Task coordination in global virtual teams

While most of the explicit coordination studies in collocated teams see Table A.1 suggest matching task coordination mechanisms to task interdependence for effective coordination, studie

TASK COORDINATION IN GLOBAL VIRTUAL

TEAMS

JULIANA SUTANTO

(B.COMP (HONS), NUS)

A THESIS SUBMITTED FOR THE DEGREE OF DOCTOR OF PHILOSOPHY DEPARTMENT OF INFORMATION SYSTEMS NATIONAL UNIVERSITY OF SINGAPORE

Acknowledgements

This thesis has been made possible thanks to the assistance and support of a number

of individuals, for which I would like to express my appreciation

I thank my supervisors Assist Prof Atreyi Kankanhalli and Prof Bernard C.Y Tan for their advice and guidance throughout the duration of this thesis Atreyi has always been accessible for discussions and for providing advice and mentoring anytime I needed it Bernard has been an invaluable source of inspiration and support throughout the study The combination of their support and guidance has been instrumental for this thesis I look forward to working with them in the future

Faculty members at the National University of Singapore and from external universities have contributed to the success of this study From the annual IS workshops in which I have participated, I received many interesting and useful suggestions for carrying out this research from both the faculty members at the National University of Singapore as well as the visiting professors Several doctoral students gave me valuable comments when a part of this thesis was discussed at the PACIS 2006 and ICIS 2007 doctoral consortiums The anonymous editors and reviewers of journals and conferences offered directions to upgrade the quality of this research

I thank my friends and colleagues for their assistance during this study In particular, I thank Mr Phang Chee Wei and Mr Tan Chuan Hoo for motivating me and for always being around to support me They have been a source of support during my PhD

study I would also like to thank my friends who helped me in contacting the companies for my research Without them, I would not be able to complete this thesis

I am very thankful to my family for their encouragements during the many years that I spent working on this thesis I thank my husband, Fadly for his constant support, valuable advice, and great patience when I would scold him to vent out my frustration I thank my mother for her constant prayers for my academic career I thank my whole family for their encouragement and forbearance during my study

Last, but not least, I thank God for his great mercy and blessings Without His wonderful blessings, I would not be what I am today Without His generous help, I would not be able to complete this thesis Thank you God, thank you Jesus Christ, and thank you dearest Mother Mary for the strength and wisdom that You gave me to pull through all the obstacles to complete this thesis

Table of Contents

ESSAY 1

1 Introduction 1

1.1 Research Motivation 1

1.2 Research Questions 3

1.3 Potential Contributions 4

2 Conceptual Background 5

2.1 Task Interdependence 6

2.2 Classifying Task Coordination Mechanisms into Task Coordination Modes 8

2.3 Matching Task Interdependence and Task Coordination Portfolios in Collocated Teams 12

3 Research Methodology 14

3.1 Team Background 14

3.2 Data Collection 18

3.3 Data Analysis 19

4 Findings… 23

4.1 Findings in Team A 23

4.2 Findings in Team B 28

4.3 Findings in Team C 32

5 Discussions of the Findings 37

5.1 The Effect of Perceived Time Constraint on Group Task Coordination Modes 37

5.2 The Opposing Effects of Member Dispersion on Group Task Coordination Mode 40

5.3 Matching GVT Contingencies to GVT Task Coordination Portfolio 42

6 Contributions and Implications 51

6.1 Contributions to Research 51

6.2 Implications for Practitioners 54

7 Conclusion 56

ESSAY 2 1 Introduction 57

1.1 Research Motivation 57

1.2 Research Questions 59

1.3 Potential Contributions 60

2 Conceptual Background 62

2.1 Task Coordination Modes and Communication Complexities 62

2.2 IT-mediated Task Coordination Modes 64

2.3 Ideal Profiles of GVT Task Coordination 68

3 Research Model 76

3.1 Defining Fit 76

3.2 Research Hypotheses 77

4 Research Methodology 80

4.1 Survey Instrument Development 83

4.1.1 Operationalization of the Main Variables 83

4.1.2 Operationalization of the Control Variables 89

4.1.3 Conceptual Validation 93

4.2 Pilot Study 96

4.3 Field Study Description 99

5 Data Analysis and Findings 101

5.1 Instrument Validation 101

5.2 Interrater Agreement 103

5.3 Hypotheses Testing 103

6 Discussions of the Findings 111

7 Contributions and Implications 114

7.1 Contributions to Research 115

7.2 Implications for Practitioners 118

8 Conclusion 123

Summary…… 124

Appendix A – Summaries of Previous Studies 146

Appendix B – Operationalization of the Variables 150

Appendix C – Modified Operationalizations of Team Efficiency and Effectiveness Variables 155

Appendix D – Power Analysis 156

Appendix E – Interrater Agreements of Common Variables 157

Appendix F – Cluster Center Values 160

Abstract

Global virtual teams (GVT) are known to afford benefits to organizations such as reduced costs and improved performance by utilizing expertise from around the globe The potential benefits have provided an impetus for the adoption of GVT work structures in organizations However, realization of the benefits is contingent on GVT’s ability to coordinate their tasks effectively This dissertation documents a qualitative exploratory study (essay 1) and a quantitative explanatory study (essay 2)

of GVT task coordination This dissertation examines how to design effective GVT task coordination, proposes optimal GVT task coordination designs, and tests the impact of the proposed designs on GVT work performance

Essay 1 – Designing Task Coordination Portfolios in GVT

Unlike collocated teams working in a common context, GVT have added task coordination challenges that can range from more obvious differences in members’ time zones and working hours, to less controllable differences in members’ IT infrastructure Thus there is a need to align GVT work structure and GVT task coordination design Through an in-depth study of 13 tasks in three GVTs, this first essay seeks to address how to design effective GVT task coordination We observe that by designing a set of task coordination mechanisms (i.e., task coordination portfolio) that fits the team’s contingencies in the form of task interdependence, members’ dispersion, and perceived time constraint, GVT’s task coordination is likely

to be effective

Essay 2 – Task Coordination in GVT: An Empirical Study

in the Context of Software Development

GVT are proliferating in today’s work environment, and are particularly popular in software development where over half of the Fortune 500 firms actively engage in global software development practices While globally dispersed software development teams offer advantages of time zone and labor-cost differences to slash software development costs, poor task coordination has been shown to lead to substantial loss of software development speed as well as erase much of the cost savings In this second essay, we first extend the concept of optimal task coordination portfolios introduced in the first essay to optimal IT-mediated task coordination portfolios By theoretically matching GVT contingencies (i.e., task interdependence, members’ dispersion, and perceived time constraint) to IT-mediated task coordination portfolios, we propose six ideal profiles of GVT task coordination These profiles are then statistically validated by surveying 112 globally dispersed software development teams The results support the hypothesized positive relationship between the ideal profiles of GVT task coordination and task coordination effectiveness, which in turn leads to better GVT performance

