Manufacturing Technology Management - Jan.ucc.nau.edu

Journal of

ManufacturingTechnology

Management

Modeling relationship dynamics in GM'sresearch-institution partnershipsGülcin H. Sengir

Based at General Motors Research and Development, Warren, Michigan, USARobert T. TrotterReagents' Professor of Anthropology, Northern Arizona University, Flagstaff, Arizona, USAElizabeth K. BriodyBased at General Motors Research and Development, Warren, Michigan, USADevadatta M. KulkarniBased at General Motors Research and Development, Warren, Michigan, USALinda B. CatlinPresident of Claymore Associates, Colorado Springs, Colorado, USATracy L. MeerwarthBased at General Motors Research and Development, Warren, Michigan, USA

Journal of Manufacturing Technology Management, Vol. 15 No. 7, 2004,© Emerald Group Publishing Limited, 1741-038X

Emerald

Journal of ManufacturingTechnology ManagementISSN 1741-038X©2004 Emerald GroupPublishing Limited

Journal of Manufacturing TechnologyManagement aims to give a broadinternational coverage of subjectsrelating to the management ofmanufacturing technology and theintegration of the production, design,supply and marketing functions ofmanufacturing enterprises. Emphasis isplaced on the publication of articleswhich seek to link theory with applicationor critically analyse real situations withthe objective of identifying good practicein manufacturing. Papers accepted forpublication are double blind refereed toensure academic rigour and integrity.

EditorDr David BennettAston Business School, AstonUniversity, Aston Triangle,Birmingham B4 7ET, UKTel: +44 (0) 121 35936 11

Fax: +44 (0) 121 359 5271

E-mail: [email protected]

Reviews EditorDr Bart MacCarthyBusiness School, University ofNottingham, Nottingham, UKE-mail: [email protected]

Managing EditorRosie Knowles

Indexed and abstracted inEmerald Reviews (formerly Anbar)INSPECResearch Trends in Advanced

Manufacturing

Editorial Review BoardHarry BoerUniversity of Aalborg, DenmarkNourredine BoubekriNorthern Illinois University, USAMike ByrneUniversity of Nottingham, UKAfonso FleuryUniversity of Sao Paulo, BrazilIan GibsonThe University of Hong Kong,Hong KongA. GunasekaranUniversity of Massachusetts,Dartmouth, USAAbdel-Aziz HegazyHelwan University, EgyptBob HollierUM/ST, UKHiroshi KatayamaWaseda University, JapanTarek KhalilUniversity of Miami, USAAshok KochharUniversity of Aston, UKDoug LoveUniversity of Aston, UKDouglas K. MacbethUniversity of Glasgow, UKLes MitchellUniversity of Hertfordshire, UKShunji MohriUniversity of Hokkaido, JapanAndy NeelyCranfield University, UKKul PawarUniversity of Nottingham, UKRoy SnaddonUniversity of Witwatersrand,South AfricaAmrik SohalMonash University, AustraliaDon TaylorVirginia Tech, USAWang Xing MingRenmin University, ChinaPeter WrightCelestica, UK

J ~England BD8 9BY INVESTOR IN PEOPLE

Tel +44 (o) 1274 7777 0 0Fax +44 (0) 1274 785200E-mail [email protected]

Regional offices:For North AmericaEmerald, 875 Massachusetts Avenue,7th Floor, Cambridge, MA 02139, USATel Toll free +1 888 622 0075;Fax +1 617 354 6875E-mail [email protected] JapanEmerald, 3-22-7 0owada, Ichikawa-shi,Chiba, 272-0025, japanTel +81 47 393 7322;Fax +8147 393 73 2 3E-mail [email protected] Asia PacificEmerald, 7-2, 7th Floor, Menara KLH,Bandar Puchong Jaya, 47100 Puchong,Selangor, MalaysiaTel +6o 3 8076 6009;Fax +6o 3 8076 6007E-mail [email protected] helpdesk:Tel +44 (0) 1274 7 85 2 7 8 ;Fax +44 (o) 12 74 7 8 5 204;E-mail [email protected] www.emeraldinsight.com/customercharterOrders, subscription and missing claimsenquiries:Tel + 44 (0) 12 74 777700;Fax +44 (0) 1274 785200

Missing issue claims will be fulfilled ifclaimed within four months of date ofdespatch. Maximum of one claim perissue.Reprints and permissions service:John EggletonE-mail [email protected] www.emeraldinsight.com/reprints

www.emeraldinsight.com/permissions

This journal is also available online atwww.emeraldinsight.com/jmtm.htmInternet services availableworldwide atwww.emeraldinsight.comEmerald Group Publishing Limited6o/62 Toiler Lane,Bradford,West Yorkshire,

Awarded in recognition ofEmerald's productiondepartment's adherenceto quality systems andprocesses whenpreparing scholarlyjournals for print

GIP

UKAS

ism) urQUALITYASSURED

FIRMQ~}

u26

Gnifiram nu:nroa

No part of this journal may be reproduced, stored in aretrieval system, transmitted in any form or by anymeans electronic, mechanical, photocopying,recording or otherwise without either the prior writtenpermission of the publisher or a licence permittingrestricted copying issued in the UK by The CopyrightLicensing Agency and in the USA by The CopyrightClearance Center. No responsibility is accepted for theaccuracy of information contained in the text,illustrations or advertisements. The opinionsexpressed in the articles are not necessarily those ofthe Editor or the publisher.

Emerald is a trading name of EmeraldGroup Publishing LimitedPrinted by Printhaus Group Ltd,Scirocco Close, Moulton Park,Northampton NN3 6HE

Emerald

Modeling relationshipdynamics in GM'sresearch-institutionpartnershipsGiilcin H. Sengir, Robert T Trotter,Elizabeth K. Briody,Devadatta M. Kulkarni,Linda B. Catlin andTracy L. Meerwarth

The authorsGulcin H. Sengir, Devadatta M. Kulkarni, Elizabeth K. Briodyand Tracy L. Meerwarth are based at General Motors Researchand Development, Warren, Michigan, USA.Robert T. Trotter is Reagents' Professor of Anthropology,Northern Arizona University, Flagstaff, Arizona, USA.Linda B. Catlin is the President of Claymore Associates,Colorado Springs, Colorado, USA.

KeywordsOrganizations, Partnership, Relationship management

AbstractGM has initiated partnerships with firms and researchinstitutions at a rapid pace. One effort of the multi-disciplinaryresearch team involved the construction of a relationshipdynamics model to assist in partnership planning andmanagement. Earlier research on private-sector partnershipsindicated that partnership success is largely dependent upon thedevelopment and maintenance of strong, productiverelationships between the partners. Therefore, modeling effortsfocused on the relationship itself. To increase the likelihood thatthe resulting model is realistic, valid and representative,empirical data was combined with a systems-dynamicsapproach, and the model is being validated with feedback fromstudy participants.

Electronic accessThe Emerald Research Register for this journal isavailable atwww.emeraldinsight.com/researchregister

The current issue and full text archive of this journal isavailable atwww.emeraldinsight.com/1741-038X.htm

Journal of Manufacturing Technology ManagementVolume 15 • Number 7 . 2004 . pp. 541–559(t_ Emerald Group Publishing Limited • ISSN 1741-038X001 10.1108/17410380410555790

1. Introduction

Many organizations and institutions today areengaged in a wide variety of partnerships. GM, forexample, has been involved in partnering for over30 years. Initially, these relationships involvedalliance partners in which some form of equityrelationship was common[l]. While this trendcontinues, other types of partnerships have beencreated including those with private-sector firms inwhich GM has no equity, and with a variety of USand non-US research institutions. Over time, thetype and number of these partnerships hasdramatically increased

Each of these partnership types is perceived tohave numerous benefits. GM R&D relationshipswith research institutions are designed to enhanceGM's competitiveness. R&D seeks to leveragepartner knowledge and expertise. Partneringenables GM to tap into rapidly expandingtechnical fields and bring that knowledge to bearon corporate problems. One GM executivecommented: "Innovation networks, such as thispartnership, will function as incubators for newideas. The fastest way to deliver innovation is towork together" (GM Press Release, 2001). ForR&D's university partners, known as CollaborativeResearch Labs (CRLs), the partnerships affordopportunities for university faculty and students tolearn about and contribute to "real-world"problem-solving. In exchange, they receiveresearch funding for a specified number of years,and can often take advantage of internships andother work-related opportunities for both studentsand faculty.

Because of the significant investment of R&Dtime and resources into this increasingly popularbusiness strategy, R&D management launched amajor program to study these partnerships. Ourmulti-disciplinary research team developed aproposal, and subsequently began a multi-phasestudy to understand partnership functioning.

Received: March 2004Revised: May 2004Accepted: June 2004This paper won the JMTM award for BestManufacturing Paper at IAMOT 2004. This awardwas sponsored by Emerald.

The authors appreciate the willingness of so manyGM and research-institution participants to interviewthem, respond to their social-network survey andprovide them with feedback during their validationsessions. Without their help, they would not havebeen able to develop the relationship dynamicsmodel. They also appreciate their reviewers' time andeffort. In particular, they thank Dan Reaume, MarcRobinson, Phil Keenan and Kurt Godden for theirhelpful and insightful comments.

541

Modeling relationship dynamics in GM's research-institution partnerships

Giilcin H. Sengir et al.

Our goals were to investigate the structure anddynamics of these R&D partnerships, develop amodel to diagnose and predict partnershipdifficulties, and identify ways of enhancingpartnership effectiveness.

We know from past studies that strong, effectivepartnerships are rare because of the high rates ofpartnership failure -- upwards of 60 percent withsome as high as 80 percent (Duysters et al., 1999;Ertel et al., 2001; Gulati et al., 1994; Spekman andLamb, 1997). These failures are largely associatedwith the social and cultural differences between thepartnering organizations. For example, partnersmay have different goals for the partnership,different ideas about how to achieve those goals, ordifferent expectations for the time and resourcecommitment necessary to accomplish those goals.When there are significant differences in alignmentbetween the partners, conflict typically results,leading to damaged relationships and few tangiblework products. It is not trivial to create apartnership culture in which the participantsget along, innovation thrives and outcomes areproduced.

