concurrent engineering - lori.academicdirect.orglori.academicdirect.org/cites/200801.pdf ·...

http://cer.sagepub.com

Concurrent Engineering

DOI: 10.1177/1063293X07084638 2008; 16; 37 Concurrent Engineering

Anja M. Maier, Matthias Kreimeyer, Clemens Hepperle, Claudia M. Eckert, Udo Lindemann and P. John Clarkson Development

Exploration of Correlations between Factors Influencing Communication in Complex Product

http://cer.sagepub.com/cgi/content/abstract/16/1/37The online version of this article can be found at:

Published by:

http://www.sagepublications.com

can be found at:Concurrent Engineering Additional services and information for

http://cer.sagepub.com/cgi/alertsEmail Alerts:

http://cer.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://cer.sagepub.com/cgi/content/refs/16/1/37SAGE Journals Online and HighWire Press platforms):

(this article cites 39 articles hosted on the Citations

at Uni Tehnica on March 23, 2008 http://cer.sagepub.comDownloaded from

CONCURRENT ENGINEERING: Research and Applications

Exploration of Correlations between Factors Influencing Communicationin Complex Product Development

Anja M. Maier,1,* Matthias Kreimeyer,2 Clemens Hepperle,2 Claudia M. Eckert,1

Udo Lindemann2 and P. John Clarkson1

1Engineering Design Centre, University of Cambridge, England2Technical University of Munich, Institute of Product Developement, Germany

Abstract: Designing complex products, such as jet engines, cars or certain types of software, necessitates the coordination of activities

of many participants during the design process. Communication is seen as the vehicle by which this coordination could be achieved.

Communication itself is influenced by many different factors that are connected. This study presents an exploration of correlations between

these factors based on statistical analyses of empirical data. The research uses data collected via the ‘Communication Grid Method’, (CGM)

a structured maturity grid method to assess the perception of communication within and across team-interfaces. Five empirical studies in the

aerospace, automotive, and IT industries where concurrent engineering is practiced are used. The results offer insights for researching and

managing communication across inter-departmental interfaces. It has been shown in particular, how directly and indirectly linked factors

influencing communication in product development form a network of correlations. Mutual trust and collaboration exhibit thematic centrality.

Key Words: communication, concurrent engineering, research and development management, product development organizations,

case studies in industry, collaboration, maturity grid.

1. Introduction

Products and services result from interactions amonga multitude of people who work across functional,organizational, cultural, temporal, and geographicalboundaries [1,2]. In concurrent engineering, tasks aredistributed among individuals and whenever possibleexecuted in parallel, increasing the need for effectivecommunication. Concurrent collaborative design pro-cesses are challenging because strong interdependenciesbetween design decisions make it difficult to converge ona single design solution that satisfies these dependenciesand is acceptable to all participants [3]. Typically,the different participants in the design process possessdifferent competences, skills, responsibilities, interests,and inhabit different ‘object-worlds’ [4]. Everyone hasdifferent ‘viewpoints’ [5] which can lead to conflictsthat need to be resolved through negotiations [4].For example, the early phases of designing the body ofa car require bringing together car styling and coreengineering. Core engineering encompasses, Forexample, the product’s performance and properties,such as stiffness or crashworthiness as well as manu-facturability [6]. In particular, collaboration between

embodiment design (CAD) and simulation departments(CAE) is a key element to a highly efficient designprocess [7,8].

To illustrate, in automotive sheet metal design,a number of welding spots need to be specified. Ata particular German automotive manufacturer, thewelding spots are designed in collaboration withmanufacturing engineers from the production planningdepartment in order to place the spots within the reachof welding robots. Coordinates of each welding spot arerecorded in a spreadsheet that is parametrically linked tothe CAD data files of the sheet metal design of thecar body. All files are available through the company’sproduct data management (PDM) system. However,when assembling input files for numerical simulation,this information is not used by the engineers compilingthese files. Assumptions of the connectivity betweendifferent components might differ from the originaldesign. After simulation, the welding spots are no longercontained in the CAD model of the individual partreducing its value in the context of the actual bodydesign. Lack of coordination, obvious in this example,reflects the different understandings and goals ofdifferent groups as well as lack of adherence to processsteps and use of different tools. A better understandingof each other’s intentions, different forms of representa-tion, and information needs could improve the process.*Author to whom correspondence should be addressed.

E-mail: [email protected]

Volume 16 Number 1 March 2008 371063-293X/08/01 0037–23 $10.00/0 DOI: 10.1177/1063293X07084638


The example shows that there is a multitude ofinfluences on how engineers communicate to coordinatetheir efforts. Communication is often perceived directlyas a problem but can be difficult to recognize because ofinteractions with other factors, such as planning orproduct complexity. Managers of design processes needto have a sense of where processes can be influenced.This requires understanding of how factors are con-nected, so that effort is not misdirected by attending tothe symptoms, yet possibly ignoring underlying causes.In practice, it is often possible to analyze a specificsituation, however little theoretical understanding of thecorrelation of factors influencing communication hasbeen published.This study presents a network of factors affecting

communication. It is structured as follows: Aftera description of the background and purpose of thisresearch, the procedure of data acquisition and analysisis presented, encompassing four stages:Acquisition of empirical data: Empirical data on

factors influencing communication was acquired byapplying the ‘Communication Grid Method’ (seeSection 3 in this study, [9–11]) in industrial case studies.It provided numerical scores of the engineers’ perceivedcurrent and desired states of factors influencingcommunication. Only data elicited on the current stateprovides input for all subsequent analyses.Exploration of inter-variable correlations: In order

to elicit correlations between values for factors (termedvariables for statistical analysis purposes) statisticalcorrelation analyses were performed, using Kendall’stau-b measure of association.Exploration of network of linkages: The outcome

of inter-variable correlation was further processed inorder to find structural characteristics, such ascycles and clusters, using a software tool based ongraph-theory [12].Literature exploration of direction of correlations:

Findings from previous stages were brought inconjunction with literature from various relevantdisciplines in order to look for possible directions ofinfluence.Further, conclusions and implications for engineering

research and practice are drawn and the results arecritically reflected. The study closes with suggestions forfurther research.

2. Background: Current Research

As highlighted by many researchers and practitioners(e.g., [13–20]) communication is one of the criticalsuccess factors of collaborative design. This is especiallyrelevant in complex product development due to thelarge number of both components and design partici-pants involved in the process.

Allen [21,22] has done pioneering work on the roleof effective communication in product developmentprocesses since the early 1970s. He suggests that thedegree of interdependence between engineers’ work isdirectly related to the probability that they engage infrequent technical communication. Smith and Eppinger[23] and Sosa et al. [18] use task interdependencyto identify the activities that require higher effortto coordinate. Loch and Terwiesch [24] present ananalytical model to study the coupling of uncertainty,dependence, and communication, suggesting thataverage communication frequency increases with thelevel of uncertainty and dependence. In these studies,communication is mostly defined as informationtransmission. Consequently, the frequency and flow ofinformation is measured.

As the example in the introduction shows, thereis more to communication than information flow.Eckert and Stacey [25] concluded in their researchthat there are several interaction scenarios, such as‘handover’ or ‘joint-designing’ where communicationmisalignments tend to occur due to, for example, a lackof overview of the sequence of tasks by the individualdesign engineers, lack of information flow, or misinter-pretation of information [26].

The importance of communication in collaborativedesign is indisputable, yet it raises questions of how toresearch communication in collaborative design in orderto improve the design process. Hales [27,28] looks atseveral levels of influence from the macroeconomic,microeconomic, corporate, management, project team,and personal level and draws a comprehensive list offactors influencing the design process. Badke-Schauband Frankenberger [29,30] examine a comprehensive listof ‘prerequisites’ influencing the design process in orderto arrive at the characterization of critical situations.‘Prerequisites’ range from individual, team, organiza-tional, design task, and outcome to boundary condi-tions. In their analyses, communication is a key factor,but no description is given of what communicationincorporates. Indications point to different groupcharacteristics, such as group climate or quality ofleadership as influencing communication in collabora-tive design. In researching whether the level of techno-logical complexity affects project group communication,Roberts et al. [31] come to the conclusion that inmoderately complex projects information sharing isgreater than in highly complex projects despite thefact that there is a higher need of information. Chiu [32]concludes that the type and structure of team organi-zation impacts communication. Tiernan et al. [33]detected that changes in organizational structure affectcollaborative design. While the importance of commu-nication is generally acknowledged, there is littleconsensus on how it can be directed or, at best,systematically improved.

38 A. M. MAIER ET AL.


For the purpose of this study, communication isdefined as the cognitive and social process by whichmessages are transmitted and meaning is generated.It can thus be seen as the vehicle by which behavior iscoordinated [9,33–37]. Communication is influencedby a number of factors, which are assumed to beassociated. How these factors are interrelated, directlyand indirectly, remains has to be seen. The relationalinterdependence between the elements of a systemdetermines a change of some parts, which areoriginally not affected [38].

It is postulated that understanding and managing ofcommunication might be aided by gaining insight intoits structure. This might come about through reflectionin the form of assessment. Busseri and Palmer [39]concluded that self-assessment during the design processhas a positive effect on how well a team functions, asit increases reflection. The beneficial effect of self-reflection in teams has been emphasized by severalauthors [40–43]. Badke-Schaub et al. [44] advocateeducation and training of engineers to increase reflec-tion. Maier et al. [9,10] developed a practical methodof self-assessment – the ‘Communication Grid Method’– to reflect on communication in engineering design. Themethod uses the designers’ perceptions on the currentand desired states of multiple factors influencingcommunication during a certain project phase as theassessment index (see next section) which is the basis forempirical data in this research project.