List of Tables

ESSAY 1

Table 1 Examples of Task Coordination Mechanisms 10

Table 2 Team Characteristics 15

Table 3 Data Sources 18

Table 4 Cross-Case Findings 38

Table 5 Proposed GVT contingency—Task Coordination Portfolio Fit 42

ESSAY 2 Table 1 Proposed Ideal Profiles of GVT Task Coordination 70

Table 2 Some of the Well-Known Software Development Lifecycle Models 82

Table 3 Unstructured Sorting Results 95

Table 4 Structured Sorting Results 96

Table 5 Results of Factor Analysis and Reliability Check 98

Table 6 Descriptive Statistics 100

Table 7 Results of Factor Analysis and Reliability Check 102

Table 8 Descriptive Statistics 104

Table 9 Normality Tests 106

Table 10 Correlation Matrix of the Independent and Control Variables 106

APPENDIX Table A.1 Summary of Explicit Coordination Studies in Collocated Teams 146

Table A.2 Summary of Implicit Coordination Studies in Collocated Teams 147

Table A.3 Summary of Explicit Coordination Studies in Distributed Teams 148

Table A.4 Summary of Implicit Coordination Studies in Distributed Teams 149

Table B.1 Operationalization of IT-mediated Task Coordination Portfolio 150

Table B.2 Operationalization of Task Interdependence 151

Table B.3 Operationalization of Member Dispersion 151

Table B.4 Operationalization of Time Constraint 151

Table B.5 Operationalization of Task Coordination Effectiveness 152

Table B.6 Operationalization of Team Efficiency 152

Table B.7 Operationalization of Team Effectiveness 152

Table B.8 Operationalization of Team Size 153

Table B.9 Operationalization of Task Novelty 153

Table B.10 Operationalization of Task Analyzability 153

Table B.11 Operationalization of Task Variability 154

Table B.12 Operationalization of IT Accessibility 154

Table C.1 Modified Operationalization of Team Efficiency 155

Table C.2 Modified Operationalization of Team Effectiveness 155

Table D.1 Power Analysis Result 156

Table E.1 Interrater Agreements for Common Variables 159

Table F.1 Cluster Center Values 160

List of Figures

ESSAY 1

Figure 1 Typology of Task Interdependence 6

Figure 2 Task Interdependence-Task Coordination Portfolio Fit in Collocated Teams .13

ESSAY 2 Figure 1 Communication Complexities of Different Task Coordination Modes 64

Figure 2 Research Model 76

Figure 3 Hypotheses Testing Results using Summated Scales 107

Figure 4 Hypotheses Testing Results using Factor Score Scales 108

Figure 5 Hypotheses Testing Results in Requirement Analysis and Design Phases 110 Figure 6 Hypotheses Testing Results in Coding and Testing Phases 110

Essay 1

Designing Task Coordination Portfolios in

Global Virtual Teams

1 Introduction

Global virtual teams (GVT) refer to groups of people with a common purpose, who

carry out interdependent tasks across countries and sometimes across organizational boundaries, using information and communication technology (ICT) to communicate much more than face-to-face meetings (Cramton 2001, Maznevski and Chudoba 2000) GVT offer a multitude of benefits to organizations (Dube and Pare 2004, Powell et al 2004) Among them, GVT allow organizations to swiftly respond to customer needs and utilize expertise from around the globe to effectively perform organizational tasks GVT deployment can lead to cost savings, higher productivity, and improved organizational performance (Ahuja et al 2004)

However, GVT entail certain challenges that are less evident in collocated teams

Particularly, task coordination has been observed as a key concern in GVT Task

coordination is the management of task interdependencies across geographically

dispersed members For example, Carmel (2006) noted that task coordination hurdles across time zones negated some of the cost savings possible via GVT Studies comparing collocated and distributed teams, observed greater task coordination hurdles (Hinds and Mortensen 2002) and poorer coordination outcomes (Cummings

and Kiesler 2003) in distributed teams Sharma and Krishna (2003) reported the case

of a globally dispersed software development team that suffered a delay of one year and budget overruns due to poor task coordination When teams are primary mechanisms for accomplishing organizational work, effective coordination of teamwork becomes an important organizational issue (Faraj and Sproull 2000) Effective task coordination enables team members to enhance task performance by establishing shared understanding of members’ work and avoiding redundant and duplicate work (Hoegl et al 2004, Kraut and Streeter 1995) Thus it is vital to understand how to design effective GVT task coordination in order to realize potential benefits of GVT

Previous studies of GVT task coordination proposed frequent communications, besides standardization and scheduling coordination mechanisms, for effective GVT task coordination (e.g Cummings and Kiesler 2003, Herbsleb and Grinter 1999, Warkentin et al 1997) Surprisingly, a survey by Hinds and McGrath (2006) revealed that GVT with relatively more cross-site communication experience more (not less) coordination problems Subsequently Hinds and McGrath (2006) highlighted the need

to properly align GVT communication and work structure The need to match GVT task coordination mechanisms and GVT work structure is also echoed by Massey et

al (2003) While their experimental study suggested scheduling and synchronization

is associated with higher GVT performance, they noted that "Future research is

needed to examine the effects of alternative coordination mechanisms and how to match an appropriate mechanism to the task" (p 151) The practical challenges of