In an earlier analysis of collaborativerelationships, we found that study participantsrepeatedly stressed the importance of "trust andmutual respect, " "being open to suggestions " and"working together" in their relationships (Catlinet al., 2003). In a study of GM's private-sectorpartnerships, we found that study participantsoffered prescriptive statements about partnershipbehavior to ensure partnership success( Meerwarth et al., 2002). Indeed, partnership rules,as they became articulated, accepted and followed,enabled these partnerships to flourish. From thesefindings we conclude that even among bright,highly-trained individuals engaged in technicalresearch, partnership success entails relationshipbuilding and maintenance.

In this paper, we take the position thatpartnering relationships are part of a dynamicsystem that can be modeled. We saw in our earlierpartnership studies that individual and partnering-organization relationships can be influenced andshaped by certain factors (e.g. "goodcommunication," "trust," etc.), and that there arecommon themes and patterns, includingpredictability in relationships. Modeling permitsus to study the inter-connectivity among thesefactors in partnership relationships. We prefer toview partnering relationships as data inputs thatcontribute to overall partnership functioning,rather than isolated entities (Winder and Judd,1996). Viewing partnering relationships as havingat least some recurring features positions us todescribe, explain and learn from the system as awhole. We are then able to sort through patterns,

Journal of Manufacturing Technology ManagementVolume 15 • Number 7 • 2004 • 541—559

identify best practices and recommend strategiesfor assimilating that learning into the entirepartnership system. This new view permits us totap into and observe partnership dynamics thathave been heretofore unknown, unrecognized orunexplained.

Such modeling requires a complex tool setbecause of the difficulty of capturing andmeasuring these kinds of collaborativerelationships. Our tool set, combining qualitativeand quantitative analyses with a systems-dynamicsapproach, enabled us to conceptualize therelationship dynamics among partnershipparticipants. We define relationship dynamics asthe interactions associated with the collaborationprocess. Partnering, by its nature, involvesrelationships between individual counterparts, aswell as relationships established betweenpartnering organizations and/or institutions. A keydimension of partnership relationships andinteractions is their reciprocal nature sincerelationships involve the mutual exchange of ideas,favors and the like (Sahlins, 1972). Reciprocity,the process for establishing and maintainingrelationships through the exchange of goods andservices, is the glue holding these partnershiprelationships together.

Modeling relationship dynamics incollaborative ventures makes interactions visibleand thus understandable. Once understandable,managers are positioned to make judgments aboutthe observed patterns and intervene, asappropriate, to increase the likelihood ofpartnership success. In our model, we seek toaddress the following questions about partneringrelationships:(1) What are the critical components of

partnering relationships and the dependenciesbetween components?

(2) How do these components change over time?(3) How are these partnerships structured?(4) How do these partnership networks evolve

over time?(5) How can partnership success be measured?

We begin by reviewing the literature pertaining tomodeling organizations. Next, we discuss our datasets and methods. Working inductively, we usedsalient patterns from the qualitative data as inputto the model, expanding the conceptualization ofthe model based on the social-network survey.Third, we describe the key components ofrelationship effectiveness – communication, jointwork, quality of interaction and connectivity ofsocial structure. We illustrate the linkages amongthese key components and their subcomponents,defining the basic building blocks (i.e. concepts,stocks, flows, etc.) and representing the interactiverelationships. Fourth, we outline the simulation

542


Gulcin H. Sengir et al.

process we used. Finally, we generate somehypotheses from the qualitative data as we begin tovalidate the utility of the model.

2. Modeling relationship dynamics2.1 Modeling organizationsComputer modeling has been popular inengineering, mathematical, computer, physicalsciences and domains that are inter-disciplinarywith these areas. However, using computersimulation in the social-sciences has remained anemerging area of development for some time.Historically, the first organizational models werebased on administrative decision-making theoriesof Herbert A. Simon, from the 1940s and 1950s.This work led to the analysis of how organizationalobjectives are formed, how strategies evolved andhow decisions are made (Cyert and March, 1963;Simon, 1944). Starting in the late 1990s, thecomputational modeling of social organizationsand institutions became a topic of growing interestin both the computer science and social-sciencecommunities. Artificial intelligence techniques,such as distributed and learning models, wereadopted to accommodate the ecological orcontextual complexity of organizationalrelationships. Complexity theory helped in theunderstanding of organizational complexity from ahistorical perspective; organizations are affected byevolutionary events, feedback-loops and pathdependencies that are critical to understandingand formalizing organizational culture andbehavior (Kauffman, 1995; Waldrop, 1992).

Simultaneously, social scientists have beenfocusing on formalizing theories in relationships ine-business, information exchange, changes inorganizational structure, authority structures,delegation, reciprocity and cooperation,empowerment, organizational evolution, and roles(Prietula et al., 1998). Computational modeling ofphenomena such as cooperation, coordination,dynamics of organizations and consensusformation has helped in understanding thecontributing factors (Lomi and Larsen, 2001).There is now good evidence that feedbackprocesses of mutual influence operate in socialgroups at multiple levels and ways, including theevolution of complex cultural behavior (Small,1999). Understanding the relationship betweenorganizational structures, control and culturalvalues is key to long-term cooperative activity(Turpin, 1999). At the same time, the dynamics ofcultural influences across a social-network, wherethe nodes represent individuals communicatingand working jointly, represents a dynamicmicrocosm of a knowledge network; over time,

Journal of Manufacturing Technology ManagementVolume 15 . Number 7 • 2004 • 541-559

such knowledge networks can evolve into close-knit clusters of expertise (Carley and Hill, 2001).

2.2 Using systems-dynamics approach tomodel organizationsThe term systems thinking is widely applied todescribe a range of tools and methods used to helpmanagers understand the inter-relatedness oforganizational issues. Many of the ideas and toolsof systems thinking derive from the application ofsystem dynamics (SD) models. SD is a method fordescribing, modeling and simulating dynamicalsystems. SD was originally developed in the 1950sand 1960s at MIT by Jay W. Forrester as a set oftools for relating the structure of complexmanagerial systems to their performance overtime, via the use of simulation.

What makes using SD different from otherapproaches, is the use of feedback-loops, such asthose encountered in electrical and mechanical-control systems. SD concepts such as "stocks andflows " [2] describe the primary system structuresand processes, and how they are connected byfeedback-loops. Such loops create the non-linearcharacteristics of social interactions (Forrester,1989) that are part of a systems approach tomodeling. Focusing on flows and stocks ofinformation, people and other resources has led tothe exploration of complex dynamics and temporalcharacteristics of organizations (Sastry, 2001).When these models are applied to the socialdomains of organizations, the target is always adynamic entity - changing and reacting to itsenvironment (Richmond, 1993); it has bothstructure and behavior. The model itself can berepresented as an equation, a logical statement or acomputer program.

Basic SD models can be constructed using asimple spreadsheet program. The variables arearranged in columns and each time-step isrepresented in a single row. Spreadsheet modelscan be very helpful for understanding the linearaspects of SD models[3]. However, the grouparound Forrester developed the simulationsoftware Dynamo, the ancestor of a number ofmodern simulation languages, to handle morecomplex simulation conditions. More modern SDsoftware products followed, allowing the design ofsystem models in a graphical mode as flowdiagrams[4]. In our research and modeling, wehave taken advantage of these advances to proposeand test a more complex model than would bepossible through spreadsheet analysis.

2.3 The gap this research fillsOur approach fills in a number of gaps in thesocial-science, social-network and modelingliteratures. All three approaches are commonly

543


Gofdn H. Sengir et al.

used to describe business environments orsimulate business conditions, but they are virtuallynever combined or applied simultaneously.

One contribution involves a systematic processfor linking the qualitative data to baselinequantitative data at several key juncturesthroughout the research. Qualitative datastructured the design of the quantitative researchtools such as the social-network survey, theconstruction of the cultural model and thevalidation of the model. This process produced anempirically grounded set of components for themodel. The findings pertaining to key behaviorpatterns from the qualitative analysis correlatedwell with the statistical analysis of survey data andsociometric analysis of the social structure data.The process of triangulating the quantitative andqualitative data is also supported by the emergenceof cultural themes from study-participant quotes.Triangulation (i.e. qualitative reliability andvalidity processes) enabled us to conduct aninternal-consistency check of the findings fromone source of data (e.g. the social-network survey)by linking and cross-referencing them with datathat were not collected in the same time frame orwith the same technique.

We extended this iterative process when wecombined social-network analysis, statisticalanalysis, with simulation of the model using SDsoftware to address the structure and the dynamicaspects of organizations simultaneously.

Another gap is addressed by focusing on theevolution of the collaborative relationships overtime. Most previous investigations have typicallyignored the longitudinal aspects of networks.

Our model is also innovative in terms of itsability to represent the culture of an inter-organizational entity. It is able to combinegenuinely diverse populations rather than focus ona single homogeneous organization. Most studies,to date, have been restricted to single-organizationissues, rather than the more complicated inter-organizational collaborative partnerships.

Our creation of the concept of relationshipdynamics is probably the most important andunique feature of the research. This new conceptresults from and affects collaborative partnerships.We created this concept to help explain the themeof reciprocity found in our qualitative data, as wellas the rules, roles and collaboration themes foundin an earlier data set involving GM's relationshipwith private-sector firms. Our research shows thatmany different socio-cultural componentscontribute to the partnering experience. Wedescribe the four most salient components in detailas they relate to these GM research partnerships.The current version of the model focuses onrelationships. In the future, we plan to expand the

Journal of Manufacturing Technology Management

Volume 15 . Number 7 . 2004 • 541-559

model to include such factors as the allocation ofhuman resources and funding to the partnerships.

3. Data and methods

3.1 Background on sample selectionMuch of the empirical foundation for our model isa study of GM R&D parmerships with fiveresearch institutions. The model comprehends theperspectives and experiences of the GMparticipants combined with the research-institution participants who collaborated withthem.

When we began our study, four CRLs wereworking with GM R&D in the US. Agreementswere drawn up between GM and the universitiesabout the structure and content of the joint work,the funding designated to support universitypersonnel, and the length of the agreement. Weapproached each of the Co-Directors of the CRLsto gauge their interest in participating in our study;all indicated their willingness to take part.