Insights into associations between factors impactingon communication is seen as a lever to increaseunderstanding and make the complexity of communica-tion more transparent.

3. Acquisition of Empirical Data

Data acquisition proceeded by using the‘Communication Grid Method’ [9–11] which has beendeveloped with the focus of assessing communication incollaborative design. The method aims at reflection inboth senses of the word: to mirror back perceptionsabout communication in industrial practice as well as totrigger thinking about communication practices in asystematic way. It is based on the idea of processassessment via maturity grids in quality management [45]and software development [46]. Both have so far beenadapted for instance to areas including technical innova-tion [47], project management [48], and design [49], butnot yet to the assessment of communication in colla-borative design. Therefore, Maier et al. [9,75] focused ona maturity grid-based approach applied in a participatorygroup workshops. For the purpose of this study, themethod functions solely as acquisition method forempirical data. In order to place the results in context,in what follows, the set up of the method is described.

Generally, in a maturity grid, levels of maturity areallocated against certain key aspects. Text descriptionsinside the cells of the grid express different levels ofmaturity which show the development from an initial toa more advanced state for the considered subject [50].In the ‘Communication Grid Method’ (CGM), �50factors influencing communication – elicited through acombination of ethnographic case studies in industryand analysis of relevant literature prior to this researchproject – are selected to which four levels of maturityare allocated [10]. The maturity levels are based on theidea of learning types from Argyris and Schon [51]. TheCGM is administered in an interview or a structuredgroup workshop to elicit the design engineers, percep-tion on the current (as-is) and desired states (to-be) offactors influencing communication at a certain teaminterface. Paper-based grid sheets are distributed andscores by the design engineers on the current and desiredstates are collected (see Figure 1 as an excerpt of the fullset of grid sheets. The set of grid sheets with definitionsof factors and questions is listed in table format inAppendix Table A1).

When applying the CGM different influences(represented by a set of �50 factors) are allocated tofive levels of influence (product, information, individualteam member, project team, and organization) which aresubdivided into 11 areas of influence on communicationin product development. Each area acts as a category,under which individual factors are subsumed. Twoexamples for subdividing areas into more detailedfactors are shown in Table 1.

The CGM sheds light on both the current and thedesired states of factors influencing communication asassessed by participants in the case studies. Thus, bycorrelating the values it is the assumption that aninformative picture about the interdependencies amongthe selected factors can be drawn, based on the existingcommunication situation. Values for the desired states arenot considered in this research.

In total, 38 engineers and managers completed thegrid sheets in either individual interviews or groupworkshops. All case studies were concerned withcommunication at a certain team-interface withregard to a specific project at a specific design phase.All projects contained routine as well as non-routinedesign elements in order to design the product, whereasroutineness is expressed on an axis, with the possibilityof different degrees of routineness [52]. The definitionof factors was given to all participants in the studiesin order to provide a common reference point.

For the purpose of analyzing communication inproduct development, data from the following fivecompanies was included:

. A large, globally operating company in informationtechnology, developing and servicing software and

Exploration of Correlations between Factors Influencing Communication 39


providing technical consultancy. Communicationbetween service support and development of oneparticular code-base of a particular software projectwas to be analyzed. The work incorporated routine aswell as innovative solutions to a problem customersfound in the company’s software. Participants of thecommunication audit knew each other well and mosthad been in the company for more than 10 years.

. A small company producing electron beam andfurnace technology, based in the UK. The ‘worksorder’ process was to be improved through ananalysis of how a ‘works order’ flows through the

company from its initial order to an offer. The focusof the audit lay on communication among themanagement team (engineering, manufacturing/spares, sales, service, and finance). All members hadbeen in the company for more than 15 years.

. A large aerospace company, based in the UK,designing, manufacturing, and servicing aero engines.Two studies within two distinct branches (civiland defense) of the company were conducted. Inone project, the communication audit conductedinvestigated the state of communication at theinterface between design and a service team of one

TeamworkPlease answer for your project team

Levels of maturity

Factors A B C DCurrent

score

Desired

score

Collaboration

Everyone looks

solely after his

or her tasks

Collaboration happens

only if asked for in

order to fulfil tasks

Collaboration happens

proactively in order to learn

from others and improve

own approaches

Collaboration is constructive, happens

regularly whenever necessary and

there is continuous effort to improve it[ ] [ ]

Common goals and

objectives

Not known. No

thinking about it

Known but everyone

follows just his or her

own goals

Known and sometimes

consideration about the way

common goals can be

reached through working

together

Entirely clear and identification with it

which is expressed in communication

and continuous effort to assess and

adjust goals and objectives and the

way to each them

[ ] [ ]

Team identity

There is none

and it is not seen

as necessary

Small groups form

depending on the task

and these groups get

their identity through

the tasks

Attitudes with respect to

team identity are

continuously reflected upon

in order to find a common

denominator

There is a strong sense of belonging to

the team and continuous reflection on

how team identity can be kept up and

strengthened for the project

[ ] [ ]

Comments

Figure 1. Example grid sheet ‘teamwork’.

Table 1. Categorization of Factors in the ‘Communication Grid Method’.

Level of influence Areas of influence Examples of individual factors

z



particular aero engine. The other project aimed todiagnose the state of communication at the interfaceof preliminary design for one of the civil engines andthe business unit designing IP turbine blades. In bothcases, the participants knew each other well and mostof them had been in the company for many years.

. A large automotive manufacturer, based in Germany.The audit looked at communication at the interfacebetween embodiment design (CAD) and simulation(CAE) during the early phases of designing the bodyof a car for one particular vehicle series. The studyfocused on the design of the so-called ‘trimmed body’for the serial development of one of the company’svehicle series. The observed interface to simulationwas that of engineers involved in developing thefunction ‘noise vibration harshness’ (NVH).The majority of participants had been in the companyfor over 15 years.

4. Exploration of Inter-variable Correlations

Statistical correlation analysis was used in order toexplore connections between factors influencingcommunication.

4.1 Set Up

Answers given in the grid sheets on the currentstates of 27 factors common across all industrialstudies formed the basis of exploring correlationsbetween factors influencing communication in productdevelopment. The population of people asked aboutselected factors was at least N¼ 30. For statisticalcalculation purposes, factors are referred to as vari-ables. As the empirical data acquired has the standardof an ‘ordinal scale’, Kendall’s tau-b was used forcalculating the correlation [50,51]. For visualizationpurposes, the correlation matrix across all theparticipants was processed manually (see AppendixTable A2).

Correlations were categorized according toBrown [52]: Moderate low correlations range from0.4 to �0.5, moderate high from 0.51 to �0.6, andhigh correlations from 0.61 to �0.8. Only correlations ata significance level of at least p50.05 and correlationswith an absolute coefficient value of equal to andmore than 0.4 were chosen. All of the correlations forfurther analyses, meeting the selection criteria in thisstudy, are characterized by a positive correlationcoefficient. The correlation coefficients are symmetrical.Consequently, the correlation matrix does not showwhich variable drives a particular correlation.

4.2 Selected Results

Four ‘high’ and eleven ‘moderate-high’ correlationswere found, as well as 18 correlations having a‘moderate low’ coefficient. Constrained by the selectioncriteria mentioned above (Section 4.1), Figure 2 andAppendix Table A2 show the complete set of correla-tions found.

As examples of the findings, representative instancesare listed below. Variables that correlate with at leastfour other variables and show at least one ‘high’ corre-lation coefficient are termed ‘core variables’. The resultsthus highlight:

. ‘collaboration’,

. ‘mutual trust’,

. ‘overview of sequence of tasks in the design process’,and

. ‘autonomy of task execution’.

‘Collaboration’ shows correlations with nine othervariables. ‘Mutual trust’ displays correlations with sixother variables, and ‘overview of sequence of tasks in thedesign process’ is related to four other variables (bold inFigure 2). The statistically inferred importance of thesecorrelations for design management is supported byliterature (see Section 6).

Factors selected in this research project are a basis togenerate hypotheses for further research on whetherthose four factors can be depicted as the core influenceson communication in product design.

Findings in this section shed light on singularlinkages, function as input for exploration of linkagesbeyond correlations between two variables, andform the basis for the literature analysis and criticalreflection.

5. Exploration of Clusters and Cycles

In addition to the exploration of correlations betweentwo factors, graph-theoretical analyses were performedto detect linkages between several factors.

5.1 Clusters of Factors

The correlation matrix was manually manipulatedin order to visualize clusters among the 27 selectedfactors. Clusters in this context consist of factors that arecompletely linked amongst each other. Clustersthat stand out based on the correlation matrix arehighlighted in the rearranged matrix (Figure 3).



In total, eight complete clusters were detected. Asthe clusters overlap, for visualization purposes,only Clusters A–D are highlighted in the matrix inFigure 3.Clusters A–H encompass the following:

Cluster A: ‘Availability of information about productspecifications’; ‘Handling of technical conflicts’; ‘Rolesand responsibilities’.Cluster B: ‘Handling of technical conflicts’; ‘Roles and

responsibilities’; ‘Mutual trust’; ‘Collaboration’.Cluster C: ‘Mutual trust’; ‘Collaboration’; ‘Do you

know what information the other party needs’.Cluster D: ‘Collaboration; ‘Do you know what

information the other party needs’; ‘Autonomy of taskexecution’.Cluster E: ‘Collaboration’; ‘Autonomy of task

execution’; ‘Overview of sequence of tasks in thedesign process’.Cluster F: ‘Collaboration’; ‘Overview of sequence of

tasks in the design process’; ‘Project reviews’.Cluster G: ‘Collaboration’; ‘Project reviews’; ‘Activity

at interface with the other party’.