GVT task coordination and the lack of understanding of the phenomenon motivate us

to investigate the optimal design of GVT task coordination mechanisms

1.2 Research Questions

With this objective, we build on coordination studies from the collocated team context As interdependencies lie at the heart of coordination (Malone and Crowston 1994), most studies in collocated teams proposed to match the type of task

interdependence to the task coordination mechanisms Task interdependence denotes

the extent to which team members are dependent upon one another to perform their individual jobs (Van de Ven et al 1976) In collocated teams, specific sets of task coordination modes are found to effectively coordinate different types of task interdependence (Andres and Zmud 2001/2002, Kraut and Streeter 1995, Kumar and van Dissel 1996); the rationale for which can be explained by information processing theory This theory suggests that a higher interdependence task requires more information processing for coordination, which can be satisfied by a more complex task coordination mode with higher information processing capacity In this study, we

refer to a set of task coordination modes used for a task as a task coordination

portfolio Unlike collocated teams working in a common context, GVT has added task

coordination challenges that can range from more obvious differences in time zones and working hours, to less controllable differences in members’ IT infrastructure (Riopelle et al 2003) Thus, how this concept of fit in collocated teams can be extended to design effective GVT task coordination is of interest in this study Therefore the research questions for this essay are:

1 Are there any other factors besides task interdependence that influence the design

of task coordination portfolios in GVT?

2 How can these contingencies be fit to task coordination portfolios to achieve effective task coordination?

The paper starts by explaining the fit between task interdependence and task coordination portfolio in collocated teams Subsequently, we use a qualitative case study approach for in-depth investigation of task coordination in three GVTs As team tasks change over the duration of a project, task interdependencies and task coordination portfolios used vary (Adler 1995) For each GVT under study, we first identify the tasks as our unit of analysis (in total we identify 13 tasks) For each task

we identify the coordination portfolio used, other contingencies influencing the design

of the coordination portfolio besides task interdependence (if any), and coordination outcomes Based on the data analysis, we then derive the optimal GVT task coordination portfolios to fit the contingencies that result in effective task coordination

This essay aims to contribute to the understanding of GVT task coordination in several ways First, as a stepping stone for the study of GVT task coordination, this essay offers an in-depth investigation regarding this matter Second, in investigating GVT task coordination, the essay builds on the concept of fit between task interdependence and task coordination portfolios in collocated teams; thus creating an opportunity for future research to empirically compare and contrast task coordination portfolios designs in collocated teams and GVT Third, the essay aims to propose optimal GVT task coordination portfolios to fit GVT contingencies to address the need to match GVT task coordination mechanisms and GVT work structure Findings

in this essay can also guide organizations in handling GVT task coordination to fully reap the benefits offered by such work structures

2 Conceptual Background

To better position the study, it is important to recognize that most previous task coordination studies highlight either explicit or implicit coordination, while a few emphasize both (e.g Hinds and McGrath 2006) Explicit or administrative coordination refers to formal and interpersonal mechanisms used to manage dependencies in the team to accomplish its task; whereas implicit or social cognition coordination refers to socially shared cognitive processes to meet the demands of task-based skill and dependencies Summaries of these studies are presented in Appendix A

Effective task coordination is defined as the institution of shared understanding of

members’ work and the avoidance of redundant and duplicate work (Hoegl et al

2004, Kraut and Streeter 1995) While most of the explicit coordination studies in collocated teams (see Table A.1) suggest matching task coordination mechanisms to task interdependence for effective coordination, studies of implicit coordination in both collocated and distributed teams (see Table A.2 and A.4 respectively) appear to claim more effective coordination when team members have shared work experience

or familiarity with one another1 Explicit coordination studies in distributed teams (see Table A.3), however, do not provide clear guidelines on how to achieve effective task coordination With various possible task coordination mechanisms such as communications, standardization, and scheduling, there is a lack of guidelines on how

to design an appropriate task coordination mechanism for a distributed team

As the root of many team performance problems can be attributed to the mismatch

1

Shared work experience or familiarity cannot always be controlled through member selection or socialization

between the required and actually used coordination mechanisms (Gerwin 2004), in the distributed team context there is a need to examine the effects of alternative coordination mechanisms and how to match an appropriate mechanism to the task (Hinds and McGrath 2006, Massey et al 2003) Therefore although explicit and implicit coordination are both important, this paper will focus on investigating explicit coordination in GVT In particular, we aim to identify optimal task coordination design under different GVT contingencies To investigate the phenomenon, we start with a review of explicit coordination studies in collocated teams which suggest matching task interdependence to task coordination mechanisms for effective task coordination

Four types of task interdependence have been proposed (Grandori 1997, Van de Ven

et al 1976), i.e., pooled, sequential, reciprocal, and team interdependence (see Figure 1)

Figure 1 Typology of Task Interdependence

In a pooled interdependence task each member completes their part independently

followed by aggregation With minimal interactions between members, a pooled interdependence task has low information processing need for coordination For

Task enters team

Task leaves team

Task enters team

Task leaves team

Task enters team

Task leaves team

Task enters team

Task leaves team

Pooled Sequential Reciprocal Team

example, in software development, each programmer may code and test their part

independently followed by system integration A sequential interdependence task

involves the completion of one part before the other begins In this type of task, there

is greater information processing need for coordination compared to a pooled task because members must adjust their work if any downstream member fails to meet the task expectations An assembly line format is an example of a sequential interdependence task As opposed to sequential interdependence tasks that move in

one direction, reciprocal interdependence tasks flow in a “back and forth” manner

between members The level of information processing needs for coordination is higher in reciprocal interdependence compared to the previous two types because each member’s action must be adjusted according to the actions of other interacting members For example, in software bug fixing, there can be a “back and forth” flow

of work between a software tester and a software coder In a team interdependence

task, all members concurrently diagnose, problem-solve and collaborate as a group to

deal with the task There is no measurable temporal lapse in the flow of work among team members The level of information processing need for coordination is the highest in team interdependence tasks because members must adjust their actions in the process of simultaneously working on the same task at the same point in time, e.g during brainstorming for new product development

It is important to note that when the taxonomy of task interdependence was first introduced, it was considered as an inherent property of the task at hand (Van de Ven

et al 1976) Subsequent research has revised this view by suggesting that team members may construct and modify their task interdependence (Grant 1996) For instance, they can change the sequencing of work from linear (sequential) to parallel

(pooled)