Because we wondered about the degree ofsimilarity across all types of research institutions,we included a research laboratory in our sample.The employees of this research laboratory engagedin research projects, as did the universityparticipants. In addition, the same concentratedattention in particular technical or "thrust" areas,and the designation of leadership roles within eachthrust area, were also points of similarity with theCRLs. Yet, at the same time, this researchlaboratory added diversity to the sample because itwas not a university. Moreover, GM has an equityinterest in this research laboratory. We believedthat crafting a partnership model from twodifferent types of research partnerships wouldmake our model more comprehensive.

3.2 Research designOur research design (Figure 1) incorporated amulti-method approach to data collection andanalysis; one of its outcomes was the relationshipdynamics model.

Since we began our research with littleknowledge of partnership structure and dynamicsin the research-institution partnerships, we workedinductively. First, we created research goals withGM managers that targeted their concerns.Second, we identified those data collectiontechniques that we believed would yield the mostuseful insights. For example, we developed open-ended questions for our interviews and focusgroups and built rapport with a cross-sectionalsample. Third, we used salient patterns from thequalitative data as a framework for the questionson the social-network survey, and as input to an

I

544


Gukin H. Sengir et al.

Journal of Manufacturing Technology ManagementVolume 15 • Number 7 • 2004 . 541—559

Figure 1 Partnership research design

InterviewsFocus Groups Qualitative Statistical

AnalyzeData

DevelopResearch

Goals

Social-NetworkSurvey

Sociometric NetworkVisualization

CaptureSystems

Dynamics

ConstructBaselineModel

early version of the model. Fourth, we combinedthe statistical and sociometric analyses from thesurvey data with network-visualization programsto explore the dynamics and evolutionary patternsassociated with partnerships. Fifth, we validatedour qualitative and quantitative results innumerous open seminars at GM R&D, the fouruniversities, and the research laboratory.

Our next step involved constructing a baselinemodel to show the inter-dependencies andevolutionary potential. We used the results fromboth the qualitative data and the social-networksurvey to create the key components of the model,calculating overall relationship effectiveness as aweighted average of all components. We also reliedon our data to create a hypothetical time lineconsistent with the stage-based insight from thequalitative data. Next, we simulated the modelusing a systems-dynamics approach. We generateda number of initial hypotheses from our data setand are now in the process of validating the modelby re-examining the empirical data for evidence ofthe simulated results.

3.3 Data collectionWe initiated the project by conducting in-depthinterviews with 83 individuals, 47 from GM and36 from the five research institutions; in eachpartnership, there were only two partners (i.e. theGM side and the research-institution side). Ourinterview questions focused on the nature of theparticipants ' past and current relationships withtheir counterparts, perceived success factors forand obstacles confronting the partnership,institutionaliorganizational cultures of thepartners, and expectations about the future of the

partnership. Our ten focus groups had nineparticipants on average, combining the views andexperience of over 90 individuals. The focus-groupquestions explored partnership goals andexpectations, the participants ' current assessmentof the partnership, recipes for an ideal partnershipand ideas for strengthening these long-termrelationships.

The social-network survey was subsequentlydistributed via e-mail to all those associated withthe five partnerships. We designed the survey togather data on partnership structure, dynamics androles. The survey contained questions about therespondents' roles, communication, joint work,trust, cooperation and conflict between therespondents and their counterparts, and status anddecision-making of the respondents relative to theircounterparts. The survey response rate rangedfrom 50 to 100 percent across the five partnerships.The 173 respondents produced an overall responserate of 60.5 percent, about twice the expectedreturn rate. They provided information on 505unique alters (i.e. people named), as well as a totalof 1,597 relationship pairs (see Appendix 1 for thelist of survey questions).

In Fall 2002, we conducted ten seminars orvalidation sessions (Kirk and Miller, 1986),attended by both those who were part of the initialdata collection process and those who had notparticipated in the study. These sessions allowedus to present what we had learned to date, andgather input on the validity and saliency of ourfindings. As such, they were an opportunity to testthe soundness of our analyses and interpretations,and to integrate new insights into our work. Wealso viewed these sessions as a way to provide

545


GOkin H. Sengir et al.

timely feedback to those who participated or wereinterested in the study.

3.4 Data analysisWe followed an empirical-analysis strategy thatwas both inductive and comparative. For thequalitative analyses, we focused on themes andpatterns that emerged from the interview andfocus group data (Bernard, 1998; Schensul andLeCompte, 1999; Trotter and Schensul, 1998).These analyses entailed the• discovery of key content elements (e.g. the

structure of partnerships, the collaborationprocess, relationships, values, ideas, etc.);found in the interviews and focus groups;

• identification of key quotes or verbatim dataillustrating important themes or content areas;and

• linkages pertaining to themes and patternsacross all the interviews and focus groups.

With the qualitative analyses as a foundation, weextended our understanding of partnershipstructure and dynamics through severalquantitative analyses based on the social-networksurvey. We then compared the results from ourqualitative analyses with the quantitative analyses(e.g. sociometric, network-visualization, statistical,etc.) from the social-network survey.

3.4.1 Qualitative analysesOur interview and focus-group data from bothsides of the five partnerships indicated that mostaspects of the collaborations were working verywell. Therefore, we made an initial assumptionthat the structure and dynamics of the partnershiprelationships were generally tending towards a"typical" partnership structure, and that thenetwork structure of each partnership could beused to represent that structure for the particulartime frame or stage. We used this projectedstructure to create the baseline model of theexisting partnership patterns.

For the CRLs, our initial qualitative analyses ofthe structure of the partnerships led us to create asix-stage partnership cycle. We labeled the stagesSelection, Courtship, Start-Up, Mid-Term,Mature and Transition. We found that three ofthese stages were well matched with our social-network data – Start-Up, Mid-Term andMature[5]. We also found that the partnershipcycle associated with the research laboratorypartnership was compressed compared with theCRL partnership cycle. Because the projectsbetween GM and the research laboratory werebudgeted and reviewed on an annual basis, thoseparticipants experienced all six stages in any givenyear. As a result of the qualitative analysis, we

Journal of Manufacturing Technology ManagementVolume 15 • Number 7 • 2004 • 541—559

constructed the model with a time-basedframework [6] .

3.4.2 Sociometric and network-visualization analysesNext, we analyzed the quantitative data from thesocial-network survey using egocentric networkanalysis, sociometric analysis and network-visualization techniques (Wasserman et al., 1994).The egocentric analysis allowed us to compareindividuals and groups with respect tocommunication, trust and conflict, among othervariables. With the sociometric data [7), we pooledall the responses and analyzed the structure of thepartnerships and the role dynamics. Graphicdisplays of the sociometric data revealed evidenceof the time-based framework noted in thequalitative data. Through network-visualizationprograms, which utilize both two- and three-dimensional visualizations, we were able tocompare the structural analysis of the partnershipnetworks both within and across partnerships[8].In particular, these programs illustrated aprogression of structural complexity as wecompared partnerships by age (see Appendix 2 fora complete listing of the software programs).We integrated these findings into our model bymaking the assumption that collaborativerelationships pass through increasingly-complexstages in a partnership cycle.

3.4.3 Statistical analysesWe also analyzed the survey data statistically togenerate quantitative input for the systems-dynamics elements of the cultural model[9].We performed cross tabulations on the surveyresponses. Next, we pooled the GM and theresearch-institution participants separately andlooked for patterns to characterize the partners.Then, we clustered the responses from each set ofpartnership participants to create a profile of thefive partnerships[10]. As with the qualitative data,we made an initial assumption that eachpartnership represented an appropriate examplefor a generic partnership at that particularstage[l1]. We also examined correlations betweenkey issues surfacing between the partneringorganizations. Then we used the correlations tocalibrate the relationships between the componentsof the model, and the frequencies computed aspercentages as input to simulate the model.

4. Relationship dynamics model

4.1 Rationale for the modelOur model specifies relationship effectiveness as acomposite of four basic system-level components:communication, joint-work, quality of interactionand connectivity of social structure. Our rationale

546


GOlcin H. Sengir et al.

for the selection of these components is based onour knowledge of the social-science literature, andanalyses from our data sets on GM partnerships.

The primary input into the conceptualization ofthe model was the initial qualitative analysis of theinterview and focus-group data from these fiveresearch-institution partnerships, along withearlier analyses of the GM's relationships withprivate-sector partnerships (Meerwarth et al.,2002) and GM's global product programrelationships (Briody, 1998; Briody et al., 2004forthcoming). For example, past observations ofGM relationships with internal GM partners, andthe scientific literature more broadly, indicate thatpartnerships routinely face conflict. Conflictemerges under a variety of circumstances includingthe existence of divergent technical and businessgoals, lack of agreement on work processes andpractices, and differences in expectations aboutroles. Therefore, we created a subcomponent ofthe model that took level of conflict into account.

Following the thematic leads provided by thequalitative analysis, we then searched the businessand social-science literatures for secondary dataand theoretical confirmation of the componentsthat were emerging for the model. We suspectedthat our preliminary selection of componentswould be confirmed as key concepts or concerns ina substantive keyword search of the existingbusiness research literature. The search results[12]provided strong support for our basic assumptions,as well as individual citations on the keycomponents highlighted in this paper. As anexample, there were 236,046 citations available inthe business literature on the concept ofcommunication, 50,208 on trust, 39,133 onconflict, 36,278 on cooperation, 2,976 onorganizational structure, and approximately 1,000corresponding to our concept of joint work (basedon the combination of several categories). Therewas also a substantive literature on the use ofsocial-network analysis in business contexts. Thecontent of selected articles from this search furtherconfirmed the centrality of the components of theemerging model in the creation and maintenanceof successful partnerships.

A third input into the conceptualization of themodel involved the analyses generated from thesocial-network survey. The range of responses toeach question suggested that the questions weretapping into critical partnership variables. Weasked individuals to rate their relationship witheach person they named in their network fromlowest (or poorest) to highest (or most positive),using a six-point scale. The fact that the wholerange of responses was used by at least some of therespondents indicates that we were measuringconditions that both varied within and across the


Volume 15 . Number 7 . 2004 . 541—559

partnerships, and could be critically judged by therespondents. Clearly, all relationships were not thesame, or based solely on some kind of social-desirability condition. In addition, the fact thateach respondent varied his/her ratings, rather thansimply duplicating the ratings for all individualsthey named, also indicates that we were tappinginto conditions that were measurable variables inpartnership relationships.

