonomy. Three items measured the extent towhich team members felt they had decision-making authority and accountability (1, "not at all,"5, "a great deal"). The planned counterpart, direc-tive control, signifies close supervision. Threeitems gauged whether management was becomingmore involved in project details (for instance, alter-ing the budget and priorities) and providing feed-back and technical and commercial information (1,"never," 5, "often").

Project performance. We measured project per-formance using the averaged judgments of theproject manager and his or her department head.Perceptual measures of performance were used toreflect expectations rather than final outcomes. Useof these measures enabled us to gauge perceptionsof project performance during each panel of datacollection and to assess performance regardless ofwhether or not the project was completed by theend of the two years.

To tap the multifaceted nature of performance,we studied project innovation and efficiency. Theresulting measures are similar to those used in sur-vey research (e.g., Souder, Sherman, & Davies-

Cooper, 1998) and to conceptualizations posed bycase studies [e.g., Bowen, Clark, Holloway, &Wheelwright, 1994). Two measures addressed al-ternate dimensions of innovation. Answering thesequestions required respondents to assign probabil-ities between 0 and 100 percent. A seven-item tech-nical knowledge measure gauged the probabilitythat a project would yield valuable technicalknowledge, suggest new markets, promise a majorbreakthrough or an incremental advance, and helpin ongoing projects. Four items assessed the prob-ability that the project would meet its commercialobjectives—that is, develop a product with reason-able manufacturing costs that could gain its pro-jected market share.

We also employed two measures of project effi-ciency. The measure within budget consisted ofthree items that assessed the extent to which aproject was progressing within projected cost levels(1, "way under budget," 5, "way over budget"). Atwo-item measure labeled on time gauged whetherthe project was meeting its schedule (1, "way be-hind schedule," 5, "way ahead of schedule").

Project uncertainty. We examined technical andcommercial forms of project uncertainty. As inmeasures used in recent studies [e.g.. Green, Gavin,& Aiman-Smith, 1995; Souder et al,, 1998), bothforms were deemed a function of familiarity. Fortechnical uncertainty, we reverse-coded ten itemsindicating the degree to which team members were


familiar with the science and technology used in aproject and the division's level of expertise in thisarea. For commercial uncertainty, five items, alsoreverse-coded, assessed a firm's experience com-mercializing (manufacturing, packaging, advertis-ing, selling, and distributing) products like the de-veloping product.

Our plan for assessing change in uncertainty wasto assign projects to conditions of decreasing andincreasing uncertainty on the basis of their scores attim es 1, L - 1, and L. Repeated-measures analyses,however, indicated that, although technical uncer-tainty did change over time (F^ 75 - 5.67, p < .01),commercial uncertainty remained stahle (F2 76 =0.09, p < .91). Considering these findings, and theabsence of a compelling rationale for choosing oneform of uncertainty over another, we created twoseparate measures. For change in technical uncer-tainty, we followed our initial plan, creating me-dian splits based on a project's scores at each panel.Projects with decreasing technical uncertaintywere coded as - 1 [n ^ 26); projects with unchang-ing uncertainty were coded as 0 (/i = 27); projectsfacing rising uncertainty were coded as 1 [n = 26).For magnitude of commercial uncertainty, weagain developed three categories. Yet given thegreater stability of this variable, categories signifylow, unchanging {- 1, n = 28); varied or changing(0, n = 17); and high, unchanging (1, n = 33)commercial uncertainty. Assessing both the changein uncertainty and the magnitude of uncertaintyhas a long history in organization studies (e.g.,Duncan, 1972; Perrow, 1967).

RESULTS

Hypotheses l a - l c concern the convergentand discriminant validity of the measurementscheme—developed in accordance with our con-ceptual framework (Figure 1). To test these hypoth-eses, we employed confirmatory factor analysis(CFA) using EQS (Bentler, 1995). For purposes ofconstruct development, we used the responses ofthe 213 project team members who participated inat least one panel of data collection. We estimateda CFA on the 20 items measuring project manage-ment styles. The model included six factors repre-senting emergent and planned styles of monitoring,evaluation, and control. Each item was allowed toload only on the factor for which it was a proposedindicator, and no correlations were permitted inthe error structure. Factors were allowed to covarywithin managerial activities (for example, partici-pative and directive control were allowed to co-vary), because we proposed emergent and plannedstyles as offering interrelated approaches to each

activity. Consistent with our conceptual frame-work, the three planned activities were allowed tocovary, reflecting their tight coupling. Results indi-cate that this six-factor model was a good fit to thedata ix^ = 267.21, df = 164. goodness-of-fit index[GFI] = .89, comparative fit index [CFI] = .90,root-mean-square error of approximation [RMSEA]= .055). Item loadings were as proposed and sig-nificant [p < .01). Factor covariations were alsosignificant (p < .01). Table 1 gives results.