Overall a hierarchical relationship is suggested among the four types of task interdependence with pooled interdependence being the least complex and team interdependence being the most complex (Kumar and van Dissel 1996) While moving up the hierarchy, more information processing is required to coordinate the task (Andres and Zmud 2001/2002, Kraut and Streeter 1995) To fulfill the information processing needs of tasks with different types of interdependence, various task coordination mechanisms have been suggested

2.2 Classifying Task Coordination Mechanisms into Task Coordination Modes

Researchers have identified specific task coordination mechanisms, as shown in Table

1

Author(s) Context of

the Study

Task Coordination Mechanisms

Adler (1995) ̇ Standards (e.g compatibility standards, design rules

or tacit product/process fit knowledge, and manufacturing flexibility)

̇ Plans or schedules (e.g capabilities development schedules, sign-offs, and exceptions resolutions plans)

̇ Mutual adjustments (e.g coordination committees, producibility design reviews, and producibility engineering changes)

̇ Teams (e.g joint development, joint teams, and transition teams)

Andres and Zmud

(2001/2002)

̇ Informal, cooperative (e.g horizontal communication)

̇ Formal, controlling (e.g vertical communication) Curtis et al (1988) ̇ Project members’ interaction through team meetings

and informal, ad hoc communication between team members

Gresov (1998)

Collocated teams

̇ Standardization

̇ Supervisory discretion

̇ Employee discretion

̇ Personnel specialization

̇ Vertical communication

̇ Horizontal communication Kraut and Streeter

(1995)

̇ Formal impersonal (e.g project documents and memos, project milestones and delivery schedules, modification request and error tracking procedures and data dictionaries)

̇ Formal interpersonal (e.g status review meetings, design review meetings and code inspections)

̇ Informal interpersonal (e.g group meetings, collocation of requirements and development staff)

̇ Electronic communication (e.g e-mail and e-bulletin boards)

̇ Interpersonal network (e.g number of supervisors from outside the project talked to in the previous 2 years)

Nidumolu (1995) ̇ Vertical coordination (coordination through project

manager, or steering committees)

̇ Horizontal coordination (coordination through meetings, or one-on-one discussions)

̇ Plans

̇ Mutual adjustments Van den Ven et al

(1976)

̇ Rules (policies and procedures)

̇ Predetermined work plans or schedules

̇ Mutual adjustments through vertical communication channels (unit supervisor, assistant unit supervisor), or horizontal communication channels (formally

designated work coordinator, simply contacting another member who is likely to have the desired info)

̇ Group meeting (scheduled and unscheduled) Sabherwal (2003) Outsourced

project to collocated teams

̇ Rules or standards (e.g compatibility standards, data dictionaries, design rules, error tracking procedures)

̇ Plans or schedules (e.g delivery schedules, project milestones, requirement specifications, sign-offs, test plans)

̇ Formal mutual adjustment (e.g code inspections, coordination committees, design review meetings, hierarchies, liaison roles, reporting requirements, status review meetings)

̇ Informal mutual adjustment (e.g collocation, impromptu communications, informal meetings, joint development, transition teams)

Bordetsky and

Mark (2000)

Virtual teams

̇ Feedback (communication between collaborating partners)

Cummings and

Kiesler (2003)

GVT ̇ Face-to-face meetings

Kumar et al (2005) ̇ Sequencing or scheduling

̇ Real time remote communication

̇ Boundary spanning (communication through boundary spanners)

Massey et al

(2003)

̇ Scheduling

̇ Synchronization Montoya-Weiss et

al (2001)

̇ Scheduling Prikladnicki et al

(2004)

̇ Face-to-face meetings

̇ Standards Sharma and

Table 1 Examples of Task Coordination Mechanisms

The various task coordination mechanisms in Table 1 can be categorized as different task coordination modes depending on the type of communication and the communication partners Coordination through rules, policies, procedures, standards, plans, and schedules relies on prior specification of codified blueprints, action programs, or specific targets (Sabherwal 2003) As codified blueprints of actions, the application of these coordination mechanisms does not require much one-on-one

communication between participants Hence, they correspond to the impersonal mode

of task coordination (Van de Ven et al 1976), which requires minimal direct communication between task performers

In contrast, coordination through mutual adjustments uses interpersonal interaction to make changes based on information obtained during the task (Sabherwal 2003) The interpersonal interaction may take place between superior and subordinate (vertical communication) or between peers (horizontal communication) (Nidumolu 1995),

which corresponds to the personal mode of task coordination (Van de Ven et al

1976)

While team meetings have been considered as task coordination by mutual adjustments in a few studies (e.g., Kraut and Streeter 1995), they differ from mutual adjustments through vertical and horizontal communications by the simultaneity of multilateral interactions They facilitate interaction among team members, and offer them an opportunity to coordinate their tasks directly with one another (Gittell 2002)

Hence team meetings correspond to the group mode of task coordination (Van de Ven

et al 1976)

Impersonal coordination requires predetermined task coordination information (e.g., rules and plans) to be conveyed to task performers Personal coordination involves communicating task coordination information from supervisor to subordinate or from peer to peer, perhaps altering or processing the information to some extent During group coordination, there are multiple exchanges of task coordination information among all task performers

2.3 Matching Task Interdependence and Task Coordination Portfolios in Collocated Teams

According to information processing theory, a higher interdependence task requires more information processing for coordination, which can be satisfied by a more complex task coordination mode with higher information processing capacity Previous studies in collocated teams suggest that pooled and sequential interdependence tasks should be primarily coordinated by impersonal task coordination mode (i.e., standards and plans) (Gresov 1989, Thompson 1967)

Reciprocal interdependence requires back and forth interactions among task performers whose form, direction and content of interaction often cannot be anticipated in advance Hence, it should be primarily coordinated by personal task coordination mode (i.e., mutual adjustments) Since situations with higher level of task interdependency also contain lower level of task interdependency (Thompson

1967, Kumar and van Dissel 1996), besides personal coordination mode, optimal task coordination portfolio for reciprocal interdependence tasks should also contain impersonal coordination mode

Team interdependence, also known as intensive interdependence, refers to the situation in which people work simultaneously on the same task at the same point in time (Grandori 1997) In such highly interdependent work, team meetings can enhance work performance by facilitating interaction among members, and offering them an opportunity to clarify doubts and coordinate their tasks directly with one another Thus, optimal task coordination portfolio for a team interdependence task consists of not only impersonal and personal coordination modes, but also group

coordination mode

Figure 2 shows the fit between task interdependence and task coordination portfolio (i.e combination of coordination modes used for a task) in the collocated team context How this fit can be extended to the GVT context is of interest in this study