4.2 Description of the modelThe complementary sources of input to the modelconfirmed the appropriateness of culture-levelcomponents rather than a focus on thecharacteristics of the respondents. Ourcomponents are necessarily composite Conditionstargeting system-level dynamics rather thanindividual dynamics. Individuals have an impacton the system on a person-to-person basis.However, these individual effects are both shapedby and filtered through cultural norms,expectations, roles and reciprocity rules. Effectivesystem-level changes must take into account andaddress the culture of the entire system, ratherthan simply rely on individual-level change ordynamics. Therefore, the most powerful modelingand intervention opportunities for thecollaborative ventures occur at the system- orcultural-level, and not at the individual- orpsychosocial-level. We now turn to a description ofour four key components and their criticalsubcomponents.

4.2.1 CommunicationOur qualitative data indicate that goodcommunication is the most frequently-citedideal component that both parties want out ofa partnership. The importance of communicationis reflected in numerous statements in theinterviews, and most succinctly summarizedby one of the CRL thrust leaders when hestated, "Communication is the key. Weneed to know the GM requirements"[13].Consequently, we collected survey data on twoaspects of communication: the importanceof the communication and the frequency ofcommunication between each set of individualsengaged in the partnerships.

4.2.2 Joint workJoint work, or the process of achieving the primarygoals of the thrust areas, is both the raison d'etreholding the relationships together, and the drivingforce justifying the establishment of thepartnership in the first place. As one CRL thrustleader stated, "The most important ingredient [ina collaborative partnership] is working jointly tokeep these guys interested in the program". Jointwork also encourages the continued participation

547



of GM and the partners in the collaborationdespite difficulties during the partnership cycle.There were numerous indications in the qualitativedata that the time invested in the joint work wascritical, and sometimes a pressure point on theoverall relationship. As one GM research managerexpressed, "We should put more effort into it onour side. We set it up and put some effort in, butwe aren't engaged as much. That's the danger. Youcan't get much out of these kinds of things unlessyou put the time in. Money is no substitute".Consequently, we decided to measure theperceived importance of joint work between all ofthe individuals surveyed and their counterparts byasking about the relative importance of joint-workactivities. We also asked about the frequency of thejoint work conducted between collaboratingpartners.

4.2.3 Quality of interactionThe qualitative data indicated that there were atleast three crucial cultural themes – trust,cooperation and conflict – that have a directimpact on the quality of the interaction. Oursocial-network survey measured the perceivedlevels of trust, cooperation and conflict that existedbetween the survey respondents and each of theindividuals they named in the social-networksurvey. These data provide both an individual-leveland a partnership-level measure of the quality ofinteraction.

Trust. Participants often emphasized theimportance of building and maintaining theirrelationships with their partners to enhance trust.One GM researcher stated, ... "there has to beconfidence and trust between the two groups –trust that you'll get results, and trust that therelationship will develop into a good one". In fact,trust was the most frequently-mentioned featurewhen quality of interaction was discussed.Therefore, we constructed a question in the social-network survey that directly measured the level oftrust between each of the respondents and thosethey named in their partnership.

Cooperation. The concept of cooperation alsoemerged frequently in the qualitative data. Itappeared both separately and in conjunction withthe concept of trust, and therefore, became one ofthe key themes in the data. As one CRL researchersaid, "Cooperation must be nurtured", anotherstated that, "The most important thing is peoplehave to feel comfortable with each other becausethat certainly makes it a lot easier. You need to beable to talk to each other even if you are unhappy,and then you can try to fix it". The general use ofthe term cooperation, and the emphasis on positiverelationships in the qualitative data, caused us toinclude a survey question designed as a measure ofperceived person-to-person cooperation levels.


Volume 15 - Number 7 • 2004 • 541-559

Conflict. The qualitative data generally focusedon what was going well within these research-institution partnerships. However, our studyparticipants did provide examples of periodic,temporary, or sporadic conflict. In one example, aGM researcher indicated that, "It's a forcedrelationship. What makes you think you can workwith all those people?" On another occasion, oneof the research laboratory researchers stated,"Sometimes we have those conflicts with technicaldirections of what's to be achieved and what's not.Then I think [of] those conflict issues, [and] wejust have to have a conference ... and we try toresolve those issues". Some of the conflict was dueto the cultural differences between GM and theresearch institutions (e.g. issues of intellectualproperty rights, publication, patents, etc.). Inother instances, conflict arose from problemsassociated with either structure (e.g. differences inhierarchies, institutional processes, etc.) orindividual interactions. Differences between thepartners can result in conflict that then impactscommunication, joint work and even the structureof the relationship. The qualitative indicatorssuggesting that some conflict is present even in thebest of collaborations motivated us to ask aboutthe level of perceived partnership conflict.

4.2.4 Connectivity of social structureKey individuals in the partnerships recognized theneed for a complex social structure that maximizedthe positive components of communication, jointwork and the quality of interaction, whileminimizing the competition among these threecomponents. One GM researcher explained thathis partnership 's success was due, in part, to hiscounterpart's efforts to establish and maintainconnections with him and his GM colleagues. Hecommented, "We've gotten to know each otherbetter. [One of the CRL leaders] does an excellentjob of keeping us informed, and involved, and hisfaculty involv ed". Following this qualitative lead,we decided to examine two role variables thatexpress the key structural components of therelationship. The components chosen to constructthis segment of the model include role dynamicsexpressed as fragmentation and reach, andstructural components expressed as networkdensity, transitivity and betweenness-centrality.

Role dynamics. The role dynamics portion of themodel consists of one positive and one negativeforce within the sociometric data.

(1) Fragmentation. We define fragmentation as ameasure of the amount of dislocation of individualconnections in the network caused by the removalof "key players" (i.e. central figures in apartnership) and their connections to others.Fragmentation can be conceptualized as thecreation of individual islands or clusters of

548


Giilcin H. Sengir et al.

relationships with no bridge between them; assuch, it acts as a negative force in the sociometricdata. Researchers and managers in thesepartnerships acknowledge the tendency of thecollaboration to fragment over time. Such apattern makes the issue of fragmentation central tothe structure and dynamics of collaborativepartnerships. One CRL research managercommented, "There's a very natural tendency fortwo institutions to set up a collaborative projectand then have that collaboration naturallyfragment or naturally segment". We were able tomeasure the impact of fragmentation in the modelby removing key players from the overall structureof the partnership. We could then determine theimpact of that removal on other components in themodel.

The impact of fragmentation is shown inFigure 2 when one or more key players areremoved from the partnership. It shows a smallnetwork (Configuration A) initially connected bytwo key players (1 and 8). The loss of key player 8(Configuration B) means that the subunitrepresented by individuals 9-12 is no longerconnected to the whole; the network isfragmented. In contrast, the loss of key player 1(Configuration C) translates into a much lesseffective configuration of the partnership at thecore. This condition leads to longercommunication lines between all individuals,though all of the individuals in the network are stillconnected, that is, they are not fragmented. Bothof these conditions are important in understandingthe impact of the loss of key players in any givenpartnership.

(2) Reach. A second role-based dynamicexpressed in the qualitative data is the need for

Journal of Manufacturing Technology ManagementVolume 15 • Number 7 . 2004 • 541—559

individuals who can communicate and reinforcethe primary goals of the partnership to themaximum number of network partners. Reach is apositive force in the sociometric data. The need forgood reach is alluded to in the following CRLresearcher statement:

He sees his job as director to clearly articulate thelong-term goals, to clearly articulate the directionsthat we want people to go on, but then, to not tellpeople what to do [which might restrict creativity]– to essentially tell them where we want to get andprovide the freedom for them to actually get therein the best way they can.

Consequently, we used a measure of reach toidentify the key individual or individuals linked toas many distinct partnership participants aspossible. Reach provides a measure of theproportion of the total network that a singleindividual is either in direct or indirect contactwith at a given point of time. In Figure 2(Configuration A), key player 1 can easily contactand/or influence either the whole network, ormajor parts of it, directly or through a minimumnumber of intermediates. The partnership-network data indicate that participants havedifferent levels or spreads of reach, depending onthe size and complexity of the partnership as itchanges over time.

Structure. We used basic sociometric analysis ofthe social-network survey data to compute threesubcomponents of the partnership networks thathelp define the structure of connections andconnectivity in the partnerships. Thesecomponents include density, transitivity andbetweenness-centrality.

(1) Density. Density provides a measure for theoverall amount of "connectedness" in a network.

Figure 2 Examples of fragmentation potential of key playersll4]

Configuration A Configuration B Configuration C

Source: Adapted from Borgatti (2003, p. 242)

549

Modeling relationship dynamics in GM's research-institution partnershipsGulcin H. Sengir et al.

This measure is derived by counting the totalnumber of connections that exist between peoplein a group, and then counting the total number ofpossible connections. Dividing the existingconnections by the potential connections producesa number that identifies the proportion of allexisting tics among individuals to all potential ties.The higher the number, the denser is the network.When the network is dense, it provides moreopportunities for alternate routing through thenetwork in case of a failure of one or more links.One GM researcher stated it this way,"The partnership has become much stronger andit's more synergistic and intertwined. We considerit our responsibility that the work (of the differentCRLs) is complementary and synergistic".Density translates into a more stable structure forthe whole network. It also reduces the problemsthat occur when someone is removed from thenetwork, thereby resulting in a loss of connections.In Figure 2 (Configuration A), the core groupconnected by the key player 1 has a higher numberof actual relationships compared to possiblerelationships. Therefore, this core group is denserthan the subgroup of key player 8; the latter hasfewer total connections compared with potentialconnections.