To test whether emergent and planned styles ofeach activity were distinct, we compared thismodel to a nested alternative. Following Bentler(1995), we developed a nested model by addingmeaningful constraints to our original model. Spe-cifically, we set the covariations between activitiesequal to one. These constraints signify that con-trasting approaches comprise a single construct—for instance, that participative and directive controlare two ends of the same continuum—as some re-searchers have suggested (e.g., McDonough & Barc-zak, 1991). We conducted a CFA to assess themodel i)^ = 472.74, df = 167, GFI = .84, CFI = .71,RMSEA ^ ,093). A chi-square difference testshowed that the fit of the nested model was signif-icantly worse than that of our original six-factormodel (A/ = 205,56. df=3,p< .001). In sum. ourCFA results supported Hypotheses la- lc : emergentand planned styles offer distinct, yet interrelated,approaches to project monitoring, evaluation, andcontrol.

Examining the dynamics and impacts of manage-ment styles required moving to the project level ofanalysis. Because responses reflect the judgmentsof multiple informants, we assessed interrateragreement [r^J. Using Glick's (1985) heuristic of .6.the median r ,, values for our measures (whichrange from .75 to .98) supported aggregation ofthemeasures. We developed scales using the average ofthe respondents' judgments. In light of the samplesize and the need to do multivariate analyses, wereplaced missing data with means at the item level.Examination of means, standard deviations, andcorrelations before and after replacement revealedonly minor differences. We used responses col-lected at time L - l and time L to conduct multipleregression analyses. Table 2 provides complete de-scriptive statistics and correlational informationfrom both data panels.

Repeated-measures analyses helped assesswhether managerial activities fluctuate. We com-pared the project management data generated dur-ing time 1 (the first data panel) with the data gen-erated at time L - 1 and time L. Analyses offeredmixed support for our hypotheses, highlighting thedynamic nature of management styles. Hypothesis

CM

tii

tt<H

ou•ac(Su

\ \ \ \

O T - - O O M O O M

o c o o o o

C O O O O O

o o o o o o

j ^ --H OT > CJ

^ -P-H ^

c c .2 £ --' M "G •5 "- ™

11 11O O ^ 5 "S

•^ ^ 2 O cd^ p ^ --H ^ p I

O O J U I O

r r

CT^lnr-^sO N t n o

O O O I - H

r r r r

in V CO ind fM 6 6

ia

con

U

5

'cu

li

m

U

S0

I

tU (J

'> .X.n -^

>> - * l o

'3 .y oi

tJ U GC m 0

udge

••= • £

0 U

c "33 G

" ^ (U ^ G

•- I - - I 4- ki CO U

S g « ^ o. .>

O O ( ^ fcj t3 '^

S £ tin a. Q

u J

S E' o " ' •"

•5 '-'O I

Cd

60

Q.

aCO' '

- Jdi

e

G_O

"co

b0u

-aIUDIDI

JOinSi

-a

o o

CO

V

r UlC O G -•2 E -2 -c— ^ 3 00

u 5 c u


2a posits that the use of emergent activities de-clines as a project progresses. In support, we foundthat managers encouraged less monitoring of un-derstandings (F2, 7fi = 5.37, p < .01) and informa-tion gathering (F = 3.29, p < .04) over time, hut theuse of participative control remained relativelyconstant [F = 0.17, p < .84). In Hypothesis 2b, wepropose that the three planning activities are morestahle, In each case, however, decreasing use wasreported. Emphasis on monitoring milestones [F =4.54, p < .01) and directive control [F = 15.44, p <.001) declined in all three time periods. Formalreviews decreased significantly only during the lastsix months of data collection (time L — 1 to L; F =4.14, p < .04).