Figure 2 Task Interdependence-Task Coordination Portfolio Fit in Collocated Teams

Task Interdependence Task Coordination Portfolio

Impersonal, personal and group task coordination modes (standards, plans, mutual adjustments, team meetings)

interdependence

3 Research Methodology

Positivist exploratory case study methodology was considered appropriate for this study for several reasons First, the phenomenon is complex, and cannot be studied outside its context (Benbasat et al 1987, Yin 2003) GVT task coordination is intertwined with GVT environment in which the task coordination takes place Second, while finding in collocated team studies may provide useful pointers, the idiosyncratic structural and contextual issues surrounding GVT call for specific research attention (Powell et al 2004) Third, our main objective is to propose optimal task coordination portfolios in GVT context Although we referred to explicit coordination studies in collocated teams to guide our investigation, we allowed for new findings to emerge that could refine our initially adopted theoretical perspective (Eisenhardt 1989) To ensure rigor in our methodology, we closely followed the suggested guidelines for positivist exploratory case study research (Dube and Pare 2003)

The three GVTs in this study, identified as teams A, B and C, consisted of Masters level students located at three universities in Asia, North America, and Europe Having globally dispersed members and relying mainly on information and communication technology (ICT) to communicate and collaborate, they fulfilled the necessary characteristics of GVT These teams differed from typical student GVT in previous studies in terms of project realism All participants had at least two years of work experience and were selected by organizational sponsors to work on specific projects based on their skills and expertise Further, each GVT was assigned a project that was proposed by an organization and monitored by a project manager from the

sponsor organization At the end of the five-month projects, the teams had to present

their results to the organizational sponsors and faculty members who together

evaluated their performance The background information of the three teams is shown

in Table 2 As can be seen in the table, the presence of different types of task

interdependence in these projects provided us the opportunity to investigate the fit

between task interdependence and task coordination portfolios

Team A Team B Team C Team size 7 members 11 members 9 members

2 in Asia, 4 in North America, 3 in Europe

Sponsor Global computer company Global consulting company Global telecom company

Project

objective

Re-engineer financial

analyst business unit

within the sponsor

company for more

effective structure

Collect information about risk assessment and management in different industries using a pre- designed questionnaire

Identify emerging mobile applications, charging mechanisms, and technologies from Internet Service Providers (ISPs)

(pooled interdepend.)

2 Interview and administer risk assessment survey in each location

(pooled interdepend.)

3 Weekly report writing of interview progress

(sequential interdepend.)

* added due to conflict

during previous task

1 Discuss the type of mobile application to focus on and create questionnaire and interview questions

(team interdepend.)

2 Distribute interview job

Team A worked on a global project sponsored by an international computer company and supervised by a project manager who resided in Asia Their objective was to draft

a proposal to re-engineer the Financial Analyst (FA) business unit within the sponsor company for a more effective structure The team was required to interview the FAs and their managers from around the world to identify the problem areas and to propose solutions in the form of a report to the sponsor company Almost all team A’s task such as questionnaire creation, distributing interview jobs, identification of the problem areas, and writing the recommendation report required input from all team members All members needed to brainstorm to design the questionnaire, decide on who should interview whom, identify and classify the problems across the interviews, and generate the recommendation proposal Thus these tasks could be categorized as

team interdependence tasks The only remaining task in the project i.e., carrying out

interviews, was a pooled interdependence task The member dispersion in team A was

relatively lower (across 2 continents, with 7 hours common time frame) than in teams

B and C (across 3 continents, with 1 hour common time frame)

The global project that team B worked on was sponsored by an international consulting company, and supervised by a project manager who was stationed in North America Their goal was to understand risk assessment procedures in businesses in a variety of industries Members needed to collect information about risk measurement, risk monitoring, and risk management in global businesses through interviews with top executives from around the world The collected information would then be compiled by the project manager The team was provided with a pre-designed risk assessment questionnaire and a list of industries to investigate Since each member could almost independently identify the target companies in the suggested industry

sectors and interview the executives before the aggregation of results, team B’s tasks

could be a priori classified as pooled interdependence tasks However, conflict during

the course of the project led the project manager to split the team into two subgroups and add another task to each subgroup, which was to write weekly reports about their interview progress Subgroup members needed to send their weekly reports to subgroup leaders who would then summarize the reports and send the summaries to the project manager Since this last task required sequencing, it could be classified as

a sequential interdependence task

Team C worked on a global project sponsored by an international telecommunications company, and supervised by a project manager who resided in Europe The purpose

of the project was to understand the global market for mobile applications from the Internet Service Provider (ISP) perspective Members had to interview ISPs from around the world, and identify significant and emerging mobile applications, charging mechanisms and underlying technologies Almost all of team C’s tasks (i.e., brainstorming the type of applications and generating the questionnaire, distributing interview jobs, analyzing the data, and writing the final report) required input from all members To accomplish the tasks, members had to concurrently diagnose, problem-

solve and collaborate as a group Hence, these tasks were categorized as team

interdependence tasks The only remaining task in the project i.e., carrying out

interviews, was a pooled interdependence task

All teams were provided with technology for emailing, teleconferencing, video conferencing, bulletin board discussion, and instant messaging (ICQ) The bulletin board was linked to the course website where members of all teams could post text

messages Team members could teleconference from their personal computers However, for videoconferencing they had to reserve and use a separate videoconference room There was only one videoconference room in each location

Different roles for multiple investigators encourage the development of distinct views that can then be compared (Dube and Pare 2003, Eisenhardt 1989) In our study, two authors were directly involved in the data collection process while one author remained a detached outsider to challenge objectivity of the study Table 3 shows our multiple data sources over the five-month duration of the global projects Each source adds richness and strengthens theory grounding by triangulation of evidence The data were gathered mainly through non-obtrusive methods such that they were not created

as the result of the presence of researcher during team interaction Moreover, bulk of the data came from communication logs that are stable and exact All e-mail logs, ICQ logs, and bulletin board postings were recorded Some teleconference meetings were videotaped and detailed meeting minutes were available for the rest Meeting minutes of videoconference meetings were obtained All project documentation was archived, including project reports, and members’ lesson-learned papers