(2) Transitivity. Transitivity is a sociometricmeasure that identifies the proportion of triples(i.e. three people all connected to each other) thatare connected, compared with the potential totalnumber of these triples. It provides a measure ofthe connections between the individuals who areconnected to a central person, rather than a simplemeasure of all connections. The need for thismeasure was captured in a quote from a GMparticipant who praised the ability of one of theCRL partners to make these kinds of connectionsbetween individuals: "During the [joint meetings],he [the CRL partner] does an excellent job ofpresenting to us, bringing in others from outsidehis department, and that has led to somerelationships [with those other departments andindividuals]". Transitivity is similar to density inthat it provides another way of examining theoverall connectedness and stability of a network.Referring again to Figure 2 (Configuration A), keyplayer 1 is part of several triples (e.g. 2-1-7, 4-1-5)in which three people are connected to each other.These triples are examples of transitiverelationships. By contrast, key player 8 has onetransitive relationship (i.e. 1-8-7), and thatsubgroup in the partnership has a lower transitivityvalue than the core group of key player 1.

(3) Betweenness-centrality. Some individualsoccupy central positions in a collaborativenetwork. Consequently, they become key linkagepoints for good communication flow to either part

Journal of Manufacturing Technology ManagementVolume 15 . Number 7 . 2004 • 541—559

or all of the network. These individuals exhibit acondition that we label "centrality." One GMleader working with the research laboratorypartnership described this type of centrality ininformation flow in the following statement:"The most frequent contact is the PI [PrincipalInvestigator] to the [GM] Champion, and a lot ofthe communication goes along there (and out toothers)". There are several different sociometricforms of centrality. We chose to use betweenness-centrality, which is formally defined as the numberof times a vertex occurs on a geodesic [15]. Figure 2shows that key players 1 and 8 exhibit the type ofcentrality that is expected of a GM Champion orPI. Individuals in these types of positions typicallycontrol the flow of information, influence othersand/or hold the network together.

4.2.5 The relationship dynamics modelThe overall connections between the four keycomponents of our relationship dynamics modelcan be illustrated as an interactive web. Each of thefour components, in turn, is a composite ofrelationship subcomponents that further definebehavior and interactions. We illustrate the modelin Figure 3.

5. Simulating relationship dynamics

5.1 Basic building blocks to represent themodelWe hypothesize that each of the four system-levelcomponents – quality of interaction, socialstructure, joint work and communication –contributes differentially to relationshipeffectiveness in these collaborative partnerships.Representing relationship effectiveness in a modelrequires taking into account the varying levels ofinteractions and feedback. At any particular time,relationship effectiveness is an aggregated outcomeof these four components and any subcomponents.In turn, each component impacts and is affectedby the composite relationship effectiveness and thevarying levels of the other composite components.We are in the process of validating the model withexisting qualitative data, reviews with studyparticipants and results from a follow-up survey.

Systems-dynamics is a standard way to examinethe interactions among the components of themodel and the evolutionary change at the system-level. In a systems-dynamics framework, each ofthe components has both positive and negativeeffects on each part of the system and the system asa whole. This framework enables an investigationof the non-intuitive dynamics among the keycomponents. A systems-dynamics approach hasthe ability to represent the emerging behavior of

550

Modeling relationship dynamics in GM's research-institution partnershipsakin H. Sengir et al.

Joint WorkImportance

Journal of Manufacturing Technology ManagementVolume 15 • Number 7 . 2004 • 541-559

Joint WorFrequency

Figure 3 Relationship dynamics model

Connectivity ofSocial Structure

Densit

Role Dynamics

Reach Fragmentation

Structure

TransitivitBetweenness-

Centrality

interacting loops (i.e. balancing, reinforcing ordraining feedback), the ability to represent non-linear effects, and the use of continuous-timerepresentation. In addition, by tracing through thefeedback-loops, we can explore the evolutionaryprocesses over time (Sastry, 1997, 2001). Thus, ithas the potential to lead to new hypotheses,research questions and extensions of the currentmodel.

In Figure 4, we show how we visualize quality of

interaction. We represent it using cooperation,trust and conflict as the building blocks.

These three subcomponents are represented asrectangles (or "stocks " in systems-dynamics

terminology), which are "stocked" by(i.e. accumulate or contain) the survey data thatflow into them. Levels in these stocks vary overtime based on various direct and indirect forces."Flows" represent the actions or activities overtime. We can change the levels of stocks via"converters" (depicted as circles) that representrelationships as mathematical equations. Trust,conflict and cooperation are linked (indicated bythin arrows, and referred to as "connectors" insystems-dynamics terminology) to form acomposite component in the diagram – in thiscase, quality of interaction. Quality of interaction isrepresented as a "converter," an element that

Figure 4 Example of representing components of the model as stocks and flows[l61

quality ofinteraction

change incooperation level

cooperation levelfrom survey data

effect of conflictchange inconflict level

effect ofquality ofinteraction

conflict lccelfrom survey data

effect of conflict change intrust level

trust levelfrom survey data

551


Gufcin H. Sengir et al.

converts these inputs into an output, which thenleads to the other components of the model.

There are also feedback-loops (shown as thinloops) that represent feedback or causality fromincreasing and decreasing levels of conflict oncooperation and trust to the other components.Quality of interaction, which represents the inter-twined set of interdependent relationships betweencooperation, conflict and trust, in turn affects thelevels of these and other key components. Werepresent these cause-and-effect relationships withthe loops and converters. Because relationships areneither static nor linear, we illustrate the model asa closed-loop; dependent and interdependentvariables become part of a web ofinterrelationships.

5.2 Simulating the modelOur first round of qualitative and survey dataenabled us to conceptualize the model, identify thefour key components, and conduct some initialvalidation. As such we feel that we have a proof ofconcept. However, because of our current datalimitations (e.g. the current data represent onlyone point in time), we make some initialassumptions in developing the model. We describethese assumptions below, anticipating that somewill be relaxed and/or revised based on ourfollow-up survey to participants in thesepartnerships. In particular, this follow-up surveywill provide us with data at a second point in time,and increase the sample size from five partnershipsto nine. In later analyses, we also will examine thepartnerships by technical or "thrust" areas.Thus, the second survey responses added to theinitial data will enable us to estimate the modelparameters more accurately and increaseconfidence in the simulations involving dynamics.

The partnerships from our first round of datacollection were at different stages in theirpartnership life cycle when we collected the surveydata. As indicated earlier, we used the survey datato create a hypothetical five-year timeline, witheach partnership representing an example of ageneric partnership at that particular stage ofdevelopment[17]. We then used this assumption todevelop a baseline composite partnership over afive-year time frame. We normalized the data toaccommodate differences in the number ofrespondents associated with each partnership.

Since our youngest partnership was in its firstyear, the initial stage depicted in the model is thefirst year; we label it the Start-Up stage (Section3.4). Organizational theory indicates that at thebeginning, organizations have low inertiameaning young organizations can change easilybut the performance levels are low (Sastry, 2001).This baseline condition matches well with the


Volume 15 • Number 7 • 2004 • 541–559

qualitative data that we collected about the start-up processes for each of the collaborations. Datafrom a partnership 1.5 years into the partnershipcycle represent partnerships towards the end of theStart-Up stage. The Mid-Term stage isrepresented by a partnership that was two years oldat the time of the survey. Finally, two partnershipswere at years 3 and 4, which were associated with

the Mature and Late Mature stages, respectively.We know from the qualitative data that eachpartnership stage is depicted with particularattributes and issues.

5.2.1 Simulation procedureWe initialized the stocks in the model with datafrom the social-network survey to representdifferent levels of interaction among participants.We picked the data values at the 5th, 50th and 95thpercentile ranges to represent low, average andhigh levels for all of the components [ 18] . We alsocalculated the survey values as percentages for therole dynamics and connectivity components(i.e. density, transitivity, centrality fragmentation;reach, etc.) using network role programs(e.g. Key Player), and basic sociometric programs

(e.g. UCINET X); we calibrated the stockscorresponding to these components.

To model the dependencies in the system, weconstructed the underlying pattern from thecorrelations between the key componentsobserved in the statistical inferences in the survey

data. We calculated the composite values as aweighted average of the contributing componentsthat include these correlations. Because themodel's components interact with each otherdirectly, the components cannot be viewedindependently (Reaume and Alden, 2002). Forexample, if a component like trust is increased,overall relationship effectiveness is improved;indirectly, other components and subcomponentsare also likely to be enhanced, further improvingthe entire relationship.

Despite the sampling limitations, we wereinterested in the potential use of the model as apredictive tool. Therefore, we calibrated the modelby setting the appropriate parameter values andfunctional relationships consistent with thechanges observed in the partnership cycle. Wechose Euler's method to estimate the change instocks over the interval dt.

:stock = dt • flow

where flow is the rate of change correspondingto the stage in the partnership cycle.We calculated a new value for stocks based on thisestimate:

stock, = stock,_ d , + Astock

552



We compared the approximations with ourbaseline partnership behavior. Then we set thesimulation time-step (dt) to 0.25, where the unitof time is a year corresponding to the stage inthe partnership cycle. This time-step sets theinterval of time between calculations to 1/4th ofa year, which seemed feasible and sufficientto simulate the pattern observed from theempirical data.

Using regression, we calculated the flow rates(i.e. a temporal feature derived from the surveydata) to reflect how composites might behave overtime. This rate of change was determined as theslope of the best-fit line between the baseline-model data points. We initialized the model tosteady-state, representing an organization that isnot experiencing change. Then, we started varyingthe levels of stocks one by one, while observing theeffects on the composite components. We thengenerated several hypotheses to begin to test themodel.

5.2.2 Understanding effects: examples of trust andconflictIn this last section of the paper, we wanted to beginto examine the utility of the relationship-effectiveness model. Therefore, we returned to thequalitative data to generate an initial list ofhypotheses about the components and the ways inwhich they were connected. It struck us that someof the more interesting aspects of the partneringrelationship pertained to trust and conflict. Weheard repeatedly, for example, that without trust,the relationship would not be successful in thelong-term. Indeed, interviewees indicated thattrust was a necessary condition for partnershipeffectiveness. By contrast, interviewee storiessuggested that when conflict erupted, it could bequite potent, though not necessarily long-lasting.Moreover, when we examined the correlationbetween the level of trust and conflict respondentsexpressed for the counterparts they named, wefound a negative correlation (- - 0.4) at theindividual-level. Thus, in this section, to illustratethe model's utility, we will focus our discussionon these two subcomponents of quality ofinteraction.