To examine relationships between project man-agement and performance, we used multiple re-gression analyses. Table 3 gives these results. Giventhe dynamic nature of management styles, analyseswere conducted twice, at times L - l and L. Wealso tested a more stringent, lagged model (L - 1 —*•L) to assess whether managerial activities at timeL — 1 were predictive of performance at time L. Wepropose that emergent activities positively impactproject innovation (Hypothesis 3a). whereasplanned activities enhance efficiency (Hypothesis3h). However, we included all six activities in eachequation to explore the possibility of more intricatetensions between the styles.

Results offered mixed support for Hypothesis 3a.Contrary to our expectations, only planned activi-ties significantly impacted the building of technical

knowledge. Formal review and directive controlwere positively related to technical knowledge, andmonitoring milestones was negatively related inthe lagged model. The relationship hetween projectmanagement styles and commercial objectives wasmore in line with prediction. In particular, partic-ipative control had a positive effect across all threemodels, and information gathering was beneficialat time L - l . Interestingly, formal review, aplanned activity, positively impacted commercialobjectives in the lagged model.

We also found mixed support for Hypothesis 3h.As predicted, a planned approach helped projectsremain on schedule. Directive control was posi-tively related to on-time performance at time L - l .Yet surprisingly, participative control—the emer-gent counterpart—appeared to speed product de-velopment in all three models. The results of testsof within-budget performance were contrary to ourhypothesis. Formal review, the planned approachto evaluation, was significantly hut negatively re-lated to within-hudget performance both at time Land in the lagged model.

To examine the impact of uncertainty on projectmanagement-performance relationships, we firstconducted moderated regression analysis (MRA;Aiken & West, 1991). Because the uncertainty cat-egories reflect measurements from time 1 throughL, we tested interaction effects at time L. Conduct-ing a total of eight MRAs, we regressed each per-formance indicator on change in technical uncer-tainty or on magnitude of commercial uncertainty.

TABLE 3Results of Multiple Regression Analyses: Project Management-Performance

Relationships at Different Points in Time

Variable

Adjusted R^FP

Monitoringunderstandings

Monitoring milestonesInformation gatheringFormal reviewParticipatory controlDirective control

Technical Knowledge

TimeL - l

0,220.153.34*0.00

-0.07

-0.17-0.14

0.47***0.000.24^

TimeL

0.200.204,11*0.00

0.18

0.17-0.17

0.13-0.16

0.26*

TimeL - 1 ^ L

0.140.071.98^0.08

-0.01

-0.26^-0.21

0.32*-0.08

0.21

Commercial Ohjectives

Time. L - l

0.250.183.860.00

-0.20^

0.130,29"0.110,24*0,02

TimeL

0,140,071,920,08

-0,06

0.140,090.040,29**0,03

TimeL - 1 -»

0,200,132.980.01

-0,17

0.090.090.31*0.30*

-0.13

TimeL L - l

0.220.153.340.01

0.07

0.01-0.03-0.11

0.28**0,37**

On Time

TimeL L

0.160.092.240,05

-0.20

0,210.05

-0,110.32**0,03

Time- 1 ^

0,150,082.140,06

-0,09

-0.020.070,200.36*

-0.09

Time•L L - 1

0,110.041.500.19

-0.18

0,030.03

-0,32* -0.09

0,13

Within Budget

TimeL

0.300.244,97**0.00

-0.17

0,070.02

-0.44**-0.12-0.16

TimeL - 1 ^ L

0,180.112.64*0.03

-0.02

-0,160.21

-0.43**-0.08

0,02

"" p < ,10* p < .05

'* p < .01* p < .001

2002 Lewis, Welsh. Dehier, and Green 559

on the six managerial activities, and on the set ofsix interactions (each activity by technical/com-mercial uncertainty). Four models, shown in Table4, offered support for Hypothesis 3c. Models in-cluding technical uncertainty were significant forboth technical knowledge and within-budget per-formance. Commercial uncertainty, in contrast,moderated the relationships between the manage-rial activities and commercial objectives and on-time performance. Interestingly, interaction effectswere most pronounced for planned activities. Forinstance, monitoring milestones displayed signifi-cant interaction effects in ail four models, suggest-ing that the influence of this activity is affected byboth the change in technical uncertainty and themagnitude of commercial uncertainty.