Data Sources Team A Team B Team C

Synchronous

Communication

logs

̇ 1 teleconference meeting transcription

̇ 3 teleconference meeting minutes

̇ 3 ICQ meetings logs

* didn't use videoconference

̇ 3 teleconference meeting transcriptions

̇ 3 teleconference meeting minutes

* didn't use videoconference and ICQ

̇ 3 teleconference meeting transcriptions

̇ 7 teleconference meeting minutes

̇ 3 videoconference minutes

* didn't use ICQ Asynchronous

3.3 Data Analysis

Each task in a GVT project constitutes our unit of analysis (see Table 2) In total we identified 13 tasks i.e., 5 in team A, 3 in team B (originally team B had 2 tasks which were restructured to 3), and 5 in team C Our analysis followed four broad steps: (1) for each task we identified its task coordination portfolio, (2) we also identified additional contingencies besides task interdependence (if any) that influenced the task coordination portfolio design, (3) we assessed each task's coordination effectiveness, and finally (4) derived the fit between GVT contingencies and task coordination portfolios by comparing portfolios used for specific contingencies and resultant outcomes across all tasks

Step 1 – Identifying Task Coordination Portfolios

As defined in this study, a task coordination portfolio may consist of one or more task

coordination modes used for a task To identify use of the impersonal task

coordination mode, we looked for predetermined rules, standards, plans or schedules

Below is an example of team A’s use of the impersonal task coordination mode:

1st draft: Saturday, 15 May, 12 pm Asian time Asian and North American teams send out each team's proposed solutions (as detailed as possible) Each team then reviews other team's proposal—for input and addition to own team's 2nd draft proposal

To identify use of personal task coordination mode, we looked for mechanisms by

which a leader (vertical) or a team member (horizontal) directed coordination of the task with mutual adjustments i.e., subsequent adjustments to the work arrangements

by others Below is an example of team B’s use of the personal, vertical task coordination by its project manager:

I need to know which of you will be contacting which of the contacts If you are already

on holiday or gone for the weekend I will just assign them by the industries …

Based on the reply of team B’s members, the project manager coordinated the works

An example of Team A’s use of the personal, horizontal task coordination is evidenced in the following email message of one of its members:

Although A1 started this first, maybe it is a good idea to continue with the questionnaire that A2 sent to avoid confusion later when the questionnaire for managers is needed

Based on a positive response from A1, this member then further coordinated the works

To identify use of group task coordination mode, we looked for team meeting

mechanisms Below is an extract from team C’s teleconference meeting illustrating the use of this mode:

C1 (Member 1 in team C): … for tomorrow, do we want to give the project manager a full application or a narrowed down application?

C2: Full, I think

C1: What do the rest think?

C3: I agree

C4: I think we give him a full one and if he wants to cut it … just let him

C5: Anyway he has to cut it otherwise it is too much for us

C6: … when we talk to the ISP, are we asking the ISP what the applications are?

C7: … I think, … we ask the ISPs which applications they think are upcoming, …

C3: … at first we need to ask what they think is the best application … then we make a comment

C8: … Right now I think all the top applications are wireless e-commerce, video, web pages

C9: We can see other applications also by reading I think it is good to just make a summary

of the applications to give to the project manager

At the end of Step 1, we had identified the task coordination portfolios used for coordinating each of the 13 tasks in the three GVTs

Step 2 – Identifying Additional Contingencies

In this step, we identified additional contingencies besides task interdependence that influenced the design of each task coordination portfolio found in Step 1 For example, when a member in team A used the impersonal task coordination mode as part of the coordination portfolio for their interview job distribution task, she wrote:

… Due to time constraint, I have taken the liberty to assign the people who will interview the FAs and their business managers … I hope you guys do not mind … Here is the assignment:

In this example, perceived time constraint led the member to utilize impersonal

coordination mode As another example, with a one-hour common time frame, team B was restricted in its use of team meetings A member from North America wrote in his lesson learnt paper:

Being dispersed across 3 continents, we did not have much common time for team meetings The team has done a poor job of keeping others up to date about what they are doing …

Thus, the use of the group task coordination mode was dependent on members’

dispersion

Step 3 – Assessing Task Coordination Effectiveness

Based on the previous conceptualization of task coordination effectiveness (Hoegl et

al 2004, Kraut and Streeter 1995), we noted down the incidents of duplicate and redundant (extra) work for each task The presence of duplicate or redundant work indicated ineffective task coordination For example, team A’s concurrent use of personal and group task coordination modes for interview job distribution task was

considered as ineffective due to the existence of duplicate work Two people

mistakenly contacted the same interviewee as evidenced in the following email log:

Just off the phone with Carol … It turns out that A3 had contacted her and had sent her the questionnaire …

Another example of ineffective coordination occurred during team B’s first task (i.e., identify companies to be interviewed) Task coordination was considered as

ineffective due to the existence of redundant work The impersonal task coordination

message was as follows:

Here are some of things that I will need from each of you by Tuesday

… List industries which are local to your region List companies in these industries which are local List any companies in which you personally know an employee Format: Name (of the employee), How do you know them (friend, classmate, relative, etc.), Position

Misinterpreting the impersonal task coordination message, members in one continent did extra work Instead of performing the work individually as specified, they redundantly discussed about completing the work together, as shown in the email log

below:

Shall we meet one of these days? … We can discuss about the company lists …

We did not find duplicate or redundant work in any of team C’s tasks The effectiveness of team C’s task coordination was apparent in member’s lesson learnt paper:

… my team has been working effectively and harmoniously throughout the time

Step 4 – Deriving GVT contingencies—task coordination portfolios Fit

After identifying task coordination portfolio and contingencies, as well as assessing

coordination effectiveness for each task, we derived the fit between GVT

contingencies and task coordination portfolios This was done by comparing portfolios used for specific contingencies and resultant outcomes across all tasks Optimal portfolios are those that resulted in effective coordination for specific contingency combinations

4 Findings

The findings are presented by team and task Each task in a team constitutes our unit

of analysis (see Table 2) In total there are 13 tasks i.e., 5 in team A, 3 in team B, and

