University of Southern California Dissertations and Theses

Technology support for virtual collaboration for innovation in synchronous and asynchronous interaction modes

(USC Thesis Other)

Technology support for virtual collaboration for innovation in synchronous and asynchronous interaction modes

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
TECHNOLOGY SUPPORT FOR VIRTUAL COLLABORATION FOR
INNOVATION IN SYNCHRONOUS AND ASYNCHRONOUS
INTERACTION MODES

by

Nathan David Yates

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(INFORMATION SYSTEMS)

August 2007

Copyright 2007 Nathan David Yates

ii
Acknowledgements
I would first like to express my heartfelt thanks to my chair, Ann Majchrzak,
for the extraordinary effort she has made on my behalf through all stages of the
dissertation process.
I have been grateful for the love and support of my family, especially my
parents David and Cassie Yates, my in-laws Albert and Jackie Harris, and my sons
Justin and Ryan who have provided both comic relief and the resolve to finish.
Finally, and most importantly, I would like to thank my wife Heather for her
unyielding support, love, and understanding throughout this process. My family, my
success, and my happiness are all due to Heather and I enjoy sharing them with her
now and always.

iii
Table of Contents
Acknowledgements

ii
List of Tables

vii
List of Figures

x
Abstract

xii
Chapter 1: Executive Summary
1.1 Rationale: Understanding Individual’s Cognitions in Synchronous and
Asynchronous Interaction Modes
1.2 Summary of the Research Questions
1.3 Summary of Findings
1.4 Summary of Implications of the New Theory of Virtual Collaboration
for Innovation

1
2

4
5
9

Chapter 2: Virtual Collaboration for Innovation
2.1 Collaboration Outcomes of Virtual Collaboration for Innovation
2.1.1 Innovation Outcomes in Virtual Collaboration for Innovation
2.1.2 Learning Outcomes in Virtual Collaboration for Innovation
2.1.3 Both Learning and Innovation Outcomes Realized in Virtual
Collaboration for Innovation
2.2 Factors Which Positively Influence Learning and Innovation in Virtual
Collaboration for Innovation
2.3 Difficulties Achieving Sharing and Reflection in Virtual Collaboration
for Innovation
2.4 Virtual Collaboration for Innovation Occurs Over a Series of
Interaction Modes
2.4.1 Differences in Interaction Mode Impact Sharing and Reflection
2.4.2 Prior Theories of Virtual Collaboration for Innovation Have Not
Incorporated Interaction Mode Differences
2.5 Multi-Dimensional Role of CT Support for Virtual Collaboration for
Innovation
2.5.1 Inconsistent Reports of the Utility of CT Support
2.5.2 Multi-dimensional CT Support
2.6 Summary and Research Questions
2.7 Summary of Findings

11
15
15
17
18

20

23

26

28
31

33

33
35
39
40

iv
Chapter 3: Theory Development
3.1 Virtual Collaboration for Innovation Requires Perspective Making and
Perspective Taking
3.2 Differential Impact of Sharing and Reflection on Learning and
Innovation
3.2.1 Impact of Sharing and Reflection in Synchronous and
Asynchronous Interaction Modes on Learning
3.2.2 Impact of Sharing and Reflection in Synchronous and
Asynchronous Interaction Modes on Innovation
3.2.3 Summary of Impacts of Sharing and Reflection on Learning and
Innovation
3.3 Differential Impact of CT Support on Sharing and Reflection
3.3.1 CT Support for a Testing and Adjusting Strategy
3.3.2 CT Support for an Attention Focusing Strategy
3.3.3 CT Support for a Contextualization Strategy
3.3.4 CT Support for a Perspective Taking Strategy
3.4 Differences in CT Support for the Cycle of PM/PT
3.4.1 Impact of CT Support for a Testing and Adjusting Strategy on
Sharing in the Synchronous and Asynchronous Interaction Modes
3.4.2 Impact of CT Support for an Attention Focusing Strategy on
Sharing in the Synchronous and Asynchronous Interaction Modes
3.4.3 Impact of CT Support for a Contextualization Strategy on
Reflection in the Synchronous and Asynchronous Interaction Modes
3.4.4 Impact of CT Support for a Perspective Taking Strategy on
Reflection in the Synchronous and Asynchronous Interaction Modes
3.4.5 Summary of Hypothesized CT Support
3.5 Control Variables
3.5.1 Control for Number of Participants
3.5.2 Control for Number of Participant Locations
3.6 Summary: Combined Research Model

42
43

46

47

50

52

53
56
57
58
60
61
63

64

66

67

69
71
71
73
74

Chapter 4: Case Study: Individual’s Use of CT to Support Their
Interaction With Others as Part of The Millenial Project
4.1 Case Study Setting
4.2 Case Study Methodology
4.3 Findings: Evidence of Differential Impact of CT-Supported PM/PT on
Collaboration Outcomes in the Synchronous and Asynchronous Interaction
Mode
4.3.1 Impact of Sharing and Reflection on Learning
4.3.2 Impact of Sharing and Reflection on Innovation

77

78
79
80

82
83

v
4.4 Findings: Evidence of CT Support
4.4.1 Initial Lack of CT Support Negatively Impacted Collaboration
Outcomes
4.4.2 Evidence of CT Support for Sharing and Reflection in the
Synchronous Interaction Mode
4.4.3 Evidence of CT Support for Sharing and Reflection in the
Asynchronous Interaction Mode
4.5 Summary of Case Study Findings

84
84

87

91

95

Chapter 5: Research Methods
5.1 Research Setting
5.2 Data Collection
5.2.1 Survey Methodology
5.2.2 Survey Responses
5.2.3 Respondent Demographics
5.3 Survey Measures
5.3.1 CT Support Measures
5.3.2 Sharing Diverse Knowledge and Reflection on others’ shared
knowledge
5.3.3 Collaboration Outcomes
5.3.4 Control Variables
5.4 Overview: Data Analytic Techniques for Analysis of the Structural
Model
5.5 Summary

97
97
99
100
103
105
107
107
115

118
121
123

124
Chapter 6: Analysis Results
6.1 Construct Validity, Reliability, and Preliminary Data Analysis
6.1.1 Inter-Construct Correlations
6.1.2 Analysis of Construct Correlation, Convergent Validity and
Reliability
6.1.3 Discriminant Validity
6.1.4 Appropriateness of Individual Level Analysis of Participant Data
Nested in Groups
6.1.5 Assessment of Common Method Variance
6.1.6 Questions of Non-Response Bias in the Survey Responses
6.2 Analysis of the Structural Model
6.3 Summary of Analysis

125
125
126
128

130
133

137
144
146
152

vi
Chapter 7: Discussion and Conclusions
7.1. Discussion of Results
7.1.1 Impact of Sharing and Reflection on Learning and Innovation
7.1.2 Effect of CT Support on Sharing and Reflection
7.2 Limitations
7.2.1 Limitations Concerning Validity
7.2.2 Limitations Concerning Generalizability
7.3 Research Implications
7.3.1 Implications for Theory Development on Innovation
7.3.2 Implications for Theory Development on Virtual Collaboration
7.3.3 Implications for Theory Development for CT Design
7.4 Practical Implications for Managers
7.5 Future Research
7.5.1 Future Research in the Area of Innovation
7.5.2 Future Research in the Area of Virtual Collaboration
7.5.3 Future Research in the Area of CT Design
7.6 Conclusion

155
158
158
164
169
170
172
173
174
177
180
181
183
184
185
185
186

References 188

vii
List of Tables

Table 1-1: Summary of CT Support Effects

Table 2-1: Summary of Interaction Mode Effects on Sharing and Reflection

Table 3-1: Summary of Hypothesized Impacts of Sharing and Reflection on
Learning and Innovation

Table 3-2: Definitions and CT Support for Te’eni Communication Strategies

Table 3-3: Summary of CT Support Utilized Based on Need for Sharing or
Reflection, and Interaction Mode

Table 3-4: CT Support Hypotheses and Rationale for Situation Where CT
Support Effect Not Anticipated to Occur

Table 5-1: Respondent Demographic Statistics

Table 5-2: CT Support for a Contextualization Strategy

Table 5-3: CT Support Identified in Prior Research

Table 5-4a: CT Support Items, Synchronous Interaction Mode

Table 5-4b: CT Support Items, Asynchronous Interaction Mode

Table 5-5a: CT Support Item Loadings, Synchronous Interaction Mode

Table 5-5b: CT Support Item Loadings, Asynchronous Interaction Mode

Table 5-6: CT Support Constructs

Table 5-7: Sharing Diverse Knowledge

Table 5-8: Reflection on Others’ Shared Knowledge

Table 5-9: Innovation

Table 5-10: Learning

Table 5-11: Skewedness – Combined Learning Items
7

30

53

55

62

70

107

109

110

111

112

112

113

114

116

117

118

119

121

viii
Table 5-12: Number of Participants

Table 5-13: Number of Participant Locations

Table 6-1: Correlation of Latent Constructs, Synchronous Interaction Mode

Table 6-2: Correlation of Latent Constructs, Asynchronous Interaction Mode

Table 6-3: Item Loadings, AVE, and Composite Reliability, Synchronous
Interaction Mode

Table 6-4: Item Loadings, AVE, and Composite Reliability, Asynchronous
Interaction Mode

Table 6-5: Cross Loadings, Synchronous Interaction Mode

Table 6-6: Cross Loadings, Asynchronous Interaction Mode

Table 6-7: Individual Responses Per Project

Table 6-8: Inter-Group Agreement, Synchronous Interaction Mode

Table 6-9: Inter-Group Agreement, Asynchronous Interaction Mode

Table 6-10: Inter-Construct Correlations, Synchronous Interaction Mode

Table 6-11: Dis-Attenuated Partial Correlations, Synchronous Interaction
Mode

Table 6-12: T-Statistics, Inter-Construct Correlations, Synchronous Interaction
Mode

Table 6-13: T-Statistics, Dis-Attenuated Partial Correlations, Synchronous
Interaction Mode

Table 6-14: Inter-Construct Correlations, Asynchronous Interaction Mode

Table 6-15: Dis-Attenuated Partial Correlations, Asynchronous Interaction
Mode

Table 6-16: T-Statistics, Inter-Construct Correlations, Asynchronous
Interaction Mode
122

122

127

128

130

130

132

133

134

137

137

141

141

142

142

143

143

143

ix
Table 6-17: T-Statistics, Dis-Attenuated Partial Correlations, Asynchronous
Interaction Mode

Table 6-18: Comparison of Innovation for Respondents vs. Non-respondents

Table 6-19: Comparison of Respondents vs. Non-respondents for Second
Round Survey

Table 6-20: Model Checking – Investigation of Alternative Paths for
Significance

Table 7-1: Communication Strategies for Reducing Cognitive Complexity
From Te’eni (2001)
144

145

146

150

165

x
List of Figures
Figure 1-1: Summary of Findings: Impact of Sharing and Reflection on
Learning and Innovation

Figure 1-2: CT Support Effects on Individual’s Cognitive Needs for Sharing
and Reflection

Figure 1-3: Model of CT Support for Virtual Collaboration for Innovation
Across Synchronous and Asynchronous Interaction Modes

Figure 3-1: Diagram of PM/PT Process

Figure 3-2: Hypothesized Effects of Sharing/Reflection on Learning and
Innovation in Synchronous and Asynchronous Interaction Modes

Figure 3-3: Summary of Hypothesized CT Support Effects in Synchronous and
Asynchronous Interaction Modes

Figure 3-4: Combined Research Model – Virtual Collaboration for Innovation
Across a Cycle of a Synchronous and Asynchronous Interaction Mode

Figure 4-1: Nominal Extent of Sharing and Reflection in Virtual Collaboration
for Innovation with Regular Face to Face Interaction Compared to
Completely Virtual

Figure 5-1: Probability Plot for Item 1 of Learning Scale

Figure 5-2: Probability Plot for Transformed Item 1 of Learning Scale

Figure 6-1: Results of the Structural Model

Figure 6-2: Breakout of Results for Synchronous Interaction Mode Only

Figure 6-3: Breakout of Results for Asynchronous Interaction Mode Only

Figure 6-4: Final Model, CT Support for PM/PT in Virtual Collaboration for
Innovation in the Synchronous and Asynchronous Interaction Mode

Figure 7-1: General PM/PT Process

6

8

9

44

52

71

74

81

120

121

148

149

149

154

159

xi
Figure 7-2: Findings – Effects of Sharing/Reflection on Learning and
Innovation

Figure 7-3: CT Support Effects on Individual’s Cognitive Needs for Sharing
and Reflection

Figure 7-4: Final Model: Virtual Collaboration for Innovation in Synchronous
and Asynchronous Interaction Modes

Figure 7-5: Nominal Extent of Sharing and Reflection in Virtual Collaboration
for Innovation with Regular Face to Face Interaction Compared to
Completely Virtual

163

167

169

182

xii
Abstract
This dissertation presents a new theory of collaborative technology (CT)
support for virtual collaboration for innovation, that explains the differential effects
CT support has on learning and innovation in synchronous vs. asynchronous
interaction modes. Virtual collaboration for innovation is defined as the interaction
of individuals from distinct knowledge domains, separated by time and space, and in
which they share and combine their knowledge to develop and implement creative
ideas. Synchronous and asynchronous are the two virtual interaction modes in which
individuals collaborate, and are differentiated by the immediate or delayed ability to
share knowledge and give and receive feedback. Each interaction mode, I argue, has
different characteristics which increase the cognitive complexity associated with
sharing diverse knowledge (sharing) and reflecting on others’ shared knowledge
(reflection), and in each mode individuals have different opportunities for realizing
learning and innovation outcomes from sharing and reflection.
Building on the theory of perspective making and perspective taking (Boland
& Tenkasi 1995), I developed a new theory based on these interaction mode
differences which explains how collaboration across a cycle of both interaction
modes leads to learning and innovation and which accounts for the tension between
sharing diverse knowledge and attaining common ground needed to understand
others’ diverse knowledge; it explains the multi-dimensional role of CT support for
enabling four strategies from Te’eni (2001) for reducing cognitive complexity in

xiii
virtual collaboration for innovation, each of which is useful in different interaction
modes to facilitate the different cognitions of sharing and reflection ; it clarifies
existing paradoxes regarding the role of CT support in virtual work, such as whether
or not contributions should be anonymous or attributable; and it justifies the value of
CT support for creating successful innovation in organizational contexts.

1
CHAPTER 1

EXECUTIVE SUMMARY

This dissertation presents a new theory of collaborative technology (CT)
support for virtual collaboration for innovation, that explains the differential effects
CT support has on learning and innovation in synchronous vs. asynchronous
interaction modes. Virtual collaboration for innovation is defined as the interaction
of individuals from distinct knowledge domains, separated by time and space, and in
which they share and combine their knowledge to develop and implement creative
ideas. Synchronous and asynchronous are the two virtual interaction modes in which
individuals collaborate, and are defined by the ability to share knowledge and give
and receive feedback in real time. Learning is defined as an individual building a
new mental model based on her understanding of others’ knowledge, and innovation
is defined as an individual building innovative work products by creatively
combining her own knowledge with others’ knowledge.
Finding that existing theories had not taken interaction mode differences into
account, I developed a new theory based on these differences that I tested through a
multi-method analysis including a four-month case study of 15 individuals and a
two-round survey of 106 and 67 individuals respectively. The new theory explains
how collaboration across a cycle of both interaction modes leads to learning and
innovation; it explains the role of CT support in virtual collaboration for innovation;
it clarifies existing paradoxes regarding virtual work; and it justifies the value of CT

2
support for creating successful innovation in organizational contexts. In the
following paragraphs I briefly explain the rationale for developing this new theory; I
summarize the important findings; and I briefly describe the implications of this
research for theory and practice.

1.1 Rationale: Understanding Individual’s Cognitions in Synchronous and
Asynchronous Interaction Modes
I focused on the cognitive processes in virtual collaboration for innovation
rather than the motivational processes, because past literature has suggested that an
individual’s ability to successfully learn and innovate during virtual collaboration for
innovation with others depends on two cognitions performed simultaneously:
reflecting on others’ shared knowledge (hereafter called “reflection”) and sharing
one’s own diverse knowledge (hereafter called “sharing”). In addition, I focused on
the differences between synchronous vs. asynchronous interaction modes because,
when collaborating virtually, each interaction mode presents different cognitive
demands to individuals and affords them different opportunities for sharing and
reflection. Further, sharing and reflection are in tension since sharing introduces
new, diverse knowledge while reflection underscores attempts to develop common
ground; thus simultaneously attempting both adds greater complexity. Unless these
cognitive demands are resolved, however, individuals may have difficulty attempting
sharing and reflection, and subsequently learning and innovation may not occur.

3
Past theories may not have accounted for the additional cognitive demand
from coordination inefficiencies in synchronous and asynchronous interaction
modes that effects individual’s cognitions, because they only examined virtual
collaboration for innovation in one interaction mode, synchronous (e.g.
teleconference) or asynchronous (e.g. email exchange, use of virtual workspace).
Additionally, in past research synchrony has often been confounded with co-
presence (i.e. face to face is synchronous, virtual is asynchronous) which makes it
difficult to understand differences between synchronous and asynchronous virtual
interaction modes. As a result we do not understand how differences in the
interaction mode affect the opportunity for sharing and reflection, and how that
subsequently impacts learning and innovation. In the synchronous interaction mode,
dialogue is fast-paced and continuous and there are greater cues available from
verbal expression; however, knowledge is processed serially and cognitive resources
are involved in following synchronous dialogue, thus it is harder to obtain and give
meaningful feedback. In the asynchronous interaction mode, by contrast, individuals
have the time and cognitive resources to obtain and give meaningful feedback but
they experience greater cognitive complexity making their knowledge accessible
(and likewise having others’ knowledge accessible) and obtaining timely feedback,
since there is no ongoing synchronous dialogue.
Thus synchronous and asynchronous interaction modes provide different
opportunities for sharing and reflection and impose certain cognitive demands which

4
make sharing and reflection harder to accomplish. Theories of CT support have not
investigated the different ways CT support might help reduce cognitive demands in
synchronous and asynchronous interaction modes, although evidence suggests that
individuals relying on CT can successfully accomplish these cognitions. Thus, I
conclude that new theory is needed which explains the effects of CT support in
synchronous and asynchronous interaction modes, and how that support facilitates
the sharing and reflection needed for learning and innovation.

1.2 Summary of the Research Questions
Given the different interaction capabilities afforded by asynchronous vs.
synchronous interaction modes when collaborating virtually, I asked whether these
interaction modes affected the two cognitions of sharing and reflection, and the role
of CT support. In particular, I asked the following two research questions: 1) Does
an individual’s sharing and reflection effect learning and innovation differently
depending on whether the individual is interacting synchronously vs.
asynchronously? And 2) Are there different kinds of CT support that facilitate an
individual’s sharing and reflection differently when the individual is interacting with
others synchronously vs. asynchronously?

5
1.3 Summary of Findings
In addressing the first research question, I found that learning and innovation
were dependent on the need for attending to self and/or others’ knowledge coupled
with the opportunity to do so in each interaction mode. My theory predicted that
learning is enabled by sharing in both modes, and also by reflection in the
asynchronous interaction mode (but not the synchronous), because individuals need
to explain their own knowledge as well as understand others’ knowledge; however,
in the synchronous interaction mode they have less opportunity for reflection.
Unexpectedly, I found that the effect of sharing in the asynchronous mode was
completely mediated through reflection, suggesting the theory of a direct effect was
not entirely accurate. By contrast, innovation is the result of reflection during both
interaction modes, but not directly from sharing in either interaction mode, because
individuals must focus on others’ knowledge whenever it is shared rather than their
own knowledge. Thus when innovating, sharing supports reflection, since
understanding one’s own knowledge better helps with understanding others. Figure
1 shows a summary of these findings.

6

Figure 1-1: Summary of Findings: Impact of Sharing and Reflection on Learning and
Innovation

In addressing the second research question, I found that CT support affects
both sharing and reflection in different ways depending on the different cognitive
demands of the interaction modes. I found differential value of four different types
of CT support, based on a framework of strategies to overcome cognitive complexity
from Te’eni (2001): CT support for testing and adjusting, attention-focusing,
contextualization, and perspective-taking. Table 1-1 summarizes these strategies and
gives examples of how CT supports each strategy.

7

Table 1-1: Summary of CT Support Effects
Communication
Strategy
Definition How CT supports communication strategy
Testing and
Adjusting
Spontaneously changing
what knowledge is
shared based on
feedback from self and
others
Displaying and editing shared knowledge in real-
time in a visible, tractable format, such as through
document sharing features or whiteboards; instant
feedback available and recorded
Attention
Focusing
Directing knowledge
specifically to others to
initiate dialogue and
obtain feedback
Text-based communication such as email, instant
messaging, discussion forums that permit two-way
asynchronous dialogue
Contextual-
ization
Provision of explicit
context in a message
helps individual
understand why others
made a contribution
Contextualization mechanisms such as displaying
names/profiles of collaborators, attributing verbal
and written contributions, file and document
sharing tools that permit back and forth
comparison of different versions
Perspective
Taking
Cognitively focused on
receiver’s view because
individual can access
their knowledge in their
own words
Access to stored knowledge in repositories or
virtual workspaces when needed for reflection or
use

To facilitate sharing, individuals need to quickly understand others’ feedback
in the synchronous mode and they need a means for obtaining feedback in the
asynchronous mode – which CT support for testing and adjusting and attention-
focusing, respectively, provides. For reflection, individuals need contextual
information in both modes and access to others’ knowledge in the asynchronous
mode which CT support for contextualization and perspective-taking provides. Thus
different features of CT support reduce cognitive demands in synchronous vs.
asynchronous modes and facilitate sharing and reflection, which ultimately leads to
learning and innovation. Figure 1-2 depicts the effects of CT support on sharing and
reflection.

8

Figure 1-2: CT Support Effects on Individual’s Cognitive Needs for Sharing and Reflection

Figure 1-3 shows the final model of CT support for virtual collaboration for
innovation that incorporates a complementary cycle of both synchronous and
asynchronous interaction modes.

9

Figure 1-3: Model of CT Support for Virtual Collaboration for Innovation Across Synchronous
and Asynchronous Interaction Modes

1.4 Summary of Implications of the New Theory of Virtual Collaboration for
Innovation
My research has a number of implications for theories of innovation, virtual
collaboration and CT design. I show that inconsistent findings from past research
may be attributed to the failure to account for interaction mode and CT support
differences; but taking these differences into account helps to resolve these
inconsistencies. For example, theory suggests that sharing and reflection occur
simultaneously in dialogue and that both are needed for learning and reflection, yet
past research has found that sharing occurs only synchronously and reflection only
asynchronously. I resolve this inconsistency by showing that sharing and reflection

10
occur both synchronously and asynchronously, but sharing in both modes has an
indirect effect on innovation; whereas synchronous sharing directly impacts learning
but asynchronous sharing indirectly impacts learning, mediated through
asynchronous reflection which directly impacts learning. This also helps resolve the
tension between sharing and reflection regarding whether common ground (or
mutually held knowledge) positively or negatively impacts sharing and reflection.
Cramton (2001) and others argue that greater common ground with others helps
individuals share and reflect across boundaries, but Carlile (2004) argues that too
much common ground constrains innovation since individuals reuse existing
common knowledge. I show that individuals develop common ground through
reflection, but at the same time through sharing they interject the diverse knowledge
needed for innovation into dialogue. Third, I resolve the question of whether
contributors in virtual collaboration for innovation should be anonymous to reduce
social evaluation pressures, by showing that awareness of others and identity are
needed to enable context and feedback needed for sharing and reflection. Finally for
theories of CT design, I show that CT should be designed as a bundle of features
since each feature is used in concert with others not independently, but that a
rationale is needed for why each feature is useful in each interaction mode, thus
requiring a multi-dimensional understanding of CT support.
There are also a number of practical implications from this research. Most
importantly, the research shows how individuals in virtual collaboration for

11
innovation rely on CT support to maintain more consistent sharing and reflection
over time, as opposed to during face to face interaction where individuals share and
reflect greatly while meeting but very infrequently when not face to face. The
findings also show managers how to assess the value of CT support in their
organization by pointing to gains in innovation and individual learning.
In conclusion, the development of a single theory that explains how
individuals learn and innovate in virtual collaboration for innovation provides a
robust framework for understanding collaborative success and the role of CT support
in realizing that success.