Cluster H: ‘Collaboration’; ‘Project reviews’; ‘Do youknow what information the other party needs’.

Clusters can be viewed as groupings that are compactin comparison to the rest of the network, i.e., they havea higher coherence than their environment. In productarchitectures, they are, for example, used to findpurposeful boundaries for modularization of aproduct [56]. For the purpose of this research, factorscontained in a cluster could be read as inseparable.To shed light on the particular communication pheno-menon inherent to each cluster, a more coarse leveldenomination might be applicable for each cluster.

5.2 Indirect Linkages Between Factors

Cycles occurring in the network of linkages refer toloops of linked nodes. However, as the chains of causeand effect cannot be concluded from the correlationmatrix, cycles cannot be termed feedback loops. As it iscommon in graph-theory, the term ‘node’ is used for afactor influencing communication in product develop-ment. The length of the cycle refers to the number

Mutualtrust

Collaboration Autonomy of taskexecution

Overview of sequence oftasks in the

design process

Figure 2. Network of all Correlation Coefficients.



of nodes a cycle is consisting of. Accordingly, a linkagebetween two factors is also called ‘edge’. Cycles depictnetwork structures in which all nodes are circularly andindirectly linked to each other. The more often a nodeappears in the overall number of cycles, the more(indirectly) it is linked to other elements present in thosecycles (Figure 4(b)). Equally, a high occurrence ofa certain edge emphasizes its importance in respect tothe number of existing cycles (Figure 4(c)).

As the chains of cause and effect are not defined forthe linkages, the explanatory power of the cycle analysisis limited. For example, the build-up of resonancesthrough feedback to the starting node cannot beobserved in this kind of model. Yet, analyses point tothe importance of looking at the web of linkages as awhole, especially to recognize the existence and impor-tance of indirect linkages between factors (e.g., [57]).Furthermore, the more frequent a single node occurs inthe overall network, the higher the importance andimpact for the overall structure.

In total 467 cycles were found in the network oflinkages. The occurrence of cycles per cycle lengthis illustrated in Figure 4(a). The minimum cycle lengthis 3 as a cycle must at least contain three differentfactors. Cycles with this length also stand for thesmallest possible clusters as the three nodes in thesecycles are fully linked with each other.

The 467 cycles show that changing the status of a factoris likely to cause many unintended changes on factors notdirectly linked with the initially changed factor. Certaincycles might possibly qualify as feedback loops; forinstance, a good establishment of mutual trust typicallyenhances the understanding and implicit repartition ofroles and responsibilities, thus enhancing collaboration.In turn, good collaboration positively influences trustamong the people collaborating.

Only 14 of the 27 considered factors (nodes) are partof all 467 cycles as visualized in Figure 4(b). The factor‘collaboration’ appears most often in the detected cycles.In 445 of the 467 cycles, this factor contributes to theexistence of the respective cycle. Thus, ‘collaboration’must be seen as a factor of major importance in thenetwork of factors influencing communication inproduct development. However, as communication assuch is about collaborating, it is also evident that thesubjective perception of the quality of collaboration is akey factor towards evaluating communication. Anotherimportant factor in this context is ‘mutual trust’ as beingpart of 395 of the 467 cycles.

As indicated earlier, the two factors ‘collaboration’and ‘mutual trust’ play a major role in almost all cycles.Only two (Examples 1 and 2) can be set up withouteither one. As the linkage between the factors‘autonomy of task execution’ and ‘representation’

.

.

. .

.

.

.

. .

. . .

.

.

. ..

.

. .

. ..

. .

.

.

.

.

.

.

.

.

. .

. . .

.

.

.

.

.

.

.

.

.

..

....

. . .

. .

.

.

.

.

.

.

.

.

Figure 3. Clustures A to D in correlation matrix.



and between ‘autonomy of task execution’ and ‘avail-ability of information about product specifications’occurred both in 264 of the 467 cycles (seeFigure 4(b)), Example 3 is of interest. All three examplescontain the four most frequently occurring nodes(see Figure 4(c)).

. Example 1 (cycle length¼ 4): ‘Project reviews’ – ‘Doyou know what information the other party needs’ –‘Autonomy of task execution’ – ‘Overview of sequenceof tasks in the design process’ – ‘Project reviews’.

. Example 2 (cycle length¼ 3): ‘Handling of technicalconflicts’ – ‘Availability of information about

264

264

182

180

179

179

179

178

177

177

175

175

170

165

123

123

111

104

90

90

90

83

83

83

83

83

0 50 100 150 200 250 300

(a)

(c)

(b)

Figure 4. Top left (a) Number of cycle per cycle length; Top right (b) Occurrence of nodes in all detected cycles and (c) Occurrence of edgesin all detected cycles.



product specifications’ – ‘Roles and responsibilities’ –‘Handling of technical conflicts’.

. Example 3 (cycle length ¼ 5): ‘Collaboration’ – ‘Rolesand responsibilities’ – ‘Availability of informationabout product specifications’ – ‘Representation’ –‘Autonomy of task execution’– ‘Collaboration’.

Analyzing cycles leads to detection of indirectlinkages. These linkages can be brought into the formof a hierarchy. ‘Mutual trust’ and ‘collaboration’ act ascentral starting points for hierarchies. The example inFigure 5 shows nodes that are related and ordered ina way that their relation to ‘mutual trust’ is depictedin a hierarchical structure. Several layers of impact areshown. The further away from the starting point, thelower the impact.

Following indirect linkages, some factors, not explicitlyconnected in the graph shown, are strongly linkedindirectly [58], e.g., via a third one (Figure 5): ‘Doyou know what information the other party needs’and ‘overview of sequence of tasks in the design process’is the most important indirect linkage, as these twoare connected via a number of other pathways.

At a lower level, the following factors form indirectlinkages, such as:

. ‘Activity at interface with the other party’ and‘overview of sequence of tasks in the design process’.

. ‘Do you know what information the other partyneeds’ and ‘education/training’.

. ‘Availability of information about product specifica-tions’ and ‘mutual trust’.

. ‘Availability of information about product specifica-tions’ and ‘collaboration’.

This, again, highlights the importance of understand-ing factors as an interconnected network.

5.3 Preliminary Summary: Network of Factorsas a Whole

Exploring the network of factors based on thecorrelation analysis resulted in eight complete clustersafter rearranging the correlation matrix. The analysis ofnodes and edges resulted in linkages which were notdetected by the correlation analysis, given the selection

Mutual trust

Figure 5. Impact of ‘trust’ on other factors (ordered as hierarchy across several layers).



criteria used. They were manually added to therearranged correlation matrix in order to view largercompleted clusters, in which only one or two linkageswere missing. By this procedure, nine factors with a highdegree of linkage among each other were elicited:

. ‘availability of information about product specifica-tions’,

. ‘handling of technical conflicts’,

. ‘roles and responsibilities’

. ‘mutual trust’,

. ‘collaboration’,

. ‘do you know what information the other partyneeds’,

. ‘autonomy of task execution’,

. ‘project reviews’, and

. ‘overview of sequence of tasks in the design process’.

Cycles, cycle lengths, occurrences of nodes and edges inthe network of factors were analyzed. In the 467 detectedcycles the nodes ‘collaboration’ and ‘mutual trust’ play animportant role for the existence of cycles. Changing thestatus of a factor might have influence on many otherfactors, if the changed factor belongs to the group offactors showing a high linking degree among each other.Further, changing a certain factor which occurs inmany cycles might have unintended effects on not directlylinked factors that are part of the respective cycles.

6. Literature Analysis: The Search for Direction

As elicited correlations and cycles are undirected,further evidence is needed in order to determine

a possible causal direction. To this end, literature froma number of disciplines, such as new product develop-ment, management science, and psychology wasconsulted.

6.1 Correlations Supported by Literature

Table 2 summarizes evidence from the literaturedrawn to support correlations explored in this study.

Determination of possible directions of cause andeffect is not conclusive. Yet, evidence for each possibledirection has been found. With regard to the firstexample of correlating factors ‘collaboration’ and ‘teamidentity’, references in the literature seem to suggest that‘team identity’ is one of the drivers of ‘collaboration’[29,59,60]. Concluding from the reviewed literature adirection seems to be apparent. The factor ‘roles andresponsibilities’ seem to be another driving factor of‘collaboration’ [61–63]. In terms of ‘mutual trust’ and‘collaboration’, evidence in the literature can be foundwhich supports both directions [64–67]. Interviewingengineers from 34 medium-sized manufacturingcompanies with regard to their business relation withthe customer and supplier, Bstieler [67] concludes thata higher level of trust is positively related to perceivedcontinuity of collaborative development projects.This supports a bidirectional influence. With regard to‘autonomy of task execution’ and ‘representation’,Eckert et al. [68] suggest that if you do not have clearrepresentations you need to negotiate and cannot carryyour tasks out autonomously.