5 in team C For each of the 13 tasks, we identified its task coordination portfolio, additional contingencies besides task interdependence (if any) that influenced the task coordination portfolio design, and the effectiveness of its task coordination

Task 1: Questionnaire Creation (Team Interdependence)

At the beginning of their project, team A faced a team interdependence task They needed to create a questionnaire, which required all members to coordinate and work simultaneously Team A’s task coordination began with the project manager posting

impersonal task coordination information in the bulletin board, which stated among

other things the standards and rules to follow while creating the questionnaire, as indicated below:

The questionnaire should gauge how to attain the most cost effective organization model that will increase the effectiveness and reduce cost of the finance organization while:

1 Providing at least the same level of support (if not better) for the country business management…

2 (Maximizing) knowledge sharing, back-ups, and career paths for the analysts …

Following the posting of impersonal coordination information, two members in different locations (Asia and North America) simultaneously formulated the questionnaire and emailed their versions of the questionnaire to the rest of the team This finding confirmed that impersonal mode is indeed not sufficient to coordinate a

team interdependence task Following the duplicate work, one member took charge to

coordinate the work by personal coordination mode

Although A1 started this first, maybe it is a good idea to continue with the questionnaire that A2 sent to avoid confusion later when the questionnaire for managers is needed

Based on a positive response from A1, this member then arranged for ICQ meetings

(group coordination mode) to further coordinate the task Email was then utilized to

disseminate the work plan discussed during the ICQ meeting

Coordination design in this task was found to be contingent on the type of task interdependence No other influences on the choice of task coordination modes were indicated Coordination in this task was initially ineffective because of the existence

of duplicate work when the project manager coordinated the task with the impersonal coordination mode alone at the start

Task 2: Interview Job Distribution (Team Interdependence)

After the questionnaire creation, team A needed to decide who should interview whom This was again a team interdependence task because all members needed to

discuss and agree on the assignment together However members’ perception of time

constraint imposed by the project manager added to the difficulties in gathering all

members for group coordination Consequently, team A coordinated this task by

alternating between ICQ-mediated group coordination and email-mediated personal

coordination between representatives from North America and Asia Alternating between group and personal coordination modes however proved to be detrimental There was a misunderstanding about who should interview a particular financial analyst In an email sent before it was clear that it was the responsibility of the members in Asia However, in a subsequent ICQ meeting session, some members in North America volunteered to interview the same financial analyst No one corrected

this duplicate assignment When a member from Asia contacted the financial analyst,

he found out that she had been interviewed by his teammates in North America, as

indicated in the email log below:

Just off the phone with Carol… It turns out that A3 had contacted her and had sent her the questionnaire Have I missed an email on this? Isn’t Asia supposed to interview Carol?

During this time, the project manager would frequently send new lists of interviewees Perceiving time pressure, a member in Asia was extremely worried She then

formulated an interview job plan (impersonal task coordination mode) and sent it to

all other members through email:

… Due to time constraint, I have taken the liberty to assign the people who will interview the FAs and their business managers … members in North America and Asia will have a chance to interview both FAs and business managers so as to have a better view of their jobs Here is the assignment:

As indicated above, coordination design in this task was influenced not only by the type of task interdependence, but also by the perceived time constraint Coordination was considered as ineffective when there was a duplicate interview job assignment while alternating between group and personal modes Subsequently, team A had a meeting to select leaders for their remaining tasks A member wrote in her email:

We felt that everyone is working with little coordination We are afraid that some times a lot of duplications in efforts are being spent on some tasks and at other times not enough efforts are allocated to other tasks Therefore, we thought that we could increase our efficiency and effectiveness by assigning one or two of us to be the "driver" for those tasks The "driver" will be responsible for coordinating, monitoring, and controlling the team's effort on the assigned tasks

Task 3: Interview (Pooled Interdependence)

The selected leader for the interview task coordinated the task with the impersonal

task coordination mode (plan/schedule), as indicated in the email log below:

… arrange your interviews as early as you possibly can I think May 14th is a little bit late … try to finish the interviews by May 12 th (around one month before the project deadline)

Task interdependence and perceived time constraint seemed to influence the coordination design Task coordination was considered effective because of the lack

of duplicate or redundant work and the interviews being completed satisfactorily

Task 4: Problem Identification and Generate Solutions (Reciprocal Interdependence)

Based on their interview results, team A needed to discuss the problems faced by the financial analysts and devise solutions At this time, some members kept stressing about the approaching deadline in their email, and occasionally used capital letters when they wanted their remote partners to take immediate action In this situation, we observed a case where the inherent task interdependence was modified to suit the

team contingency The selected leader reduced the interdependency of this task from

team interdependence where everyone needed to problem solve together to reciprocal

interdependence (back and forth sending of solutions) between members in Asia and North America The leader coordinated the task with the impersonal task coordination

mode (plan/schedule), as indicated in the email log below:

1 1st draft: Saturday, 15 May, 12 pm Asian time Asian and North American teams send out each team's proposed solutions (as detailed as possible) Each team then reviews other team's proposal—for input and addition to own team's 2nd draft proposal

2 2nd draft: Tuesday, 18 May, 12 pm Asian time Asian and North American teams send out the 2nd draft proposed solutions

3 Final: Thursday, 20 May (around three weeks before the project deadline), 12 pm Asian time Asian and North American teams send out final proposed solutions

Besides the above impersonal task coordination mode, we also observed personal,

horizontal coordination between representatives from Asia and North America For

example, after each continent exchanged their second draft, the following personal task coordination was noted:

Hi (to North American members):

We (the Asia members) suggest to refocus the solutions to say why we make certain recommendations to answer directly to the project manager’s questions of cost effectiveness, customer's needs, and employee's needs…

Based on the response from the North American representative, both continents

standardized the structure of their proposed solutions

Perceived time constraint caused the team to reduce the inherent interdependency of this task and influenced their task coordination design in addition to task interdependence The task coordination design was considered effective as there was

no instance of duplicate or redundant work We observed work routines in this task, i.e., collocated members would problem solve together and then passed the draft of their suggested solutions to their remote teammates If there were ambiguities, a member would liase with members in another location

Task 5: Final Report Writing (Team Interdependence)

It was observed that in contrast with the previous three tasks, in this last task, there was no tension in the team with regard to the approaching deadline This could be because in the previous task, they had not only identified problems and generated solutions but also written them down; thus reducing the time needed to complete their final report writing task To coordinate this last task, team A utilized a teleconference

meeting (group mode) Below is an extract from the meeting transcript:

A2: … I think chapter 5 should be done by …

A4: It is on “Limitations and other factors”

A2: Yes right, so the person doing the final editing will take a look at chapter 5 But A5 said that we may need to bring it forward and put it after chapter 1 to preset the stage A6: Why don’t we move the limitations under methodology?