Next, we used split-sample analyses to detail theimpact of managerial activities under varied condi-tions of uncertainty. Table 5 gives these results,which extended our previous regression analy-ses—in some cases by providing greater support forhypothesized project management-performance re-lationships, in others by elaborating our surprising

findings. For example, as predicted, emergent ac-tivities significantly affected technical knowledge.The influence of these activities, however, fluctu-ated with changes in technical uncertainty. Underconditions of decreasing uncertainty, monitoringunderstandings facilitated technical learning,whereas participative control had a negative im-pact. When technical uncertainty was rising, infor-mation gathering proved detrimental. Planned ac-tivities, in contrast, consistently enhanced theprobability that a project would build technicalknowledge, even under conditions of rising techni-cal uncertainty. This finding supports Devaux's(1999) paradoxical claim that planning is most vitalwhen a situation is difficult to plan. When aproject's science and technology is nebulous, for-mal reviews and directive control may provideteams with vital direction.

Split-sample analyses also supplemented previ-ous findings regarding commercial objectives.Monitoring understandings continued to displayan unexpected negative impact when commercialuncertainty was high. This emergent activity, how-

TABLE 4Results of Moderated Regression Analyses: Project Management-Performance

Relationships, Time L, as Moderated by Project Uncertainty

R^ base

f..73

PR^, with interactions^\3. 64

p

Afl after interactions

' 'fi. 64

P

Change in technical uncertaintyMagnitude of commercial uncertaintyMonitoring understandingsMonitoring milestonesInformation gatheringFormal reviewParticipatory controlDirective controlUncertainty X monitoring understandingsUncertainty X monitoring milestonesUncertainty X information gatheringUncertainty X formal reviewUncertainty x participatory controlUncertainty X directive control

TechnicalKnowledge" •

.204.11

.00

.38

3.03.00

.193.26

.01

0.42**

0.190.11

-0.180.180.070.33**0.80

-1.23*0.001.01*

-0.76-0.22

CommercialObjeclives*'

.141.92

.08

.00

.132.27

.0^

0.31**-0.03

0.080.000.160.27*0.14

-0.21+0.37-*

-0.140.06( f e l l0.19

On Time''

.162.24

.05

.362.74

.00

.162.77

.05

0.19+-0.19

0.160.09

-0.030.32**0.020.070.34*

-0.210.010.060.33*

WithinBudget^

.304.97

.00

.382.90

.02

.081.55n.s.

0.05

-0.09-0.01-0.03-0.41**-0.14-0.10

0.21-0.30*-0.07-0.09

0.16-0.17

" Models including change in technical uncertainty were significant for both technical knowledge and within-budget performance.'• Models including magnitude of commercial uncertainty were significant for both commercial objectives and on-time performance.

+ p < .10* p < .05

**p < .01

Ite

au

d d d d

c o TO a

a; <

c-i0

d

CDCC

d1

i n

d1

i n

qd

!N COq csj

d d

enCO

d1

n

d

( DO

d

CO

od

d ( D O COr-< CO • 3 ' CSJ

d d d d

en

CO

£"'CS)

C

sub

OJ10!D0)L I .

CJ

d

—a,

toJ33

S-5

E a.,,C 3 ==

a; d ^^ ' d

COin

g ° 3. ^ d 00

(D

CDDC

l/i .—I

^ a, c

I S ^ V V V

2002 Lewis. Welsh. Dehler. and Green 561

ever, appeared beneficial under conditions of lowuncertainty. Interestingly, the interaction effectwas reversed for the planned approach to monitor-ing. Monitoring milestones had a positive impactwhen commercial uncertainty was high, but thisactivity reduced the likelihood of achieving com-mercial objectives when uncertainty was low.Control activities demonstrated a similar pattern.Participative control was advantageous under con-ditions of high uncertainty, and directive controlhad a positive impact when uncertainty was low.

Changes in uncertainty appeared to alter the im-pact of managerial activities on project efficiency.We proposed that planned activities speed productdevelopment. As in previous analyses, participa-tive control unexpectedly and consistently fosteredon-time performance. In the MRA, however, wefound significant interaction effects for twoplanned activities. Examining the split-sampleanalyses suggests that monitoring milestones anddirective control slowed the development processwhen commercial uncertainty was high. Yet underconditions of low uncertainty, these activitiesraised the likelihood that a project would stay onschedule.