Although evidences have been found whichindicate trends, results have to be read with a note

Table 2. Evidence from literature

[60]

[55]

[29]

[69][63]

[61]

[62]

[64]

[67][66]

[65]

[70][71]

[72]

[73][3]

[61]

[68]



of caution. Studies have used different researchmethods and people participating in the studies dealtwith different tasks. Consequently, results presented inthis research provide directions as to where suggestionsseem apparent.

6.2 Correlations not Detected in ReviewedLiterature

Some statistically suggested correlations could not befound in the literature:

. ‘Roles and responsibilities’ $ ‘Mutual trust’;

. ‘Roles and responsibilities’ $ ‘Handling of technicalconflicts’;

. ‘Overview of sequence of tasks in the design process’$ ‘Autonomy of task execution’;

. ‘Education/training’ $ ‘Project reviews’;

. ‘Terminology’ $ ‘Application of corporate visionand values’.

Several reasons might contribute to the fact that noevidence in the literature was found. Some literaturemight have escaped the authors’ radar, some might havenot been appropriate, definitions used in the originaldata acquisition phase for the individual factors(see Appendix Table A2) might not have concurredwith the definitions used in other research projects, or,factors might simply not be linked directly. Despite thesignificance level chosen, correlations might just bestatistical correlations.

6.3 Potential Indirect Linkages

Another aspect why no reference was found fora detected correlation might be that the two respectivefactors are only linked indirectly. This could apply, for

example, to the correlation between the factors ‘rolesand responsibilities’ and ‘mutual trust’ (see Figure 6).

Correlation between ‘collaboration’ and ‘roles andresponsibilities’ is supported by three references in thereviewed literature and correlation between the factors‘collaboration’ and ‘mutual trust’ is supported byfour references (see Table 2). However, no referencesfor the correlation between the factors ‘roles andresponsibilities’ and ‘mutual trust’ were found. Withinthis context, it leads to the assumption that they couldbe indirectly linked, bearing the data acquisitionprocess in mind.

7. Conclusions

The overall objective of this study was to explore inter-relations between factors influencing communicationin product development emerging from data acquiredthrough empirical studies. A communication gridanalysis in five case studies in industry yielded empiricaldata from engineers and managers scoring their percep-tion on the current state of factors influencingcommunication.

As correlations statistically inferred are non-directional, possible directions were elicited throughintensive and extensive literature reviews. In addition toillustrating inter-variable correlations, cycles usinggraph-theory are presented.

Results presented in this study are seen as initialattempts to exhibit interconnections between factorspertinent to design and communication. As in any otherempirical study that collected data from empiricalstudies with small samples within each study, thegenerality of the findings cannot be claimed beforecompleting similar studies. However, the authors wouldexpect to obtain analogous results in other projects

Collaboration

Roles and responsibilites

Mutual trust

?

[63][61]

[62]

[64][67]

[66][65]

Figure 6. ‘Roles and responsibilities’ and ‘trust’ indirectly linked by ‘collaboration’.



developing complex systems. This study is explorativein nature and as such, drawing explicit normativeconclusions is avoided.The findings in this study enrich the understanding

of communication as a major determinant in collabora-tive design by showing different factors and theirinterrelations impacting on communication and on thedesign process. This analysis has yielded a number ofinteresting results:

(1) Core factors: The factors ‘mutual trust’, ‘collabora-tion’, ‘roles and responsibilities’, ‘project reviews’,‘availability of information about product specifica-tions’, ‘handling of technical conflicts’, ‘do youknow what information the other party needs’,‘autonomy of task execution’, and ‘overview ofsequence of tasks in the design process’, possessingof a high linking degree, are portrayed as corefactors that influence communication in productdevelopment.

As can be seen in Figure 2, the analysis shows thefactors ‘hierarchies’, ‘availability of informationabout the company’, ‘transparency of decisionmaking’, and ‘common goals and objectives’ arebelow the selected criteria of at least ‘moderate-high’correlation coefficient and statistically significantvalues. They might act as ‘other variables’ whichalso cause communication in design, as any covar-iance they share with the selected independentvariable might falsely have been attributed to thatindependent variable.

(2) Clusters of factors, such as Cluster A (‘availability ofinformation about product specifications’, ‘handlingof technical conflicts’, and ‘roles and responsibil-ities’) as shown in Section 5.1 represent groupsinside which every node correlates with every othernode. For example, Cluster A could be named‘finding compromises’, given that technical conflictscan only be purposefully resolved if data on productspecifications are available and if roles and respon-sibilities in particular for the process of decisionmaking is clear.

(3) Cycles: Analyzing cycles led to the detection ofindirect linkages. These linkages can be brought intoa hierarchy. ‘Mutual trust’ and ‘collaboration’ act ascentral starting points for hierarchies. Changing thestate of a factor is likely to cause many unintendedchanges on factors not directly linked with theinitially changed factor.

(4) Direction of which way causation flows throughliterature: Concluding from the reviewed literature,directions, such as the following are suggested: ‘teamidentity’ and ‘roles and responsibilities’ seem to bedrivers of ‘collaboration’. These analyses provideexamples to be compared with results of otherresearch projects.

8. Implications

8.1 Implications of Findings for Research

Contributions of this work to current researchin concurrent product development are highlightedbelow:

(1) Theoretical exploration into connections of factors:It presents a theoretical exploration into connectionsof factors influencing communication in collabora-tive design based on empirical studies in differentindustry sectors. It might open a stream of researchinto eliciting further evidence with regard to possibleand plausible chains of multiple cause and effects.Whereas the literature points toward importance ofindividual factors, correlations with other factorswhich point to trends or possibly regularities were sofar rarely found. This research project contributestowards complementation.

(2) Generating hypotheses and questions: The findings inthis study provide empirical support for a set ofrelationships between factors influencing commu-nication in product development. The range offindings – in particular exposition of indirectlinkages – support the general statement that thereare a number of influences on communication inengineering design which need to be viewed incontext with other influencing factors. The resultslead towards generating the following exemplaryhypotheses and questions, falsification of which issubject to further research.(a) Correlations which have been found in this

research but not yet supported by literature,such as the link between: ‘overview of sequenceof tasks in the design process’ and ‘autonomy oftask execution’. It could be hypothesized thatthe greater the overview, the greater the abilityto collaborate, thus a reduced need for clear taskseparation. Likewise, it could be possible thatthe greater the overview, the greater the abilityof a team to draw clear distinctions betweentasks and the greater the ability to assessexpertise of colleagues and judge when tocontact them and when best not to.

(b) ‘Overview of sequence of tasks in the designprocess’ is influenced by ‘education/training’ orvice versa as they are indirectly linked. Could achange in one have a knock-on effect on the other?Or would it depend on the nature of the linkage?

(3) Assessment of communication: Although communi-cation is inherently subjective and boundaries ofdefinition of factors are drawn differently; thisresearch provides an attempt towards makingcommunication more ‘tangible’ and accessible forresearch and management. It supports the quest for



a taxonomy permitting an assessment of collabora-tive design communication.

8.2 Implications of Findings for Designersand Managers

This research emphasizes the importance ofcommunication for concurrent engineering in termsof communication as coordinating behavior towardsa common goal. Studies show that it is beneficial forthe individual engineer and the design managers to beaware of how communication can be structured(e.g., using the factors as influences applied in thisresearch) and how those factors are interconnected.

The example, described in the introduction abouta lack of coordination of using available informationon welding spots during the design of an automotivebody. The clear establishment of common goals, forexample, by working towards one common datamodel reachable through a common access portal touse available data best possible, could be a purposefulentry point. Knowing that, furthermore, a bettercoordination of preceding and following processsteps is necessary, this could be consolidatedintroducing a project review of recent projects toestablish ‘lessons learned’ for all engineers involved,thus improving their knowledge about each othersinformation needs.

The results can be a basis to expose patternsof connections between factors. This knowledge couldthen be used in support for design management. It isbelieved that the presentation of interrelationsencourages recognition of patterns. Recognition ofpatterns is thus facilitated through prior expositionvia analysis. Following Meijer [74], in the world ofmanagement, the word complex is often a synonym forcomplicated or difficult, involving many factors anduncertainty. The results presented above could functionas a starting point for preparation of managerialdecision making which could thus also reduceuncertainty.

In other words, this research has important implica-tions for managers from two different perspectives.From a strategic viewpoint, exposition of the network offactors influencing communication in product develop-ment based on empirical data might provide someinsight into the types of factors to consider in designingan environment where design teams collaborate mosteffectively. From a tactical or operational viewpoint,findings in this study might provide decision-supportand a starting point for intervention actions. Referringback to what has been said earlier with regard to seeingcommunication as a complex system, increasing trans-parency into the network of connected factors mightreduce uncertainty.

9. Critical Reflection of Results

Some concerns to be addressed when exploringcorrelations between factors influencing communicationstatistically are the following:

(1) Number of participants: Limitations of this workconcern the number of participants in these studies.Further studies should be made to validate thegenerality of the results. Extension of empiricaldata might show if the assumed linear behaviorof the correlations can be confirmed, a topicdiscussed in Section 10 (Future Research). A largernumber of responses would also allow for con-sideration of more than 27 factors in future researchprojects.

(2) Conceptual similarity: High correlations (40.60) andsignificant correlations (min p� 0.05) were theselection criteria and existed between the differentfactors. High correlation between factors mayindicate an overlap in the actual factor beingevaluated from a conceptual standpoint. Whileindividual analyses may support specific correlationsbetween selected factors, the factor itself may not beconceptually distinguishable.