12
CHAPTER 2
VIRTUAL COLLABORATION FOR INNOVATION
Competitive advantage in knowledge-based organizations is dependent on
innovation (Carlile 2004; Cummings 2004; Grant 1996). Innovation is defined as
“the successful implementation of creative ideas within an organization” (Amabile
1988, p.126) and refers to how ideas are developed and put to use. When innovating
with others, individuals typically must synthesize knowledge from various domains
and create new ideas through dialogue with others (Alavi and Leidner 2001; Boland
and Tenkasi 1995; Carlile and Rebentisch 2003; Leonard and Swap 1999; Malhotra
et al. 2001), since no one individual has all of the knowledge and expertise needed to
solve innovative tasks (Bruner 1990; Stasser and Titus 2003). Each individual has
what is called diverse knowledge, that is unique (i.e. not known by others pre-
discussion), and domain-specific (knowledge of a particular area based on the
individuals’ experience rather than generalized knowledge) (Boland and Tenkasi
1995). Sharing diverse knowledge with others helps individuals make implicit
knowledge explicit and easier to apply toward successful accomplishment of
innovative tasks (Griffith et al. 2003; Leonard and Sensiper 1998; Nonaka 1994). By
sharing diverse knowledge with others who have a greater range of perspectives (i.e.
individual’s approaches to solving innovative tasks, how they view the importance
and utility of what they know, and how they believe their diverse knowledge ought
to be used), individuals think about and discuss ideas of which they would have

13
taken for granted or been otherwise unaware (Dougherty 1992; Boland and Tenkasi
1995). For example, Boland and Tenkasi (1995) cite an example from Bradshaw
(1992) of how the Wright brothers innovated a working aircraft where others failed
by discussing metaphors of kite flying. Incorporating new perspectives helped them
realize success where others who focused on the purely mechanical problems failed.
Increasingly, individuals are working together in ‘virtual’ collaborations
(Desanctis and Monge 1999; Gibson and Gibbs 2006; Griffith et al. 2003). In virtual
collaboration, individuals who are separated by time and space exchange and
combine their knowledge in order to accomplish innovative tasks, as face to face
collaboration is impractical or impossible. They are able to engage with a more
varied and numerous set of knowledge experts situated in their own environments or
knowledge domains, with access to localized resources not typical with face to face
meetings (Griffith et al. 2003). When these collaborations are ‘purely’ virtual (thus
no face to face interaction), all interaction is completely mediated by collaborative
technology (CT) (Cohen and Gibson 2003; Griffith et al. 2003). How individuals
adapt various features of CT to help meet their collaborative needs is referred to as
CT support. For example, Majchrzak et al. (2005b) found that individuals rely on
CT support for contextualization to achieve collaborative know-how, or knowledge
of how to communicate one’s ideas or coordinate actions with others.
Virtual collaboration for innovation is therefore defined as the interaction of
individuals from distinct knowledge domains in which they share and combine their

14
knowledge to develop and implement creative ideas; and in which individuals are
separated by time and space and all of their interaction is mediated by CT support.
Because individuals rely on CT support in virtual collaboration for innovation, with
different means to share knowledge and feedback than those individuals in face to
face collaboration (Clark and Brennan 1991; Cramton 2001), theories are needed
which explain how and why CT support leads to successful collaboration outcomes
(Griffith et al. 2003; Jarvenpaa and Leidner 1999).
The remainder of this chapter will review existing literature on virtual
collaboration for innovation, arguing that while research has explained many of the
difficulties of virtual collaboration for innovation it has not explained how CT
support helps individuals overcome those difficulties to successfully realize
collaboration outcomes. The review will show there are two important collaboration
outcomes, learning and innovation, yet individuals experience difficulties realizing
these outcomes in virtual collaboration for innovation. The literature suggests
learning and innovation may depend on individual’s cognitions of sharing diverse
knowledge with others (Stasser and Titus 2003) and reflection on others shared
knowledge through dialogue (Carlile 2004; Boland and Tenkasi 1995), but it is not
clear precisely how these cognitions lead to learning and innovation. In addition, the
literature suggests that virtual collaboration for innovation occurs over a cycle of
synchronous and asynchronous interaction modes (Hollingshead and McGrath 1995;
Montoya-Weiss et al. 2001), which have different characteristics that have not

15
previously been taken into account. I conclude that theories which do not take these
factors into account cannot explain how CT support helps individuals to learn and
innovate successfully; and a theory is needed which explains learning and innovation
in both synchronous and asynchronous interaction modes.

2.1 Collaboration Outcomes of Virtual Collaboration for Innovation
Innovation literature has focused on two individual-level collaboration
outcomes, innovation and learning, but separately using different theories under the
assumption that they reflect different aspects of knowledge use, such as application
and integration (Alavi and Leidner 2001; Dillenbourg 1999). In this section I review
theories that explain each outcome from virtual collaboration for innovation
separately, then theory which suggests both are independent but related outcomes
that result from the same social-cognitive processes and as such a theory is needed
which investigates both learning and innovation as collaboration outcomes.

2.1.1 Innovation Outcomes in Virtual Collaboration for Innovation
The word innovation is used to describe how individuals understand how to
combine others shared knowledge with their own to build innovative work products
that are needed to accomplish innovative tasks (Amabile et al. 1996; Barki and
Pinsonneault 2001; Burke and Chidambaram 1999) In their review of innovation in
organizations, Brown and Eisenhardt (1995) state that successful communication and

16
knowledge sharing between individuals in project teams, especially across
knowledge domains or boundaries, is critical for organizational innovation. Thus
with the majority of research on organizational innovation has focused on group
level or higher processes, numerous theories of innovation also agree that knowledge
must be applied ‘creatively’ and ‘productively’ by individuals to successfully
innovate (Amabile 1988), and that the knowledge individuals need to build these
creative and productive work products comes from collaborating with others (Carlile
2002; Dougherty 1992; Leonard and Swap 1999; Nonaka and Takeuchi 1995;
Suedfeld et al. 1992).
In virtual collaboration for innovation, theories have stressed that innovation
refers to the quality of ideas (Barki and Pinsonneault 2001) rather than as idea
quantity, recognizing that technology must help individuals combine their
knowledge and not just share it (Alavi and Leidner 2001) as was the focus of earlier
GSS research. For example, Barki and Pinsonneault (2001) note that fewer
contributions that are ‘novel,’ ‘effective,’ and ‘creative’ are more valuable than a
greater amount of ideas that are less novel effective or creative. Similarly, Burke
and Chidambaram (1999) assess innovation in terms of ‘creative,’ ‘realistic,’ and
‘comprehensive’ outcomes, reinforcing that contributions must not only be novel and
creative, they must also be practical and useful. This suggests that when individuals
in virtual collaboration for innovation combine their knowledge in creative yet useful
ways, they will be able to innovate as a result.

17
2.1.2 Learning Outcomes in Virtual Collaboration for Innovation
Innovation theories have also suggested that learning is an important outcome
from innovation, from the organizational to the individual level (Edmondson 2002),
since it reflects the extent individuals have been able to acquire and integrate of new
knowledge (Cross and Sproull 2004; Gray and Meister 2004; Majchrzak and Beath
2006).
Individuals strive to learn from others in virtual collaboration for innovation
(Alavi and Leidner 2001; Cramton 2002; Cross and Sproull 2004; Dillenbourg 1999;
Gray and Meister 2004; Majchrzak and Beath 2006; Robey et al. 2000) by
integrating diverse knowledge shared by others with their own. When individuals
are not only exposed to diverse knowledge from others but have the ability and
opportunity to integrate that knowledge, they are more likely to develop new
perspectives and consider a broader range of knowledge when formulating new ideas
(Boland et al. 1994; Dillenbourg 1999).
Norman (1982) describes this process of developing new perspectives as a
type of cognitive learning known as ‘structuring’ new schemata, where schemata are
mental structures that organize knowledge and experience. Vandenbosch and
Higgins (1996) refer to this as building new mental models since overcoming the
limits of existing mental models is what drives individuals to engage in collaborative
work (Bruner 1990; Gray and Meister 2004; McGrath 1991). Mental models are
cognitive representations of one’s world that are used to organize knowledge in

18
simple yet robust ways. When individuals need to think or act they reference their
mental models to decide how to proceed (Leahey and Harris 1989). Vandenbosch
and Higgins (1996) distinguish building new mental models from maintaining
existing mental models. When individuals maintain their existing mental models in
the face of new knowledge it reinforces their tendency to think and act in the same
way, thus they have difficulty valuing others knowledge and, more importantly, they
have difficulty applying their own knowledge in creative ways. When new
knowledge is experienced, however, and individuals can internalize and use it to
build new mental models, it helps them think and act in new ways. In virtual
collaboration for innovation individuals experience diverse knowledge shared by
others and thus have the opportunity to build new mental models.

2.1.3 Both Learning and Innovation Outcomes Realized in Virtual Collaboration
for Innovation
In virtual collaboration for innovation the technologies and interactive
processes which influence individual cognitions may be the same for both learning
and innovation (Dillenbourg 1999; Majchrzak and Beath 2006), since individuals
develop insight as well as deliverables when they synthesize knowledge shared
across different domains. Organizations rely on collaboration to foster both learning
and innovation at individual and organizational levels to develop flexible,
sustainable, and inimitable knowledge resources (Goodman and Darr 1998; Levitt

19
and March 1988). Yet sometimes it is difficult to isolate individual success within
the group context of virtual collaboration for innovation. Often these outcomes are
assessed at the shared or group level (c.f. Benbunan-Fich et al. 2002; Edmondson
2002). However, McGrath (1991) and Boland et al. (1994) argue that even in group
contexts, individual contributions are what make and define success. Goodman and
Darr (1998) similarly argue that individual level learning is a necessary antecedent
for organizational or group level learning.
Dillenbourg (1999) suggests that these two individual level collaboration
outcomes must be measured distinctly from group level outcomes, or otherwise,
group characteristics can overshadow individual level influences. He also argues
that innovation and learning must be measured as separate and unique outcomes
rather than one as a consequence or byproduct of the other, noting that learning does
not begat innovation or vice versa. Instead, learning and innovation should be
examined as related outcomes that can reinforce each other in cycles over time
(Leonard and Sensiper 1998), but also independent outcomes since it is not always
clear at any particular stage how individuals are learning vs. innovating. In virtual
collaboration for innovation, therefore it is important to understand how CT support
will help individuals learn and innovate, recognizing that while the same processes
lead to both outcomes, there are differences at any particular time in how learning
and innovation are realized.

20
2.2 Factors Which Positively Influence Learning and Innovation in Virtual
Collaboration for Innovation
Greater breadth of diverse knowledge sharing can positively impact learning
and innovation if individuals are able to integrate and apply that diverse knowledge
(Brown and Eisenhardt 1995; Cummings 2004; Gibson and Gibbs 2006; Leonard
and Swap 1999), since by definition individuals in virtual collaboration for
innovation do not already possess all the knowledge needed to solve innovative tasks
(Bruner 1990). In virtual collaboration for innovation the potential access to diverse
sources of knowledge is greater than in co-located settings, since individuals are
located in different spaces and have access to local resources (Griffith et al. 2003).
But innovation theorists have also recognized that the mere transfer of knowledge
between individuals, without the meaning and feedback individuals need to interpret
that knowledge because of their unique backgrounds and expertise, is not likely to
help individuals achieve successful collaboration outcomes (Carlile 2004; Dougherty
1992; Leonard and Swap 1999). Thus collaboration involves discussing diverse
knowledge with others to help individuals understand both their own knowledge
better, and understand others knowledge and perspectives better, which in virtual
collaboration for innovation may mean understanding others’ knowledge and
perspectives with little information about others’ situation and/or expertise (Boland
et al. 1994; Boland and Tenkasi 1995; Carlile 2004; Cramton 2001; Griffith et al.
2003).

21
Learning and innovation from virtual collaboration for innovation therefore
depend on a dialogic process in which individuals share diverse knowledge with
others, and simultaneously attempt to understand, and make sense of others’ shared
knowledge (Boland and Tenkasi 1995; Carlile 2004). These two activities are often
framed as cognitive decisions, since knowledge is tacit and situated in practice
(Nonaka 1994; Wenger 1998) and each individual must consciously decide what of
their knowledge and experience to share with others - recognizing that their diverse
knowledge adds value to the extent that it is unique yet the more unique the greater
the difficulties others may have understanding (Stasser and Titus 2003; Wittenbaum
et al. 2004). Therefore the cognition of sharing diverse knowledge (hereafter
referred to as ‘sharing’) can be defined as the decision of how to explain one’s own
knowledge in a way that others can understand it. Simultaneously individuals try
and make sense of and interpret others knowledge, through a cognitive process of
reflecting on others’ shared knowledge (hereafter referred to as ‘reflection’) (Boland
and Tenkasi 1995; Carlile 2004; Weick 1995). The cognition associated with
reflection is first understanding what others are sharing, and then comparing it to
what the recipient already knows, so that the individual can offer feedback that the
knowledge was received and understood (Boland et al. 1994; Clark and Brennan
1991).
Thus learning and innovation appear to depend on two cognitions, sharing
and reflection, since creative combination of diverse knowledge is needed to resolve

22
innovative tasks, yet it is not enough for diverse knowledge to be present - it must be
understood and its value appreciated as well. However, it is not clear exactly how
sharing and reflection lead to learning and innovation, since innovation theories have
focused on the value of diverse knowledge and its combination as inputs to
innovation, but not the process of transforming diverse knowledge into learning and
innovation (Carlile 2004). Without a clearer understanding of the dynamics of
sharing and reflection the literature has fragmented into theories which predict both
sharing and reflection are simultaneously necessary for learning and innovation (e.g.
Boland and Tenkasi 1995; Carlile 2004; Leonard and Swap 1999) and theories which
predict sharing is more optimal at certain times (such as during synchronous
dialogue) and reflection during other times (such as when individuals are not
engaged in synchronous sharing) (Dennis et al. 1997; Kerr and Murthy 2004; Massey
et al. 2002). This dichotomy has been further complicated by disagreement over the
need for common ground or mutual knowledge (Cramton 2001; Krauss and Fussell
1990) which theories suggest are necessary to share knowledge across boundaries,
but which Carlile (2004) and Stasser and Titus (2003) argue causes individuals to
discuss commonly-held knowledge instead of diverse knowledge, impeding
innovation. Jehn et al. (1999) suggest individuals can maintain knowledge diversity
while developing value and social commonalities, which would resolve this
disagreement, but offer no clear strategies for how to share diverse knowledge while
simultaneously understanding others’ knowledge and perspectives.

23
Thus theories are needed in particular for virtual collaboration for innovation
that can more fully explain this process, since the interplay between greater access to
diverse knowledge, in contrast to the potential difficulties sharing and reflection that
knowledge across boundaries, yields unique circumstances that have been difficult to
explain with past innovation theory (Chidambaram and Tung 2005; Malhotra and
Majchrzak 2004). These circumstances unique to virtual collaboration for
innovation that potentially limit collaboration success are reviewed next.

2.3 Difficulties Achieving Sharing and Reflection in Virtual Collaboration for
Innovation
Sharing and reflection are not solely cognitive activities; individuals must be
motivated to share their knowledge (Fulk et al. 1996; Kalman et al. 2002; Szulanski
1996). This is especially the case when sharing diverse knowledge, where they are
concerned about how others will interpret their knowledge and what knowledge
would be valuable to others (Brandon and Hollingshead 2004; Thomas-Hunt et al.
2003; Wasko and Faraj 2005). Ensuring alignment between sharing and individual
and organizational incentives could help to alleviate some of this concern
(Wittenbaum et al. 2004; Yuan et al. 2005). In virtual collaboration for innovation,
for example, if individuals recognize that each others’ knowledge is needed for the
innovation to take place, they are more motivated to decide how to share their

24
diverse knowledge in a form usable to others, and correspondingly are more likely to
reflect on others shared knowledge.
Even assuming that people are motivated to share doesn’t mean that they are
cognitively able to share, however, since characteristics of the virtual environment
can impose cognitive demands that make sharing and reflection less efficient, and
further constrain collaboration by reducing opportunities for sharing and reflection.
Individuals in virtual collaboration for innovation may be less aware of
others’ situation and/or location (Cramton 2001; 2002; Maznevski and Chudoba
2000), which may lead them to misunderstand others knowledge or actions or make
assumptions about others that impact what knowledge they share (Cramton 2001;
Kiesler and Cummings 2002; Walther 1996). These are coordination inefficiencies
(Baron et al. 1992; Steiner 1972) which are types of process loss that explain why
individuals in groups are each less effective than they would be as individuals
working separately, although overall they are better off working together.
Inefficiencies related to situational context are resolved by generating trust or
familiarity or routines for working together, which can be managed with group
norms, leadership, or the chance to develop familiarity through personal contact such
as in face to face meetings (Hinds and Bailey 2003); in fact, some virtual team
research has concluded that regular face to face interaction is required for a virtual
team to develop trust and norms (Kiesler and Cummings 2002; Maznevski and
Chudoba 2000; Olson et al. 2002) suggesting only hybrid teams (not purely virtual)

25
can be successful. However Malhotra et al. (2001) and Walther (1995) have reported
instances of these outcomes in purely virtual circumstances, suggesting that
contextual information is available virtually but without specific strategies for
communicating context it is not easily understood (Te’eni 2001).
Individuals in virtual collaboration for innovation may also be less aware of
others’ cognitive context (Majchrzak et al. 2005b; Te’eni 2001), which can be a
more sustained problem since it is not always apparent even to the individual how
their diverse knowledge should be explained to others (Boland and Tenkasi 1995;
Stasser and Titus 2003). Since virtual collaboration for innovation often involves a
greater range of knowledge areas/domains than are typical in face to face
collaboration (Griffith et al. 2003), and mediated communication offers fewer cues
and less direct feedback than what individuals are used to with face to face
communication (Clark and Brennan 1991; Hinds and Bailey 2003; Kiesler and
Cummings 2002; Maznevski and Chudoba 2000), deciding what to share and how to
reflect on others knowledge can be cognitively demanding. Unless individuals can
resolve this cognitive complexity they are not likely to share diverse knowledge in a
way that others can make sense of and use toward accomplishing successful
collaboration outcomes.
Additional cognitive complexity may not have an equally negative impact on
both sharing and reflection, since different cognitions are used for each, one focused
on what an individual knows and one focused on what others know. For example,

26
information exchange models in GSS research have similarly found that individuals
in virtual collaboration for innovation often realize a ‘productivity paradox’
(Chidambaram and Tung 2005) because a greater amount of knowledge is shared but
individuals have a more difficult time in reflection on that shared knowledge.
Chidambaram and Tung attribute this to legacy GSS design for features that help
individuals overcome sharing difficulties but leave individuals with few options for
reflection. Thus additional cognitive complexity makes sharing and reflection more
difficult in virtual collaboration for innovation, yet resolving this complexity may
require different strategies for sharing vs. reflection that must be accounted for in
theories of virtual collaboration for innovation.

2.4 Virtual Collaboration for Innovation Occurs Over a Series of Interaction
Modes
Even if cognitive complexity is resolved, individuals must have the
opportunity for the cognitions of sharing and reflection to culminate in learning and
innovation outcomes. Individuals in virtual collaboration for innovation interact in a
series of both synchronous and asynchronous interaction modes (Marks et al. 2001),
which are substantively different from face to face interaction (Burke and
Chidambaram 1999; Griffith et al. 2003; Malhotra et al. 2001; Montoya-Weiss et al.
2001) and offer different opportunities for sharing and reflection. Interaction mode
is defined as an instance of virtual collaboration for innovation and are synchronous

27
or asynchronous categorized by the availability of instant or delayed sharing and
feedback (Clark and Brennan 1991; Finholt et al. 1990), rather than being
categorized by the type of CT support used (Zack 1993). Hollingshead and McGrath
(1995) state that virtual collaboration for innovation is a series of complementary
synchronous and asynchronous interaction modes and that virtual collaboration for
innovation occurs during both (Marks et al 2001; Montoya-Weiss et al. 2001), which
implies synergies and tradeoffs between the two modes. In addition, individuals (in
groups) accomplish virtual collaboration for innovation in a succession of
interactions and not in a single point in time (McGrath 1991), implying that
collaboration occurs over a repeated cycle of synchronous and asynchronous
interaction modes. The synchronous and asynchronous interaction modes are
referred to as complementary therefore because individuals regularly expect virtual
collaboration in one mode to follow virtual collaboration for innovation in the other,
which becomes a repeating cycle, and as a result learning and innovation outcomes
depend in part on collaboration from each interaction mode. This cycle is not unlike
that discovered by Maznevski and Chudoba (2000) wherein individuals interacted in
a repeating cycle of face to face interaction and virtual interaction. However the
difference is that as defined in this research, virtual collaboration for innovation does
not involve any face to face interaction.
Malhotra et al. (2001) provides evidence of how a purely virtual team
interacts in both these interaction modes rather than substituting one for the other.

28
Most research however has only looked at one interaction mode or the other. More
often, the two interaction modes have been compared or substituted for the others as
a method of isolating effects useful for building theories on CT support
characteristics (Chidambaram and Tung 2005; Finholt et al. 1990; Hightower and
Sayeed 1996; Kerr and Murthy 2004; Nowak et al. 2005). However doing so makes
it difficult to explore how theories of virtual collaboration for innovation operate
across both interaction modes. Thus theories of virtual collaboration for innovation
are needed which incorporate both interaction modes as a comprehensive set rather
than as isolated interaction modes, but which also recognizes that in each interaction
mode there are differences in how individuals overcome cognitive complexity to
accomplish sharing and reflection, and how they have different opportunities for
sharing and reflection which impacts learning and innovation.

2.4.1 Differences in Interaction Mode Impact Sharing and Reflection
Because knowledge is sent and received in real time in the synchronous
interaction mode, there is rapid multi-cue feedback available, individuals quickly
iterate through constraints/dependencies as knowledge is evaluated, and all
individuals are typically involved in the interaction, research has suggested that the
synchronous interaction mode is better suited for fast and iterative dialogue where
individuals are sharing quickly and frequently and with a number of individuals at
once (Dennis et al. 1997; McGrath 1991). However, knowledge is processed

29
serially in the synchronous interaction mode and cognitive resources are at a
premium, leaving little time for reflection on shared knowledge (Clark and Brennan
1991; Finholt et al. 1990; Montoya-Weiss et al. 2001). Thus individuals may
experience a greater potential for sharing in the synchronous interaction mode, but
they may also experience coordination inefficiencies which can increase cognitive
complexity of trying to obtain meaningful feedback from others and with trying to
understand and compare others knowledge as it is shared in the ongoing dialogue
(Clark and Brennan 1991).
Individuals find that reflection on knowledge that is available for review and
revising at their own pace, such as in asynchronous interaction modes, is easier to
accomplish providing they have the time and cognitive resources available (Boland
and Tenkasi 1995; Clark and Brennan 1991; Nowak et al. 2005). If individuals may
review messages before sharing, after carefully considering what others know and
with whom they are communicating, they may also find sharing to be more
productive in asynchronous interaction modes (Finholt et al. 1990). However,
individuals may find sharing diverse knowledge difficult since feedback is delayed; a
lack of cues make miscommunication more likely; and communication seldom
involves all individuals in virtual collaboration for innovation (more often one to one
transmissions – one to many is also possible but sharing diverse knowledge can be
difficult with one to many communication since others’ may respond with feedback
at several different times) (Nowak et al. 2005). Table 2-1 summarizes the

30
differences found in prior research on how interaction mode characteristics impact
sharing and reflection.
Table 2-1: Summary of Interaction Mode Effects on Sharing and Reflection
Interaction Mode Sharing Reflection
Synchronous Potential for sharing
knowledge quickly with
others, rapid feedback
available, individuals may
quickly iterate through
constraints/dependencies as
knowledge is evaluated,
and all individuals are
typically involved in the
interaction; However,
difficult to get meaningful
feedback from others
because they are focused
on sharing their own
diverse knowledge
Knowledge is processed
serially and cognitive
resources are at a
premium, leaving little
time for reflection on
shared knowledge, thus
attempting reflection
during synchronous
dialogue can be
cognitively complex
Asynchronous Sharing diverse knowledge
difficult since feedback is
delayed; a lack of cues
make miscommunication
more likely; and
communication seldom
involves all individuals in
virtual collaboration for
innovation making
deciding what to share
more difficult
Others knowledge may be
reviewed at the
individual’s own pace,
provided it is available;
without synchronous
dialogue however cues
needed to help interpret
shared knowledge may be
missing, increasing the
cognitive complexity of
reflection

Therefore in each interaction mode, there are coordination inefficiencies
which increase cognitive complexity associated with sharing and reflection (Boland
and Tenkasi 1995; Finholt et al 1990), which negatively impact learning and
innovation since both cognitions are essential for virtual collaboration for innovation

31
(Carlile 2004). Collaboration across both interaction modes might help individuals
overcome the cognitive complexity from these inefficiencies (Montoya-Weiss et al.
2001), yet theory is needed to explain how the differences in interaction mode
impact sharing and reflection when individuals interact across both modes.