Contrary to our predictions, formal review con-tinued to show a significant and negative impact onwithin-budget performance. In addition, bothforms of control displayed negative effects in thesplit-sample analyses. Participative control proveddetrimental when uncertainty was rising, and di-rective control appeared inefficient under more sta-ble conditions. We found, however, that oneplanned activity behaved more as predicted. Mon-itoring milestones increased the probability that aproject would remain within budget when techni-cal uncertainty was stable.

DISCUSSION

This study is a response to calls for finer-grainedconstructs that capture the tensions and build firomthe activities of project management (e.g., Dough-erty, 1996; Jelinek & Schoonhoven, 1990). Ourconceptual framework juxtaposes emergent andplanned styles to accentuate their differences. Yet,rather than depict these styles as mutually exclu-sive, we propose that each promotes a distinct, butpotentially complementary, approach to projectmonitoring, evaluation, and control. The resultingconstructs may help researchers and practitionersmove beyond either/or thinking to examine the in-terplay between these varied styles. Tests suggestthat the corresponding measures are reliable andvalid, providing useful means of investigating com-plicated project management issues, such as how

combinations of managerial activities affect teamcommitment and firm support for a project.

This exploratory study offers a basis for ongoingconceptual development. In particular, some itemswere customized for this research site. The infor-mation-gathering measure, for instance, gaugedteam interactions with marketing and regulatorypersonnel. Firm managers stressed that these exter-nal groups dealt with issues most problematic fortheir R&D efforts. Alternative sites might suggestother sources of pertinent information [such as sup-pliers, customers, quality assurance, or manufac-turing). Tests on different samples also would fur-ther assess the validity, reliability, and factoranalytic structure of this new instrument. Althoughour research focused on R&D, extended studiesmight examine the role of management stylesacross additional phases of product development(such as transfer from R&D to production to even-tual product launch).

Project Management Dynamics

A notable strength of this research is its design.Survey studies of product development often ex-amine a single cross-section of data, asking respon-dents, in hindsight, how a project was managed(Brown & Eisenhardt, 1995). In contrast, we col-lected data in real time, following 80 projects overa series of six-month intervals. Our analyses revealthe dynamism of managerial activities, as well asvariations in uncertainty and project managementand in the relationships between performance.These results demonstrate the need to purposefullyincorporate time within future research. If innova-tion is marked by periods of stability and clarityand by bursts of creativity and ambiguity, then akey issue is how managers respond to such fluctu-ations (Dougherty, 1996). Retrospective studiesmay offer limited, or inaccurate, insights into theuses and effects of management styles.

Our findings indicate that, regardless of style, theuse of most project management activities declinesover time. The one exception was participativecontrol, suggesting the unique nature of this emer-gent activity. The work of Van de Ven and PoUey(1992) offers a possible explanation. In their longi-tudinal study, managers tended to retain, if notincrease, team members' autonomy. These authorstheorized that as a project progresses, its goals andpotential for success will become clearer and teamswill become increasingly focused and routine-oriented. As these changes occur, managers mayfeel more secure maintaining the current level ofparticipative control. Extending this claim, we pro-pose that managers also may reduce planned activ-


ities for fear that such intervention will impedeprogress.

Furthermore, despite declining usage, emergentand planned activities significantly impacted per-formance even 24 months into a project. This find-ing suggests that project management styles remainvaluable, but their roles may change over time.Initially, emergent approaches may help teamswork through the novelties of a project, whereasplanned activities provide a sense of direction andorder. Later in the process, more subtle managerialactivities may serve a supporting role, helpingteams to cope with varying levels of uncertaintyand to push projects toward completion. Replicat-ing this study in other sites would assess the gen-eralizahility of these findings and elaborate thechanging role of management styles. By examiningmultiple data panels, our study contributes valu-able, but partial, insights into the process by whichmanagers adjust their activities to varying projectconditions. More continuous longitudinal researchmight detail how and when managers engage inthese activities, thus responding to Pennings's(1992) call for greater process understandings.