After each meeting, members would email the meeting minutes as a work plan,

leading to concurrent utilization of group and impersonal coordination modes When there was ambiguity during report writing, personal coordination was utilized to resolve the confusion as shown below:

A6: … Are we to put all the tables in our report ?

A4: Put them in the appendix

A6: All the details are put in appendix We need to be selective

A4: I think all important things need to put in the executive summary

Coordination in this task was contingent on task interdependence and was considered effective Everyone knew who would write which chapters and who would do the final editing No duplicate or redundant work was observed

Summary of Team A findings

Besides task interdependence, perceived time constraint was also found to influence task coordination design by limiting the use of group coordination Task coordination was ineffective when impersonal mode was used alone to coordinate the first team interdependence task and when the team alternated between personal and group modes (without the use of impersonal mode to schedule/plan the coordination) in the second team interdependence task The rest of the tasks (i.e task 3 with impersonal task coordination mode, and tasks 4 and 5 with impersonal, personal, and group task coordination modes) were well coordinated

Task 1: Identify Target Interviewees (Pooled Interdependence)

Team B’s first task required members to identify and list their target interviewees, which would then be pooled together by the project manager Coordination design in this task was contingent on task interdependence The project manager coordinated

the task by emailing an impersonal task coordination (plan/schedule) message:

Here are some of things that I will need from each of you by Tuesday

… List industries, which are local to your region List companies in these industries, which are local List any companies in which you personally know an employee Format: Name (of the employee), How do you know them (friend, classmate, relative, etc.), Position

There was confusion in the team following the above message Instead of performing

this task individually, members in Asia thought that they had to work together to complete the task As a result, there were redundant discussions among members in Asia as evidenced in the following email message, which signaled ineffective task coordination:

Shall we meet one of these days? I am wondering whether the submission of the company list in our country is to be done together or is it individual?

Task 2: Interview (Pooled Interdependence)

The next task was to conduct interviews and administer the questionnaire to the target interviewees previously identified by each member The project manager sent the compiled list of interviewees to the team At this point, members in North America were worried about their remote partners’ work progress because they could not gauge what was going on in the other two continents To fulfill their need to be aware

of other members’ work progress, members in North America initiated team meetings

(group mode) After each team meeting, a member would email the meeting minutes, which contained their work plan (impersonal mode) until their next meeting

However, the team experienced difficulty in arranging team meetings due to their time differences as indicated in the following email log:

Since members in Europe has problem with US 10pm, Europe 7am, Asia 1pm video conference group meetings, how about considering another day instead of Wednesday …

or how about 1am to 2am US time??? I know that it is difficult for my teammates in US

I myself have a class at 8 in the morning after that, but why not give it a try instead of ruling out the various possibilities

Simultaneously time pressure from the manager also added to the complexity of scheduling team meetings Even when a team meeting was tentatively scheduled, some members could not attend it While some members questioned the importance

of the team meeting for coordination, some other members grumbled about not having enough team meetings for sharing local information A member from North America

wrote in his lesson learnt paper:

The team has done a poor job of keeping others up to date about what they are doing and

it came to confrontation levels Individuals thought that others were not doing anything when they had actually completed their own work and had done work for others…

Due to lack of sharing of local task expectations, there was confusion about who was supposed to deal with each interviewee The situation is exemplified by an incident of duplicate work when a North American member and a European member interviewed the same person, as indicated in a member’s lesson-learnt paper:

I felt that it is not clear who should be responsible for interviewing someone … B1 (a member in North America) and B2 (a member in Asia), both contacted L (an interviewee

in Asia) B2 was confused on what is the “latest” arrangement

To remedy the confusion due to lack of awareness of what was happening at other

locations, personal task coordination was exercised by the project manager An email

was sent by the project manager to team B’s members:

I need to know which of you will be contacting which of the contacts I need this as soon

as possible If you are already on holiday or gone for the weekend I will just assign them

by the industries …

Based on members’ responses, the project manager re-allocated the remaining interview jobs Unfortunately, despite the project manager’s efforts, the seed of group friction was already sown Members clashed and subsequently, the project manager split the team into two subgroups (across locations) Coordination in this task was considered ineffective with the existence of duplicate work and confusion about who should do what

Task 3: Weekly Report Writing of Interview Progress (Sequential Interdependence)

Following the split, the project manager added another task for each subgroup, which was to write weekly reports about their interview progress The project manager

coordinated this task with the impersonal task coordination mode in the form of rules

as shown below:

rules: … each team can go after companies in any of the 7 industries However, if one team member has already made contact, that is their company … Exceptions - Billion dollar business units …

Assignments for Team Captains:

… I need a summary of activities from each captain about what their team has done …

This task required sequencing, i.e., first subgroup members would send their weekly reports to subgroup leaders, then subgroup leaders would summarize the reports and send the summaries to the project manager To coordinate the work, each subgroup

leader utilized impersonal task coordination mode (plan) as shown in the following

email from a subgroup leader to his subgroup members:

As you may have seen in the project manager’s email, she wants a weekly report from the two sub teams, written by the captain To put this together as she wants, I need to know from all of you every week: Companies pursued, scheduled, done

Personal task coordination via vertical communication was also utilized in this task

For example, when a subgroup member could not submit her interview progress report on time, the subgroup leader personally coordinated with the particular member

as indicated below:

B3, since you have an exam to focus on, I can add your missing information next week

Coordination design in this task was contingent on task interdependence and was found to be effective in terms of reporting interview progress No duplicate or redundant work was observed in this task

Summary of Team B findings

Coordination of team B’s first pooled interdependence task was ineffective even though impersonal task coordination mode was utilized The main reason was because the dispersed members interpreted the impersonal coordination information differently and they did not know about this until there was redundant work In their