2.4.2 Prior Theories of Virtual Collaboration for Innovation Have Not
Incorporated Interaction Mode Differences
A limitation with previous work that might otherwise explain some of these
differences is the problem of confounding of synchrony with co-presence, which has
made it difficult for theories of virtual collaboration for innovation to show clearly
what leads to successful collaboration outcomes (Barki and Pinsonneault 2001;
Dennis 1996; Dennis et al. 1997; Kerr and Murthy 2004). Only recently have
researchers addressed this issue and made the distinction clear in the literature
regarding synchronous and asynchronous interaction modes vs. face to face and
mediated communication (Griffith et al. 2003; Nowak et al. 2005). Nowak et al.
report that this confound has caused confusion as to whether cue rich/cue lean media
are more/less effective or not and if synchronous/asynchronous media are more
effective or not. More importantly, since individuals adapt available CT support to
their needs (Postmes et al. 2000; Walther 1996) this problem has made it difficult to
isolate how the CT support has had an impact in synchronous and asynchronous
interaction modes. In addition, task complexity often varies across studies; and

32
individuals are typically evaluated on a single task without reoccurrence or any time
to reflect on shared knowledge (Hollingshead and McGrath 1995); these factors
impact the extent to which sharing and reflection are necessary for learning and
innovation (in fact, in these situations, more routine outcomes such as knowledge
transfer or access may be expected), thus they also complicate the difficulty of
isolating impact of synchrony vs. co-presence.
Another difficulty understanding interaction mode differences stems from
comparisons with face to face collaboration that highlights negative aspects of
mediation in general such as misunderstanding and conflict (Cramton 2001; Hinds
and Bailey 2003). It is evident that in virtual collaboration for innovation there are
additional process losses that individuals in face to face collaboration avoid;
however, the recommended strategy calls for compensating for these inefficiencies
with face to face interaction as a remedy, which is not always possible or desirable
(e.g. Malhotra et al. 2001).
Thus theory has not incorporated the differential impacts of collaborating in
these interaction modes, particularly when both interaction modes are examined
together as a complementary set of virtual interactions rather than as substitutes
(Marks et al. 2001; Montoya-Weiss et al. 2001). A theory is needed which explains
how different characteristics of the interaction mode impact individual’s
collaborative needs in virtual collaboration for innovation.

33
2.5 Multi-Dimensional Role of CT Support for Virtual Collaboration for
Innovation
Because all interaction in virtual collaboration for innovation is mediated by
CT support, theories are needed which address how CT support facilitates sharing
and reflection in both synchronous and asynchronous interaction modes. Two major
issues seem to have prevented the development of a coherent theory of CT support in
the past – first, inconsistent results regarding whether or not CT support is a positive
or negative factor in supporting sharing and reflection; and second, a tendency to
treat CT support as a black box or as a single factor that impacts collaboration rather
than a multi-dimensional bundle of significant features. Both issues are reviewed in
the following sections.

2.5.1 Inconsistent Reports of the Utility of CT Support
Research has reported both positive and negative impacts of CT support.
Hollingshead and McGrath (1995) explain that CT support, while it imposes some
barriers to communication, it helps overcome others by altering the “informational,
temporal, and interactional” conditions in which individuals interact. Barriers to
communication from CT support use include difficulty receiving feedback and
interpreting knowledge (Kirkman et al. 2004) and difficulty experimenting with new
knowledge, impeding innovation (Gibson and Gibbs 2006). Theories of conflict
resolution and information processing have cast doubt on whether or not CT support

34
can help individuals overcome cognitive complexity, instead noting that CT support
tends to increase cognitive complexity (Gibson and Gibbs 2006; Hinds and Bailey
2003) by adding to process conflict, distracting individuals attention from dialogue,
reducing individual’s control over their communication, and reducing availability of
informal feedback.
Others have found that CT support can in some circumstances enhance
collaboration. Walther’s (1996) hyper-personalization model, based on social
information processing theory, shows that individuals in mediated contexts – such as
virtual collaboration for innovation – can achieve outcomes such as relationship
formation that are as good as or better than in face to face settings. Walther argues
that individuals may seek out additional information about others through low-cue
media, despite prior theories of CMC design which predicted otherwise (Ramirez et
al. 2002). While Walther and colleagues have focused on relational outcomes rather
than cognitive or innovative ones, their work demonstrates that individuals are able
to overcome coordination inefficiencies thought to be inherent in virtual
collaboration for innovation, when relying on CT support. In addition, CT support
has also been found to improve collaboration know-how (Majchrzak et al. 2005b) by
enhancing contextualization, improve the acquisition of social information about
others (Ramirez et al. 2002), provide access to external information (Sole and
Emdondson 2002) to jumpstart creativity, and support iterative, interactive
collaboration (Malhotra and Majchrzak 2004).

35
Finally, theories of motivation including expectancy theory (Kalman et al.
2002) and social impact theory (Chidambaram and Tung 2005) have been applied to
explain motivations for sharing diverse knowledge, and both have identified CT
support as useful for helping individuals decide how to make their sharing more
valuable to others, increasing motivations to share. However, the distinction
between synchronous and asynchronous interaction mode was left unclear; thus it is
uncertain if these theories apply in both interaction modes or only one or the other
since motivation may be tied to feedback which is differentially available in
synchronous and asynchronous interaction modes.
Examining differential CT support use in synchronous and asynchronous
interaction modes may help address this discrepancy in the literature by offering
more specific strategies for how CT support aids sharing and reflection in
synchronous and asynchronous interaction modes, which clarifies the role of CT
support. Additionally, this approach demonstrates that CT support is not effective in
all situations, which may help explain inconsistent results from past research.

2.5.2 Multi-dimensional CT Support
Different theories have provided rationale for dimensions of CT support but
none has specifically focused on differential CT support in virtual collaboration for
innovation.

36
Research focused on CT support in information use contexts, such as GDSS
research, has focused on the use of CT support to exchange information and make
decisions (Dennis 1996; Dennis et al. 1997; Pinsonneault et al. 1999). This research
has found that CT support for such conditions as anonymity, parallel processing, and
group memory help motivate individuals to share knowledge and facilitate more
knowledge transfer, however outcomes from using these CT support features have
been mixed. Pinsonneault and Heppel (1998) for example note that anonymity may
free some from social evaluation thus leading to sharing, but that feedback is less
meaningful in anonymous situations. Much of this research has been conducted with
face to face-mediated groups, or by comparing nominal (not CT supported) groups
with CT-supported groups, which treats CT support as a black box (supported or not
supported). As previously noted, innovation theories regard knowledge transfer as
only a building block, a necessary but not sufficient condition for learning and
innovation, because the type of knowledge shared is as important as the act of
transferring knowledge (Boland and Tenkasi 1995; Carlile 2004). As a result, it is
difficult to explain the role of CT support for virtual collaboration for innovation
using GDSS theories unless they focus on what knowledge is being transferred,
rather than the process of transfer (Te’eni 2001).
Other work has focused on CT support matched with task type (Kerr and
Murthy 2004; Zigurs and Buckland 1988), which is useful for identifying CT support
that might be needed to accomplish innovative tasks (identified as high in

37
communication support and high in information processing support), but does not
specify how that CT support should be used in the synchronous or asynchronous
mode. Because individual’s perception of their collaborative need - which can vary
among individuals accomplishing different parts of the same innovative task - is
what drives their perception of needed CT support, it is difficult to generalize that
certain CT support is needed for certain tasks without examining in more detail the
impact of CT support on sharing and reflection for that type of task (Hollingshead
and McGrath 1995).
More recently, other theories of interpersonal interaction have been applied to
explain CT support in virtual collaboration for innovation, which have taken a multi-
dimensional view of CT support as a bundle of capabilities useful for facilitating
different needs. Carte and Chidambaram (2004) draw on social categorization
theory to explain how CT support provides reductive and additive capabilities which,
in different points of time in a group’s lifecycle, aid in cohesion and performance
(Gersick 1988). However, despite identifying differences in CT support, they do not
specify any differences related to the interaction mode, and they also confound
interaction mode with CT support. For example, they recommend visual anonymity
as a reductive attribute during synchronous teleconference (but do not specify how
anonymity should be supported in the asynchronous interaction mode). Malhotra et
al. (2001) present ample evidence of CT support in both modes, drawing on the
theory that individuals need to develop mutual understanding in virtual collaboration

38
for innovation (Krauss and Fussell 1990), but focus on group level interaction thus
they do not investigate the impact of CT support on individual cognitions. While
Malhotra et al. were among the first to show that CT support in both modes impacted
successful collaborative outcomes, because their focus was on interpersonal work
processes rather than individual social-cognitive processes the collaborative needs
particular to each interaction mode were not addressed. Lastly, Majchrzak et al.
(2005b) draw on Te’eni (2001) to explain how CT support can help provide
cognitive context individuals need to develop collaborative know-how, showing how
CT support aids in reflection on others shared knowledge for innovative tasks;
however, their theory does not specify differences in interaction mode, partly
because the CT support studied was useful in both interaction modes and partly
because they did not isolate use in different interaction modes in their
operationalization of the theory.
Thus numerous theories have been applied to explain how CT support aids
individuals in virtual collaboration for innovation, but reports of the effects of CT
support have been inconsistent, only a small proportion of research has examined the
multi-dimensional role of CT support for individual’s collaborative needs, and very
rarely in both the synchronous and asynchronous interaction modes. More work is
needed to address the impact of these factors to resolve past discrepancies and to
discover how CT support in virtual collaboration for innovation helps individuals
realize successful collaboration outcomes.

39

2.6 Summary and Research Questions
In summary, theories of virtual collaboration for innovation do not explain
how individuals use CT support to realize learning and innovation over a cycle of
complementary synchronous and asynchronous interaction modes. Prior theory has
identified that individuals must overcome cognitive complexity to support sharing
and reflection, but have not provided a rationale for how CT support helps resolve
cognitive complexity differently in synchronous and asynchronous interaction
modes. In addition, prior theories have for confounded or ignored the difference
between the two interaction modes when it comes to explaining how sharing and
reflection impact learning and innovation differently in each interaction mode, since
these theories have not accounted for the different opportunities for sharing and
reflection. As a result we do not know if the differences between these two
interaction modes are significant in virtual collaboration for innovation, nor do we
understand how learning and innovation result from collaboration in either and/or
both of these interaction modes.
Thus given the different interaction capabilities afforded by asynchronous vs.
synchronous interaction modes when collaborating virtually, I ask the following two
research questions:

40
1) Does an individual’s sharing and reflection affect learning and innovation
differently depending on whether the individual is interacting synchronously vs.
asynchronously? And,
2) Are there different kinds of CT support that facilitate an individual’s
sharing and reflection differently when the individual is interacting with others
synchronously vs. asynchronously?

2.7 Summary of Findings
In addressing the first research question, I expanded on the theory of
perspective making/perspective taking (Boland and Tenkasi 1995) which explains
how sharing and reflection leads to learning and innovation, and I found that
outcomes were dependent on the need for attending to self and/or others’ knowledge
coupled with the opportunity to do so in each interaction mode. Thus learning is the
result of sharing in both interaction modes and also reflection in the asynchronous
interaction mode (but not the synchronous) because individuals need to explain their
own knowledge as well as understand others’ knowledge which they have less
opportunity to do synchronously. In contrast, innovation is the result of reflection
during both interaction modes, but not directly from sharing in either interaction
mode, because individuals must focus on others’ knowledge whenever it is shared
rather than their own knowledge. Thus reflection consistently contributes to
innovation in either interaction mode.

41
In addressing the second research question, I found that CT support affects
both sharing and reflection in different ways depending on the different cognitive
demands of the interaction modes. I found differential value of four different types
of CT support, based on a framework of strategies for reducing cognitive complexity
in virtual collaboration from Te’eni (2001): CT support for testing and adjusting,
attention-focusing, contextualization, and perspective-taking. To facilitate sharing,
individuals need to quickly understand others’ feedback in the synchronous mode
and they need a means for obtaining feedback in the asynchronous mode – which CT
support for testing and adjusting and attention-focusing, respectively, provides. For
reflection, individuals need contextual information in both modes and access to
others’ knowledge in the asynchronous mode which CT support for contextualization
and perspective-taking provides. Thus different features of CT support reduce
cognitive demands in synchronous vs. asynchronous modes and facilitate sharing and
reflection, which ultimately leads to learning and innovation.

42
CHAPTER 3
THEORY DEVELOPMENT
To build a theory of CT support for virtual collaboration for innovation I first
address the research question: Does an individual’s sharing and reflection affect
learning and innovation differently depending on whether the individual is
interacting synchronously vs. asynchronously? I chose the theory of perspective
making and perspective taking (hereafter PM/PT) (Boland and Tenkasi 1995), to
explain how individuals in virtual collaboration for innovation realize learning and
innovation. Boland and Tenkasi argue that in a collaborative forum individuals are
able to accomplish two cognitions for sharing and reflection, and that as a result they
combine others knowledge with their own in a way that leads to learning and
innovation. Since their theory makes no acknowledgement of the tension between
sharing and reflection due to resolving conflict between diverse knowledge and
common knowledge, and because it does not incorporate interaction mode
differences, I theorize an extension to PM/PT theory which does account for these
influences. This part of the theory development, section 3.2, is labeled ‘Differential
Impact of Sharing and Reflection on Learning and Innovation.’
After theorizing the contribution of sharing and reflection, I next address the
second research question: Are there different kinds of CT support that facilitate an
individual’s sharing and reflection differently when the individual is interacting with
others synchronously vs. asynchronously? Boland and Tenkasi speculate in PM/PT

43
theory that CT support creates the forum for dialogue needed to support sharing and
reflection, however, their theory does not specify how CT support facilitates the
cognitions of sharing and reflection, and it has not taken into account characteristics
of the interaction mode which impact how CT support is used. I present an
extension to their theory for how CT support enhances the forum for dialogue based
on strategies for resolving cognitive complexity in virtual collaboration for
innovation from Te’eni (2001), which predict how individuals accomplish sharing
and reflection that leads to successful learning and innovation. This section, section
3.3, is labeled ‘Differential Impact of CT Support on Sharing and Reflection.’
The final model will show how virtual collaboration for innovation over a
cycle of synchronous and asynchronous interaction modes leads individuals to
successfully achieve learning and innovation.

3.1 Virtual Collaboration for Innovation Requires Perspective Making and
Perspective Taking
In this section I review the theory of perspective making and perspective
taking (Boland and Tenkasi 1995) which I argue is useful for understanding how
sharing and reflection lead to learning and innovation and which includes a
framework for incorporating CT support, but which also has clarified neither the role
of CT support nor differences in sharing and reflection across different interaction
modes.

44
Boland and Tenkasi (1995) argue that in virtual collaboration for innovation
individuals must engage in a cycle of perspective making and perspective taking
(PM/PT) – a social-cognitive process that explains how they transform others shared
knowledge into a form useable for themselves – in order to realize successful
learning and innovation. Perspective making they define as “communication that
strengthens the unique knowledge (of an individual).” Perspective taking is defined
as “communication that improves an individual’s ability to take the knowledge of
others into account” (p.351). They argue that PM/PT is necessary for individuals to
learn and innovate because individuals who come from different ‘communities of
knowing’ or ‘thought worlds’ (Dougherty 1992) maintain different perspectives,
which correspond to diverse knowledge and differences in how individuals believe
that innovative tasks ought to be accomplished. Not only must these differences be
communicated, but they must be acknowledged and resolved if individuals are going
to make use of others’ knowledge (Boland and Tenkasi 1995; Carlile 2004).
Figure 3-1: Diagram of the PM/PT Process (Adapted from Boland and Tenkasi 1995)

45
Figure 3-1 depicts a diagram of the PM/PT process, which first shows that
there are three inputs to the PM/PT process, conditions that are necessary for PM/PT
to take place: diverse knowledge from numerous knowledge workers; an innovative
task for which they must collaborate (Carlile 2004; Leonard and Swap 1999; Stasser
and Titus 2003); and a forum for sharing diverse knowledge through dialogue. In
this forum individuals exchange and interpret diverse knowledge with others.
Boland and Tenkasi contend this forum is supported by organizational elements such
as lateral organization structures, relational elements such as peer to peer
collaboration, and technical elements such as CT support for enabling virtual
collaboration for innovation.
Through the PM/PT process, individuals turn inputs into collaboration
outcomes, most notably learning and innovation, by simultaneously: a) sharing their
diverse knowledge with others from different communities of knowing (thus
requiring cognitions for deciding what to share and how it will be received by others
with little common knowledge); and b) reflecting on others’ shared knowledge (thus
requiring cognitions for attempting to understand what others are sharing and then
comparing it to what is already known). Learning occurs as individuals create for
themselves and others new ‘meanings,’ new ‘routines,’ and ‘new knowledge’ that
form the basis of a new mental models and builds “interpretive competence” (Boland
and Tenkasi 1995, p.355).

46
Boland and Tenkasi likewise argue that through PM/PT individuals are able
to make comparisons and draw conclusions from sharing and reflection in a dialogic
forum, which promotes innovation. Unlike debate, which highlights differences in
shared knowledge but forces participants to rigidly focus on pre-discussion
preferences, dialogue encourages each successive contributor to use their own
perspective to build on previously shared knowledge (Boland et al. 1994) which
when codified as work products (e.g. documents) are integrative and introspective.

3.2 Differential Impact of Sharing and Reflection on Learning and Innovation

The first research question asks how sharing and reflection differentially
impact learning and innovation depending on the interaction mode. Although the
theory of PM/PT identifies both sharing and reflection as necessary for learning and
innovation, it does not specify how these cognitions contribute differently to learning
and innovation nor does it account for interaction mode differences. I argue that for
learning and innovation, individuals may experience different opportunities for
sharing and reflection in the synchronous vs. asynchronous interaction mode.
Therefore sharing and reflection should impact learning and innovation differently
depending on the interaction mode unless the same opportunities are afforded in each
mode. Below I first discuss my extensions to the PM/PT theory for explaining
learning and then discuss my extensions for explaining innovation.

47
3.2.1 Impact of Sharing and Reflection in Synchronous and Asynchronous
Interaction Modes on Learning
Learning, according to Boland and Tenkasi (1995) is an outcome of the
PM/PT process by which individuals build new mental models through sharing and
reflection. Sharing, they argue, helps individuals re-construct meaning and
rationalize what they know so that they can evolve and enhance their knowledge.
Individuals literally explain what they know to themselves (Vygotsky 1978), a
process known as self-elaboration in social-cognitive learning theory (Chi et al.
1989; Majchrzak and Beath 2006; Webb and Palincsar 1996). An individual’s self-
elaboration of what she knows forms associations between her existing mental
models and new knowledge which is used to build a new mental model (Lindsay and
Norman 1977).
Sharing may not be necessary for other learning outcomes that may result
from collaboration besides building new mental models, such as the following
others’ directions or joint problem solving (Dillenbourg 1999), because for these
outcomes collaboration is less about dialogue and more about coordination of
thought and action. That is, what individuals know has little bearing on their
coordinated action. However, in virtual collaboration for innovation successful
learning depends on individuals focusing on sharing since innovative tasks require
more attention to promoting one’s own view during dialogue with others (Boland
and Tenkasi 1995; Dillenbourg 1999; Gray and Meister 2004). Otherwise,

48
individuals will not have a frame of reference for using others’ knowledge for
building new models since they do not have an appreciation of how to make use of
what they already know (Gray and Meister 2004; Vandenbosch and Higgins 1996).
Thus sharing is likely to contribute to learning whenever individuals are involved in
dialogue, which includes both synchronous and asynchronous interaction modes.
By focusing on ongoing dialogue in synchronous interaction modes and by
initiating or responding to coordinated dialogue in asynchronous interaction modes
individuals have the opportunity for sharing in both modes. Motivational and
cognitive complexity factors may limit the opportunity for sharing in both modes,
however if individuals recognize the importance of their sharing in dialogue (as was
proposed in chapter 2), and the cognitive complexity with doing so is resolve (which
CT support will be shown to resolve) then in both modes individuals have the
capability to decide what diverse knowledge to share and the means to share it with
others. Thus, I hypothesize:
H1: Sharing in the synchronous interaction mode and the asynchronous
interaction mode will contribute to learning.

However, learning requires more than just sharing; individuals also need to
reflect on others’ shared knowledge for learning because it is others’ knowledge that
leads to building new mental models rather than reinforcing existing models (Boland
and Tenkasi 1995; Carlile 2004; Vandenbosch and Higgins 1996). Unless

49
individuals understand new knowledge they are more likely to maintain existing
models, which causes them to rely on pre-discussion preferences because their
thinking and perspective is unchanged (Greitemeyer and Shulz-Hardt 2003;
Vandenbosch and Higgins 1996). Through reflection on others knowledge when it is
shared individuals decide how to understand and integrate that knowledge, so they
can think about innovative problems in new ways.
Gray and Meister (2004) note that more controlled, effortful processing of
others knowledge is required for learning in virtual collaboration for innovation.
Through reflection, individuals are able to make sense of and integrate others
knowledge to build new mental models and thus increase learning (Boland and
Tenkasi 1995); however, reflection in the synchronous mode may not have this
impact because individuals have to stay focused on sharing in ongoing dialogue
(which also includes soliciting feedback from others), and thus may not be able to
engage in the effortful processing of others knowledge needed to integrate it for
building new mental models (Gray and Meister 2004; Petty and Cacioppo 1984). In
the asynchronous mode, not only do individuals have more opportunity for reflection
since they are not focused on synchronous sharing in dialogue, they also have the
opportunity to take time for reflection to understand how to integrate others’
knowledge. Thus reflection in the synchronous interaction mode is not likely to lead
to learning, but reflection in the asynchronous mode is anticipated to contribute to
learning.

50
H2: Reflection in the asynchronous mode will contribute to learning but not
reflection in the synchronous interaction mode.