The Paradox in ProjectManagement-Performance Relationships

This study explores the impact of contrastingmanagement styles on multiple dimensions of per-formance. Our extensive regression analyses com-plicate existing understandings but also reveal anoverarching paradox in project management-performance relationships. Results suggest thatmeeting tenuous demands for innovation and effi-ciency requires a blend of emergent and plannedapproaches. As Quinn [1988: 90) proposed, highperformers may "transcend style," breaking awayfrom a preferred mode of thinking and behaving toapply a more flexible and complicated repertoire ofactivities. Specifically, we find that successfulproduct development may require managers (1) touse emergent and planned activities concurrently,(2) to go back and forth between styles as changesin project uncertainty occur, and (3) to make trade-offs between contending demands. We now brieflydiscuss these three managerial capabilities.

First, the ability to combine emergent andplanned activities may improve certain facets ofproject performance. For example, both approachesto control helped teams use their time more effi-ciently, particularly under conditions of low com-mercial uncertainty. Like McDonough and Barczak(1991). we found that participative control spedproduct development. Providing teams with deci-sion-making discretion and accountability may fos-

ter spontaneity, encouraging members to improviseand explore market opportunities as they arise. Yetunder the same conditions, close managerial super-vision also enhanced on-time performance. By ex-amining participative and directive control as dis-tinct activities, our study clarifies the frequent butambiguous calls for subtle control: effective man-agers provide strong leadership to keep teams fo-cused and on schedule, while empowering teammembers to foster motivation and creativity (e.g.,Clark & Fujimoto, 1991; Wheelwright & Clark.1992). As Jelinek and Schoonhoven proposed, ac-celerating innovation requires the sharing of con-trol, hecause "discipline is exercised best in a freerprocess" (1990: 442).

Second, successful managers use iteration be-tween styles in response to changes in project un-certainty. For instance, significant interaction ef-fects suggest that monitoring milestones andunderstandings hoth enhance the likelihood thatprojects will meet their commercial objectives, butunder different conditions. When commercial un-certainty is low, monitoring understandings is ben-eficial, possibly encouraging teams to step back,assess their work, and integrate their evolvingknow-how with existing organizational competen-cies (for instance, expertise marketing that type ofproduct). In contrast, when an organization is un-familiar with the developing product, monitoringmilestones becomes vital. Some researchers haveproposed that milestones offer a sense of structure,reducing team members' anxiety as they experi-ment with alternative designs (e.g., Eisenhardt &Tabrizi, 1995). Mirvis (1998), however, suggestedthat milestones play a very different role whenuncertainty is high. Frequent milestones actuallymay stimulate anxiety, pushing team memhers tolook ahead—previewing possible crises (like poortest results)—and to think through their potentialresponses.

The third paradoxical pattern exposes the possi-ble need for trade-offs. Product development effortsdemand hoth innovation and efficiency, hut certainactivities may foster one, while impeding the other.Formal review provides a case in point. Acrossvaried conditions of uncertainty, this planned styleof evaluation enhanced the likelihood that projectswould build technical knowledge. At the sametime, formal review activities consistently and neg-atively impacted within-budget performance. AsRosenau and Moran (1993: 166) proposed, frequentand systematic reviews may help top managementstay informed, provide teams with valuablefeedback, and adjust project resources as needed.Such evaluations, however, consume a tremendousamount of resources themselves. Hence, this activ-

2002 Lewis. Welsh. Dehler, and Green 563

ity may bolster innovation, but at the expense ofbudgetary efficiency.

In sum, our results support Dougherty's (1996:433) claim that successful product developmentrequires the capacity to manage tensions—not bysticking to the tried and true, but by applying dif-ferent aggregates of action at different times. Yetour findings also accentuate the need for furtherresearch. Building on our conceptual framework,contingency studies might investigate complex in-teractions between project contexts (for instance,product novelty, team characteristics, environmen-tal uncertainty) and management styles. Such workmay help researchers comprehend how managersrespond to numerous and fluctuating contingenciesand offer practical insights into the conditions un-der which certain managerial activities becomecritical.