Comparison of trust and conflict. From ourqualitative interviews we learned that trust was oneof the most frequently-mentioned elementsneeded in a collaborative relationship. One GMresearch manager remarked that, "There is alwaysthis trust hurdle you have to get by. In an idealpartnership, you would achieve a workable-level oftrust very early on". Figure 5 summarizes thefrequencies of the survey respondents ' ratingspertaining to trust. We use a scale of 0 to 6, where 0indicates the lowest level of trust and 6 indicatesthe highest level of trust. Most respondents' trust


Volume 15 • Number 7 • 2004 • 541—559

Figure 5 Levels of trust on a scale from 0 to 6

0 Trust at Start-Up Stage80 ■ Trust at Late Start-Up Stage

70 0 Trust at Mid-Term Stage0 Trust at Mature Stage■ Trust at Late Mature Stage

Levels

rate high, either as 5 or 6 on the rating scale.However, at least some participants in thepartnerships have very low levels of trust with atleast some of the people they named. Indeed,during the Start-Up phase, some significant trustissues appear (e.g. intellectual-property rights,number and type of reviews, etc.).

The survey data also indicate that conflict levelsare low in these partnerships. Figure 6 shows howthe overall ratings of conflict vary over thepartnership cycle. As with trust, we used a scale of0 to 6, where 0 indicates the lowest level ofconflict and 6 indicates the highest level of conflict.Low conflict may be a sign of the potential forsuccess in these partnerships – at least from thestandpoint of the relationships amongparticipants. Conflict, while low, is an importantdimension of these partnerships. Two quotes –one from a university researcher and one from

Figure 6 Conflict level ratings on a scale from 0 to 6

0Conflict at Start-Up Stage

■ Conflict at Late Start-Up Stage

0 Conflict at Mid-Term Stage

0Conflict at Mature Stage

■ Conflict at Late Mature Stage

50

40

30

20

10

c 60mco 50a.

ce 400

30v

20a.

10

01 40 2 3

r15 6

80

90

70

60

0 IIIIUr,rIfl-1 ►ti_n _

0 1 2 3 4 5 6

553



a GM researcher – illustrate two differentperspectives on the direction given to andmanagement of the CRL work. The universityresearcher said, "We can't have GM corning intelling us what to do. The final decision on thescope of the work has to be with the faculty". TheGM researcher expressed it differently: "I get thefeeling from [the university] that this CRL is apart-time endeavor for them... They are willing todo it but on their time... You [GM] have to talk orhold back money to get accountability from theuniversities". Comments like these clearly indicatea difference of opinion or a difference inexpectations between the partners. If such adifference is not discussed and addressed, it couldexpand to encompass other issues and result inincreasing conflict.

Figure 7 compares average trust and averageconflict over the partnership cycle. This figureshows some interesting patterns. First, surveyparticipants' scaled responses show the remarkablecontrast between trust and conflict in these fivepartnerships: trust is very high and conflict is verylow. This pattern confirms much of our qualitativedata that these partnerships, as a whole, areworking well. A second pattern that emerges is thattrust is largely the mirror image of conflict. Whilethe differences between the partnerships are small,the correlation between trust and conflict holdsnot just at the individual-level, but at the partner-level as well.

Finally, the scaled responses suggest that thepatterns for these two subcomponents arerelatively stable over the course of the partnershipcycle with only some variation by stage. Forexample, conflict is somewhat higher at the startand the end of the partnership cycle comparedwith the middle stages of the cycle. We know fromthe qualitative data that the initial stage of the cycleis stressful for participants because they are tryingto develop relationships, agree on their projectgoals and create a process for working together.

Figure 7 Average trust and conflict in partnership cycle

6

s

Start-Up Late Start- Mid-Term Mature LateStage Up Stage Stage Stage Mature

Stage


Volume 15 • Number 7 • 2004 • 541—559

Similarly, the end of the cycle is associated withsome increased conflict because there isuncertainty about renewal – whether of individualprojects as in the case of the equity partnership, orcollaborative labs as in the case of the universitypartnerships. Yet, even with this slightly elevatedlevel on either end of the partnership cycle, therelative stability is an indicator of evenly-balancedrelationships.

Hypotheses. While we believe that both trust andconflict can have wide-sweeping effects on theoverall partnering relationship due to ourqualitative results, we are currently unable to testthese effects due to data limitations. However, weare able to examine how trust and conflict canimpact other components or subcomponents inthe model directly by looking at individualdifferences within each partnership. For example,from the qualitative data, we found that some ofthe relationships between individual counterpartsgot off to a rocky start during the Start-Up phase.For example, issues emerged as goals andprocedures for joint work were established. In onecase involving the late Start-Up partnership, somewithin the CRL expressed concerns that GM wasattempting to direct their research focus. Sincethese concerns continued to be expressed overmany months, we hypothesized that low trustresulted, which then led to low communicationfrequency. A cross-tabulation of trust andcommunication fztquency from the survey dataindicated that this hypothesis was supported.

In another example from the qualitative datainvolving an early Start-Up partnership, we learnedthat there were concerns on both sides of thepartnership that GM researchers had insufficienttime to devote to their partnership work.Therefore, we hypothesized that those reportinglow levels of joint-work frequency would reporthigher levels of conflict during Start-Up. When weperformed cross-tabulations on the survey data, wefound support for this hypothesis as well.

To complement these Start-Up hypotheses, wedecided to focus on the quality of interaction in alater stage of the partnership cycle. We thought itwould be interesting to explore the relationshipbetween conflict and cooperation in GM'sresearch laboratory partnership compared with thefour university partnerships. We suspected thatthose in the GM research laboratory partnershipwould have to take, and be more willing to take,direction from GM due to the equity relationshipthan would those in the university partnerships. Ofcourse, in any of these five partnerships, conflictcould arise (e.g. due to differences in expectations,disagreements over technical issues, etc.).However, we suspected that cooperation was likelyto vary between the equity (i.e. research

4N

m 3-

2

0

-4— Average Trust—I— Average Conflict

554



laboratory) and non-equity (i.e. university)partnerships. Therefore, we hypothesized thateven if conflict were high for the participants in theequity partnership, their cooperation would behigh. By contrast, if conflict were high forparticipants in the CRL partnerships, theircooperation ratings would be low. Cross-tabulations of conflict and cooperation from thesurvey data showed a distinct difference in thepattern: those working on GM and its equitypartner's projects who reported high conflict stillreported good cooperation, while partnershipcooperation was low or non-existent among thosewho reported high conflict in partnershipsinvolving GM and the CRL universities.

In our follow-up survey, we will continue tofocus on the structure and dynamics of GMresearch-institution partnerships. We will use thefollow-up survey data to explore whether or notthe differences we observe across and withinpartnerships are due to distinctive aspects of agiven partnership, or due to the evolutionaryprocess inherent in the partnership cycle. We alsowill add recently-created partnerships to thesample which will both increase our sample sizeand add more diversity.

6. Conclusions

6.1 Enhancing partnership success

• Partnering, an increasingly-popular businessstrategy, links organizations and institutionstogether to collaborate on projects, buildcompetencies and enhance competitiveness.Yet partnerships exhibit a high rate of failure,typically caused by organizational and culturaldifferences between the partners. Identifyingways of reducing the incidence of failure andincreasing the likelihood of success will haveenormous benefits for the partneringorganizations.

• Partnership success is largely dependent uponthe development and maintenance of strong,productive relationships between the partners.Without strong relationships, there is neither acommitment to the partner nor the likelihoodof achieving partnership goals.

6.2 Modeling relationships

• Partnering relationships are part of a dynamicsystem that can be modeled. Relationships canbe influenced and shaped by certain factors,and analyzed for common themes andpatterns. The modeling results can be used todiagnose and predict partnership differences.Interventions can then be designed and

Journal of Manufacturing Technology ManagementVolume 15 • Number 7 • 2004 . 541—559

implemented to improve partnershipeffectiveness.Modeling partnering relationships iscomplicated because they are complexcultural interactions, and because theychange over time. To increase the likelihoodthat the resulting model is realistic, validand representative, a systems-dynamicsapproach can be combined with empiricaldata, and then validated with feedback fromstudy participants. Systems-dynamics isused to study the interactions among thecomponents of the model and to exploretheir evolution.Partnerships go through stages in predictableways that can be captured in a dynamic model.Although the concept of a stage-basedpartnership cycle came from the qualitativedata, partnership stages can be applieddirectly to the social-network survey data,supporting our initial findings of partnershipevolution.

6.3 Developing model components andhypotheses

(1) Four key components of relationshipeffectiveness emerged from the analysis of thequalitative data and social-network survey –communication, joint work, quality ofinteraction and connectivity of socialstructure. These components and theirassociated subcomponents became thefoundation for the relationship dynamicsmodel.• Joint work is composed of both the

frequency and importance of jointactivities.

• Quality of interaction is constructed fromthe values associated with three keycultural themes and processes: trust,cooperation and conflict.

• Communication is composed of thefrequency and the importance ofcommunication with each partner.

• Connectivity of social structure includestwo subcomponents: structure and roledynamics.

Independent theoretical justification for thesecomponents and subcomponents, derivedfrom a search of the business researchliterature, confirmed their centrality in thecreation and maintenance of partnerships.At any particular time, relationshipeffectiveness is an aggregated outcome ofthese four components and subcomponents,their interaction effects, and any change overtime.

555


Gblcin H. Sengir et al.

(2) In examining two subcomponents - trust andconflict - over the course of the partnershipcycle, the pattern was relatively stable bystage, though the start and end of the cycletended to be more stressful than the middlestages. Interestingly, trust and conflict werenegatively correlated: trust was very high andconflict was very low. This pattern confirmsmuch of our qualitative data that thesepartnerships, as a whole, are working well. Inaddition, trust was largely the mirror image ofconflict, indicating that the correlationbetween trust and conflict holds at thepartner-level as well.