3.2.2 Impact of Sharing and Reflection in Synchronous and Asynchronous
Interaction Modes on Innovation
The act of innovation through creating innovative work products requires an
individual to combine others’ knowledge with her own diverse knowledge,
involving, as with learning, both sharing and reflection (Boland and Tenkasi 1995;
Gray 2000). Individuals must combine diverse sources of knowledge such that
irrelevant information is discarded, and relevant information synergistically
combined. Not only do individuals, by definition, lack all of the requisite knowledge
to innovate (Bruner 1990), their own knowledge is insufficient to evaluate if what
other information they might find is relevant or superfluous (Dillenbourg 1999; Gray
2000). Thus while in some situations it is conceivable for individuals to innovate
without participating in virtual collaboration for innovation, when others’ expertise is
needed it is more easily obtained and understood through sharing and reflection
(Carlile 2004).
Sharing may not directly contribute to innovation for two reasons: first, one’s
knowledge may be applied without first sharing it, unlike others’ knowledge which
must first be sourced (Gray and Meister 2004); second, increased time analyzing
one’s own point of view can detract from understanding others (Gray 2000), forcing

51
greater specialization which detracts from innovation (although in the long-term it
may benefit innovation, hence the reason specialists are picked for virtual
collaboration for innovation in the first place). Thus sharing represents a tradeoff – it
may not directly lead to innovation, but it may help individuals make sense of others
knowledge since greater awareness of what one knows can help an individual
prepare themselves for receiving new knowledge. Thus no direct contribution to
innovation is anticipated from sharing, but an indirect effect through reflection is
possible.
The cognition of reflection is a significant factor in innovation, however,
since it generates recognition of the differences and dependencies between others’
knowledge and one’s own, which clarify how diverse knowledge should be
combined when building innovative work products (Boland and Tenkasi 1995;
Carlile 2004). Reflection in both modes may be helpful for innovation as long as
others’ knowledge is accessible and the individual can devote her cognitive resources
to make sense of that knowledge in each interaction mode. In the synchronous
interaction mode, others’ knowledge is accessible through back and forth sharing
during dialogue; reflection on that knowledge can help identify how it can be
combined and synthesized for innovation, even if individuals approach the problem
of innovation from their own frame of reference since new knowledge does not
immediately contribute to learning. In the asynchronous interaction mode, access to
others’ knowledge may be more problematic (Gray 2000), but when it is accessible

52
individuals also have the time and cognitive resources for reflection, and they are
better able to innovate through consideration of how to make use of their own and
others’ knowledge. Thus, I hypothesize,
H3: Reflection in both the synchronous and asynchronous interaction mode
will contribute to innovation.

Figure 3-2: Hypothesized Effects of Sharing/Reflection on Learning and Innovation in
Synchronous and Asynchronous Interaction Modes

3.2.3 Summary of Impacts of Sharing and Reflection on Learning and Innovation
In summary, I hypothesize that learning is differently impacted by sharing
and reflection in synchronous vs. asynchronous interaction modes, while innovation

53
is impacted by reflection in both synchronous and asynchronous interaction modes.
Figure 3-2 depicts these hypothesized relationships. Significant differences are
hypothesized between synchronous and asynchronous interaction modes regarding
the contribution of sharing and reflection to learning, which is a departure from the
theory of PM/PT which suggests that sharing and reflection both are needed for
learning and innovation. Table 3-1 summarizes the rationale for these differences.
Table 3-1: Summary of Hypothesized Impacts of Sharing and Reflection on Learning and
Innovation
Collaboration
Outcomes

Cognition Learning Innovation
Sharing Diverse
Knowledge (Synchronous
Interaction Mode)

H1
Sharing Diverse
Knowledge (Asynchronous
Interaction Mode) H1
No hypothesis in either mode:
Concentrating on sharing one’s own
knowledge removes focus from
reflection on others knowledge;
although, an indirect effect may occur
since sharing makes one’s own
knowledge more accessible for
combination with others through
reflection while innovating
Reflection on Others
Shared Knowledge
(Synchronous Interaction
Mode)
No hypothesis: individuals
lack time and resources to
incorporate others shared
knowledge with what they
know if they are focused on
sharing in ongoing dialogue
H3
Reflection on Others
Shared Knowledge
(Asynchronous Interaction
Mode)
H2 H3

3.3 Differential Impact of CT Support on Sharing and Reflection
While the first research question focused on how the differential impact of
sharing and reflection on learning and innovation depends on the interaction mode,

54
in the second research question I ask what differential effect CT support has in the
synchronous vs. asynchronous mode on individual’s sharing and reflection. No prior
theory of virtual collaboration for innovation has yet been developed that explains
the differential effect of CT support on critical cognitions relevant to learning and
innovation in synchronous vs. asynchronous interaction modes, although Boland and
Tenkasi (1995) suggested the need for CT support to facilitate these cognitions.
While there is no comprehensive theory, Te’eni (2001) offers a framework
for reducing cognitive complexity during communication with others through the
application of four different communications strategies which I theorize might be
supported by CT. Previous models, according to Te’eni, are too simplistic and fail to
incorporate what is communicated as well as how it is communicated (emphasis in
original). As a result, they focus too narrowly on the transmission of knowledge
without taking into account why that knowledge was shared and how it will be used.
Different strategies are therefore needed to support different cognitions such as
sharing and reflection since trying to accomplish each one in virtual collaboration for
innovation results in increased cognitive complexity that the strategies help to
resolve. Te’eni suggests use of one (or more) of five strategies (four listed below –
the affectivity strategy was dropped based on the difficulty enacting this strategy
with CT support according to Te’eni 2001 and Boland and Tenkasi 1995) that
individuals use that may explain the impact CT support has on sharing and reflection
by reducing the cognitive complexity. Because individuals rely on CT support in

55
virtual contexts to create the forum for dialogue they need for sharing and reflection,
and the strategies, listed in Table 3-2, explain how individuals reduce cognitive
complexity to achieve their communication goals of sharing and reflection, I theorize
that CT support for each strategy will help individuals in virtual collaboration for
innovation resolve the cognitive complexity associated with sharing and reflection,
differently in each interaction mode, since the conditions which lead to greater
complexity differ by interaction mode. Each strategy is explained in greater detail
below, and for each one I theorize how individuals may adopt CT support to
accomplish each strategy.
Table 3-2: Definitions and CT Support for Te'eni Communication Strategies (adapted from
Te'eni 2001)
Communication
Strategy
Definition How CT supports communication strategy
Testing and
Adjusting
Spontaneously changing
what knowledge is
shared based on
feedback from self and
others
Displaying and editing shared knowledge in real-
time in a visible, tractable format, such as through
document sharing features or whiteboards; instant
feedback available and recorded
Attention
Focusing
Directing tailored
knowledge specifically
at others to initiate
dialogue and obtain
feedback
Text-based communication such as email, instant
messaging, discussion forums that permit two-way
asynchronous dialogue
Contextual-
ization
Provision of explicit
context in a message
helps individual
understand why others
made a contribution
Contextualization mechanisms such as displaying
names/profiles of collaborators, attributing verbal
and written contributions, file and document
sharing tools that permit back and forth
comparison of different versions
Perspective
Taking
Cognitively focused on
receiver’s view because
individual can access
their knowledge in their
own words
Access to stored knowledge in repositories or
virtual workspaces when needed for reflection or
use

56

3.3.1 CT Support for a Testing and Adjusting Strategy
Testing and adjusting refers to individuals using feedback from others to
decide what diverse knowledge to share and planning their pattern of sharing as the
virtual collaboration for innovation unfolds (Clark and Brennan 1991; Street and
Capella 1985). Boland et al. (1994) explain that sharing involves not only what we
know, but how we perceive our own knowledge and the knowledge of others. Thus
knowing what to share, and how it will be received, will help individuals rationalize
and organize their sharing diverse knowledge.
In synchronous dialogue it can be difficult for individuals to obtain directed
feedback from others about what they have shared since feedback competes with
ongoing sharing and individuals focused on sharing may find it difficult to perceive
how others have made sense of their diverse knowledge (Clark and Brennan 1991;
Te’eni 2001). This is particularly so in larger group settings (as opposed to dyads),
where voice communication is less effective as the number of individuals increases
(Chidambaram and Tung 2005). It is also more difficult for individuals to assess
their own sharing and offer themselves feedback (Boland et al. 1994) because
sharing, especially in synchronous interaction modes, is not reviewable; thus once
shared individuals have no mechanism for evaluating their own knowledge and
adjusting accordingly (Clark and Brennan 1991).

57
I argue that CT support provides interactive access to feedback through
editing work products such as documents, spreadsheets, or illustrations. For
example, seeing one’s own knowledge added to a document in real time, a document
simultaneously viewed and editable by others, helps an realize how their diverse
knowledge appears to others, and it gives them the opportunity to receive immediate
and meaningful feedback either verbally or through the same media. Thus CT
support can help individuals decide what diverse knowledge to share and understand
how others will understand that knowledge.

3.3.2 CT Support for an Attention Focusing Strategy
Attention focusing is a strategy of directing the receiver’s attention toward
knowledge shared by the sender, which helps individuals reduce the cognitive
complexity associated with sharing due to difficulties alerting others to their
contribution. The cognition of sharing requires an individual to know who they are
communicating with so the individual can decide how best to share in a way that
others will understand; however, in virtual collaboration for innovation, particularly
in asynchronous interaction modes, others may not be aware of what the individual
has to share and as a result can not offer any meaningful feedback. By calling
attention to their sharing diverse knowledge, Te’eni states that individuals are able to
direct and manipulate others’ processing of the diverse knowledge. If others are not
paying attention or are not aware of individual’s sharing then it is not likely that the

58
individual will obtain the feedback needed to verify how others’ received their
diverse knowledge (Cowan 1988).
CT support for an attention focusing strategy, I argue, is those features which
individuals use to let others know they are sharing, such as emails or other messages
or flagging mechanisms directing recipients to forums to access others’ diverse
knowledge. Typically verbal cues are used for attention focusing in synchronous
interaction modes (Clark and Brennan 1991; Walther 1995); however if verbal
bandwidth is saturated with sharing diverse knowledge already it may be difficult for
others to break in and focus attention on their own sharing diverse knowledge.

3.3.3 CT Support for a Contextualization Strategy
Contextualization refers to the inclusion of explicit context information in
messages which gives a recipient clues as to how that knowledge should be
interpreted. Majchrzak et al. (2005b) explain that Te’eni’s idea of context refers to
cognitive context – identifying alternative perspectives, details and changes in shared
content – rather than situational differences (Cramton 2001; Weisband 2002).
Situational context seems to be more important for conflict resolution and avoiding
misattribution (Cramton et al. in press; Hinds and Mortensen 2005) while cognitive
context is more important for reflection on others shared knowledge.
Majchrzak et al. (2005b) building on Boland et al.’s (1994) design theory for
distributed cognition outlines five aspects of a contextualization strategy. The first

59
of these is ownership. Ownership refers to visibility through CT support as to who is
sharing. CT support can list individuals by name or organization, describe their
background and expertise, and even display a picture or other icon. These features
can help contextualize the diverse knowledge these individuals share. As another
example, when sharing in discussion threads or task lists individual’s contributions
are automatically stamped with identifying information.
The second aspect is called easy travel, which refers to switching back and
forth among different topical or situational layers, or switching from general
summary to details. Easy travel contextualizes diverse knowledge by linking it to
other shared knowledge, helping individuals track how others’ sharing relates to
previous (or future) shared knowledge.
The third aspect is support of multiple perspectives. This means that the CT
support facilitates simultaneous sharing by more than one individual. CT support
might use multiple windows to display documents and charts, even from separate
desktops, and can maintain versions of documents that facilitate comparison.
Multiple perspectives help individuals contextualize shared because they can easily
highlighting consistencies and inconsistencies between different diverse knowledge.
The fourth aspect of contextualization is indeterminacy. Indeterminacy
recognizes that interpretation is not precise and that it occurs in piecemeal fashion.
Thus CT support must allow for partial messages, ongoing discussion, and open
relationships (anyone can interact with anyone else). Indeterminacy is particularly

60
important in virtual collaboration for innovation since grounding references built into
the face to face interaction mode are not present (Clark and Brennan 1991). Thus
open communication lines and structures help individuals contextualize others’
shared knowledge since misunderstandings can be immediately clarified.
The final aspect is emergence. Emergence means the CT support facilitates
understanding of new topic areas, links to new topic areas from old ones, or allows
for higher level (summaries) to be created. Emergence recognizes that as new
knowledge is created, if it is not added to existing knowledge in the right manner
then it is likely to lead to confusion rather than clarity. Emergence also recognizes
that sharing can often lead dialogue into new areas and it is not clear if these are
worthwhile or not until after they have been at least partially developed. Thus CT
support must help individuals foster both existing topics and developing ones.

3.3.4 Collaborative Technology Support for a Perspective Taking Strategy
A perspective taking strategy refers to the ability to see things from another’s
point of view, based on what they know (Te’eni 2001; Weick 1995). It supports
individual’s reflection by giving them access to others shared knowledge in their
own words, so that the individual not only discovers ‘what they know’ but also the
beliefs and motivations that accompanied the sharing of their knowledge (Boland
and Tenkasi 1995; Cross et al. 2001; Krauss and Fussell 1991). Time and cognitive
resources must be available to employ a perspective taking strategy (Swan and Shea

61
2005). To access the words and meanings of others shared knowledge, individuals
can rely on CT support to access knowledge codified in knowledge repositories
connected to virtual workspaces. These repositories contain documents, notes,
meeting minutes, schematics, presentations, and other materials shared during virtual
collaboration for innovation; they contain external knowledge added to help explain
or elaborate on others shared knowledge; and they contain documents created
together with others. Referencing these documents may help an individual make
sense of others shared knowledge.

3.4 Differences in CT Support for the Cycle of PM/PT
I have theorized that individuals use CT to support these strategies to help
overcome cognitive complexity experienced during virtual collaboration for
innovation. In this section I hypothesize how CT support is differentially needed for
reducing cognitive complexity in each interaction mode. Table 3-3 summarizes the
rationale for differential use of CT support for sharing and reflection across
synchronous and asynchronous interaction modes. It shows that testing and
adjusting and attention focusing strategies are those used by individuals to support
sharing. Both of these strategies support sharing by reducing the cognitive
complexity of making diverse knowledge available to others and permitting
meaningful feedback to be obtained. In contrast, contextualization and perspective
taking strategies support reflection by helping individuals reduce the cognitive

62
complexity of accessing others diverse knowledge in their own words and providing
information and meaning that helps them make sense of that knowledge (Boland and
Tenkasi 1995). Discussion and predictions for the impact of CT support for each
combination of interaction mode and collaborative need are presented below.
Table 3-3: Summary of CT Support Utilized Based on Need for Sharing or Reflection, and
Interaction Mode
Interaction Mode
→

Cognition
Supported ↓
Synchronous Interaction
Mode

CT Support Utilized
Asynchronous Interaction
Mode

CT Support Utilized
Sharing Diverse
Knowledge
Testing and Adjusting Strategy
utilized: Dialogue is continuous and
ideas and feedback can be shared
rapidly – however it can be difficult
to know what to share in relation to
what has already been shared, and
also difficult to offer meaningful
feedback. CT support can provide
the means to display and edit shared
knowledge in real time so it’s easier
to discuss further.
Attention Focusing Strategy utilized:
CT support needed to extend the
benefits of dialogue beyond normal
constraints of space and time to
asynchronous mode and increase
sharing when otherwise it would not
occur. Not needed in synchronous
interaction mode since sharing is
prevalent through synchronous
dialogue
Reflection on
others shared
knowledge
Contextualization Strategy utilized:
Understanding how to interpret
others’ knowledge means
appreciating the senders’ intent
when sharing, and seeing how that
knowledge fits with other shared
knowledge to address the complex,
innovative task at hand. CT support
can help individuals contextualize
knowledge shared in real time since
many cues relied upon in F2F
collaboration to contextualize are
missing.
Contextualization and Perspective
Taking Strategies utilized: Like
synchronous interaction mode
contextualization mechanisms are
needed to help individuals interpret
others’ knowledge. Unlike the
synchronous interaction mode,
individuals have the time and
cognitive resources to reflect more
extensively, but require access to
others’ knowledge. CT support can
create an environment for searching
for, accessing, and reading others’
knowledge in their own words.

63
3.4.1 Impact of CT Support for a Testing and Adjusting Strategy on Sharing in the
Synchronous and Asynchronous Interaction Modes
CT support for a testing and adjusting strategy facilitates ‘knowing what we
know’ and how to share it by presenting one’s knowledge in a format which makes it
easy for individuals to share knowledge in real time and in their own words, and
makes that knowledge visible so others can offer direct feedback (Boland et al. 1994;
Clark and Brennan 1991). In the synchronous interaction mode, sharing occurs
continuously and rapidly, however in virtual collaboration for innovation individuals
need to be able to receive meaningful feedback if their sharing is going to be of
sufficient quality and clarity to help themselves and others learn and innovate. I
argue that CT support for a testing and adjusting strategy helps individuals receive
and process feedback in real-time, making the decision of what to share easier to
make. Thus CT support for a testing and adjusting strategy complements the rapid
multi-cue feedback available in the synchronous interaction mode so that individuals
know what knowledge to share and how it is being received.
In the asynchronous interaction mode, the pace of dialogue is slower, thus
individuals are able to review and revise messages before they are sent and after they
are received thus the problem of incorporating feedback is less of an immediate
concern. In addition, numerous individuals trying to edit and comment on a visible
document without synchronous dialogue is likely to contribute added cognitive
complexity that overcome any potential gains, rendering this an ineffective way to

64
receive feedback useful for adjusting individuals sharing. Thus CT support for a
testing and adjusting strategy is not anticipated to be useful in the asynchronous
interaction mode.
H4: CT support for a control of feedback strategy will positively impact
sharing in the synchronous interaction mode but not the asynchronous
interaction mode.

3.4.2 Impact of CT Support for an Attention Focusing Strategy on Sharing in the
Synchronous and Asynchronous Interaction Modes
Text-based communication systems such as email, instant messaging, and
discussion forums support the attention focusing strategy by providing a medium of
knowledge transmission when voice (synchronous medium) is not feasible or
efficient (Finholt et al. 1990; Te’eni 2001); importantly, these mechanisms are
interactive, unlike a knowledge repository, which makes sharing and receiving
feedback more valuable (Millen et al. 2002; Qureshi and Zigurs 2001; Te’eni 2001).
Text-based communication is ideal for an attention focusing strategy since it is an
interactive medium effective in the asynchronous interaction mode for sharing
diverse knowledge with others in a way that makes it easy for them to respond.
Unlike the synchronous interaction mode, individuals have time to think
about what knowledge and feedback they are going to share before doing so in the
asynchronous interaction mode—however sharing diverse knowledge occurs less

65
rapidly. Individuals in the asynchronous interaction mode typically do not have the
ability to focus others’ attention on their knowledge, since the pace of sharing is set
by message receivers not senders (Clark and Brennan 1991). Knowledge recipients
may not expect to be involved in dialogue at any particular time. Thus individuals
can only focus others’ attention by sending it to them directly and in a manner which
facilitates easy response. Unlike a telephone call, however, text-based
communication systems permit sharing even if the recipient is not immediately
available, since knowledge shared remains visible until others attend to it (Cowan
1988).
Conversely, it is not likely that text-based communication would be
particularly valued in the synchronous interaction mode for sharing, since the
advantages of text-based communication in the asynchronous interaction mode (e.g.
message reviewability and revisability) are conversely limiting in the synchronous
interaction mode. Drafting or reading text-based messages would distract
individuals from ongoing sharing diverse knowledge (Clark and Brennan 1991;
Dennis and Kinney 1998), adding cognitive complexity. Thus CT support for an
attention focusing strategy is only expected in the asynchronous interaction mode,
since in the synchronous mode attention is already focused on dialogue and text-
based communication might actually distract others from sharing or offering
feedback to others.

66
H5: CT support for an attention focusing strategy will positively impact
sharing in the asynchronous interaction mode but not the synchronous
interaction mode.

3.4.3 Impact of CT Support for a Contextualization Strategy on Reflection in the
Synchronous and Asynchronous Interaction Modes
For reflection individuals have to understand the complexities of how
knowledge must be fit together to accomplish innovative tasks – individuals need to
understand the intent of why knowledge from others was shared and how they might
act on it (Boland and Tenkasi 1995). CT support for a contextualization strategy
uses mechanisms which help individuals build interpretations of others’ knowledge
by detailing who contributed knowledge and other information might relate to the
context in which that knowledge was shared (Boland et al. 1994; Majchrzak et al.
2005b; Te’eni 2001). CT support for a contextualization strategy aids in reflection
by identifying the differences between alternative perspectives, framing each
perspective to identify overlap and applicability given different conditions, and
ultimately helping individuals to make sense of them (Majchrzak et al. 2005b; Weick
and Meader 1993).
CT support for a contextualization strategy can help reduce cognitive
complexity by provided cues that are missing from dialogue in both synchronous and
asynchronous interaction modes. Even though some cues are readily transmitted

67
through verbal dialogue, the greater volume and pace of sharing in the synchronous
interaction mode often makes these cues difficult to interpret. Individuals have
greater difficulty have tracking, remembering, and attributing diverse knowledge
shared by others in the synchronous interaction mode. In the asynchronous
interaction mode, CT support for a contextualization strategy can help individuals
link different sources of diverse knowledge and identify common themes or
inconsistencies, despite the inability to immediately discuss the diverse knowledge
with the sender. Thus I argue CT support for a contextualization strategy is also
useful for reducing cognitive complexity for reflection in the asynchronous
interaction mode.
H6: CT support for a contextualization strategy will positively impact
individual’s reflection on others shared knowledge in both the synchronous
and asynchronous interaction mode.

3.4.4 Impact of CT Support for a Perspective Taking Strategy on Reflection in the
Synchronous and Asynchronous Interaction Modes
While CT support for a contextualization strategy aids reflection by
providing cues and other information needed to interpret others knowledge, to enable
that strategy individuals must have access to others’ shared knowledge in a useable
form. Te’eni (2001) refers to accessing others’ knowledge when needed, in their
own words as a perspective taking strategy. Individuals have the capability to store

68
shared knowledge in team workspaces or knowledge repositories for later reference
when trying to interpret meaning from others’ knowledge. When available, these
repositories give individuals flexibility to access and read others shared knowledge
when they have time and cognitive resources available for doing so, and they may
scan for knowledge particular helpful for reflection on others shared knowledge.
CT support for a perspective taking strategy is less effective in the
synchronous interaction mode, because although diverse knowledge is available, the
medium of synchronous dialogue is not adaptable (Carlile 2004); thus knowledge
cannot be manipulated or accessed as needed as easily in the synchronous mode as
individuals try and make sense of it in their own way. Individuals may have the
ability to access stored knowledge in a repository or virtual workspace
synchronously, but by doing so change the focus of their attention from the ongoing
dialogue to the asynchronous source of diverse knowledge, thus incurring immediate
collaboration inefficiencies. Unless individuals were not already involved in
ongoing dialogue it is doubtful that CT support for a perspective taking strategy
would be utilized during the synchronous interaction mode for reflection.
By contrast, individuals in the asynchronous interaction mode will not
necessarily be able to access others knowledge when needed since they are not
engaged in ongoing dialogue. However, individuals do have the time and cognitive
resources available for reflection because they can review others’ knowledge at their
own pace, rather than at the pace of ongoing dialogue. Therefore, CT support for a

69
perspective taking strategy is only hypothesized to impact reflection in the
asynchronous interaction mode:
H7: CT support for a perspective taking strategy will positively impact
reflection in the asynchronous interaction mode but not the synchronous
interaction mode.