Furthermore, although we find that project man-agement style matters, existing research demon-strates the importance of numerous product devel-opment methods (cf. Brown & Eisenhardt, 1995;Montoya-Weiss & Calatone, 1994; Wheelwright &Clark, 1992). A greater understanding of projectmanagement styles may extend this work. Emer-gent and planned styles offer internally consistentmodes of thinking and behaving that help managerstranslate structural mechanisms and standardizedengineering tools into action. For instance, futurestudies might assess how contrasting styles influ-ence the use and effectiveness of such popularmethods as concurrent engineering, cross-func-tional teams, or quality function deployment(QFD).

Concluding Note

As global competition intensifies and technolog-ical change accelerates, comprehending the chal-lenges of product development is increasingly vi-tal. This study explored the nature, dynamics, andimpacts of contrasting management styles in di-verse product development projects. We found thatmanagers may be well served by developing a com-plicated and paradoxical repertoire of practices.Similarly, Mirvis praised the blending of ap-proaches as "planned serendipity"; the notion is"that order arises out of chaos and that freedomproduces control" (1998: 590). Managing tensionsbetween loosely and tightly coupled activities, bot-tom-up and top-down processes, and flexibilityand discipline may provide a key to high perfor-mance. We hope that, by capturing the tensions andbuilding from activities of project management,this conceptual fiamework provides a potentially

valuable foundation for future research. As Quinnexplained:

Just as organizations are usually not without ten-sion, neither are managers. There are no clear mapsfor problem detection or solution. Many diversekinds of behaviors are expected, and these are, at aminimum, distant, at a maximum, competing. [Aninsightful conceptual] framework allows for the dy-namic tensions that can sometimes be at the heart ofmanagerial life, It also allows for the fact that behav-iors may change. A manager may engage in a set ofbehaviors reflecting one set of values at one pointand in an entirely different set of values at anotherpoint. (1988: 88)

REFERENCES

Aiken, L., & West. S. 1991. Multiple regression: Testingand interpreting interaction. Newbury Park, CA:Sage.

Ancona. D. G.. & Caldwell, D. F. 1992. Bridging theboundary: External activity and performance in or-ganizational teams. Administrative Science Quar-terly, 37: 634-666.

Barrett, F, ). 1998. Creativity and improvisation in jazzand organizations: Implications for organizationallearning. Organization Science, 9: 605-622.

Benghozi, P. 1990. Managing innovation: From ad hoc toroutine in French Telecom. Organization Studies,11: 531-554.

Bentler. P. M. 1995. EQS: Structural equations programmanual. Encino. CA: Multivariate Software.

Bowen. H. K., Clark, K. B.. Holloway, C. A., & Wheel-wright, S. W. 1994. The perpetual enterprise ma-chine. New York: Oxford University Press.

Brown, S. L., & Eisenhardt, K. M. 1995. Product devel-opment: Past research, present findings, and futuredirections. Academy of Management Review, 20:343-379.

Bums, T., & Stalker, CM. The management of innova-tion. London: Tavistock.

Clark, K. B,, & Fujimoto, T. 1991. Product developmentperformance. Boston: HBS Press.

Crossan. M. M. 1998. Improvisation in action. Organiza-tion Science, 9: 593-599.

Devaux, S. A. 1999. Total project control. New York:Wiley.

Dougherty, D. 1992. Interpretive barriers to successfulproduct innovation in large firms. Organization Sci-ence, 3: 179-202.

Dougherty, D. 1996. Organizing for innovation. In S. R.Clegg, C. Hardy, & W. R. Nord (Eds.). Handbook oforganization studies: 424-439. Thousand Oaks,CA: Sage.

Duncan, R. B. 1972. Characteristics of perceived environ-


ments and perceived environmental uncertainty.Administrative Science Quarterly, 17: 313-327.

Eisenhardt, K. M., & Tabrizi, B. N. 1995. Acceleratingadaptive processes: Product innovation in tlie giobalcomputer industry. Administrative Science Quar-terly, 40: 84-110.

Glick, W. H. 1985. Conceptualizing and measuring or-ganizational and psychological climate: Pitfalls inmulti-level research. Academy of ManagementReview, 10: 601-616.

Green, S. G., Gavin. M. B.. & Aiman-Smith, L. 1995.Assessing a multidimensional measure of radicaltechnological innovation. IEEE Transactions in En-gineering Management, 42/3: 203-214.