(3) An initial list of hypotheses was generatedfrom the qualitative data to begin testing themodel's utility. One hypothesis, for example,was that conflict and cooperation would varyby type of research-institution partnership(i.e. equity vs university). Cross-tabulations ofconflict and cooperation from the survey datashowed a distinct difference in the pattern:those working on GM and its equity partner'sprojects who reported high conflict stillreported good cooperation, while partnershipcooperation was low or non-existent amongthose who reported high conflict inpartnerships involving GM and the CRLuniversities.

Notes

1 The GM-Isuzu Motors Ltd relationship, dating to 1971,began with "simple technology exchanges" but now"extends to all phases of automaking - fromdevelopment through sales " (Isuzu, 1998, p. 26).

2 Stocks are reservoirs for the values that define the primaryelements in the model. Flows are the connections thatindicate the impact or relationship between one stock andanother.

3 I-Think, a web-based survey program, also can be used forinitial analysis of data in tabular formats.

4 STELLA from High Performance Systems, Inc. is a SDsoftware which enables creating models of structuralstock-flow-diagrams and time-series analysis. Powersimby Business Simulation Company, and Vensim by VentanaSystems, Inc. are other available SD modeling software.

5 We did not have social-network data on the Selection,Courtship or Transition stages.

6 Currently, we are collecting sociometric data on theCourtship and Transition stages in our follow-up survey,and will extend the model into these stages. The Selectionstage will be based only on the qualitative data.

7 We used several network programs, including ego-network programs (MULTINET, FATCAT), network roleprograms (Key Player) and basic sociometric programs(UCINET X) to conduct the egocentric and sociometricanalyses.

8 Three programs were particularly valuable for thevisualization analysis: PAJEK, NETDRAW and MAGE.

Journal of Manufacturing Technology ManagementVolume 15 • Number 7 • 2004 • 541-559

9 We used SAS and SPSS Statistical Software to obtain thedescriptive statistics.

10 A report on the main findings of the survey analysis is inprocess.

11 We acknowledge the need to separate the effects ofindividual partnership differences from stage-basedeffects of the partnership cycle (and are conducting afollow-up survey to address that issue).

12 We used an online search engine, EBSCO's BusinessSource Premier, for the basic key word search.

13 The italicized statements in the paper represent directquotes from our qualitative data.

14 This figure is adapted from Borgatti (2003, p. 242).15 In this case, the vertex is a key player with strong

communication capability since he/she can reach so manyso quickly in the network.

16 This figure is adapted from Richmond (1993).17 We felt reasonably comfortable making this assumption

because our qualitative data suggested that the olderpartnerships, at one time, experienced the kinds ofprocesses and tensions currently facing the youngerpartnerships.

18 We plan on adding more granularity in specifying inputlevels as we develop extensions to the model, which willcomprehend components that are not relationshipspecific.

References

Bernard, H.R. (Ed.) (1998), Handbook of Methods in CulturalAnthropology, Altamira Press, Walnut Creek, CA.

Borgatti, S.P. (2003), "The key player problem", DynamicSocial Network Modeling and Analysis: WorkshopSummary and Papers, The National Academy of Sciences,pp. 241-52.

Briody, E.K. (1998), "Meetings, Besprechungen, and meetings:dealing with differences in cross-cultural meeting styles",American Anthropological Association, 2-6 December,Philadelphia, PA.

Briody, E.K., Cavusgil, S.T. and Miller, S.R. (2004 forthcoming)," Why cross-unit corporate ventures may fail: the case ofthe delta small car program at general motors", LongRange Planning.

Carley, K.M. and Hill, V. (2001), "Structural change and learningwithin organizations", in Lomi, A. and Larsen, E.R. (Eds),Dynamics of Organizations, AAAI Press, Menlo Park, CA,pp. 63-92.

Catlin, L.B., Briody, E.K., Trotter, R.T. II, Meerwarth, T.L. andKulkarni, D.M. (2003), " Building in reciprocity: creating asolid foundation for R&D partnerships", Society forApplied Anthropology, 19-23 March, Portland, OR.

Cyert, R.M. and March, J.G. (1963), A Behavioral Theory of theFirm, Prentice-Hall, Englewood Cliffs, NJ.

Duysters, G., Kok, G. and Vaandrager, M. (1999), "Craftingsuccessful strategic technology partnerships", R&DManagement, Vol. 29 No. 4, pp. 343-51.

Ertel, D., Weiss, J. and Visioni, L.J. (2001), Managing AllianceRelationships: A Cross-Industry Study of How to Buildand Manage Successful Alliances, Vantage Partners,Brighton, MA.

Forrester, J.W. (1989), "The beginning of system dynamics",International Meeting of the System Dynamics Society,Stuttgart, 13 July.

GM Press Release (2001), "General motors to fund S3 millionwork-systems lab at Stanford University", 22 January.

556

Modeling relationship dynamics in GM's research-institution partnershipsGOlcin H. Sengir et al.

Gulati, R., Khanna, T. and Nohria, N. (1994), "Unilateralcommitments and the importance of process in alliances",Sloan Management Review, pp. 61-9.

1"Isuzu", company brochure], (1998).Kauffman, S.A. (1995), At Home in the Universe: The Search for

Laws of Self-Organization and Complexity, OxfordUniversity Press, Oxford.

Kirk, J. and Miller, M.L. (1986), Reliability and Validity inQualitative Research, Sage Publications, Beverly Hills, CA.

Lomi, A. and Larsen, E.R. (Eds) (2001), Dynamics ofOrganizations: Computational Modeling and OrganizationTheories, AAAI Press, Menlo Park, CA.

Meerwarth, T.L., Briody, E.K. and Kulkarni, D.M. (2002), "Thediscovery and exploration of partnership rules: amethodological perspective " , Society for AppliedAnthropology, 6-10 March, Portland, OR.

Prietula, M., Carley, K. and Gasser, L. (Eds) (1998), SimulatingOrganizations: Computational Models of Institutions andGroups, AAAI Press, Menlo Park, CA.

Reaume, D.J. and Alden, J.M. (2002), Decision Analysis UsingInfluence Diagrams, GM R&D Center Publication 9269,Warren, MI.

Richmond, B. (1993), "Systems thinking: critical thinking skillsfor the 1990s and beyond", System Dynamics Review,Vol. 9 No. 20, pp. 113-33.

Sahlins, M.D. (1972), Stone Age Economics, Aldine de Gruyter,Berlin.

Sastry, M.A. (1997), "Problems and paradoxes in a model ofpunctuated organizational change", AdministrativeScience Quarterly, Vol. 42 No. 2, pp. 237-75.

Sastry, M.A. (2001), "Understanding dynamic complexity inorganizational evolution: a system dynamics approach", inLomi, A. and Larsen, E.R. (Eds), Dynamics ofOrganizations, AAAI Press, Menlo Park, CA, pp. 377-404.

Schensul, J.J. and LeCompte, M.D. (1999), Designing andConducting Ethnographic Research, Vol. 1, Altamira Press,Walnut Creek, CA.

Simon, H.A. (1944), "Decision-making and administrativeorganization", Public Administration Review, Vol. 4 No. 4,pp. 16-31.

Small, C. (1999), "Finding an invisible history: a computersimulation experiment (in Virtual Polynesia)", Journal ofArtificial Societies and Social Simulation, Vol. 2 No. 3.

Spekman, R. and Lamb, C.J. (1997), "Fruit fly alliances: the rise ofshort-lived partnerships", The Alliance Analyst, Vol. 15,pp. 1-4.

Trotter, R.T. II and Schensul, J.J. (1998), "Methods in appliedanthropology", in Bernard, H.R. (Ed.), Handbook ofMethods in Cultural Anthropology, Altamira Press, WalnutCreek, CA, pp. 691-736.

Turpin, T. (1999), "Managing the boundaries of collaborativeresearch: a contribution from cultural theory",International Journal of Technology Management, Vol. 18Nos 3-4, pp. 232-46.

Waldrop, M.M. (1992), Complexity: The Emerging Science at theEdge of Order and Chaos, Simon and Schuster, New York,NY.

Wasserman, S., Faust, K. and lacobucci, D. (1994), SocialNetwork Analysis: Methods and Applications, CambridgeUniversity Press, Cambridge.

Winder, R.E. and Judd, D.K. (1996), "Organizational orienteering:linking deming, covey, and senge in an integrated fivedimension quality model", available at: www.ldri.com/articles/96orient.html

Journal of Manufacturing Technology ManagementVolume 15 • Number 7 - 2004 - 541—559

Appendix 1. Social-network survey

Your NameYour OrganizationYour Position TitleDate

Q1. Please name your GMPartnershipQ2. What is your role on thepartnership?Q3. What is the name of your project(s) in thispartnership?Q4. Have you had experience working withother collaborative partnerships (anyorganization)? Yes NoQ5. Please name all of the people you have arelationship with as part of your partnership(named in Q1). Use full name if possible.Q6. Put GM if the person works for GM. Putpartner if the person works for the partner.Q7. Please indicate the formal relationshipbetween you and the person named.

1 = peer2 = someone who has lower rank orstatus compared to you3 = someone who has higher rank orstatus compared to you

Q8. Does your relationship with the namedperson pre-date the initiation of thepartnership?

1 = Yes, it predates the start of thepartnership.2 = No, it began with or after thepartnership

Q9. Using a scale of 0 to 6, where0 = none1 = lowest6 = highest

Please rate the frequency that youcommunicate with this person compared withthe others on your list.Q10. Using a scale of 0 to 6

0 = none1 = lowest6 = highest

Please rate the importance of thecommunication with this person.Q11. Using a scale of 0 to 6


Please rate the level of trust you have for thisperson.Q12. Using a scale of 0 to 6


557


Glick? H. Sengir et al.

Please rate the frequency that you work withthis person compared with the others on yourlist.Q13. Using a scale of 0 to 6


Please rate the importance of work you dowith this person.Q14. Using a scale of 0 to 6


Please rate the level of cooperation that existsbetween you and this person.Q15. Using a scale of 0 to 6


Please rate the level of conflict you have withthis person.Q16. When decisions are made, which of thefollowing describes the normal pattern, inrelation to the named person?

0 = No decisions need to be made inthis relationship.1 = You normally make the decisions inthe relationship.2 = The decisions are normally jointdecisions.3 = The other person is normallyresponsible for making the decisions.