3.4.5 Summary of Hypothesized CT Support
Table 3-4 summarizes the hypothesized impacts of CT support based on
theory of how CT support creates a forum for dialogue needed for individuals’
PM/PT, and in that forum how CT support reduces the cognitive complexity
associated with sharing and reflection in synchronous and asynchronous interaction
modes. Grayed out boxes indicate relationships not expected based on Te’eni’s
(2001) explanation of the communication strategies; enabling certain strategies can
actually increase cognitive complexity in some situations, thus not all strategies are
useful in all circumstances. The table also gives a rationale where CT support for a
strategy is not hypothesized to have an impact on sharing or reflection even if that
strategy itself is a useful one in that particular circumstance.
Different CT supported strategies are hypothesized to be useful for sharing
and reflection; in addition, different CT supported strategies are needed in the
synchronous vs. the asynchronous interaction mode. The only exception is for CT
support for a contextualization strategy. This strategy is hypothesized to impact

70
reflection on others shared knowledge in both interaction modes. Hypothesized CT
support impacts are depicted in Figure 3-3.
Table 3-4: CT Support Hypotheses and Rationale for Situation Where CT Support Effect Not
Anticipated to Occur
CT Support for
Communication
Strategy
Sharing Diverse
Knowledge –
Synchronous
Interaction Mode
Sharing Diverse
Knowledge-
Asynchronous
Interaction Mode
Reflection on
others shared
knowledge-
Synchronous
Interaction Mode
Reflection on others
shared knowledge –
Asynchronous
Interaction Mode
Attention Focusing Since pace of sharing
diverse knowledge is
controlled by
knowledge sender
and dialogue is real-
time. No CT support
is needed to
implement an
attention focusing
strategy
H5

Contextualization

H6 H6
Testing and Adjusting
H4
Not as critical to
obtain immediate
feedback since
sharing and receipt of
feedback are delayed.
Use of this type of CT
support may incur
extra coordination
inefficiencies

Perspective Taking Use of CT support
for perspective
taking is not an
efficient way to
garner others shared
knowledge.
Individuals are
cognitively involved
in real-time sharing
and cannot dedicate
time and cognitive
resources needed to
searching through
knowledge
repositories; in
addition, others
shared knowledge is
immediately
available through
dialogue
H7

71

Figure 3-3: Summary of Hypothesized CT Support Effects in Synchronous and Asynchronous
Interaction Modes
Input Process
CT Support for
Testing and
Adjusting Strategy
Sharing Diverse
Knowledge
Sharing Diverse
Knowledge
Reflection on Others
Shared Knowledge
CT Support for
Contextualization
Strategy
CT Support for
Attention Focusing
Strategy
CT Support for
Contextualization
Strategy
CT Support for
Perspective Taking
Strategy
H4
H6
Reflection on Others
Shared Knowledge
H5
H6
H7
Synchronous Interaction Mode
Asynchronous Interaction Mode

3.5 Control Variables
This section presents two control variables that potentially impact
individual’s sharing and reflection in the synchronous and asynchronous interaction
mode.

3.5.1 Control for Number of Participants
Social impact theory suggests that virtual collaboration for innovation
involving larger numbers of individuals makes sharing and reflection more difficult

72
(Kidwell and Bennett 1993; Latane 1981), because in larger groups individual
contributions are devalued and individuals are less motivated to share.
Chidambaram and Tung (2005), for example, found that when the number of
individuals interacting was low, individuals contributed more knowledge.
Wittenbaum et al. (2004) however theorize that individuals are more apt to self-
promote and thus are greatly motivated to share, despite group size.
It is unclear if CT support would compensate for decreased social impact
influence in larger groups, if present, but evidence suggests this may be the case.
Valacich et al. (1995) investigating production blocking, suggests that in the face to
face interaction mode individuals were sharing less, but that individuals in the
asynchronous interaction mode were sharing more. This suggests that sharing may
be greater the larger the number of other individuals in the asynchronous mode, thus
a control for number of participants is included in the analysis.
It is unclear if the number of participants will impact reflection, although
prior research suggests that as the number of participants increases the cognitive
complexity of reflection also increases (Te'eni 2001), implying a negative effect.
Chidambaram and Tung (2005) found support for this notion, showing that
individuals in smaller groups performed better, implying they were better able to
understand and integrate the perspectives of others. However, Valacich et al. (1995)
found that larger groups of individuals as a whole had more innovative outcomes.
Although analysis was completed at the team level, it suggests that the greater

73
number of participants had no detrimental impact on reflection. Given these results,
a control is included for number of participants on reflection in both the synchronous
and asynchronous interaction mode, with the expectation of a slightly negative effect
as the number of participants increases.

3.5.2 Control for Number of Participant Locations
Griffith et al. (2003) suggest that in virtual collaboration for innovation
involving individuals from a greater number of distributed locations, reflection is
more difficult since there is greater diversity of knowledge and context to be
reconciled. Similarly, Chidambaram and Tung (2005) suggest that individuals in
less immediate contexts and with greater psychological distance find sharing diverse
knowledge more difficult, because individuals have greater difficulty identifying
others’ contributions when distributed across multiple locations thus making it
harder to obtain useful feedback. However, they found that while individuals
contributed less there was no impact on collaboration outcomes, suggesting no
corresponding negative impact on reflection. Despite mixed findings, I control for
the influence of number of participant locations on both sharing and reflection, in
both the synchronous and asynchronous interaction mode.

74
Figure 3-4: Combined Research Model - Virtual Collaboration for Innovation Across a Cycle of
a Synchronous and Asynchronous Interaction Mode
Input Process Outcomes
CT Support for
Testing and
Adjusting Strategy
Sharing Diverse
Knowledge
Innovation
Sharing Diverse
Knowledge
Reflection on Others’
Shared Knowledge
CT Support for
Contextualization
Strategy
CT Support for
Attention Focusing
Strategy
CT Support for
Contextualization
Strategy
CT Support for
Perspective Taking
Strategy
H4
H6
Reflection on Others’
Shared Knowledge
H5
H6
H7
Synchronous Interaction Mode
Asynchronous Interaction Mode
H1
H3
H2
H3
Learning
H1

3.6 Summary: Combined Research Model
Figure 3-4 shows the combined research model that incorporates virtual
collaboration for innovation across a cycle of both synchronous and asynchronous
interaction modes. As the figure shows, no relationship is hypothesized between
learning and innovation; although Boland and Tenkasi (1995) describe a mutually
supporting role, they make no explicit argument for a causal relationship (Boland
and Tenkasi 1995). Learning and innovation, over time, are thought to have a
positive reinforcing relationship (Dillenbourg 1999), but in the short term it is not
clear if there are any causal influences.

75
PM/PT theory identifies both sharing and reflection as necessary for learning
and innovation, but does not specify how these cognitions contributed differently to
learning and innovation nor did it account for interaction mode differences. I argue
that learning and reflection depend on the need for attending to self and/or others’
knowledge coupled with the opportunity to do so in each interaction mode.
Therefore sharing and reflection are hypothesized to impact learning and innovation
differently depending on the interaction mode unless the same opportunities are
afforded in each mode. However, the model also shows that in each interaction
mode, both sharing and reflection are needed despite their different impacts. This
structure helps clarify previous inconsistencies that a) described sharing and
reflection as separate cognitions that were best realized in separate interaction modes
despite theory built on the premise that both were needed; and b) described sharing
and reflection as contrasting cognitions, with common ground creating a tension
between not enough context for reflection, but too much common knowledge to
inspire original thought.
Finally, Te’eni (2001) critiqued previous models of collaboration as too
simplistic for failing to incorporate the what of communication and instead focusing
on the how. The model developed here suggests that theories of virtual collaboration
also need to take into account the when since virtual collaboration for innovation
occurs over a cycle of a synchronous and asynchronous interaction mode.
Incorporating this difference creates a multi-dimensional account of CT support that

76
helps resolve inconsistent reports of the value of CT support from past research that
did not differentiate between synchronous or asynchronous interaction modes.
In the next chapter I present a case study that validates the theoretical
underpinnings of the model developed in chapter 3. Little empirical evidence could
be found of CT support for virtual collaboration for innovation over a sustained cycle
of synchronous and asynchronous interaction modes. Thus the purpose of the study
was to better understand how CT support helped individuals accomplish the
strategies needed for sharing and reflection, verify that there were actually
differences between CT support used in the synchronous vs. asynchronous
interaction modes, and evaluate the causal chain of events from use of CT support,
through sharing and reflection, to evaluation of learning and innovation outcomes.

77
CHAPTER 4
CASE STUDY: INDIVIDUAL’S USE OF CT TO SUPPORT THEIR
INTERACTION WITH OTHERS AS PART OF THE MILLENIAL PROJECT
From July-October 2005 I conducted a longitudinal case study analysis of
fifteen individuals engaged in virtual collaboration for innovation on an initiative
called the millennial project, at an organization named CentCo (described in detail in
Chapter 5). The purpose of the case study was to provide empirical evidence of CT
support for virtual collaboration for innovation over a sustained cycle of synchronous
and asynchronous interaction modes. Multiple synchronous and asynchronous
interaction modes could be compared to evaluate if differences between sharing and
reflection were consistent for each interaction mode (and not, for example, the result
of interpersonal factors unrelated to interaction mode). Also, virtual collaboration
for innovation in synchronous and asynchronous interaction modes could be matched
to specific efforts to innovate, to assess causal influences.
The following research questions were investigated through analysis of the
case study:
1) Does an individual’s sharing and reflection affect learning and innovation
differently depending on whether the individual is interacting synchronously vs.
asynchronously in sustained cycles of virtual collaboration for innovation?
2) Are there different kinds of CT support that facilitate an individual’s
sharing and reflection differently when the individual is interacting with others

78
synchronously vs. asynchronously over a sustained cycle of virtual collaboration for
innovation?

4.1 Case Study Setting
Individuals on the millennial project were given the innovative task of
applying a radical new technology to their organizations’ range of products and
services. CentCo asked individuals with varied expertise in engineering, education,
marketing, business development, and the new technology to build a number of
innovative work products: a feasibility report stating how the technology might be
used; a number of prototypes; and finally research plans for key business areas
showing how they would incorporate the technology. Thus for this particular case,
there was clearly an innovative task that had to be accomplished and CentCo
deliberately arranged for numerous experts with diverse knowledge to work together.
All of this expertise was not on hand in any one location. Individuals in the virtual
collaboration for innovation came from five main locations: two locations in
California, one in New York, one in Virginia, and one in Florida. Other individuals
from outside the company were consulted and added to the virtual collaboration for
innovation as needed, from sites in Florida and Maryland.
Virtual collaboration for innovation occurred in the synchronous and
asynchronous interaction modes but included no face to face interaction mode. The
reasons stated for this were: the budget for the project was low, and travel would not

79
be permissible; individuals had other duties and could not dedicate full-time effort to
the project; and a variety of skillsets and perspectives residing in geographically
separate locations were needed. This case therefore was an ideal situation for
studying virtual collaboration for innovation since the task was recognizably
innovative; diverse knowledge experts were included by design, and all interaction
was mediated by CT support.

4.2 Case Study Methodology
Case study methodology typically involves multiple data collection methods
employed over a period of time (Eisenhardt 1989; Yin 2002). For the four months of
the study, I observed virtual collaboration for innovation in the synchronous
interaction mode on eighteen occasions over that time, each consisting of one to one
and a half hour virtual meetings. Data concerning virtual collaboration for
innovation in the asynchronous interaction mode was gathered through observation
of those individuals I was collocated with in California; with a sampling of email
traffic and instant messages in which I was copied during the project; and from a
detailed analysis of the documents and work products in the online knowledge
repository/virtual workspace individuals used. In addition, I supplemented my
analysis of the individuals with semi-structured interviews after the project
completion (each interview lasted between 30-45 minutes) with all project
participants. Thus by the end of the 4 month period I had amassed data on

80
individuals’ use of CT support for virtual collaboration for innovation over a
repeated cycle of synchronous and an asynchronous interaction modes, and had data
on individual collaboration outcomes – evidence of learning, though the extent to
which they built new mental models and evidence of innovation, through the extent
to which they built innovative work products.

4.3 Findings: Evidence of Differential Impact of CT-Supported PM/PT on
Collaboration Outcomes in the Synchronous and Asynchronous Interaction
Mode
This section addresses the first case study research question: Does an
individual’s sharing and reflection affect learning and innovation differently
depending on whether the individual is interacting synchronously vs. asynchronously
in sustained cycles of virtual collaboration for innovation?
Previous work (Hinds and Bailey 2003; Maznevski and Chudoba 2000) has
suggested that innovative tasks require a regular face to face interaction mode
interspersed with a virtual interaction mode (synchronous or asynchronous not
specified), resulting in a cycle of interaction dependent on the re-occurrence of the
face to face interaction mode as depicted by the solid blue line in figure 4-1.

81
Figure 4-1: Nominal Extent of Sharing and Reflection in Virtual Collaboration for Innovation
with Regular Face to Face Interaction Compared to Completely Virtual
Sync
Asynch
Sync
Asynch
Sync
Asynch
Sync
Asynch
Regular Face to Face
Interaction
Completely Virtual

This figure shows a notional depiction of the cycle of interaction found by
Maznevski and Chudoba with synchronous and asynchronous interaction episodes as
they occur over time on the x axis, and the intensity of sharing and reflection on the
y axis. The graph depicts the nominal extent of sharing and reflection shown over a
series of interaction modes – alternating face to face and virtual interaction
represented by the solid blue line as described by Maznevski and Chudoba (2000);
and alternating synchronous and asynchronous interaction mode shown by the pink
dotted line. Because sharing and reflection are so low in the asynchronous
interaction mode when individuals rely on a face to face interaction mode,
individuals could not learn or innovate and were forced to hold face to face sessions
to achieve these outcomes.
By contrast, the case study group relying on CT support was more
consistently engaged over time in sharing and reflection. As time passed, sharing in

82
the asynchronous interaction mode even surpassed sharing in the synchronous
interaction mode, changing the dynamic from virtual collaboration for innovation
centered on the synchronous interaction mode to one in which collaborative
outcomes were more a result of the asynchronous interaction mode.

4.3.1 Impact of Sharing and Reflection on Learning
This section provides more insight into how individuals in the millennial
project realized a learning outcome. Evidence suggests that sharing in both the
synchronous and asynchronous interaction mode helped individuals to learn.
Individuals acknowledged that sharing helped them understand how ‘what they
know’ was important and how it should be integrated with others’ diverse
knowledge. As an example, a technology specialist sharing in the synchronous
interaction mode was first to suggest that the business model should have precedence
over the technical solution. This was notable since he was best qualified to identify
an optimal technology solution for the project and had no prior experience with
business plans.
As more sharing occurred in the asynchronous interaction mode, email and
instant message traffic dramatically increased, leading to sharing sessions that were
ongoing intermittently for hours. In addition, individuals reported that they were
regularly referencing documents stored in the virtual workspace as they attempted
reflection.

83
Individual’s (and the researcher’s) evaluations of the quality of feedback they
gave to others, which was both more thoughtful and inclusive of others shared
knowledge, provide evidence of the extent they were able to learn. Those
individuals sharing most frequently were found to be the individuals most capable of
evaluating others and providing insightful feedback.

4.3.2 Impact of Sharing and Reflection on Innovation
Evidence also indicates that greater sharing and reflection to a greater extent
contributed to innovation. Senior managers provided feedback regarding the quality
of innovative work products – feasibility studies for management, request for
funding from corporate agencies, statements of work for customers, and marketing
and engineering specifications for trade-shows. Evaluations indicate that innovative
work products were of higher quality because authors presented well-balanced,
considered opinions and recommendations that drew on diverse knowledge from
multiple knowledge domains. At the individual level, the quality of innovative work
products demonstrated how authors had succeeded in reflection (both on knowledge
shared in dialogue and in knowledge referenced from the virtual workspace) in both
the synchronous interaction mode and the asynchronous interaction mode. For
example, one individual reported that he referenced four months worth of archived
meeting notes to help craft a presentation. He combined diverse knowledge shared
in the latest synchronous interaction mode with that shared in the asynchronous

84
interaction mode, which helped him narrow his argument and focus on the most
relevant issues.

4.4 Findings: Evidence of CT Support
In this section I next investigate the second case study research question: are
there different kinds of CT support that facilitate an individual’s sharing and
reflection differently when the individual is interacting with others synchronously vs.
asynchronously over a sustained cycle of virtual collaboration for innovation? The
section is divided into three parts. The first describes how the individuals initially
relied on very little CT support, and as a consequence, they experienced great
difficulty with sharing and reflection in both interaction modes, and their
collaboration outcomes suffered accordingly. What the next two sub-sections
describe is the results of increasing adoption of CT support in the synchronous and
asynchronous interaction modes, and how as a result individuals realized greater
learning and innovation.

4.4.1 Initial Lack of CT Support Negatively Impacted Collaboration Outcomes
Early in the observation period, individuals were collaborating primarily in
the synchronous interaction mode for sharing, but most of their reflection occurred in
the asynchronous interaction mode. Initially, individuals used teleconference to
support virtual collaboration for innovation in the synchronous interaction mode and

85
email and a virtual workspace (to a limited extent) to interact in the asynchronous
interaction mode. They found that explaining complex concepts in teleconference
was extremely difficult. Others were reluctant to offer feedback after another shared
because they found it was hard to follow the thought process. Rather than offering
substantive feedback there were responses of “yeah” or “uh-huh” or “sounds good”
or just silence. There was little sharing diverse knowledge – in fact, very little back
and forth dialogue at all. Key points from what sharing did occur were summarized
and made available offline in the team virtual workspace. In the asynchronous
interaction mode, individuals tried to make sense of materials in the virtual
workspace, often by sending clarification emails to others to request feedback, but
rarely were they sharing new diverse knowledge asynchronously. They would rather
save their diverse knowledge for the next synchronous interaction mode. As an
example, at the start of one synchronous meeting one of the participants named Rick
asked John “I was reading (a research document) you posted in Livelink (the virtual
workspace used) and I wanted to see if we could use it.” Unfortunately since John
didn’t have access to it during the synchronous meeting he wasn’t able to explain it
very well and only answered direct questions from Rick. No one else joined in the
discussion about that topic.
Evidence suggested a number of factors contributed to individual’s
difficulties sharing in the synchronous interaction mode. For example, individuals
often found it difficult to remember who was sharing at any particular time, and also

86
who else was participating in the virtual collaboration for innovation. This made it
hard for individuals to tailor their own contributions and feedback when sharing. All
of this created mounting confusion and individuals found it difficult to extract any
meaning and understanding from others’ knowledge. As one member, Scott,
commented, the “problem with these things (teleconferences) is people call in and
I’m not sure who’s on the meeting or not.” Others commented that they were unsure
if they should share for a particular topic because they were having difficulty
following the pace of the dialogue and understanding which exactly was the current
topic. Conversely, individuals reported that reflection in the asynchronous
interaction mode was not helping them “understand where the others were coming
from,” as John expressed the problem. While there was little time for reflection in
the synchronous interaction mode, when individuals attempted to do so in the
asynchronous interaction mode when they did have time they experienced other
difficulties. They were successful only when they could link stored knowledge with
diverse knowledge shared during the previous synchronous interaction mode,
because otherwise the knowledge in the virtual workspace was too vague. For
example, individuals could pull up and read any document but they weren’t always
sure why they should do so – what knowledge that document would help them to
understand better. Very few emails or verbal conversations were made to help
individuals share asynchronously, and as a result their reflection suffered as well.

87
As a result, individuals reported it was difficult to learn from virtual
collaboration for innovation. There was little reflection (since it was hard to follow
in the synchronous interaction mode and difficult to dialogue for feedback in the
asynchronous interaction mode), thus knowledge did not cross boundaries between
knowledge domains and each expert maintained responsibility for her own area of
expertise when she tried to innovate. This was evident from statements made by
several team members that there was little crossflow of ideas, and that they didn’t
need to understand why others’ were doing what they did (only had to be aware of
what they were doing). In addition, what work products that had been created so far
took a frustratingly large amount of time and effort to coordinate, according to the
individuals, indicating little synthesis of others’ ideas.

4.4.2 Evidence of CT Support for Sharing and Reflection in the Synchronous
Interaction Mode

When individuals made the effort to rely on CT support to a greater extent,
both in the synchronous interaction mode and asynchronous interaction mode,
sharing and reflection improved significantly. Individuals began to rely on CT
support in the synchronous interaction mode in a number of ways. First, individuals
began to use a virtual meeting tool (Microsoft Netmeeting) in the synchronous
interaction mode. This tool gives each individual a visible virtual ‘identity,’ permits
simultaneous contribution to documents and whiteboards, includes chat, and
facilitates file sharing. Individuals started editing presentations and documents in

88
real time - sharing in text form on a jointly accessible and visible document (usually
also discussing aloud while doing so) – which prompted immediate feedback from
others and further sharing. Participants’ contributions were in their own words and
visible while verbal sharing from others ensued. An example of how this occurred
came from Mark, who started a ‘living agenda’ during the synchronous interaction
mode. An agenda shell was shared and as each point came up it was discussed and
individuals added their salient points in real time. Not only did this increase sharing,
but the agenda was also saved and archived for later offline reference. Individuals
subsequently reported that these agendas were a “tremendous resource” for
asynchronous reflection. Not only was others’ diverse knowledge visible, but it was
also organized by topic and recorded as a summary of the most salient points made
from individual’s sharing in the synchronous interaction mode. It gave them the
opportunity to contact specific individuals in the asynchronous mode to offer
feedback and continue the discussion through asynchronous sharing.
I expected that jointly editing documents in real time would be an example of
CT support for a testing and adjusting strategy. Evidence from the case study
supports this assertion. Before adopting this CT support, individuals were unsure
how others were receiving and interpreting knowledge they had shared because little
feedback was being offered. When others could see their diverse knowledge as it
was added and edited in real time, individuals received much quicker and much more
pertinent feedback. As individuals were sharing by contributing to the jointly edited

89
document, others’ were seizing on their knowledge and initiating further discussion
by sharing their own diverse knowledge. Because the text could be updated,
feedback was used to immediately focus in on important areas, or in some cases to
generalize to more abstract problems. Either way, it prompted greater sharing.
There was also evidence that CT support for a contextualization strategy was
used to help individuals with reflection in the synchronous interaction mode.
Contextualization mechanisms were used to identify participants, which helped
individuals understand who said what (and why). Features that identified
contributors helped individuals attribute diverse knowledge. Using real-time
document/desktop sharing, individuals’ were able to link contributions to other
knowledge which helped them understand the importance of others’ knowledge and
make sense of that knowledge. One individual named Brian matched displayed
identities with verbal and text sharing, which helped him discern ownership (who
shared what), a contextualization mechanism useful for reflection. Knowing who
was sharing gives individuals’ insight into the other’s perspectives and frames of
reference so that their knowledge is easier to interpret. Another contextualization
mechanism relied upon was hyperlinks, or links that connected different sources of
knowledge. Indicative of the part of contextualization Boland et al. (1994) refer to
as ‘easy travel,’ links between knowledge sources help individuals figure out how
diverse knowledge shared by multiple individuals fit together. This aided
individuals in reflection, as evidenced by statements such as “I had a much clearer

90
idea of what was said,” and that these links gave access to “more information to offer
than typically comes up in a (synchronous) meeting.”
Was there evidence of CT support for an attention focusing strategy or a
perspective taking strategy? Analysis indicates that attention focusing was not an
active strategy needed by individuals in the synchronous interaction mode. Diverse
knowledge shared verbally and through jointly editing documents was pushed
through the virtual collaboration for innovation to all other individuals. Those
sharing did not have to worry about others not receiving their diverse knowledge. In
the synchronous interaction mode it was more important to focus on one’s own
sharing than to attempt to focus others’ attention. Since the strategy was not
generally employed no CT support was needed for an attention focusing strategy.
Interestingly, on a few occasions individuals utilized the built-in chat feature to send
text messages to others, but this did not appear to support an attention focusing
strategy. On one occasion chat was used to quickly build a list of famous inventors
in the fashion of a collaborative writing experiment. Generating the text-based list
saved collaborative effort compared to how it would have been done verbally (Clark
and Brennan 1991). The list was the topic of immediate further dialogue and was
also archived to the teams’ collaborative workspace for later reference. In this case
the chat was used as a jointly edited document and was useful for helping direct
feedback and giving visibility into others shared knowledge. On another occasion,
chat was used by two members to direct feedback to another member while he was

91
sharing in a length presentation. In this case chat was used as an alternative
communication channel to offer feedback without having to interrupt the speaker.
The speaker was able to view the comments during the subsequent asynchronous
interaction mode and reported that they were extremely helpful since they had been
made ‘on the fly’ during the actual presentation. Thus a testing and adjusting
strategy was supported although interestingly it was others who initiated the testing
and adjusting for the speaker, not the speaker himself.
Reflection that occurred in the synchronous interaction mode seemed to be
completely supported by CT support for a contextualization strategy rather than CT
support for a perspective taking strategy. Individuals were used to a paradigm of
staying cognitively focused on the ongoing dialogue and were thus either hesitant or
just not used to taking the time to read through stored knowledge during the
synchronous interaction mode. Enabling a perspective taking strategy requires
individuals to be able to read and look for meaning in others’ diverse knowledge at
the individuals’ own pace, not at the pace of synchronous dialogue. Thus CT
support was not utilized for a perspective making strategy.