An example could be the linkages between‘lessons learned’ and ‘overview of sequence oftasks in the design process’ and between ‘lessonslearned’ and ‘best practices’ which are characterizedby a ‘high’ correlation coefficient. They are notpart of any cycle. Yet, they appear as strongcorrelations. They can be viewed as small clustersbeing tightly related to one another. Often, asthe studies have shown, users do not explicitlydifferentiate between the two latter ones, whichbasically document experience and knowledge inthe process. Equally, although there are more‘ingredients’ to lessons learned, knowledgeabout the sequence of process steps is for manyengineers a major purpose of documenting lessonslearned.

(3) Frequency of occurrence: The analysis was based onthe detected frequency of occurrence and strength ofcorrelation. Frequencies of occurrence have apowerful influence on later judgments of its value.The issue is to decide which correlations are causallymeaningful [75]. However, the ability seems to becompromised when there is no clear way to decidehow much and what type of causal understandingis needed. Evidence seems to be gathered andunderstood as converging to support a givencorrelation. In the present case, for example,through the various analyses (statistical, graph-theoretical, literature) ‘collaboration’ and ‘mutualtrust’ seem to indicate thematic centrality. It mightindicate a ‘valid’ tendency, it might, however, be



a skeletal set of a causal pattern that most peopleseem to use in similar ways.

(4) Causal relations: The complexity and variety ofcausal patterns will make it clear that one cannotpossibly track all causal patterns associatedwith factors influencing communication. Onecannot exhaustively assess all causal patternsthat people notice both because there are toomany domains to examine and because there is noeasy way to quantify the full range of what mightbe known and compare it to what is known.A literature review is used to detect potentialcausal relations.

(5) Lens: As much as detected correlations might beused as guidance or orientation points it must benoted that one might often use this kind of knowl-edge as a lens to interpret reality and, in lookingthrough this lens, is often unaware of the ways inwhich it guides one to track some sort of causalrelations more effectively than others [76].

10. Future Research

Part of the contribution of this study lies in providinga starting point for further research. For future work, weask ourselves whether found interrelations can be usedto generate hypotheses and questions (see Section 7,Conclusions) that can be tested in further empiricalstudies and ultimately form the basis for guidelines.In order to do that, preliminary questions need tobe answered with regard to the nature of correlationsfound.

10.1 Exploring the Nature of Correlations

It is assumed that in this network of factorsinfluencing communication there is no centralizedcontroller which means that global behavior emergesprobably as a result of concurrent local actions. Suchnetworks are typically modeled as multiple nodes, eachnode representing a state variable with a given value.Network dynamics is determined by the nature of theinfluences between nodes [3]. Therefore we need to askourselves, whether the influences are linear or not andare they symmetric or not? Yet, this assumption needsto be investigated further. The algorithm used tocompute correlations is based on linear influences thusunderstating possible nonlinear relationships. Linearnetworks are described as having a single attractor,i.e., a single configuration of node states that thenetwork converges towards no matter what the startingpoint, corresponding to the global optimum. Symmetricnetworks are ones in which influences between nodes are

mutual (i.e., if node A influences node B by amount Xthen the reverse is also true), while asymmetric networks(if they have cycles in them) add the complication ofhaving dynamic attractors, which means that thenetwork does not converge on a single configuration ofnode states but rather cycles indefinitely around arelatively small set of configurations [3].

10.2 Exploring Causal Relations of CorrelationsOver Time

Reviewing literature is one possibility to receive moreinformation about the chains of cause and effect of thecorrelations suggested by the performed data analyses.Another approach could be to perform detailed casestudies and asking the participants in concurrent designprojects directly which factor influences the other factorof a certain linkage.

It would also be interesting to see whether exposedcorrelations can be traced over time. Data gatheredfor this research project was acquired at a certain pointin time. Case studies performed at a certain timecould be repeated during different stages of the designprocess.

10.3 Incorporation of ‘Moderate-low’, ‘Weak’, and‘Negative’ Correlations

In this study, only linkages suggested by a ‘moderatehigh’ or ‘high’ correlation coefficient were comparedwith findings in the literature. In future researchprojects, ‘moderate low’ and ‘weak’ correlationscould also be part of the literature review. In addition,statistical analyses in this study are based onpositive correlations. Negative correlations were alsofound, yet not among the decision criteria applied inthis study.

Acknowledgments

The authors appreciate the assistance of the engineersand managers at all five industrial collaborators.Further, the authors are grateful to the editors andtwo anonymous reviewers for their helpful comments.

Appendices

The appendix consists of a table with a glossary ofhow the factors used in this study have been definedincluding the full set of grid sheets (Appendix Table A1)and the complete correlation matrix (AppendixTable A2).



References

1. Wheelwright, S.C. and Clark, K.B. (1992). RevolutionizingProduct Development, Quantum Leaps in Speed, Efficiency,and Quality. New York: The Free Press.

2. Friedman, T.L. (2005). The World Is Flat: A brief Historyof the Twenty-first Century. New York: Farrar, Straussand Giroux.

3. Klein, M., Sayama, H., Faratin, P. and Bar-Yam, Y.(2003). The Dynamics of Collaborative Design: Insightsfrom Complex Systems and Negotiation Research,Concurrent Engineering, 11(3): 201–209.

4. Bucciarelli, L.L. (1988). An Ethnographic Perspective onEngineering Design, Design Studies, 9(3): 159–168.

5. Detienne, F., Martin, G. and Lavigne, E. (2005).Viewpoints in Co-design: A Field Study in ConcurrentEngineering, Design Studies, 26(3): 215–241.

6. Sapuan, S.M., Osman, M.R. and Nukman, Y. (2006).State of the Art of the Concurrent Engineering Techniquein the Automotive Industry, Journal of Engineering Design,17(2): 143–157.

7. Andersson, K. (1999). A Design Process Model forMultiview Behaviour Simulations of Complex Products,In: Proceedings of DETC ’99, 1999 ASME DesignEngineering Technical Conferences, 12–15 September.

8. Großmann, T. (1998). Future Orientation of CAEApplications in the Passenger Car Development,VDI-Berichte 1411, Dusseldorf.

9. Maier, A.M., Eckert, C.M. and Clarkson, P.J. (2006).Identifying Requirements for Communication Support:A Maturity Grid-inspired Approach, Expert Systems withApplications, 31(4): 663–672.

10. Maier, A.M., Kreimeyer, M., Herfeld, U., Deubzer, F.,Lindemann, U. and Clarkson, P.J. (2006). ReflectingCommunication: A Key Factor for SuccessfulCollaboration between Embodiment Design andSimulation, In: Proceedings of International DesignConference – Design 2006, Dubrovnik, pp. 1483–1490.

11. Maier, A.M. (2007). A Grid-based Assessment Method ofCommunication in Engineering Design, Ph.D.Dissertation, Cambridge: Engineering Department,University of Cambridge.

12. Maurer, M., Boesch, N.-O., Tzonev, B. and Sheng, G.(2005). A Tool for Modelling Flexible Product Structures –MOFLEPS, In: Engineering Design and the GlobalEconomy, 15th International Conference on EngineeringDesign (ICED’05), Institution of Engineers Australia,Melbourne.

13. Hales, C. (2000). Ten Critical Factors in the DesignProcess, In: Proceedings of ASM/ASME Conference onFailure Prevention Through Education – Getting to the RootCause, Strongsville, Materials Parks (OH), pp. 49–55.

14. Morelli, M.D., Eppinger, S.D. and Gulati, R.K. (1995).Predicting Technical Communication in Product Devel-opment Organizations, IEEE Transactions on EngineeringManagement, 42(3): 215–222.

15. Van den Bulte, C. and Moenaert, R.K. (1998). The Effectsof R&D Team Co-location on Communication PatternsAmong R&D, Marketing, and Manufacturing,Management Science, 44(11): 1–18.

16. Gupta, A.K., Raj, S.P. and Wilemon, W. (1986). A Modelfor Studying the R&D/Marketing Interface in the ProductInnovation Process, Journal of Marketing, 50: 7–17.

17. Sosa, M.E., Eppinger, S.D. and Rowles, C.M. (2004). TheMisalignment of Product Architecture and OrganizationalStructure in Complex Product Development, ManagementScience, 50(12): 1674–1689.

18. Sosa, M.E., Eppinger, S.D., Pich, M., Mc Kendrick, D.G.and Stout, S.K. (2002). Factors that InfluenceTechnical Communication in Distributed ProductDevelopment: An Empirical Study in theTelecommunications Industry, IEEE Transactions onEngineering Management, 49(1): 45–58.

19. Yap, C.M. and Souder, W.E. (1994). FactorsInfluencing New Product Success and Failure in SmallEntrepreneurial High-technology Electronics Firms,Journal of Product Innovation Management, 11(5):418–432.

20. Montoya-Weiss, M.M. and Calantone, R. (1994).Determinants of New Product performance: A Reviewand Meta Analysis, Journal of Product InnovationManagement, 11(5): 397–417.

21. Allen, T.J. (1977). Managing the Flow of Technology:Technology Transfer and the Dissemination ofTechnological Information within the R&D Organization,Massachusetts: MIT Press.

22. Allen, T.J. (2000). Architecture and CommunicationAmong Product Development Engineers, In: Proceedingsof the 2000 IEEE Engineering Management Society, EMS –2000, Albuquerque, New Mexico.

23. Smith, R.P. and Eppinger, S.D. (1997). A Predictive Modelof Sequential Iteration in Engineering Design,Management Science, 43(8): 1104–1120.