Appendix 2. Social-network analysissoftware

2.1 Analysis programsANTHROPAC is a cognitive anthropology data-management and analysis program. It acceptsqualitative/nominal level data, as well as severalforms of quantitative data, including surveyinstrument construction. The primary analysisroutine utilized for this project was the free listingdata project. Data are input in the form ofindividually generated lists of items from culturaldomains – in this case, partnership relationships.These lists are converted to either similarity ordistance matrices, and can be analyzed within theprogram using a wide number of statisticalroutines. The standard output for the free listingroutine includes a saliency list of namedindividuals, simple descriptive statistics andboth person-by-person and item-by-itemoutput data sets that can be further analyzed(Table AI).


Volume 15 . Number 7 • 2004 • 541—559

UCINET is a social-network analysis tool. Datacan be imported or directly entered in a number ofdifferent formats. UCINET data analysis routinesinclude most of the standard sociometric measuresof network structure and dynamics including:

cohesion (e.g. distance, reachability, pointconnectivity, etc.);regions (e.g. components, k-cores, etc.);subgroups (e.g. cliques, k-plexes, factions,etc.);

(4) centrality (e.g. degree, closeness,betweenness, etc.);

(5) ego networks;(6) core-periphery;(7) roles and positions; and(8) whole network properties (e.g. density,

transitivity, etc.).

Key Player imports UCINET data and performsthree basic analyses. The first two analyses arebased on the removal of one or more key nodes.They provide the level of impact on the networkbased on the fragmentation caused by the removalof the key player(s), and the increase in averagedistance between nodes caused by the removal ofthe key player(s). The third analysis identifies theoverall reach of one or more key players,depending on the number of edges that connectthem to other people.

NETDRAW imports UCINET data files andprovides an optimized two-dimensional display ofthe network nodes (people) and edges(connections), including the directionality of theconnections. The program allows for a visualanalysis of several key attributes of the networkdata, including:(1) isolates;(2) components;(3) blocks and cutpoints;(4) k-cores; and(5) subgroups.

The program also allows several different kinds oftransformations of the shape of the data, includingcircle layouts, Grower-metric-scaling layouts,node-repulsion layouts, as well as deleting isolatesand pendents – isolates are those not connected toothers whereas pendents are those with only onepath to the network.

MAGE is a network-visualization program.It creates a three-dimensional kinetic image thatcan be interactively rotated from any point ofreference (node) within the matrix. The programallows different attributes of nodes and edges to becolor coded, to assist in visualization analysis.

FATCAT is an egocentric data-analysisprogram. It analyzes two-mode (actor by attribute)data using a variety of statistical routines.

(1)

(2)(3)

558

r

Author guidelines

Journal of Manufacturing TechnologyManagement

CopyrightArticles submitted to the journal should be original contributions andshould not be under consideration for any other publication at thesame time. Authors submitting articles for publication warrant that thework is not an infringement of any existing copyright and willindemnify the publisher against any breach of such warranty. For easeof dissemination and to ensure proper policing of use, papers andcontributions become the legal copyright of the publisher unlessotherwise agreed. Submissions should be sent to:

The EditorDr David Bennett, Aston Business School, Aston University,Aston Triangle, Birmingham, England B4 7ET.Tel: +44 (0) 121 359 3611; Fax: +44 (0) 121359 5271;E-mail: [email protected]

Editorial aimThe journal aims to provide a broad international coverage of subjectsrelating to the management of manufacturing technology and theintegration of the production, design, supply and marketing functionsof manufacturing enterprises. Emphasis is placed on the publication ofarticles which seek to link theory with application or critically analysereal situations with the objective of identifying good practice inmanufacturing.

Editorial scopeManufacturing technology strategy formulation.Design and implementation of manufacturing technology.Technology transfer and dissemination in manufacturing industries.Automation of manufacturing processes and systems.Computer-aided production and the various elements of computerintegrated manufacture.Information systems in manufacturing.Modelling and simulation of manufacturing systems.Design for manufacture and simultaneous/concurrent engineering.Human factors and work organization.The design of quality management systems and quality control.Purchasing and materials management.Supply chain management and logistics systems.Manufacturing performance and productivity measurement.

Articles may be of a theoretical nature, be based on practicalexperience, report a case study situation or report experimentalresults. The prime requirement for acceptance of an article will not beits form but rather that it:(1) makes a significant original contribution to the field;(z) is directly relevant to the management of manufacturing

technology;(3) contains elements which have general application;(4) is within the scope of the journal coverage;(5) has generally not been published previously except in very

li mited circulation.If it is felt that a contribution, though technical in nature, will be ofbroad interest, it may be published under a "technical paper" heading.A paper can be considered for publication as a "research note" if itreports work-in-progress on research which has not yet reached a stagewhere there are any final results or conclusions. Its value will be judgedby the extent to which it contributes to a debate on the researchproblem, methodology, techniques of data analysis, etc.

The reviewing processEach paper is reviewed by the editor and, if it is judged suitable for thispublication, it is then sent to two referees for double blind peer review.Based on their recommendations, the editor then decides whether thepaper should be accepted as is, revised or rejected.

Manuscript requirementsThree copies of the manuscript should be submitted in double tinespacing with wide margins. All authors should be shown and author'sdetails must be printed on a separate sheet and the author should notbe identified anywhere else in the article.

As a guide, articles should be between 3,000 and 6,000 words inlength. Please provide the word count on the first page of your paper. Atitle of not more than eight words should be provided. A briefautobiographical note should be supplied including full name,affiliation, e-mail address and full international contact details.

Authors must supply a structured abstract set out under 4-6sub-headings: Purpose; Methodology/Approach; Findings; Researchli mitations/implications (if applicable); Practical implications (ifapplicable); and the Originality/value of paper. Maximum is 250 wordsin total. In addition provide up to six keywords which encapsulate theprincipal topics of the paper and categorise your paper under one ofthese classifications: Research paper, Viewpoint, Technical paper,Conceptual paper, Case study, Literature review or General review. Formore information and guidance on structured abstracts visit:www.emeraldinsight.com /literaticlub/editors/editorialadm in/abstracts.htm

Where there is a methodology, it should be clearly described undera separate heading. Headings must be short, clearly defined and notnumbered. Notes or Endnotes should be used only if absolutelynecessary and must be identified in the text by consecutive numbers,enclosed in square brackets and listed at the end of the article.

All Figures (charts, diagrams and line drawings) and Plates(photographic images) should be submitted in both electronic formand hard copy originals. Figures should be of clear quality, black andwhite and numbered consecutively with arabic numerals.

Electronic figures should be either copied and pasted or saved andimported from the origination software into a blank Microsoft Worddocument. Figures created in MS Powerpoint are also acceptable.Acceptable standard image formats are: .eps, .pdf, .ai and .wmf. If youare unable to supply graphics in these formats then please ensure theyare .tif, .jpeg, .bmp, .pcx, .pic, .gif or .pct at a resolution of at least3oodpi and at least locm wide. To prepare screenshotssimultaneously press the "Alt" and "Print screen" keys on thekeyboard, open a blank Microsoft Word document and simultaneouslypress "Ctrl" and "V" to paste the image. (Capture all thecontents/windows on the computer screen to paste into MS Word, bysimultaneously pressing "Ctrl" and "Print screen".)

For photographic images (plates) good quality originalphotographs should be submitted. If supplied electronically theyshould be saved as .tif or .jpeg files at a resolution of at least 3oodpiand at least loom wide. Digital camera settings should be set at thehighest resolution/quality as possible.

In the text of the paper the preferred position ofall figures andplates should be indicated by typing on a separate line the words"Take in Figure (No.)" or "Take in Plate (No.)". Supply succinct andclear captions for all figures and plates.

Tables must be numbered consecutively with roman numerals anda brief title. In the text, the position of the table should be shown bytyping on a separate line the words "Take in Table IV".

References to other publications must be in Harvard style andcarefully checked for completeness, accuracy and consistency. This isvery important in an electronic environment because it enables yourreaders to exploit the Reference Linking facility on the database andlink back to the works you have cited through CrossRef. You shouldinclude all author names and initials and give any journal title in full.

You should cite publications in the text: (Adams,1997) using thefirst named author's name. At the end of the paper a reference list inalphabetical order should be supplied:

For books: surname, initials, (year), title of book, publisher, placeof publication, e.g. Fallbright, A. and Khan, G. (zoos), CompetingStrategies, Outhouse Press, Rochester.

For book chapters: surname, initials, (year), "chapter title", editor'ssurname, initials, title ofbook, publisher, place of publication, pages,e.g. Bessley, M. and Wilson, P. (1999), "Marketing for the productionmanager", in Levicki, J. (Ed.), Taking the Blinkers offManagers, BroomReim, London, pp. 29-33.

For journals: surname, initials, (year), "title of article", journalname, volume, number, pages, e.g. Greenwald, E. (2000),"Empowered to serve", Management Decision, Vol. 33 No. 5, pp. 6-lo.

For electronic sources: if available online the full URL should besupplied at the end of the reference.

Final submission of the articleOnce accepted for publication, the final version of the manuscriptmust be provided, accompanied by a 3.5" disk, Zip disk or CD-ROM ofthe same version labelled with: disk format (Macintosh or PC); authorname(s); title of article; journal title; file name.

Alternatively, the final version may be submitted as an attached fileto an e-mail by prior arrangement with the editor.

Each article must be accompanied bya completed and signedjournal Article Record Form available from the Editor or onwww.emeraldinsight.com/literaticlub/jarforms.htm

The manuscript will be considered to be the definitive version ofthe article. The author must ensure that it is complete, grammaticallycorrect and without spelling or typographical errors.

In preparing the disk, please use one of the following preferredformats: Word, Word Perfect, Rich text format or TeX/LaTeX.

Technical assistance is available from Emerald's Literati Club onwww-emeraldinsight.com/literaticlub or by contacting Mike Massey atEmerald, e-mail: [email protected]

Manufacturing Technology Management - Jan.ucc.nau.edu

Documents

Transcript of Manufacturing Technology Management - Jan.ucc.nau.edu