Hurley, A. E., Scandura, T. A.. Schriesheim, G. A., Bran-nick, M. T.. Seers. A.. Vandenberg. R. J., & Williams.L. 1. 1997. Exploratory and confirmatory factor anal-ysis: Guidelines, issues, and alternatives. Journal ofOrganizational Behavior, 18: 667-683.

Jelinek, M., & Schoonhoven, G. B. 1990. The innovationmarathon: Lessons from high technology firms.San Francisco: Jossey-Bass.

Katz, R. 1982. The effects of group longevity on projectcommunication and performance. AdministrativeScience Quarterly, 27: 81-104.

Laufer, A. 1997. Simultaneous management: Managingprojects in a dynamic environment. Boston:AMAGOM.

McDonough, E. F., & Barczak, G. 1991. Speeding up newproduct development: The effects of leadership styleand source of technology- Journal of Product Inno-vation Management, 8: 203-211.

Mirvis, P. H, 1998, Practice improvisation. OrganizationScience, 9: 586-592.

Montoya-Weiss. M. M., & Galantone, R. 1994. Determi-nants of new product performance: A review andmeta-analysis. Journal of Product Innovation Man-agement, 11: 397-417.

Moorman. G., & Miner, A. 1998. The convergence ofplanning and execution: Improvisation in new prod-uct development. Journal of Marketing, 62: 1-20.

Pasmore, W. A. 1998. Organizing for jazz. OrganizationScience, 9: 562-568.

Pennings. J. M. 1992. Structural contingency theory: Areappraisal. In B. M. Staw & L. L. Gummings (Eds.),Research in organizational hehavior, vol. 14: 267-309. Greenwich. GT: JAI Press.

Perrow. G. 1967. A framework for the comparative anal-ysis of organizations. American Sociological Re-view, 32: 194-208.

Quinn. R. E. 1988. Beyond rational management: Mas-tering the paradoxes and competing demands ofhigh performance. San Francisco: Jossey-Bass.

Rosenau, M. D.. & Moran, J. J. 1993, Managing the

development of new products. New York: VanNostrand Reinhold.

Shenhar. A. J., & Dvir, D. 1996. Toward a typologicaltheory of project management. Research Policy, 25:607-632.

Souder, W. E., Sherman. J. D.. & Davies-Gooper. R. 1998.Environmental uncertainty, organizational integration,and new product development effectiveness. Journalof Product Innovation Management, 15: 520-533.

Van de Ven, A. H.. & Polley, D. 1992. Learning whileinnovating. Organization Science, 3: 92-116.

Weick. K. E. 1998. Improvisation as a mindset for or-, ganizatinnal analysis. Organization Science, 9:

543-555.

Wheolwright, S. G.. & Glark, K. B. 1992. Revolutionizingproduct development. New York: Free Press.

Zirger. B, J.. & Maidique. M. 1990. A model of newproduct development: An empirical test. Manage-ment Science, 36: 867-883.

Marianne W. Lewis ([email protected]) is an assis-tant professor of management at the University of Gincin-nati. She earned her Ph.D. in management from the Univer-sity of Kentucky. Her research explores tensions, conflicts,and paradoxes that impede and enable innovation, partic-ularly in the areas of product development, advanced man-ufacturing technology, and organization theory.

M. Ann Welsh is an associate professor of management atthe University of Gincinnati. She earned her Ph.D. inmanagement from the University of Missouri. Her re-search, supported hy the National Science Foundation,the Genter for Innovation Management Stndies. and theCleveland Foundation, focuses on transforming organi-zations into high-performing and nontoxic systems.

Gordon E. Dehler is an associate professor of manage-ment at The George Washington University and theNorth American editor of Management Learning. He earnedhis Ph.D. from the University of Gincinnati. His researchinterests include new product development, critical ap-proaches to organizations and management learning, andtransformational perspectives on organizations,

Stephen G. Green is a professor of management at Pur-due University. He earned his Ph.D. degree in psychol-ogy at the University of Washington. His current researchinterests include leadership, socialization, and the man-agement of innovation. Steve's research has been sup-ported hy the National Science Foundation, the Centerfor Innovation Management Studies, Krannert's Centerfor International Business and Economic Research.and the Center for the Management of ManufacturingEnterprises.

product development tensions: exploring contrasting styles...

Documents