24. Loch, C.H. and Terwiesch, C. (1998). Communication andUncertainty in Concurrent Engineering, ManagementScience, 44(8): 1032–1048.

25. Eckert, C.M. and Stacey, M.K. (2001). Dimensions ofCommunication in Design, International Conference onEngineering Design (ICED), Glasgow.

26. Eckert, C.M., Clarkson, P.J. and Stacey, M.K. (2001).Information Flow in Engineering Companies: Problemsand their Causes, International Conference on EngineeringDesign (ICED), Glasgow.

27. Hales, C. (1987). Analysis of the Engineering DesignProcess in an Industrial Context, Ph.D. Dissertation,University of Cambridge, Cambridge.

28. Hales, C. and Gooch, S. (2004). Managing EngineeringDesign, 2nd edn, London, New York, Berlin: Springer.

29. Badke-Schaub, P. and Frankenberger, E. (1999). Analysisof Design Projects, Design Studies, 20(5): 465–480.

30. Badke-Schaub, P. and Frankenberger, E. (2004).Management Kritischer Situationen. ProduktentwicklungErfolgreich Gestalten. Berlin, Heidelberg: Springer.

31. Roberts, T.L., Cheney, P.H. and Sweeney, P.D. (2002).Project Characteristics and Group Communication:An Investigation, IEEE Transactions on ProfessionalCommunication, 45(2): 84–98.

32. Chiu, M.-L. (2002). An Organizational View of DesignCommunication in Design Collaboration, Design Studies,23(2): 187–210.

33. Tiernan, S.D., Flood, P.C. and Murphy, E.P. (2002).Employee Reactions to Flattening OrganizationalStructures, European Journal of Work and OrganizationalPsychology, 11(1): 47–67.

34. Luhmann, N. (1995). Social Systems, Stanford: StanfordUniversity Press.



35. Jonas, W. (1994). Design – System – Theorie: uberlegungenzu einem systemtheoretischen Modell von Design-Theorie.Essen, p. 299.

36. Pulm, U. (2004). A System-Theoretic Treatment of ProductDevelopment (Translated from: Eine systemtheoretischeBetrachtung der Produktentwicklung, Ph.D. Dissertation,Lehrstuhl fur Produktentwicklung), TechnischeUniversitat Munchen, Munchen.

37. Maier, A.M., Eckert, C.M. and Clarkson, P.J. (2005).A Meta-model for Communication in Engineering Design,CoDesign: International Journal of CoCreation in Designand the Arts, 1(4): 243–254.

38. Spencer, H. (1965). The Principles of Sociology, In:Parsons (ed.), Theories of Society, pp. 139–143,New York: Free Press.

39. Busseri, M.A. and Palmer, J.M. (2000). ImprovingTeamwork: The Effect of Self-assessment on ConstructionDesign Teams, Design Studies, 21(3): 223–238.

40. Dorner, D. (1996). The Logic of Failure, New York:Metropolitan Books.

41. Blickensderfer, E., Cannon-Bowers, J.A. and Salas, E.(1997). Theoretical Bases for Team Selfcorrection:Fostering Shared Mental Models, Advances inInterdisciplinary Studies of Work Teams, 4: 249–279.

42. Valkenburg, R.C. (2000). The Reflective Practice in ProductDesign Teams to Assess Product Design Capability, Ph.D.Dissertation, Delft University of Technology, Delft.

43. Stempfle, J. and Badke-Schaub, P. (2002). Thinking inDesign Teams – an Analysis of Team Communication,Design Studies, 23(5): 273–296.

44. Badke-Schaub, P., Wallmeier, S. and Dorner, D. (1999).Training for Designers: A Way to Reflect Design Processesand Cope with Critical Situations in Order to IncreaseEfficiency, In: International Conference on EngineeringDesign (ICED), Munich.

45. Crosby, P.B. (1979). Quality is Free, New York:McGraw Hill.

46. Paulk, M.C., Curtis, B., Chrissis, M. and Weber, C.V.(1993). Capability Maturity ModelSM for Software,Version 1.1 Pittsburgh, Pennsylvania, Report CMU/SEI-93-TR-024 ESC-TR-93-177.

47. Chiesa, V., Coughlan, P. and Voss, C. (1996).Development of a Technical Innovation Audit, Journalof Product Innovation Management, 13(2): 105–136.

48. Chrissis, M.B., Konrad, M. and Shrum, S. (2003). CMMI.Guidelines for Process Integration and ProductImprovement, Boston: Addison-Wesley.

49. Moultrie, J. (2004). Development of a Design Audit Tool,Ph.D. Dissertation, University of Cambridge, Cambridge.

50. Austin, S., Baldwin, A., Hammond, J., Murray, M.,Root, D., Thomson, D. and Thorpe, A. (2001). DesignChains. A Handbook for Integrated Collaborative Design.Loughborough: Amec.

51. Argyris, C. and Schon, D. (1978). Organizational Learning:A Theory of Action Perspective, Reading (Mass): AddisonWesley.

52. Brown, D.C. (1996). Routineness Revisited. In: Waldron,M. and Waldron, K. (eds), Mechanical Design: Theory andMethodology, pp. 195–208, New York: Springer Verlag.

53. Bolboaca, S.-D. and Jantschi, L. (2006). Pearson versusSpearman, Kendall’s Tau Correlation Analysis onStructure-Activity Relationships of Biologic ActiveCompounds, Leonardo Journal of Sciences, 9: 179–200.

54. Brosius, F. (1998). SPSS 8.0. Professionelle Statistik unterWindows. Bonn: mitp, 1998.

55. Brown, B.L. Correlations. Available: http://www.gpc.edu/�bbrown/psyc1501/methods/correlations.htm, 2006.

56. Lindemann, U., Maurer, M. and Kreimeyer, M. (2005).Intelligent Strategies for Structuring Products, In: Clarkson,J.P. (ed.), Engineering Design – Theory and Practice,pp. 106–115, Cambridge: Engineering Design Centre.

57. Boardman, J. and Sauser, B. (2006). System of Systems –the Meaning of System of Systems Engineering, 2006IEEE/SMC International Conference, 24–26 April.

58. Eichinger, M., Maurer, M., Pulm, U. and Lindemann, U.(2006). Extending Design Structure Matrices andDomain Mapping Matrices by Multiple Design StructureMatrices, In: Proceedings of ASME ESDA 2006,Torino, Italy.

59. Colarelli, S.M. and Boos, A.L. (1992). Sociometric andAbility-based Assignment to Work Groups: SomeImplications for Personnel Selection, Journal ofOrganizational Behavior, 13(2): 187–196.

60. Vianen, A.E.M.v. and Dreu, C.K.W.D. (2001). Personalityin Teams: Its Relationship to Social Cohesion, TaskCohesion, and Team Performance, European Journal ofWork and Organizational Psychology, 10(2): 97–120.

61. Perry, M. and Sanderson, D. (1998). Coordinating JointDesign Work: the Role of Communication and Artefacts,Design Studies, 19(3): 273–288.

62. Moenaert, R.K., Caeldries, F., Lievens, A. andWauters, E. (2000). Communication Flows in Interna-tional Product Innovation Teams, Journal of ProductInnovation Management, 17(1): 360–377.

63. Ahuja, M.K., Galletta, D.F. and Carley, K.M. (2003).Individual Centrality and Performance in Virtual R&DGroups: An Empirical Study, Management Science, 49(1):21–38.

64. Krackhardt, D. and Stern, R.N. (1998). InformalNetworks and Organizational Crises: An ExperimentalSimulation, Social Psychology Quarterly, 51(2): 123–140.

65. Wong, P.S.P. and Cheung, S.O. (2005). StructuralEquation Model of Trust and Partnering Success,Journal of Management in Engineering, 21(2): 70–80.

66. Ng, K.-Y. and Chua, R.Y.J. (2006). Do I Contribute MoreWhen I Trust More? Differential Effects of Cognition –and Affected-based Trust, Management and OrganizationReview, 2(1): 43–66.

67. Bstieler, L. (2006). Trust Formation in Collaborative NewProduct Development, Journal of Product InnovationManagement, 23(1): 56–72.

68. Eckert, C.M., Stacey, M.K. and Earl, C.F. (2003).Ambiguity is a Double-edged Sword: Similarity Referencesin Communication, In: ICED03, 14th InternationalConference of Engineering Design, Stockholm, Sweden.

69. Kuhn, T. and Nelson, N. (2002). Reengineering Identity,Management Communication Quarterly, 16(1): 5–38.

70. Susman, G.I., Gray, B.L., Perry, J. and Blair, C.E. (2003).Recognition and Reconciliation of Differences inInterpretation of Misalignments When CollaborativeTechnologies are Introduced into New Product Develop-ment Teams, Journal of Engineering and TechnologyManagement, 20(1): 141–159.

71. Newman, R.A. (1999). Issues in Defining Human Rolesand Interactions in Systems, Systems Engineering, 2(3):143–155.



72. Thamhain, J.J. (2003). Managing Innovative R&D Teams,R&D Management, 33(3): 297–311.

73. Ayas, K. (1997). Integrating Corporate Learning withProject Management, International Journal of ProductionEconomics, 51(1): 59–67.

74. Meijer, B.R. (2006). Organization Structures of Dealingwith Complexity, Ph.D. Dissertation, ManagementScience, Delft: Delft University of Technology.