4.4.3 Evidence of CT Support for Sharing and Reflection in the Asynchronous
Interaction Mode
Adoption of greater CT support prompted more sharing in the asynchronous
interaction mode, as previously individuals would wait for the next synchronous

92
interaction mode for sharing. Documents created in the synchronous interaction
mode and diverse knowledge shared by individuals on their own, with the expressed
purpose of helping others understand what they know, were rapidly populating the
project virtual workspace.
In the asynchronous interaction mode, evidence was found for CT support for
an attention focusing strategy, a contextualization strategy, and a perspective taking
strategy. The only strategy not used in the asynchronous interaction mode was a
testing and adjusting strategy. Since the pace of sharing was slower than in the
synchronous interaction mode, and as most sharing was accomplished via text means
(primarily email but also instant messaging and discussion forums), the threat of
missing needed feedback was greatly reduced.
Individuals primarily used CT support for an attention focusing strategy by
sharing through email, instant messaging, and/or discussion forums. These text
communication mechanisms are addressable, permitting one-to-one or one-to-many
sharing. Discussion forums are organized usually by topic and contributions are
threaded, so each individual can see what others have said about their contribution.
Individuals came to rely on CT support to a greater extent for sharing concerning
new topics in addition to clarifying or discussing previous topics. Dialogue would
often start off as one-to-one then branch out as more insight was needed to include
sharing from many others – sometimes the whole group. Going into the next
synchronous interaction mode, individuals were already ‘up to speed’ on others

93
perspectives. In the past, email had been used sparsely and usually only to share
documents back and forth or clarify knowledge already shared. Individuals soon
came to realize that by grabbing others’ attention through text-based communication
they could be sharing at any time – especially if they had new knowledge to share,
were confused and needed resolution, or if time-sensitive decisions needed to be
made.
As mentioned, they also found that sharing in the asynchronous interaction
mode through using CT support made their subsequent synchronous interaction
mode more productive. Analysis of synchronous meeting transcripts showed the
amount of sharing in the asynchronous interaction mode and the speed with which
rapid sharing started in the next synchronous interaction mode were related. It
appeared as if individuals had initiated a dialogue days before and had never stopped
sharing since, despite the face that sharing during the asynchronous interaction mode
was not continuous or rapid-paced.
In addition, use of the project virtual workspace (where stored knowledge
was available) became more frequent, as evidenced by the number of accesses by all
individuals on the project to read stored documents. Individuals were relying on CT
support for a perspective taking strategy when accessing the virtual workspace. For
example, one individual named John placed summaries of academic texts in the
virtual workspace, which the others deemed invaluable insight into difficult to
understand topics. They would not have known how to find the original texts

94
themselves, nor would they probably have understood them without the summaries.
A review of knowledge in the virtual workspace became some individual’s first step
before innovating to make sure they had the latest insight from other team members.
One individual referred to accessing others shared knowledge in the virtual
workspace as “reading the thoughts of people” and reported that the virtual
workspace tool, Livelink, was “an absolute must.”
All of this CT support used in the asynchronous interaction mode - email,
instant messaging, and the virtual workspace - included features that helped
individuals contextualize diverse knowledge, such as information regarding time and
date posted, links to other documents, and author information. In addition,
contextualization helped individuals’ create and maintain multiple versions of
documents that could be accessed, compared, and edited. One participant named
Scott would often check the logs to see who had accessed stored knowledge and
when, which helped him “understand what others’ understood.” As in the
synchronous interaction mode, CT support for a contextualization strategy was
influential for individuals’ reflection on others’ shared knowledge.
In summary, evidence from both the synchronous and asynchronous
interaction mode demonstrates how CT support for certain communication strategies
increased sharing and reflection. Different CT support was needed in the different
interaction modes. Interestingly, when individuals relied on CT support for sharing
and reflection, there was evidence of crossover effects from one interaction mode to

95
the next. Individuals began to restructure their work processes to take advantage of
this phenomenon, as evidenced by the content of an instant message from Rick
during one asynchronous interaction mode that announced “I’ve just posted my
summary. Have a look and I’ll bring it up on Friday (during the next synchronous
interaction mode).” The intent was for other individuals to reflect by examining a
posted document so that a livelier and productive dialogue would ensue, with greater
sharing, in the next synchronous interaction mode.

4.5 Summary of Case Study Findings
The case study has been presented as an encapsulated research project that
provides direct evidence that the PM/PT process does occur in both the synchronous
and asynchronous interaction modes, although differently in each. CT support
facilitates this process, although different CT support is needed for the synchronous
vs. asynchronous interaction mode.
Without CT support, the overwhelming majority of sharing was during the
synchronous interaction mode, but knowledge shared did not necessarily equal
knowledge understood. In the asynchronous interaction mode, individuals had more
time for reflection and had access to stored knowledge. However, there was less
sharing. As a result individuals did not learn to a great extent and had difficulty
innovating.

96
Uptake of CT support was instrumental for making virtual collaboration for
innovation more successful. CT support for a testing and adjusting strategy helped
individuals sharing in the synchronous interaction mode by structuring their sharing,
making it visible, and creating a mechanism for obtaining feedback that was
otherwise difficult to achieve in the fast-paced continuous dialogue. CT support for
a contextualization strategy helped individuals’ reflection since individuals had no
time or cognitive resources for reflection at their own pace. In the asynchronous
interaction mode, testing and adjusting was less of a concern since the pace of
communication was not immediate – but CT support for an attention focusing
strategy was needed to engage knowledge recipients in sharing. Reflection again
depended on CT support for a contextualization strategy but also for a perspective
taking strategy, since individuals had time available for reflection but needed access
to diverse knowledge stored in the virtual workspace. This gave them diverse
knowledge in a format that could be referenced at their own pace, but in the
individuals’ own words so meaning and perspective was preserved.
In the synchronous interaction mode, individuals’ sharing increased because
they were better able to make sense of their own knowledge and feedback from
others was more poignant. In the asynchronous interaction mode, sharing increased
dramatically whereas without CT support there was little substantive sharing. As a
result of CT support that impacted sharing and reflection individuals were better able
to learn and innovate.

97
CHAPTER 5
RESEARCH METHODS
This chapter introduces the main research setting for the study and presents
the case study and survey methodologies used to collect data for analysis. I describe
in detail the sample selection process, the survey measures, data collection results,
and preliminary data analysis. The chapter concludes with an overview of the
analysis method used (partial least squares) to test the hypotheses from chapter 3
using data collected from two-rounds of surveys.

5.1 Research Setting
Because individuals engaged in routine or non-equivocal tasks may not need
to engage in PM/PT to realize successful collaboration outcomes, it was essential to
find individuals engaged with others in virtual collaboration for innovation with no
face to face interaction. An additional criterion, as explained earlier, was that
individuals with diverse knowledge be involved in the virtual collaboration for
innovation, and ideally with each individual representing a different community of
knowing.
An appropriate research site was found in the form of CentCo (A
pseudonym), a Fortune 500 and CIO Agile 100 organization with over 120,000
employees, in locations in every U.S. state and in 25 countries. CentCo consists of
numerous business sectors and its products and services fall mostly into the

98
aerospace, defense, and information services industries. One of the sectors,
headquartered in southern California, was engaged in synthesizing decades of
scientific knowledge and engineering processes. Due to mergers and acquisitions the
processes were situated with several legacy corporations now integrated into one
organization. The goal was to create a common development framework for various
scientific and engineering practices. The project’s aim was much grander than
simply developing a list of best practices. Rather, they intended to create rigorously
vetted, completely inclusive processes for designing aircraft and spacecraft
subsystems. While reliable and well tested, the processes were to be innovative and
forward-looking (continuously sweeping in the latest scientific breakthroughs and
scientific advancements). They also needed to be relatively easy and practical to use.
Scientific and engineering experts from across the sectors’ multiple locations
were assigned to these projects. Each individual had expertise with different
technologies and tasks. It was mandated that each distributed location have a
representative for each project, for two reasons: one, it ensured that expertise built at
that location was integrated into the common development platform; two, it ensured
that the common platforms, standards, and routines generated would work in the
localized culture of each location. Therefore, individuals were engaged in virtual
collaboration for innovation with others in more than one distributed location besides
their own location.

99
Despite the strategic necessity of these initiatives to CentCo, they were not
able to assign individuals full time to these efforts. Individuals were already
committed to one or more other projects by contractual obligation, and even had that
not been the case CentCo did not have the slack resources in manpower needed to
staff these projects. Thus most individuals were required to participate on a part time
basis in addition to their other duties. Because of this, it was infeasible to rely on
regularly scheduled face to face meetings. The decision was made instead to rely
completely on virtual collaboration for innovation. Typically, then, individuals
participated in one synchronous virtual meeting per week and then engaged other
participants asynchronously throughout the week. Thus their virtual collaboration
for innovation consisted of a continuing cycle of synchronous and asynchronous
interaction modes.

5.2 Data Collection
CentCo provided a list of the development projects and points of contact for
each project. In addition, based on preliminary feedback from analysis of the
individuals in the case study CentCo asked to include individuals from a number of
other projects in the study who were creating new procurement procedures for the
company. These projects were not a part of the development effort described in
section 5.1; however, they also involved individuals in virtual collaboration for

100
innovation and were therefore included in the study. In all, individuals from 92
projects were identified for inclusion in the study.

5.2.1 Survey Methodology
Individuals were asked to participate in a two round survey. The first survey
concerned the synchronous interaction mode and was given to individuals
immediately following a weekly synchronous teleconference (which included voice
communication as well as other CT support). Teleconferences were chosen as the
focus of the synchronous interaction mode survey because they were universally
used across the projects and thus common to each individuals’ experience, and they
guaranteed interaction with numerous other individuals. A second survey was
distributed three days after respondents the first survey. This survey addressed the
asynchronous interaction mode. Since the synchronous interaction mode was a
regular weekly reoccurring event, the interim between successive synchronous
interaction modes was a logical and understandable period of time in which
individuals report on virtual collaboration for innovation for the asynchronous
interaction mode (Marks et al. 2001).
Surveying methods are most effective when participants consider very recent
events rather than having to reflect back on event that happened in the past (Reis and
Gable 2000). Thus it made sense to send individuals the surveys in two rounds, one
that was completed immediately following a synchronous interaction mode, and the

101
other three days later in the middle of the asynchronous mode. Using two separate
surveys separated by several days helped reduce the bias associated with using a
single survey asking about both the synchronous and asynchronous interaction mode.
Second, it was hoped that using two separate surveys would help avoid a halo effect
based on individuals’ preference for either the synchronous or asynchronous
interaction mode that might skew results.
Several alternate survey strategies were considered and rejected. One
possibility was to use a single survey to capture individuals’ responses concerning
both interaction modes for a period of time, such as one week. The advantage of this
strategy would have been avoiding non-response between the first and second
surveys, resulting in a full panel of data for the analysis. This strategy, however,
would have resulted in a very long survey, something CentCo was keen to avoid.
They insisted on shorter surveys, even if that meant multiple rounds of data
collection. Another method considered was to focus the analysis on a subset of
projects over an extended period of time, and repeatedly survey the individuals after
each interaction mode. This strategy was initially preferred by CentCo and would
have yielded longitudinal data and more in-depth analysis of virtual collaboration for
innovation over time. However, the projects identified for the study (a different set
than those actually included in the survey) were not staffed in time, thus this option
became infeasible.

102
Although most employees at CentCo were involved in some type of virtual
collaboration for innovation, the decision was made to survey individuals from the
identified projects rather than from CentCo at large. Individuals from the identified
projects all held diverse knowledge from a particular scientific or engineering field,
they were focused on virtual collaboration for innovation exclusively with other
CentCo members (thus they could share information freely), and they had available
the same types of CT support. Individuals from other parts of CentCo may have
used different CT support, may have been engaged in virtual collaboration for
innovation with external customers or suppliers, may not have been working on
innovative tasks after all (many projects concern existing initiatives and very little
sharing diverse knowledge is needed as procedures are well established), and the
individuals may not have been experts in any particular field. Thus, although
focusing on individuals exclusively from a small subset of projects limited the
possible sample size and reduced heterogeneity of the sample, it was necessary to
properly investigate the research model.
Individual participants were given the option of completing the surveys in a
web-based survey system or by completing a word document and emailing it to the
researcher. They were encouraged to use the web-based survey for ease of use –
85% did so for the first survey and 90% used the web-based system for the second
survey. In addition, project sponsors were asked to independently rate the
performance of the individual respondents based on the extent to which they build

103
innovative work products. To avoid bias, they rated every individual on the project
and not just those identified as respondents to one or more of the surveys. Sponsors
were not informed which individuals had responded and which had not.
Before administering the surveys, ten CentCo employees and two graduate
students pre-tested the surveys. Each pilot participant was given the survey
questions as they would appear to the individuals. Half of the participants completed
the survey independently, and the length of time it took to complete them to
complete the survey was recorded. The other six members completed the survey in
the presence of the researcher. Based on a technique known as cognitive
interviewing (Visser et al. 2000), they were encouraged to read each item and think
aloud as they formulated their response. Although this took longer, it provided
additional insight into how respondents interpreted the questions. All participants
were also encouraged to make notes regarding unclear items for later discussion.
After completing the survey, the pilot participant went through their survey item by
item in an exit interview, looking for abnormal responses and addressing their
concerns. Poorly worded items were fixed on the spot and the new version used for
each subsequent pilot.

5.2.2 Survey Responses
CentCo desired that individuals only participate if it was cleared ahead of
time with their project sponsor; thus sponsors were first contacted and asked to

104
participate. Ninety two project sponsors were contacted; 54 of them responded to
the survey (59% response rate). Of the 38 that did not respond, 9 never responded to
three separate email requests and a voicemail; 12 agreed to participate initially but
never completed the survey, and didn’t respond to a follow-up email and/or
voicemail; 11 replied but said their project either had not yet formed or had already
disbanded; three felt their projects were not appropriate for the study (e.g. they were
not meeting regularly at the current time); two declined to participate because they
were too busy; one could not open the attached document and decline to participate
due to the technical difficulties. The remaining 54 project sponsors passed on the
notification to the individuals working the project asking for participation.
Approximately 297 individuals working on the 54 separate projects were
asked to complete the first survey. Of those, 120 individuals responded for an
overall response rate of 40%. According to Sivo et al. (2006), 40% response is just
above average for research reported in top IS journals. Surveys were collected
continuously during the period from 1/1/06 – 2/23/06, but roughly corresponded to
two main waves of data collection. The first wave lasted from 1/1/06 – 1/31/06 and
yielded 66 of the 120 responses; the remaining 54 were received from 2/1/06 –
2/23/06 after a second call for participation was sent out. During the last week, a
third reminder for participation was sent and resulted in the last 12 responses.
Individual responses represented participation in 44 of the 54 projects and 24
separate geographical locations, demonstrating appropriate dispersion of responses

105
across the organizations distributed sites. Fourteen of the responses were missing the
majority of the data and were discarded. Thus 106 useable responses were gathered
to the first round survey for the synchronous interaction mode.
The second-round survey was administered to the respondents three days
after receipt of their first round survey. Whereas the first survey asked respondents
specifically about their latest synchronous interaction mode, the second round survey
asked respondents instead to consider their virtual collaboration for innovation in the
interim since that interaction mode. Delivery of the survey notice was timed to
coincide with the middle of the week between consecutive weekly synchronous
modes. Of the 120 initial respondents, 108 were asked to complete the second
survey. The last twelve respondents to the initial survey were not included in the
follow-up as the end of the study timeline for data collection was reached before they
could be included. Of the 108 contacted, 73 responses were received, of which 72
were useable. This represents a response rate of 67% for the second-round survey.
Individuals responding to the second round survey represented 38 different projects.
Due to a technical problem with the survey website, 5 cases were lost; as a result, 67
responses to the second survey remained and were used for further analysis.

5.2.3 Respondent Demographics
On average, the survey respondents had been working for 26 years, 14 of
these at CentCo. They were highly educated; 90.6% were college graduates and

106
52.9% had graduate degrees. On average, respondents reported 2-3 years experience
using CT support. As expected, the respondents were expert in their field; 76.4%
agreed that they were expert, while only 23.6% disagreed. These may have felt they
weren’t experts due to the innovative tasks they were trying to accomplish.
I asked individuals about their prior relationships with others in the virtual
collaboration for innovation. Some 33% reported that they had no prior familiarity
with any of the others; 45.4% had worked with a few or some in the past; only 18.8%
had any significant contact with others. Some individuals however did have current
contact with others working on other concurrent projects. 13.3% were currently
working with some of the others on other projects; 50% were working with a few
others; and 35.8% were not working with any of the others except on the project in
question. Table 5-1 displays summary statistics for the individuals’ responses.
These indicate that while some prior familiarity exists, on average it is not a
significant factor which might explain individuals’ sharing and reflection –
particularly since the innovative tasks required unique solutions that were different
from past innovative tasks.

107
Table 5-1: Respondent Demographic Statistics
Individual Characteristic Scale Average Min Max
Years Working for CentCo Years 14.09 < 1 40
Years Working Overall Years 26.329 1 48
Education 1 = some high school
To
7 = completed doctorate
degree
5.10 (some
graduate
school)
2 7
Experience with CT support 1 = 0-3 months to
7 = 3+ years
5.93 (2-3
years)
1 7
Task Expertise (I am an expert at my
work)
1 = strongly disagree to
7 = strongly agree
5.15 (slightly
agree)
2 7
Prior Familiarity with Others (worked
with in the past)
1 = none of them to
7= all of them
2.64 (1.93) 1 7
Currently work with others on
different projects
1 = none of them to
7= all of them
2.34(1.63) 1 7

5.3 Survey Measures
In this section I describe the constructs and items used to develop the
surveys. As explained earlier, surveys were administered to the individuals at two
points in time to capture responses regarding synchronous and asynchronous
interaction modes. In addition, project sponsors were asked to independently rate
every individual on the project on the extent they build innovative work products.
When possible, existing scales were utilized; however in a few cases it was
necessary to create new measures.

5.3.1 CT Support Measures
Majchrzak et al. (2005b) argue that simply providing CT support for use does
not guarantee its use as intended. Use depends on how individuals adapt the CT
support for their situation (DeSanctis and Poole 1994). Thus the impact of the CT
support depends on the fit between the need for CT support, the availability of

108
technology features, and the interaction mode (Goodhue and Thompson 1995). As a
result, Majchrzak et al. (2005b) recommend a strategy of having the individual
assess their perception of CT support used rather than assessing the availability of
the CT support for use.
This research employs the same strategy. While all individuals in the study
had available very similar CT support, what varied was how they used the CT
support for enabling communication strategies. Therefore the measures for CT
support ask individuals to assess the utility of the CT support for enabling different
communication strategies, rather than asking how often or how much they used the
CT support in general. Thus, for example, instead of asking “how frequently did you
access the project virtual workspace,” the measure for CT support for a perspective
taking strategy asks “the extent to which you relied on the virtual workspace for
access to shared documents.”

CT Support for a Contextualization Strategy: I adapted a 12 item scale from
Majchrzak et al. (2005b) that measured CT support for knowledge contextualization
with contributions to knowledge repositories. In the synchronous interaction mode,
all 12 items loaded on a single factor with loadings > 0.75 and total variance
explained by the single factor of 66.6% (Table 5-2). Due to data limitations and
respondent comments I attempted to reduce the items to a more parsimonious eight
items. The items taken away from the scale were judged repetitive or confusing by

109
the respondents. Removed items are highlighted in gray in the table. The remaining
eight items included questions regarding each of the original facets of distributed
cognition that influenced the original scale development. The more parsimonious
scale loaded on a single factor with total variance explained of 71.7%, loadings of
0.72 or greater, and a reliability coefficient alpha of 0.94. For the asynchronous
interaction mode, the items of the CT support for a contextualization strategy scale
also loaded on a single factor with loadings > 0.75, total variance explained 73.9%,
and alpha = 0.95.
Table 5-2: CT Support for a Contextualization Strategy
The information technology used to interact with others helped me to: (1=strongly disagree,
7=strongly agree)
Synchronous Asynchronous
Item ID Item Mean StDev Mean StDev
AUTH1 Identify who made a particular contribution 4.49 1.64 4.69 1.45
AUTH2 Remember specific knowledge contributed
by specific individuals 4.37 1.59 4.58 1.41
TRAV1
Link one contribution with other related
contributions Item Dropped
TRAV2
Identify past contributions relevant to
current ones 4.26 1.53 4.32 1.48
TRAV3
Summarize the topic of any particular
contribution 4.27 1.60 4.21 1.52
TRAV4
Remember details from any particular
contribution
4.39 1.48 4.37 1.50
INDT1
Interweave knowledge from notes, chat,
email, presentations and documents
4.19 1.73 4.32 1.50
INDT2
Revisit the information from the
contributions at a later time
Item Dropped
MULT1
Simultaneously consider multiple
contributions made by others
Item Dropped
MULT2
Make multiple comments using a variety of
communication tools
3.88 1.64 4.27 1.48
EMERG1
Identify when new contributions were
made in the discussion
Item Dropped
EMERG2
Identify new topics of discussion when
they emerged
4.63 1.45 4.46 1.56

110
Other CT Support Items:
While a pre-existing scale was available for the CT support for a
contextualization strategy measure, other CT support items had yet to be identified
and validated. To identify survey items that would capture CT support in use, I
employed a three step strategy from Straub (1989). First, I used Te’eni (2001) to
identify the CT features that would support each of the remaining communication
strategies. Because these items had not been previously validated, I drew on recent
literature that identified features of CT support (Malhotra and Majchrzak 2005) to
identify and further develop a candidate list of features. Malhotra and Majchrzak
provided an exhaustive list; no other detailed decompositions of CT support could be
found in the literature. However, additional items were added based on other studies
of individual knowledge sharing in organizational contexts (Hollingshead et al. 2002;
Malhotra et al. 2001; Markus 2001). Table 5-3 lists the features derived from prior
research.
Table 5-3: CT Support Identified in Prior Research
CT Support Identified From Past Research Reference
Context mechanisms Malhotra and Majchrzak (2005)
Email Malhotra and Majchrzak (2005)
Audio conferencing Malhotra and Majchrzak (2005)
Video conferencing Malhotra and Majchrzak (2005)
Virtual workspaces
- collaborative document editing
- instant messaging
- knowledge repositories
Malhotra and Majchrzak (2005)
Virtual workspaces
- document vault/repositories
Hollingshead et al. (2002)
File sharing Malhotra et al. (2001); Markus (2001)
Intranets Hollingshead et al. (2002)

111
The list of CT support identified in past research was shown to individuals in
CentCo’s Knowledge Management (KM) section as well as individuals taking part in
the study. Each of the items was validated by individuals in KM, who verified that
the CT support on the list were available to CentCo knowledge workers as part of
their suite of CT. Participating individuals then identified which CT support they
used, and when. To keep the list as inclusive as possible, CT support was kept as
long as at least one individual indicated they were used.
Pilot testing confirmed the wording of the CT support items on the survey
was acceptable. The survey for the synchronous interaction mode contained nine CT
support features that were relevant for this research; for the asynchronous interaction
mode, the survey contained seven items. Tables 5-4a and b list the CT support items
from the two surveys.
Table 5-4a: CT Support Items, Synchronous Interaction Mode (n=106)
To what extent did you rely on the following information technology to interact with others? (1=no
extent, 7=great extent)
Item ID Item Min Max Mean St.Dev
TOOL1 Voice conference call 1 7 6.17 1.56
TOOL2 Video conference 1 7 1.72 1.82
TOOL3 Email for file transfer 1 7 3.62 2.17
TOOL4 Email for communication 1 7 3.84 2.21
TOOL5 Instant messaging or other text-based chat 1 7 1.58 1.26
TOOL10 Livelink, teamcenter, or other document
repository for real-time access to shared
documents
1 7 4.45 2.37
TOOL11 A web browser for reference to
information on intranet
1 7 2.61 2.15
TOOL14 Collaborative authorship of meeting notes 1 7 1.77 1.36
TOOL15 Collaborative authorship of presentation,
spreadsheet or other work product
1 7 3.03 2.18

112

Table 5-4b: CT Support Items, Asynchronous Interaction Mode (n=67)
To what extent did you rely on the following information technology to interact with others? (1=no
extent, 7=great extent)
Item ID Item Min Max Mean St.Dev
TOOL2 Video conference 1 7 1.39 1.11
TOOL3 Email for file transfer 1 7 5.12 1.74
TOOL4 Email for communication 1 7 5.46 1.47
TOOL5 Instant messaging or other text-based
chat
1 7 1.53 1.22
TOOL10 Livelink, teamcenter, or other document
repository for real-time access to shared
documents
1 7 4.69 1.92
TOOL11 A web browser for reference to
information on intranet
1 7 2.79 2.03
TOOL15 Collaborative authorship of presentation,
spreadsheet or other work product
1 7 2.57 1.73

To generate CT support constructs with acceptable predictive capability,
items without reasonable variability were discarded. This included items with mean
values above 6.0 (indicating, in effect, that everyone used this feature) and below 2.0
(indicating that the feature was hardly ever used). The grayed features in the
preceding tables identify these features. The remaining features were put into an
exploratory factor analysis to detect underlying relationships between items (Tables
5-5a and b).
Table 5-5a: CT Support Item Loadings, Synchronous Interaction Mode

Component
1 2
TOOL3
.932 .044
TOOL4
.914 .199
TOOL10
.062 .807
TOOL11
.128 .809
TOOL15
.311 .396

113
Table 5-5b: CT Support Item Loadings, Asynchronous Interaction Mode

Component
1 2
TOOL3
.921 .024
TOOL4
.917 .107
TOOL10
.106 .821
TOOL11
.033 .847
TOOL15
.339 .265

Table 5-5a shows the loadings for CT support items in the synchronous
interaction mode after varimax rotation. Table 5-5b shows the loadings for the
asynchronous interaction mode. The items in both loaded onto 2 factors. Tool15 did
not load cleanly on either of the factors; however, for the CT support items there is
no expectation of prior convergence; that is, the items were not meant to reflect any
underlying latent construct. Loading onto the same construct merely reflects
correlation and not a theoretical relationship. Chin et al. (2003) suggest that these
items should be considered ‘formative’ rather than ‘reflective’ indicators (Chin and
Gopal 1995), which means change in each of the indicators value causes change in
the value of a latent variable, rather than change in the latent variable reflecting
change in all the items. The factor loadings gave an indication of the underlying
formative constructs. Table 5-6 shows how the items were grouped into formative
CT support constructs for the communication strategies, consistent with the theory
developed in Chapter 3.

