Revisiting Communication and Trust in Globally Distributed Teams: A Social Network Perspective

1
Revisiting Communication and Trust in Globally
Distributed Teams A Social Network Perspective
Manju Ahuja Kelley School of Business

• Coathors
• Saonee Sarker, Suprateek Sarker (Washington State
  University)
• Sarah Almbjerg (Copenhagen Business School)

2
Agenda

• State of knowledge on globally distributed teams
• The theorized relationships among communication,
  trust, and performance
• Communication and Trust from a Social Network
  Perspective
• Research Methodology
• Findings
• Discussion

3
Research on Trust in VTs

• Key areas of research in globally distributed
  teams
• Trust (e.g., Jarvenpaa, Shaw, and Staples 2004
  Piccoli and Ives, 2003 Sarker, Valacich, and
  Sarker 2003)
• Communication (e.g., Piccoli, Powell, and Ives
  2004 Galvin and Ahuja 2001 Jarvenpaa and
  Leidner 1998)
• The most widely researched of the issues
  surrounding virtual teams (Powell et al. 2004,
  p. 17)

Trust is Generally a dependent variable
4
Research in virtual teams

• Focus on group performance
• Need to investigate individual performance (Mehra
  et al. 2001)
• Need to identify the high performing team members
  (e.g., Powell et al. 2004)
• Reliance primarily on individual trait-based or
  sometimes behavior-based explanations
• Need structural/relational approach (Tichy 1981)
• Research on the structural position of
  individuals can answer why are some people
  better performers than others (Mehra et al. 2001)

5
RESEARCH QUESTION
What is the role of communication and trust
centrality in determining an individuals
performance within a globally distributed
team? The approach - networked individualism.
6
Networked Individualism

• Noted researchers have observed that ICT-mediated
  groups are moving towards networked
  individualism (Wellman et al. 2003)
• By bringing to bear measures and constructs of
  social structure, we can begin to how simple
  notions of .. autonomous individuals are
  incomplete (Rice 1994, p. 181)

7
Networked Individualism (contd.)

• If you took away my computer, my colleagues, my
  office, my books, my desk, my telephone I
  wouldnt be a sociologist writing papers,
  delivering lectures, and producing knowledge. Id
  be something quite other and the same is true
  for all of us. (Law 1992)

8
Virtual Teams as a Social Network

• We conceptualize a distributed team as a social
  network, and each individual having a structural
  position within that network.

Communication- Trust-based Stru. Position

Performance
9
Three Models

• We explore three perspectives regarding the
  nature of influence of trust and communication on
  individual performance in globally distributed
  teams
• They represent three Strands of Theorizing about
  the role of Communication and Trust
• an additive model
• an interaction model, and
• A mediation model.

10
The Additive Model

• Twin predictions concerning performance
• One preditcs a strong linkage between trust and
  performance (Hossain and Wigand 2004 Coppola,
  Hiltz, and Rotter 2004)
• Prevailing view of trust in the IS literature
  contends that trust has direct positive effects
  on .. performance (e.g., Iacono and Weisband
  1997 Jarvenpaa and Leidner 1999)
• The other predicts that Ineffective
  communications, may hinder performance
  (Scarnati 2001)

Trust
Individual Performance
Communication
11
The Interaction Model

• Model suggests that both trust and communication
  are necessary for higher individual performance
• That is, trust and communication interact to
  affect outcome (Jarvenpaa et al. 2004)
• E.g., team member may be perceived as a low
  performer by peers if he/she exhibits low
  communication and does not enjoy the trust of
  other members (Jarvenpaa et al. 2004)

Trust
Communication
Individual Performance
12
The Mediation Model

• Any effect of communication on performance is due
  to trust
• Communication leads to trust, and trust leads to
  performance (Coppola, Hiltz, and Rotter 2004).
• Trust is developed through communication
  (Handfield 1994)
• High levels of trust will cause the trustor ..
  to perceive good performance (Jarvenpaa et al.
  2004)
• Several empirical studies on the trust
  development process suggest that video and
  audio.. are nearly as good as face-to-face
  contacts provided that participants engage in
  various getting-acquainted activities.. (Hossain
  and Wigand 2004)

Individual Performance
Trust
Communication
13
Ego-centric Network View

• Communication Centrality
• The extent to which a member is communicatively
  connected with each of the other members within a
  team
• Trust Centrality
• The extent to which a member enjoys the trust of
  each of the other members within a team
  (trustworthiness)
• Degree-based

14
Communication Centrality
Legend Blue nodes Location A team members Red
nodes Location B team members Size of nodes
Communication centrality
15
Trust Centrality
Legend Blue nodes Location A team members Red
nodes Location B team members Size of nodes
Trust centrality
16
Research Methodology

• A field study of hybrid virtual teams
• Sample
• US-Norway student teams engaged in systems
  development
• Duration 1 semester
• US-Denmark student teams engaged in systems
  analysis
• Duration 6 weeks
• N111

17
Measures

• In-degree centrality
• In-degree centrality is relatively stable even at
  a low sampling level (Valente and Davis 1999)
• Freemans (1979) measure of relative in-degree
  centrality (i.e., the actual number of lines
  relative to the total number that it could
  sustain) was used
• Performance
• .. the effects of networks on performance..
  measured by supervisor ratings, may contain
  political aspects (Brass 2003)
• Consistent with the above comment, each team
  member was asked to rate the performance of every
  other team member

18
Analysis Technique

• Additive Model
• Linear Regression
• Interaction Model
• Hierarchical Regression (Mehra et al. 2001)
• Mediation Model
• Linear Regression following the guidelines of
  Baron and Kenney (1986)

19
Results - Additive Model

• Model Summary
• Effect of communication (b .001, pgt .10)
• Effect of Trust (b .519, p lt .05)
• R-square .646
• Results fail to support the Additive Model

20
Interaction Model

• Model Summary
• 1st block with communication centrality and trust
  centrality as predictors (R-square .646, 2nd
  model R-square .781)
• R-square change is .134 (F-change is significant)
• 2nd block included the above predictors and an
  additional interaction term
• The ANOVA model (1st Model (F 98.736, p lt .01),
  2nd Model (F 126.85, plt .01, Role of
  communication (b -.064, pgt .10), role of trust
  (b .562, plt .01), role of interaction (b -.444,
  plt .01)
• 2nd Model has better fit.
• However, direction of the interaction is anomalous

21
Mediation Model

• Model Summary (Baron and Kenney, 1986)
• Commun. centrality affects trust centr. (b .832,
  plt.01)
• Commun. centrality affects performance (b .432,
  plt .01)
• Trust centrality affects performance (b .519, plt
  .01) and effect of commun. centrality disappears
  (b .001, pgt .10)
• Thus, full mediation exists (Baron and Kenney
  1986)
• Results support the mediation model

22
Discussion

• Complete mediation of trust on the relationship
  between communication and performance
• That is, high levels of communication cannot lead
  to high performance until he/she is trusted by
  the other team members
• More communication is not always better
  (Krackhardt and Hansen 2003)

23
What about the anomaous Moderation Model?

• To understand anomalous moderating model, we
  split the sample into
• High trust centrality
• Low trust centrality
• In hi-trust group, the interaction effect is
  positive negative in the low-trust group
• Less trustworthy members are harmed by more
  communication

24
Possible effect of task? No!

• We split the sample into
• those involved in systems analysis tasks
  (US-Denmark),
• those involved in systems development tasks
  (US-Norway)
• Results are consistent, showing robustness

25
Continuing Research

• Continuing to qualitatively explore the three
  models
• Initial exploration supports regression results

26
Questions?