75. Murphy, G.L. andMedin, D.L. (1985). The Role of Theoriesin Conceptual Coherence, Psychological Review, 92(3).

76. Keil, F.C. (2003). Categorization, Causation and theLimits of Understanding, Language and CognitiveProcesses, 18(5/6): 663–692.

Anja M. Maier

Anja M. Maier graduatedform the University ofMunster and University ofToronto with a Mastersdegree in Political Science,Communication Science andPhilosophy in 2002 and isnow a Research Associate atthe Engineering Design Centre(EDC) of the University ofCambridge, where she com-pleted her doctoral degree on

communication in engineering design.Her interests are in the areas of design process

improvement, collaborative designing, communicationin design, system theory, system dynamics, performanceassessment systems and philosophy of design. In parti-cular, she is focusing on design as communicationbetween the designers and designers via the product.She has published a number of conference and journalpapers.

Ms. Maier has received a selected paper award at theDesign Conference 2006 in Dubrovnik for a paper thatwill soon be published in the Journal of EngineeringDesign. She has professional experience in the field ofinformation technology and is a fellow of the CambridgePhilosophical Society.

Matthias Kreimeyer

Matthias Kreimeyer, bornin Hannover, Germany, in1976, completed his MSc.degree as mechanical engineerat the German universities ofHannover and Munich in2004. He also graduated asIngenieur ECP from the EcoleCentrale Paris in 2003, finish-ing a double degree in generalengineering.

He has gathered experience in industry working forseveral major automotive manufacturers in Europe,Asia and Northern America. Currently, he works as aResearch Assistant at the Institute of ProductDevelopment at the Technische Universitat Munchen,Germany. He has published a number of papers oncollaboration and concurrent design processes, whichare the focus of his research interest. His work alsofocuses on methods for complexity management.

Mr. Kreimeyer has received a selected paper award atthe Design Conference 2006 in Dubrovnik for a paperthat will soon be published in the Journal of EngineeringDesign.

Clemens Hepperle

Clemens Hepperle born inInnsbruck, Austria, in 1980,graduated as mechanical engi-neer at the TechnischeUniversitat Munchen in 2006.A major part of his studies inmechanical engineeringfocused on systematic productdevelopment.He completed his Diploma

Thesis at the EngineeringDesign Centre at the

University of Cambridge (UK), dealing with commu-nication factors in collaborative design. He now worksas a Research Assistant at the Institute for ProductDevelopment at the Technische Universitat Munchen,Germany.

Claudia M. Eckert

Claudia M. Eckert is aSenior Research Associateand Assistant Director of theEngineering Design Centre atthe University of Cambridge.Her background is in mathe-matics, philosophy, and artifi-cial intelligence.She is interested in various

aspects of design processimprovement, in particular

process planning, change predication, and communica-tion in design. She is currently working on comparisonsbetween design practice in different domains.

Dr. Eckert has published over 100 conference andjournal papers and co-edited a book on designprocess improvement. She is a member of the editorialboard of Design Studies.



Udo Lindemann

Udo Lindemann born in1948, graduated as mechanicalengineer from the Universityof Hannover, Germany. Afterstarting his doctorate there, hefinished his work on‘A system’s perspective on thedesign process with specialregard of the production costsduring shape definition’ atthe Technische UniversitatMunchen in 1979.

During the following years, he worked in manage-ment positions in product design, production, andtesting technology for Renk AG, Augsburg inGermany. After 1992, he was CEO of MAN MillerDruckmaschinen GmbH. At the same time, he wasproduct manager for small sheet fed printing units atMAN Roland AG.Since 1995, he is full professor at the Technische

Universitat Munchen and Head of the Institute ofProduct Development. Prof Dr Lindemann is active in anumber of professional societies as for example BerlinerKreis, WGMK, VDI and The Design Society.

P. John Clarkson

John Clarkson is Professorof Engineering Design andDirector of the EngineeringDesign Centre at theUniversity of Cambridge. Hisresearch interests are in thegeneral area of engineeringdesign, in particular the devel-opment of design methodolo-gies to address specific designissues.In addition to publishing

over 400 papers in the past ten years he has written anumber of practitioner workbooks on medical equip-ment design and inclusive design. Prof. Clarkson is aChartered Engineer, a Fellow of the Institute ofEngineering and Technology, and a Member of theAmerican Society of Mechanical Engineers and theDesign Society.

Prof Clarkson is a Chartered Engineer, a Memberof the Institute of Electrical Engineers, AssociateEditor of the ASME Journal of Mechanical Design andon the editorial board of the Journal of EngineeringDesign.



Appendix

A1.Fullse

tofcommunicationgridsh

eets.

Maturity

leve

ls

Term

leve

lofInfluence

Factor

Definition

AB

CD

Current

Desired

Product

Represe

ntatio

nDegree

ofunderstand-

ing

and

adequacy

of

the

differenttypes

of

representations

of

aproduct

(e.g.,

bill

of

materials,drawings).

Each

departmenthas

itsown

way

ofrepre-

sentin

gproduct

infor-

mation

and

itis

unclearto

others

Each

departmenthas

itsown

way

ofrepre-

sentin

gproductinfor-

mation

and

itis

somewhat

clear

toothers

Each

departmenthas

itsown

way

ofrepre-

sentin

gproduct

infor-

matio

nanditis

mostly

clearto

others

Each

departmenthas

itsown

way

ofrepre-

sentin

gproductinfor-

matio

nanditis

always

clearto

others

[]

[]

Notatio

nDegree

ofunderstand-

ing

of‘‘for

exa

mple’’

drawingco

nve

ntio

ns.

Each

departmentuse

sits

own

notatio

nand

itis

unclearto

others

Each

departmentuse

sits

own

notatio

nand

itis

somewhatclear

toothers

Each

departmentuse

sits

own

notatio

nand

itismostlyclearto

others

Each

departmentuse

sits

own

notatio

nand

itis

alw

ays

clear

toothers

[]

[]

Term

inology

Degree

ofunderstand-

ingofsp

ecific

tech

nical

term

suse

d.

Each

departmentuse

sits

own

term

inology

and

itis

unclear

toothers

Each

departmentuse

sits

own

term

inology

and

itis

somewhat

clearto

others

Each

departmentuse

sits

own

term

inology

and

itis

mostly

clear

toothers

Each

departmentuse

sits

own

term

inology

and

itis

always

clear

toothers

[]

[]

Inform

atio

nDoyo

ukn

ow

which

info

rmation

the

otherpartyneeds?

Degree

ofthe

aware-

ness

of

the

oth

er

party’sneedsandpre-

ference

s.

No

Sometimes,

we

learn

from

mistake

sMostly

since

itis

docu

mentedand

communicated

which

inform

atio

nis

needed

Itis

entirely

clearto

us

what

inform

ation

we

need

and

the

inform

a-

tion

transm

ission

pro-

cess

iscontinuously

optim

ized

[]

[]

Ava

ilability

ofinfor-

matio

naboutco

m-

petitors

How

often

inform

atio

nabout

competitors

isdistributed

totheinter-

viewee.

Notnece

ssary

IfIask

forit

Itis

regularly

distribu-

tedto

me

Regularly

and

there

iscontinuous

effort

toadjust

the

distributio

nprocess

acco

rding

toneeds

[]

[]

Ava

ilability

ofinfor-

mation

about

our

company

How

often

inform

atio

naboutthe

own

com-

pany

isdistributed

totheinterviewee.

Notnece

ssary

IfIask

forit

Itis

regularly

distribu-

tedto

me

Regularly

and

there

iscontinuous

effort

toadjust

the

distributio

nprocess

acco

rding

toneeds

[]

[]

Ava

ilability

ofinfor-

matio

naboutpro-

cedures

How

often

inform

atio

nabout

pro

cedures

isdistributed

totheinter-

viewee.

Notnece

ssary

IfIask

forit

Itis

regularly

distribu-

tedto

me

Regularly

and

there

iscontinuous

effort

toadjust

the

distributio

nprocess

acco

rding

toneeds

[]

[]

Ava

ilability

ofinfor-

matio

naboutpro-

duct

specificatio

ns

How

often

inform

atio

naboutproduct

specifi-

catio

nsis

distributedto

theinterviewee.

Notnece

ssary

IfIask

forit

Itis

regularly

distribu-

tedto

me

Regularly

and

there

iscontinuous

effort

toadjust

the

distributio

nprocess

acco

rding

toneeds

[]

[]

(Contin

ued)



Appendix

A1.Continued.

Maturity

leve

ls

Term

leve

lofInfluence

Factor

Definition

AB

CD

Current

Desired

Organizatio

nHierarchies

Understanding

how

hiera

rchies

can

be

calle

duponto

ach

ieve

clearco

mmunicatio

n.

No

consideration

of

how

hierarchies

influ

-ence

communicatio

n

Hierarchies

are

only

calle

dupon

toco

rrect

amistake

Understanding

ofhow

hiera

rchies

opera

teand

could

be

calle

dupon

toach

ieve

clear

communication

pro

-ce

sses

There

isa

contin

uous

effo

rtto

thinkhow

hier-

archiesco

uld

beca

lled

upon

toach

ieve

clear

communication

pro

-ce

sses

[]

[]

Usage

of

pro

ce-

dures

Effort

toim

prove

de-

sign

procedures

and

the

usage

ofproce-

dures.

Pro

cedure

sare

not

use

dProcedures

are

used

inco

nsistently

Proce

duresare

under-

stoodanduse

dco

nsis-

tently

Contin

uouseffo

rtto

im-

prove

the

procedures

themselves

and

the

usa

geofproce

dures

[]

[]

Rolesandresp

onsi-

bilitie

sKnowledge

aboutthe

roles

and

resp

onsibil-

ities

of

oneself

and

others

and

the

use

of

itwhile

communicatin

g.