114
CT Support for a Perspective Taking Strategy: I asked individuals the extent to
which they referenced a virtual workspace, document repository or intranet/internet
for access to stored documents. Although CT support for this strategy was only
hypothesized to have an impact in the asynchronous interaction mode, I measured
the item in both the synchronous and asynchronous interaction mode to check for un-
hypothesized influences and alternative explanations.

CT Support for a Testing and Adjusting Strategy: I asked respondents the extent to
which they individual jointly edited presentations, spreadsheets, or other work
products. Although joint editing of work products was hypothesized to influence
sharing diverse knowledge in the synchronous interaction mode only, it was captured
in the asynchronous interaction mode also for the reasons outline above.

CT Support for an Attention Focusing Strategy: To assess CT support for an
attention focusing strategy I asked individuals the extent of their use of email to
transfer files or contact other individuals to communicate a message.
Table 5-6: CT Support Constructs
CT Support for an Attention Focusing Strategy
TOOL3 Email for File Transfer
TOOL4 Email for Communication

CT Support for a Perspective Taking Strategy
TOOL10 Document repository for real-time access to shared documents
TOOL11 Web browser for reference to intranet

CT Support for a Testing and Adjusting Strategy
TOOL15 Collaborative editing of work product

115

5.3.2 Sharing Diverse Knowledge and Reflection on others’ shared knowledge
Sharing Diverse Knowledge: No pre-existing scale could be found that captured
individuals’ perception of sharing diverse knowledge (or sharing). Past research that
has examined individuals’ sharing has employed a textual coding scheme (Suedfeld
et al. 1992). For the individuals in this study, written records of sharing were not
generally available. Even if transcripts had been available, coding diverse
knowledge shared by numerous individuals in the same interaction mode would have
been extremely complex, as the technique was developed to code singly-authored
texts. A self-assessment measure was acceptable given the circumstances.
Drawing on theories of personal creativity in social situations (Amabile 1982;
1988) and integrative complexity (the ability to both generate and integrate diverse
knowledge) (Suedfeld et al. 1992), I developed a 4 item scale asking individuals to
assess their sharing of diverse knowledge (Table 5-7). Since individuals were not
always sure if what they knew was ‘diverse’ (i.e. unique, unknown to others a
priori), they were asked if their contributions were creative, original, novel, and
complex. As part of scale development, I subjected the four items to a factor
analysis using SPSS (using the synchronous interaction mode responses first,
n=106). The four items of the scale in the synchronous interaction mode loaded onto
a single factor with loadings > 0.86, total variance explained of 81.6% and had an
alpha of 0.92. For the asynchronous interaction mode (n=67), the four items also

116
loaded on a single factor with loadings > 0.88, total variance explained of 80.7% and
alpha = 0.92.
Evaluating the scale in both the synchronous and asynchronous interaction
modes gives an added measure of assurance as to the face validity of the items.
However with new measurement scales, construct validity must be developed over
time with multi-methods (John and Benet-Martinez 2000). Confirmatory factor
analysis and discriminant analysis were employed to validate the scale for both
contexts (see chapter 6 for discriminant analysis with PLS Graph). These checks
give a measure of confidence in the scale, with the caveat that it has not been tested
in a variety of situations; generalizability to other situations and cultures may be a
concern (John and Benet-Martinez 2000).

Table 5-7: Sharing Diverse Knowledge
Think about the discussions that ensued. Rate your participation in the discussion on the following
criteria: (1=no extent, 7=great extent)
Synchronous Asynchronous
Item ID Item Mean StDev Mean StDev
DKS1 Creativity of ideas 3.78 1.22 3.79 1.36
DKS2 Originality of ideas 3.79 1.23 3.51 1.26
DKS3 Novel combination of ideas 3.66 1.34 3.45 1.34
DKS4 Complexity of ideas 4.12 1.46 3.79 1.40

Reflection on others’ shared knowledge: Since no previous instrument could be
found that captured reflection on others’ shared knowledge (reflection), I developed
a four item scale to capture the extent to which individuals were mentally open to
receiving others’ knowledge and trying to make sense of it. Items were developed
and validated according to Straub (1989). Pre-testing and piloting proceeded as

117
described above. Items were based in part on sensemaking literature (Weick 1995)
and on distributed cognition (Boland et al. 1994; Te'eni 2001). The items asked
individuals about how they cognitively process diverse knowledge from others and
how that knowledge impacts their understanding (Table 5-8). As with a majority of
the items in the survey, response was measured using a 7 point Likert scale.
One item was dropped as a result of feedback from the pilot testing. I
subjected the remaining three items to a factor analysis, and all three items loaded
onto one factor with loadings > 0.8 giving a total variance explained of 72.8%.
Reliability of the scale using Cronbach’s alpha was an acceptable 0.81. The scale
was similarly analyzed with responses from the asynchronous interaction mode
(n=67). The three items of the scale had loadings > 0.85 and total variance explained
of 80.0%, with alpha = 0.88. While acceptable validity and reliability were obtained,
the same caveat discussed with the sharing diverse knowledge measures applies for
this measure as well concerning the use of new measures.
Table 5-8: Reflection on others’ shared knowledge
Thinking about how I mentally process knowledge contributed by others, their contributions:
(1=strongly disagree, 7=strongly agree)
Synchronous Asynchronous
Item ID Item Mean StDev Mean StDev
KNOWL1 Support/align with my own understanding 5.09 1.17 5.09 1.04
KNOWL2 Help me understand the others’ perspectives on
the issues at hand
5.45 1.00 5.08 1.20
KNOWL3 Help me form a coherent picture of the issues at
hand
5.28 1.10 5.18 1.17

118
5.3.3 Collaboration Outcomes
Innovation: I asked project sponsors to rate individuals on the quality of innovative
work products they created, based on their participation in the virtual collaboration
for innovation rather than any prior expertise or abilities that they might have used. I
cautioned sponsors to only consider the impact of the individual’s collaboration with
others since any particular expertise they brought to the project (based on past
experience) might create a halo effect. Sponsors were interviewed via phone or
email. They were asked to gauge effectiveness using a 7 item scale ranging from
very low impact to great impact. A single item was used since project sponsors
would be rating numerous individuals at one time and use of multiple items was
judged in pre-test as unnecessarily long and repetitive. While response was given
with a single item, sponsors generally had multiple work products created by each
individual with which to form their evaluation.
Responses included the full range of the scale (1-7) with an average of 4.56
(standard deviation 1.50) indicating good variability in response (Table 5-9).
Table 5-9: Innovation
Based on their contribution during the most recent virtual meeting and contribution since the last
meeting until today, to what extent have the following individuals (all members listed) contributed
innovative work products to this project? Contribution includes verbal, written, technical support,
drawings, or any other participation. THIS IS NOT INTENDED TO BE AN ASSESSMENT OF
GENERAL PERFORMANCE OR PARTICIPATION FOR THE PROJECT: (1=no extent, 7=great
extent)
Item ID Item Mean StDev
INDPRF (INDIVIDUAL’S NAME) 4.56 1.50

While sponsors used only a single response to rate each individual
respondent, steps were taken to ensure accurate ratings. First, project sponsors were

119
warned in advance that they would be asked to rate individual innovation, and how
the rating would be structured. Those interviewed via phone were given the
instructions verbally and had the chance to ask clarifying questions. Those who
were sent the ranking task over email were similarly encouraged to reply with
questions. Sponsors were able to take the time to think about their ratings, consult
notes if they desired, and alter their rankings once listed before submitting.

Learning: Vandenbosch and Higgins (1996) developed the five items mental model
building scale which I adopted to assess learning, and reported an internal
consistency of 0.963. Majchrzak et al. (2005a) adopted the Vandenbosch and
Higgins scale and reported reliability of 0.911. The original 5 items scale was used
in the current research (Table 5-10). Learning was assessed in each interaction mode
and averaged on an item by item basis to form a combined learning measure.

Table 5-10: Learning
Knowledge shared during the discussion helped me to: (1=strongly disagree, 7=strongly agree)
Synchronous Asynchronous
Item ID Item Mean StDev Mean StDev
MMB1 Challenge my perspectives 4.73 1.31 4.59 1.35
MMB2 Foster my creativity 4.87 1.33 4.69 1.35
MMB3 Re-orient my thinking 4.70 1.40 4.59 1.37
MMB4 Expand my scope 4.94 1.37 4.87 1.38
MMB5 Question my preconceptions 4.50 1.40 4.45 1.46

Analysis of the distribution of responses for the learning items shows
considerable skewedness to the left. Hair et al. (1998) note that skewedness is
undesirable because the correlation coefficient does not accurately reflect the

120
relationship between the variable and other variables. In these situations,
transformation of the data is warranted (Hair et al. 1998; Hart and Hart 2004).
Figure 5-1 shows the un-transformed probability plot for item 1 of the learning
measure (all of the 5 items were skewed left – see Table 5-11). To correct for
skewedness, item values were squared. Figure 5-2 shows the resultant probability
plot for item 1 and the new skewedness values are shown in Table 5-11. Table 5-11
shows that after transformation the items show an acceptable level of skewedness
(ratio of skewedness/std. error).
Figure 5-1 Probability Plot for Item 1 of Learning Scale
MMB1
7.0 6.0 5.0 4.0 3.0 2.0 1.0
MMB1
Frequency
40
30
20
10
0
Std. Dev = 1.31
Mean = 4.7
N = 106.00

121
Table 5-11: Skewedness – Combined Learning Items
Descriptive Statistics
106 -.758 .235
106 -.938 .235
106 -.670 .235
106 -.866 .235
106 -.697 .235
106 -.125 .235
106 -.355 .235
106 -.145 .235
106 -.322 .235
106 -.080 .235
106
MMB1
MMB2
MMB3
MMB4
MMB5
SQMMB1
SQMMB2
SQMMB3
SQMMB4
SQMMB5
Valid N (listwise)
Statistic Statistic Std. Error
N Skewness

Figure 5-2 Probability Plot for Transformed Item 1 of Learning Scale
SQMMB1
50.0 40.0 30.0 20.0 10.0 0.0
SQMMB1
Frequency
40
30
20
10
0
Std. Dev = 11.27
Mean = 24.0
N = 106.00

5.3.4 Control Variables
Number of Participants: Number of participants was assessed with a single item
based on Valacich et al. (1995), as shown in Table 5-12. Managers at CentCo
advised that the number of individuals assigned to each project had been volatile
several months previous to the survey as they tried to find the best employees for

122
each. Thus, to avoid confusing respondents I asked for the number of individuals
participating in the current meeting. It was thought that to ask simply “how many
individuals participate in your project” would cause some to include current as well
as previous members. The number of participants at the time of the last meeting
gives a more accurate measure for the period of time covered by the study. Average
number of participants was approximately six.

Table 5-12: Number of Participants (n=106)
How many individuals participated in this virtual meeting? (1= 2 or 3, 2=4-6, 3=7-9, 4=10-12,
5=13-15, 6=16-18, 7=19 or More)
Item ID Item Min Max Mean StDev
NUMPART Number of Participants 1 7 2.98 1.26

Number of Participant Locations: Number of separate locations was used as a
measure of the psychological distance component of virtuality (Chidambaram and
Tung 2005; Chudoba et al. 2005). A single item assessed number of locations as
shown in Table 5-13. Average number of locations from which individuals
participated in virtual collaboration for innovation was between 4 and 5.
Table 5-13: Number of Participant Locations (n=106)
From how many separate places did individuals participate in this meeting? (if individuals were in
the same city but were not sitting face to face with each other, count each separately) (1-7or more)
Item ID Item Min Max Mean StDev
LOCALES Number of Locations 1 7 4.44 1.63

123
5.4 Overview: Data Analytic Techniques for Analysis of the Structural Model
In this final section of chapter 5, I introduce the analysis technique used to
verify construct validity and reliability and to test the significance of the structural
paths in the hypothesized model. Partial Least Squares (PLS) analysis (Chin 1998b;
Wold 1980) was employed to evaluate the structural model. PLS is an alternative to
covariance-based structural equation modeling that is a components-based method,
which places minimal demands on sample size and relaxes the assumption of
normality assumed when utilizing other structural equation modeling (SEM)
methods. It is also appropriate to use for complex models with a large number of
hypothesized relationships, as is the case with the current research. PLS is preferable
to SEM for more exploratory work, as is the case with the current research, while
SEM is better suited for confirming existing theory. PLS is a preferred approach
when sample size is low because it is not constrained by the number constructs;
rather, as a rule of thumb, sample size should be ten times the maximum number of
paths leading into any one construct or 10 times the number of items for any
formative construct, whichever is greater. Finally, PLS may be used to model
formative indicators, which other SEM tools cannot. The method has been used in a
number of recent studies of information systems (Agarwal et al. 2000; Bock et al.
2005; Chin et al. 2003; Gray and Meister 2004; Majchrzak et al. 2005a).
PLS analysis is accomplished in two steps. First, PLS simultaneously
estimates the structural paths and measurements paths in the model. PLS

124
incorporates each indicator separately and weights indicators based on their
influence on the latent construct, increasing reliability and robustness of the
prediction (Chin et al. 2003). Path coefficients from PLS output can be interpreted
in the same manner as standardized regression coefficients from multiple regression.
PLS reports R
2
values for endogenous constructs which are likewise interpreted as
with multiple regression and provide an indication of the goodness of fit for the
measurement model. Second, re-sampling is accomplished through the use of
bootstrapping, which generates means and standard deviations that are then used to
assess the significance of the model path coefficients. This analysis also generates
composite reliability and average variance extracted for the model constructs, which
allow for assessment of validity.

5.5 Summary
In this chapter I described the research site, the survey instrument, including
items used for the first round survey (for the synchronous interaction mode), the
second round survey (for the asynchronous interaction mode), and the project
sponsor survey (rating of extent individuals build innovative work products). I
described the sample selection and surveying process and presented descriptive
statistics about the responding individuals. Finally, I reviewed the analysis
technique, partial least squares, which will be employed to evaluate the measurement
model and structural model, testing the hypotheses developed in chapter 3.

125
CHAPTER 6
ANALYSIS RESULTS
Chapter 6 presents the results of the survey. Section 6.1 presents analysis of
the measurement model: including convergent and discriminant validity using output
and techniques with PLS; test for common method variance; test for appropriateness
of individual level response for further analysis; and test for non-response bias.
Section 6.2 presents the structural model and testing of the hypotheses. Section 6.3
presents a summary of the results.

6.1 Construct Validity, Reliability, and Preliminary Data Analysis
In this section I discuss the inter-construct correlations showing the relative
independence between variables in the model. I present analysis of construct
validity, reliability, based on output from PLS. Third, I present two methods for
analyzing discriminant validity.
It is also necessary to demonstrate that the data are appropriate for individual
level analysis; that is, for those individuals from the same project, inter-group
variability of response is no lower than between group variability. In section 6.1.4
the James index is employed to demonstrate that group level effects on individual
responses to the survey are minimal; otherwise, mixed level models should be used.
This influence is undesirable in an individual level analysis.

126
I also check for common method variance which is a systematic method bias
which, if present, threatens validity by providing an alternative explanation for the
relationships in the model. Common method variance is possible when predictor and
outcome variables are measured with a common instrument. I use a partial-
correlation approach to detect if common method variance if present, has a
significant effect. Finally, I demonstrate that non-response bias, a systematic bias
based on self-categorization of responders vs. non-responders, has no apparent
impact on the results.

6.1.1 Inter-Construct Correlations
Tables 6-1 and 6-2 show the inter-construct correlations calculated in PLS
Graph. High correlations (generally .6 or above) between predictor and outcome
variables indicate strong predictive power (Hinkle et al. 2003); whereas low
correlations (.4 or lower) indicate a predictor has little influence on an outcome
variable. Analysis of the correlations indicates that correlations between predictor
and outcomes variables are in the acceptable range of .4-.6, indicating moderate
correlation.
On the other hand, it is desirable that correlations between predictor variables
are low, since high correlations between predictor variables indicate the variables do
not have independent influence on the outcome variables, reducing the overall
predictive power of the model. Sheehan et al. (2004) suggest that correlation above

127
.6 between predictor variables indicates an unacceptably strong relationship and thus
the variables would not be considered as independent predictors; correlations
between .4 and .6 are moderately strong, and correlations below .4 are considered
weak (and thus desirable). The tables show that correlations between predictor
variables are below .4, indicating acceptable independence of the model constructs.
Table 6-1: Correlation of Latent Constructs (Synchronous Interaction Mode, n=106). Diagonal
elements are the Average Variance Extracted (AVE) estimated for each construct. Off-diagonal
elements are the correlations among the constructs. For discriminant validity, the square root of
the AVE (i.e., the diagonal elements) should be larger than the off-diagonal elements (Chin
1998b).
CONSTRUCT 1234567 8 9 10
1 INNOVATION 1
2
REFLECTION ON
OTHERS’ SHARED
KNOWLEDGE 0.296 0.734
3
SHARING DIVERSE
KNOWLEDGE 0.2190.519 0.816
4
CT SUPPORT FOR A
CONTEXTUALIZATION
STRATEGY 0.181 0.49 0.288 0.719
5
CT SUPPORT FOR A
TESTING AND
ADJUSTING STRATEGY 0.24 0.252 0.438 0.257 1
6
NUMBER OF
PARTICIPANTS -0.1-0.0390.0840.127-0.062 1
7
CT SUPPORT FOR A
PERSPECTIVE TAKING
STRATEGY 0.18 0.224 0.21 0.283 0.346 -0.035 1
8
CT SUPPORT FOR AN
ATTENTION
FOCUSING STRATEGY 0.016 0.071 0.169 0.221 0.291 -0.042 0.214 1
9
NUMBER OF
LOCATIONS -0.0080.01-0.013-0.0660.0210.307-0.043 0.025 1
10 LEARNING 0.117 0.428 0.524 0.242 0.247 0.079 0.196 0.208 0.098 0.779

128
Table 6-2: Correlation of Latent Constructs (Asynchronous Interaction Mode, n=67). Diagonal
elements are the Average Variance Extracted (AVE) estimated for each construct. Off-diagonal
elements are the correlations among the constructs. For discriminant validity, the square root of
the AVE (i.e., the diagonal elements) should be larger than the off-diagonal elements (Chin
1998b).
CONSTRUCT 123456 7 8 9
1 INNOVATION 1
2
REFLECTION ON
OTHERS’ SHARED
KNOWLEDGE 0.494 0.827
3
SHARING DIVERSE
KNOWLEDGE 0.0220.326 0.807
4
CT SUPPORT FOR AN
ATTENTION
FOCUSING STRATEGY -0.035 0.389 0.437 1
5
CT SUPPORT FOR A
CONTEXTUALIZATION
STRATEGY 0.1710.5870.3460.434 0.733
6
NUMBER OF
PARTICIPANTS -0.055-0.2070.3490.144-0.048 1
7
CT SUPPORT FOR A
PERSPECTIVE TAKING
STRATEGY 0.1230.3890.0670.1270.135-0.134 1
8 LEARNING 0.21 0.514 0.403 0.376 0.454 0.028 0.173 0.843
9
NUMBER OF
LOCATIONS 0.052-0.1130.14-0.09-0.320.495-0.091 -0.1 1
10
CT SUPPORT FOR A
TESTING AND
ADJUSTING
STRATEGY 0.078 0.352 0.072 0.324 0.358 -0.113 0.195 0.298 -0.1

6.1.2 Analysis of Construct Correlation, Convergent Validity and Reliability
Uni-dimensionality, an important assumption in PLS analysis (Gefen and
Straub 2005), cannot be completely determined in PLS (Gefen 2003). However,
construct validity can be demonstrated by verifying convergent and discriminant
validity (Churchill 1979; Straub et al. 2004). Convergent validity is indicated when
items are shown to correlate strongly with the intended construct; discriminant
validity is indicated when items are shown to correlate weakly with other constructs
while strongly with the intended construct. In PLS Graph, convergent validity,
applicable where multi-item reflective scales were used, was assessed by verifying

129
the item loadings were greater than 0.6 as suggested by Hair et al. (1998) or 0.7 as
suggested by Fornell and Larcker (1981) and recommended by Hulland (1999). The
more conservative measure has been applied consistently in IS research (c.f. Agarwal
and Karahanna (2000), Gray and Meister (2004) and Saraf and Langdon (2006)).
Tables 6-3 and 6-4 show all loadings for reflective items were greater than 0.7 for
the synchronous and asynchronous interaction modes. Single items and formative
scales are not listed. In addition, convergent validity for reflective items is indicated
when the average variance extracted (AVE) for the items exceeds 0.5 (Chin 1998b).
AVE shows the ratio of measurement item variance relative to measurement error
(Gefen and Straub 2005). Finally, as a measure of internal consistency, composite
reliability is shown for the reflective constructs. Composite reliability is interpreted
similarly to Cronbach’s alpha with 0.7 considered an acceptable lower bound for
alpha measures of reliability (Nunnally 1978) and internal consistency (Fornell and
Larcker 1981). All measures of internal consistency and AVE are above the
suggested values, indicating good convergent validity and reliability.