Notclearlydefin

ed

Mine

and

others’are

somewhatclearto

me

lam

fully

aware

ofmine

and

somewhataware

ofothers’

Iknow

min

eand

others’foreve

rytask

and

use

itco

nsc

iously

while

communicatin

g

[]

[]

Activity

atinterface

with

theotherparty

Degree

ofactivity

with

regard

tothe

interface

with

theotherparty.

Theytendto

forgetus

Theirco

mmunicatio

nis

reactive

Theirco

mmunicatio

nis

proactive

Our

communication

with

each

otheris

con-

tinuouslyproactive

de-

pendingonneed

[]

[]

Handlin

goftech

ni-

calco

nflicts

How

often

technical

conflictsare

addressed

andreso

lved.

Conflicts

are

alw

ays

addressed

Conflicts

are

some-

times

address

ed

ifit

can’t

be

avo

ided

any-

more

Conflictsare

mostly

ad-

dressed

and

solutio

ns

are

planned

Conflicts

are

alw

ays

addressed.

There

rsco

ntin

uouseffo

rtto

op-

timisehandlin

gofco

n-

flicts,

tofin

dso

lutio

ns

andto

implementthem

byusing

insights

from

previousca

ses

[]

[]

Transp

arency

ofde-

cisionmaking

Transp

arency

ofdeci-

sionmaking

andinvo

l-vement

of

the

right

people

inthe

decision

makingproce

ss.

Happensasithappens

Itis

sort

ofclearhow

decisions

are

made

and

we

correct

the

way

decisio

ns

are

made

once

amistake

occ

urs

The

decision

making

proce

ssis

transp

arent.

Myownco

ntributio

nto

the

decision

making

proce

ssis

clear

The

decision

making

process

isadequate

andthere

isco

ntin

uous

effo

rtto

reassess

trans-

parency

and

whether

the

right

people

are

invo

lvedin

theproce

ss

[]

[]

Organizatio

nApplication

ofco

r-porate

vision

and

values

Knowledge

and

appli-

catio

nofco

rporate

vi-

sionandva

lues.

Notkn

ownandapplied

Known

butnotkn

own

how

toapply

Known

and

partially

applied

Knownandapplicatio

nis

doneContin

uousef-

fortto

optim

iseapplica-

tion

[]

[]



Commongoals

and

objectives

Knowledgeandpursuit

ofco

mmon

goals

and

objectives.

Notkn

own.Nothinking

aboutit

Known

but

eve

ryone

follo

ws

just

his

orher

owngoals

Knownandso

metim

es

consideration

of

the

wayco

mmongoals

canbereach

ed

throughworking

together

Entirely

clearandiden-

tificatio

nwith

itwhichis

exp

ressedin

communi-

catio

nand

contin

uous

effort

toassess

and

adjust

goals

andobjec-

tives

and

the

way

toreach

them

[]

[]

Mutualtrust

Degreeofinterpersonal

trust

and

effort

tocreate

trust

with

inthe

project

team.

Wedon’tthinkaboutit

Wech

angetheco

urse

ofactiondependingon

whetherthere

istrust

or

not

We

are

aware

ofour

attitu

dewith

resp

ect

totrust

towardsthe

personweare

communicatin

g

There

istrust

andco

n-

tinuouseffo

rtto

create

trust

and

learn

how

totrust

[]

[]

Team

Best

practices

How

often

‘Best

prac-

tices’

are

considered

and

how

‘Bestprac-

tices’

are

communi-

cated

with

inthe

team

and

tootherteamsfor

future

task

exe

cutio

n.

No

capturing

ofbest

practices

Best

pra

ctices

are

‘stored’in

ourheads

Regular

consideration

ofbest

practices.

Docu

mentatio

nand

communicatio

nwith

intheteam

forfuture

task

exe

cutio

n

Regular

consideration

ofbest

practices.

Documentation

and

communication

within

the

team

and

other

teams

for

future

task

exe

cutio

n

[]

[]

Collaboratio

nRegularity

ofco

llabora-

tionandoftheeffo

rtto

improve

collaboratio

n.

Eve

ryone

looks

solely

afterhis

orhertask

sCollaboratio

nhappens

only

ifasked

for

inorderto

fulfiltask

s

Collaboratio

nhappens

proactively

inorderto

learn

from

others

and

improve

own

approach

es

Colla

boration

iscon-

structive,happensreg-

ula

rly

whenever

nece

ssary

andthere

iscontinuous

effort

toim

prove

it

[]

[]

Team

identity

Strength

ofbelonging

totheteam

anddegree

ofrefle

ction

how

team

identityca

nbe

strengthened.

There

isnone.Noris

itse

enasnece

ssary

Smallgroupsform

de-

pending

on

the

task

and

these

groups

get

their

identity

through

thetask

s

Attitudes

with

resp

ect

toteam

identity

are

continuously

refle

cted

uponin

orderto

finda

commondenominator

There

isastrongse

nse

of

belonging

tothe

team

and

contin

uous

refle

ctiononhow

team

identityca

nbeke

ptup

and

strengthened

for

theproiect

[]

[]

Project

reviews

Degreeofquantityand

quality

ofform

aland

inform

alreviewsto

plan

actions

and

torefle

ctongoals.

There

are

no

project

reviews

Pro

ject

reviews

are

only

held

tocorrect

mistake

s

Form

alreviewsto

cor-

rect

mistake

sandplan

future

approach

es

Regularform

aland

in-

form

alreviews

toplan

actions

and

refle

cton

project

goals

[]

[]

(Contin

ued)



Appendix

A1.Continued.

Maturity

leve

ls

Term

leve

lofInfluence

Factor

Definition

AB

CD

Current

Desired

Lessonslearned

How

often

‘Lessons

learned’

are

consid-

eredandhow

Lessons

learned’are

communi-

cated

with

inthe

team

and

tootherteamsfor

future

task

exe

cutio

n.

Noca

pturingof

lessonslearned

Lessons

learned

are

‘stored’in

ourheads

Regular

consideration

oflessonslearned.

Documentation

and

communication

within

theteam

forfuture

task

exe

cutio

n

Regular

consideration

oflessonslearned.

Documentation

and

communication

within

the

team

and

other

teams

for

future

task

exe

cutio

n

[]

[]

Individual

Overv

iew

of

se-

quence

oftask

sin

thedesignproce

ss

Degree

ofeve

rybody’s

ove

rview

ofthe

sequence

oftask

sin

thedesignproce

ssacc

ordingto

theirown

jobdesc

riptio

n.

Noove

rview

There

are

some

who

have

ove

rview

ofthe

sequence

oftasks

inthedesignproce

ssbut

they

should

generally

know

more

Most

people

have

ove

r-viewofthese

quence

of

task

sin

thedesignpro-

cess

and

know

how

these

are

related

Eve

ryonehas–acc

ord-

ingto

theirjobdesc

rip-

tion–enoughove

rview

of

the

sequence

of

task

sin

thedesignpro-

cess

and

there

isco

n-

tinuousthinkingofhow

the

sequence

oftask

scould

be

cle

arly

depicted

[]

[]

Autonomy

oftask

exe

cutio

nFreedom

inone’s

own

decisionsand

task

ex-

ecution

inalig

nment

with

one’s

resp

onsibil-

ities

and

coordinatio

nwith

others.

Itis

notthoughtof

Actions

are

changed

whenever

something

goeswrong.Nofurther

consideratio

n

Ihave

sufficientfree-

dom

inmy

decisions

tofulfilmytask

sauton-

omously

inalig

nment

with

myresp

onsibilitie

sand

incoord

ination

with

others

Ihave

sufficientfree-

dom

inmy

decisions

tofulfilmytask

sauton-

omously

inalig

nment

with

myresp

onsibilitie

sand

incoord

ination

with

others

and

this

applie

sto

the

whole

project

team

[]

[]

Generatio

nofinno-

vative/

altern

ative

ideas

How

generatio

nofin-

nova

tiveandalte

rnative

ideasis

supportedand

rewarded.

Notse

enasnece

ssary

There

isace

rtain

reluc-

tance

tomentio

nnew

ideas

Are

acc

epted

butonly

once

the

daily

tasks

have

been

accom-

plished

Contin

ously

supported

andrewarded

[]

[]

Educa

tion/training

Towhatd

egreetraining

and

education

plans

are

tailo

red

and

exe

-cu

ted.

There

isno

training

plan

There

are

trainingplans

butthey

are

only

on

paper

Tra

ining

plans

are

exe

cuted

Training

iscontinu-

ously

tailore

d,

exe-

cutedandappreciated

[]

[]

Best

use

ofmy

capabilitie

sHow

one’s

capabilitie

sare

realisedand

utilized.

Itis

notthoughtof

Itis

nota

matter

of

using

my

capabilities

inthe

best

waybutof

fulfilling

the

assigned

anddemandedtask

s

Isit

thoughtthrough

how

capabilitie

sco

uld

best

beutilised

Iam

attherightplace

where

mypotentia

land

capabilities

are

con-

tinuously

realised

and

utilisedacc

ordingly

[]

[]



Appendix

A2.Complete

correlationmatrix.



concurrent engineering - lori.academicdirect.orglori.academicdirect.org/cites/200801.pdf ·...

Documents