130
Table 6-3: Item Loadings, AVE, and Composite Reliability for Multi-item Reflective Constructs
(Single-Item and formative constructs not listed; Synchronous interaction mode, n=106)
Construct # of
items
Item
Loadings
AVE Composite
Reliability
Reflection on others’ shared
knowledge
3 .804, .846,
.917
.734 .892
Sharing Diverse Knowledge 4 .926, .902,
.925, .858
.816 .947
CT Support for a Contextualization
Strategy
8 .829, .861,
.864, .930,
.906, .853,
.743, .782
.719 .953
Learning 5 .881, .886,
.932, .879,
.811
.772 .944

Table 6-4: Item Loadings, AVE, and Composite Reliability for Multi-Item Reflective Constructs
(Single-Item and formative constructs not listed; Asynchronous interaction mode, n=67)
Construct # of
items
Item
Loadings
AVE Composite
Reliability
Reflection on others’ shared
knowledge
3 .885, .912,
.931
.827 .935
Sharing Diverse Knowledge 4 .889, .907,
.882, .914
.807 .943
CT Support for a Contextualization
Strategy
8 .704, .801,
.897, .932,
.947, .833,
.869, .843
.733 .956
Learning 5 .927, .905,
.942, .900,
.791
.800 .952

6.1.3 Discriminant Validity
Discriminant validity is assessed in PLS Graph using a two-step process
(Gefen and Straub 2005). First, the square root of the AVE’s for the reflective
constructs are compared to the inter-construct correlations and must be “an order of
magnitude” greater than the correlations (Gefen and Straub 2005) which means, for
example, 0.7 compared to 0.6, and not 0.7 compared to 0.07. AVE is not interpreted

131
for formative constructs. In all cases, AVE was at least an order of magnitude
greater than the correlations (see Table 6-1 for synchronous interaction mode, 6-2 for
the asynchronous interaction mode); if AVE is greater than the correlations then the
square root of AVE is automatically greater – previous studies have reported either
statistic. The tables show the comparison of AVE to correlations (note, AVE
appears as the diagonal elements. For single-item measures AVE is not interpreted
and the value listed in the table is 1.0).
The second step of the process to assess discriminant validity is to check that
item loadings correlate highly to their intended construct, and that they do not
correlate highly with any other constructs (Gefen and Straub 2005). Cross loadings
for each item (calculated by weighting standardized items from the outer model in
PLS Graph) were appended into a table with the cross loadings for the model
constructs and then correlated using SPSS. Results indicated that for all constructs,
correlations were high (> 0.8) for all items with their intended constructs, and low (<
0.6) for the other constructs, which indicates acceptable discriminant validity.
Tables 6-5 and 6-6 show the cross-correlations (with p values to show significance)
for the synchronous and asynchronous interaction mode, respectively.

132
Table 6-5: Cross Loadings, Synchronous Interaction Mode
Item /
p= Learning
Reflection on
others’
shared
knowledge
Sharing
Diverse
Knowledge
CT Support
for a
Contextualiza
tion Strategy
CT Support
for a
Perspective
Taking
Strategy
CT Support
for a Testing
and Adjusting
Strategy
CT Support
for an
Attention
Focusing
Strategy
MMB1 0.866 0.369 0.381 0.246 0.143 0.163 0.133
MMB2 0.897 0.351 0.612 0.135 0.171 0.28 0.2
MMB3 0.936 0.472 0.544 0.257 0.155 0.302 0.257
MMB4 0.895 0.348 0.504 0.224 0.221 0.239 0.195
MMB5 0.806 0.319 0.254 0.192 0.1 0.07 0.106
KNOWL1 0.282 0.806 0.338 0.474 0.186 0.222 0.053
KNOWL2 0.407 0.85 0.492 0.335 0.217 0.213 0.051
KNOWL3 0.398 0.912 0.49 0.454 0.18 0.215 0.077
DKS1 0.48 0.528 0.922 0.292 0.229 0.397 0.178
DKS2 0.442 0.462 0.897 0.311 0.135 0.367 0.169
DKS3 0.557 0.447 0.928 0.272 0.177 0.425 0.162
DKS4 0.496 0.43 0.865 0.153 0.176 0.394 0.095
AUTH1 0.167 0.401 0.099 0.829 0.306 0.151 0.206
AUTH2 0.272 0.422 0.25 0.861 0.257 0.216 0.23
TRAV2 0.185 0.387 0.229 0.864 0.174 0.149 0.184
TRAV3 0.192 0.424 0.224 0.93 0.233 0.17 0.147
TRAV4 0.19 0.389 0.217 0.906 0.318 0.238 0.193
INDT1 0.202 0.462 0.296 0.852 0.284 0.279 0.142
MULT2 0.263 0.348 0.359 0.743 0.278 0.336 0.259
EMERG2 0.141 0.457 0.26 0.782 0.107 0.211 0.16
TOOL10 0.201 0.225 0.21 0.283 0.995 0.346 0.214
TOOL15 0.259 0.252 0.439 0.257 0.342 1 0.291
TOOL4 0.213 0.071 0.167 0.221 0.186 0.291 1

133
Table 6-6: Cross Loadings, Asynchronous Interaction Mode
Learning
Reflection on
others’
shared
knowledge
Sharing
Diverse
Knowledge
CT Support
for a
Contextualiza
tion Strategy
CT Support
for a
Perspective
Taking
Strategy
CT Support
for an
Attention
Focusing
Strategy
MMB1 0.941 0.418 0.339 0.497 0.187 0.43
MMB2 0.926 0.527 0.349 0.505 0.292 0.41
MMB3 0.964 0.497 0.372 0.461 0.217 0.375
MMB4 0.916 0.515 0.505 0.572 0.15 0.448
MMB5 0.86 0.376 0.281 0.336 0.284 0.267
KNWL1 0.413 0.881 0.341 0.481 0.313 0.368
KNWL2 0.523 0.875 0.269 0.518 0.21 0.302
KNWL3 0.446 0.925 0.338 0.615 0.295 0.411
DKS1 0.437 0.337 0.885 0.352 0.091 0.392
DKS2 0.391 0.339 0.911 0.289 0.116 0.379
DKS3 0.381 0.199 0.878 0.162 0.061 0.327
DKS4 0.452 0.381 0.918 0.316 0.071 0.465
AUTH1 0.404 0.427 0.09 0.75 0.12 0.268
AUTH2 0.383 0.498 0.244 0.821 0.122 0.387
TRAV2 0.443 0.52 0.276 0.884 0.154 0.35
TRAV3 0.459 0.545 0.307 0.928 0.108 0.478
TRAV4 0.537 0.538 0.409 0.933 0.136 0.453
INDT1 0.548 0.519 0.391 0.796 0.097 0.246
MULT2 0.364 0.516 0.205 0.877 0.277 0.408
EMERG2 0.44 0.578 0.219 0.873 0.166 0.381
TOOL10 0.248 0.306 0.096 0.169 0.999 0.194
TOOL4 0.424 0.406 0.439 0.435 0.19 1
* p<.05, ** p<.01, *** p<.001

6.1.4 Appropriateness of Individual Level Analysis of Participant Data Nested in
Groups
Table 6-7 shows that some individuals provided the only response for the
project; for others, several of the individuals reported their interaction for the same
project. Relevant intergroup characteristics are listed.

134
Table 6-7: Individual Responses Per Project
Project #
# of Indiv.
Responses
# of Separate
Locations
Total
Number of
Indiv.
Participants
(1-7 scale)
Avg. Prior
Familiarity
With Other
Individuals
(1-7 scale)
Frequency
of Face to
Face
Interaction
(1-7 scale)
Stage of
Project
Lifecycle

(1-4 scale)
1 2 2 5 2.5 3.5 2
6 4 3 2.5 2.75 4.5 3
7 4 3 2.5 2.25 2 3
8 1 1 1 5 1 4
10 1 1 3 1 2 4
11 4 2 2 1.33 2.67 3
12 2 2 2.5 1 4.5 4
13 1 1 2 7 1 4
15 3 3 3 3.25 1.75 1
16 1 1 3 5 1 1
17 3 3 3 2.33 1 3
18 2 2 3.5 2.5 1.5 4
20 2 2 2 2 1 2
21 1 1 2 1 1 4
22 1 1 3 3 2 3
23 2 2 3 1 2 4
24 4 2 2 2.5 1.75 3
26 3 3 2.33 1 1.67 3
27 2 2 2 1.5 1 4
29 2 2 2 1 1 4
30 3 2 2.67 3.33 3 3
1 2 1 2.5 1.5 1 4
32 2 2 2 1 3.5 4
35 4 3 2.5 4.75 2.75 4
36 4 3 2 1.67 2.33 4
38 2 2 2 1.5 4.5 4
40 3 2 2.67 5 3 4
41 4 1 5.5 5.5 3.75 3
44 1 1 2 2 2 3
46 1 1 3 3 2 4
48 2 2 2 1 1.5 2
49 1 1 7 4 2 2
51 2 2 3 2 4 3
55 1 1 3 2 7 3
56 12 6 3.91 1.45 2.18 3
57 9 4 4.17 2 4 3
64 1 1 2 2 1 2
69 4 2 2 3.33 3 3
71 3 1 4.33 4.67 2.33 3
73 3 2 2.67 4.33 1.33 2
74 6 4 3.33 2.67 2.5 3
76 2 2 3 1.5 1 4
84 1 1 3 6 2 2

135
Table 6-7, Continued
Project #
# of Indiv.
Responses
# of Separate
Locations
Total
Number of
Indiv.
Participants
(1-7 scale)
Avg. Prior
Familiarity
With Other
Individuals
(1-7 scale)
Frequency
of Face to
Face
Interaction
(1-7 scale)
Stage of
Project
Lifecycle

(1-4 scale)
85 1 1 4 7 4 2
86 1 1 4 3 1 2

When multiple individuals report on their experience from the same project,
the possibility exists that the responses are not independent. Typically measures of
inter-rater agreement are utilized to substantiate claims that individual level data may
be aggregated to a group level for analysis (James et al. 1984; Kashy and Kenny
2000; Lindell et al. 1999). However when an individual level effect is hypothesized
but respondents are nested in groups, the interrater agreement measures can instead
show the relative independence of the responses. That is, low (below accepted
threshold) interrater agreement between members of the same group indicates
minimal influence of group level factors common to the individuals. If agreement is
low analysis may continue at the individual level – otherwise, hierarchical models
which model group and individual level effects must be employed. Hierarchical or
mixed-level models use a two step process to model the mutual influence of
individual and group effects. However, Kashy and Kenny (2000) state that
hierarchical models are only appropriate when group size is greater than the number
of predictors at the group level plus two. Thus due to data limitations a hierarchical
model was not permissible.

136
In addition, since the perception of CT support are individual level
phenomena according to prior theory (Majchrzak et al. 2005a; Majchrzak et al.
2005b) analysis should be conducted at the individual level. Thus to verify
appropriateness of individual level analysis, inter-rater agreement was calculated.
Tables 6-8 and 6-9 show the modified James index value calculated for each
construct in the synchronous and asynchronous interaction mode. The James index
r
wg
(j) (James et al. 1984) is calculated by pooling within-group variability across j
items of each scale divided by a factor based on the number of response categories
(e.g. a factor of 4 for a 7-point response scale), and multiplied by the number of
items in the scale. Recently Lindell et al. (1999) suggested a modification to the
James et al. formula that corrects for inadmissibility when the shared variance among
raters is greater than the variance of the uniform distribution of the response interval.
The corrected index r
*
wg
is calculated as:
2
2
*
1
EU
x
wg
s
s
r
− =
where
2
x
s is the within-group shared variance pooled across items and
2
EU
s is the
variance of the uniform distribution across the number of response categories.
Lindell et al. (1999) caution that the original and the modified index have a slight
tendency to overstate inter-rater agreement; even so, as Tables 6-8 and 6-9 show, all
measures of inter-rater agreement are below the suggested threshold of 0.7, as
suggested by George (1990). IS researchers (Majchrzak et al. 2005b; Yoo and Alavi

137
2001) have consistently applied the .7 standard, although Janz et al. (1997) suggest
.87. Innovation is included, although it was rated by each individual’s project
leader, to demonstrate no project-level bias is present.

Table 6-8: Inter-Group Agreement, Synchronous Interaction Mode
Scale – Synchronous Interaction Mode James Index
Reflection on others’ shared knowledge .60
Sharing Diverse Knowledge .69
CT Support for an Attention Focusing Strategy .15
CT Support for a Perspective Taking Strategy .00
CT Support for a Testing and Adjusting Strategy .00
CT Support for a Contextualization Strategy .52
Innovation .44
Learning .63

Table 6-9: Inter-Group Agreement, Asynchronous Interaction Mode
Scale – Asynchronous Interaction Mode James Index
Reflection on others’ shared knowledge .68
Sharing Diverse Knowledge .61
CT Support for an Attention Focusing Strategy .54
CT Support for a Perspective Taking Strategy .05
CT Support for a Contextualization Strategy .44
Learning .65

As the remaining constructs demonstrated sufficiently low inter-rater
agreement according to the guidelines listed above, the analysis proceeded at the
individual respondent level.

6.1.5 Assessment of Common Method Variance
Because the CT support constructs, controls, sharing diverse knowledge,
reflection on others’ shared knowledge, and learning were all captured with a single
instrument for each point of time, there is a possibility of common method variance

138
or CMV (Campbell and Fiske 1959). CMV results when variability in measured
items can be accounted for systematically by commonality of instrument or method.
In situations where CMV might be present, it is necessary to test if CMV accounts
for shared variance among variables.
In the past, common method variance was typically investigated using the
Harmon 1 factor test (Lindell and Whitney 2001). This involved putting all items
into an exploratory factor analysis and demonstrating that one factor did not account
for all (or a substantial part) of the variance. More recently, confirmatory factor
analysis has been used to do the same type of test. However, Podsakoff et al. (2003)
recommend against this technique, with the logic that it is not unreasonable to
assume that items would load on more than one factor in a factor analysis even with
CMV present. Instead, Podsakoff et al. recommend a number of other techniques;
each has its advantages and disadvantages. The best approach, they conclude, is a
multi-method factor approach, in which a number of latent sources of method
variance are modeled using structural equation modeling with the manifest variables
and the influence of these latent sources factored out. Alternatively, they
recommend single method factor approaches. Unfortunately, these methods rely on
structural equation modeling, thus they require large sample sizes and certain
assumptions about the data, such as normality.
When SEM methods are not feasible, Podsakoff et al. recommend the partial
correlation approach, which is explained in detail by Lindell and Whitney (2001).

139
They note it has limitations, most notably that it does not distinguish method
variance at the measurement level from the construct level, and that it only can
account for one source of method error at a time. However, these criticisms are more
of an issue with trying to determine and controlling for the source of method
variance (of which, they note, there are several types) rather than merely identifying
the presence of, and impact of method variance.
Ideally, to use the partial correlation approach the researcher should include a
scale in the survey that is theoretically unrelated to the other variables of interest
(such as liking for cheese) such that the correlation between the item and the other
variables could reasonably be expected to be zero. If no such scale is included, the
smallest correlation among the manifest variables of interest is used in place of this
correlation. The rationale is that if method variance exists, it would at its maximum
be equal to the smallest correlation, since correcting for method variance would
reduce the smallest correlation to zero and no lower. Following Lindell and
Whitney, the dis-attenuated partial correlation (which controls for reliability of the
construct) is:
⎟
⎟
⎠
⎞
⎜
⎜
⎝
⎛
−
−
=
s ii
s Yi
YiM
r r
r r
r ˆ where r
Yi
is the correlation coefficient suspected of being corrupted
by CMV, r
s
is the smallest of all the correlations in r
ij
, and r
ii
is the reliability of the
Y predictor variable. Significance of r-hat is calculated using the standard formula
for t,

140
3
) 1 (
2
3 , 2 /
−
−
=
−
N
r
r
t
YiM
YiM
N α

Significance of the dis-attenuated partial correlation is compared to the original
correlations. If any of the zero order correlations that were significant before remain
significant, this suggests that the results cannot be accounted for by CMV.
The partial correlation method was used to test for CMV for the constructs in
the first and second round surveys. Control variables were only measured on the
first-round survey thus are only assessed with the synchronous interaction mode.
Results for both the synchronous interaction mode and asynchronous interaction
mode indicate that CMV did not have a significant influence on any of the
relationships in the model. None of the inter-construct correlations that were
significant before controlling for CMV became non-significant. Tables 6-10 through
6-17 show the correlations and t-values for the synchronous and asynchronous
interaction mode.

141
Table 6-10: Inter-Construct Correlations, Synchronous Interaction Mode
Partial Correlation Approach for Assessing Common Method Variance, Synchronous Interaction Mode
(n=106)
Inter-Construct Correlations
CONSTRUCT 1 2 3456 78 9
Reliab
-ility
1 INNOVATION 1.000 1.000
2
REFLECTION ON OTHERS’
SHARED KNOWLEDGE 0.296 1.000 0.892
3
SHARING DIVERSE
KNOWLEDGE 0.219 0.519 1.000 0.947
4
CT SUPPORT FOR A
CONTEXTUALIZATION
STRATEGY 0.1810.4900.288 1.000 0.953
5
CT SUPPORT FOR A
TESTING AND ADJUSTING
STRATEGY 0.240 0.252 0.438 0.257 1.000 1.000
6
NUMBER OF
PARTICIPANTS -0.100-0.0390.0840.127-0.062 1.000 1.000
7
CT SUPPORT FOR A
PERSPECTIVE TAKING
STRATEGY 0.1800.2240.2100.2830.346-0.035 1.000 1.000
8
CT SUPPORT FOR AN
ATTENTION FOCUSING
STRATEGY 0.016 0.071 0.169 0.221 0.291 -0.042 0.214 1.000 1.000
9
BUILD NEW MENTAL
MODELS 0.1170.4280.5240.2420.2470.0790.1960.208 1.000 0.946

Table 6-11: Dis-Attenuated Partial Correlations, Synchronous Interaction Mode

Partial Correlation Approach for Assessing Common Method Variance, Synchronous Interaction
Mode (n=106)
Dis-attenuated Partial Correlations
CONSTRUCT 12 34 56 7 8 9
1 INNOVATION 1.000
2
REFLECTION ON OTHERS’
SHARED KNOWLEDGE 0.290 1.000
3
SHARING DIVERSE
KNOWLEDGE 0.213 0.577 1.000
4
CT SUPPORT FOR A
CONTEXTUALIZATION
STRATEGY 0.1740.5440.294 1.000
5
CT SUPPORTFOR A TESTING
AND ADJUSTING STRATEGY 0.234 0.274 0.455 0.215 1.000
6 NUMBER OF PARTICIPANTS -0.109 -0.056 0.076 0.069 -0.085 1.000
7
CT SUPPORT FOR A
PERSPECTIVE TAKING
STRATEGY 0.1730.2430.2110.2450.332-0.073 1.000
8
CT SUPPORT FOR AN
ATTENTION FOCUSING
STRATEGY 0.008 0.069 0.167 0.175 0.276 -0.080 0.179 1.000
9 LEARNING 0.110 0.474 0.547 0.198 0.231 0.046 0.160 0.188 1.000

142
Table 6-12: T-Statistics, Inter-Construct Correlations, Synchronous Interaction Mode
Partial Correlation Approach for Assessing Common Method Variance, Synchronous Interaction
Mode (n=106)
Inter-Construct Correlations
T-Stats 1 2 3 4 5 6 7 8
1
2 3.145
3 2.278 6.162
4 1.868 5.7053.052
5 2.509 2.643 4.945 2.699
6 -1.020 -0.396 0.856 1.299 -0.630
7 1.857 2.333 2.180 2.995 3.743 -0.355
8 0.162 0.722 1.740 2.300 3.087 -0.427 2.223
9 1.196 4.806 6.244 2.531 2.587 0.804 2.029 0.999

Table 6-13: T-Statistics, Dis-Attenuated Partial Correlations, Synchronous Interaction Mode
Partial Correlation Approach for Assessing Common Method Variance,
Synchronous Interaction Mode (n=106)
Dis-attenuated Partial Correlations
T-Stats 1 2 3 4 5 6 7 8
1
2 3.079
3 2.209 7.172
4 1.797 6.584 3.127
5 2.441 2.896 5.186 2.238
6 -1.112 -0.565 0.774 0.700 -0.864
7 1.787 2.538 2.190 2.561 3.572 -0.738
8 0.082 0.704 1.719 1.801 2.912 -0.812 1.843
9 1.122 5.462 6.633 2.055 2.408 0.463 1.644 0.000

143
Table 6-14: Inter-Construct Correlations, Asynchronous Interaction Mode
Partial Correlation Approach for Assessing Common Method Variance, Asynchronous
Interaction Mode (n=67)
Inter-Construct Correlations
CONSTRUCT 1 2 3 4 5 6Reliability
1
REFLECTION ON OTHERS’
SHARED KNOWLEDGE 1.000 0.935
2
SHARING DIVERSE
KNOWLEDGE 0.326 1.000 0.943
3
CT SUPPORT FOR AN
ATTENTION FOCUSING
STRATEGY 0.3890.437 1.000 1.000
4
CT SUPPORT FOR A
CONTEXTUALIZATION
STRATEGY 0.5870.3460.434 1.000 0.956
5
CT SUPPORT FOR A
PERSPECTIVE TAKING
STRATEGY 0.3890.0670.1270.135 1.000 1.000
6 LEARNING 0.514 0.403 0.376 0.454 0.347 1.000
.964

Table 6-15: Dis-Attenuated Partial Correlations, Asynchronous Interaction Mode
Partial Correlation Approach for Assessing Common Method Variance, Asynchronous Interaction Mode (n=67)
Dis-attenuated Partial Correlations
CONSTRUCT 1 2 3 4 5 6
1 REFLECTION ON OTHERS’ SHARED KNOWLEDGE
1.000
2 SHARING DIVERSE KNOWLEDGE 0.322 1.000
3
CT SUPPORT FOR AN ATTENTION FOCUSING
STRATEGY
0.392 0.422 1.000
4
CT SUPPORT FOR A CONTEXTUALIZATION
STRATEGY
0.612 0.318 0.374 1.000
5
CT SUPPORT FOR A PERSPECTIVE TAKING
STRATEGY
0.392 0.000 0.034 0.111 1.000
6 LEARNING
0.531 0.384 0.310 0.456 0.165 1.000

Table 6-16: T-Statistics, Inter-Construct Correlations, Asynchronous Interaction Mode

Partial Correlation Approach for Assessing Common Method Variance, Asynchronous
Interaction Mode (n=67)
Inter-Construct Correlations
T-Stats 1 2 3 4 5
1
2 2.759
3 3.378 3.887
4 5.800 2.950 3.854
5 3.378 0.537 1.024 1.090
6 4.794 3.523 3.246 4.076 1.405

144
Table 6-17: T-Statistics, Dis-Attenuated Partial Correlations, Asynchronous Interaction Mode
Partial Correlation Approach for Assessing Common Method Variance, Asynchronous
Interaction Mode (n=67)
Dis-attenuated Partial Correlations
T-Stats 1 2 3 4 5
1
2 2.719
3 3.409 3.728
4 6.198 2.688 3.225
5 3.409 0.000 0.274 0.890
6 5.016 3.323 2.606 4.100 1.338

6.1.6 Questions of Non-Response Bias in Survey Responses
Before investigating further the responses of those individuals who
participated in the study, it is necessary to show that no systematic response bias
exists in the sample (Sivo et al. 2006). A number of steps were taken to avoid non-
response bias. First, a pre-survey announcement was sent to all eligible individuals
by the manager in charge of the initiative. The announcement informed individuals
of the purpose of the study, the organization’s commitment to the survey, reassured
individuals their responses would be anonymous, and assured them that participation
was voluntary (Schaefer and Dillman 1998). Having the manager send the email in
conjunction with the researcher helped demonstrate organizational cooperation,
which can reduce non-response (Dillman 1999; Sivo et al. 2006). Lastly, follow-up
notices were sent to project sponsors as recommended by Dillman (1999) asking
them again to pass on the request for participation to the individuals on the project.
The follow-up notices contained links to the web survey as well as an attached word
document format of the survey.

145
The only data available about non-responders to the first survey is project
sponsor ratings of innovation for each individual. Therefore, using analysis of
variance I compared this rating of the 120 respondents to the first survey to the 177
who did not respond. Due to missing data, 30 of the non-responders were not rated,
leaving a non-respondent sample size of 147 available for the ANOVA. Table 6-18
shows the results. Based on results of the innovation measure, there was no
significant difference between respondents to the survey and non-respondents.

Table 6-18: Comparison of Innovation for Respondents vs. Non-respondents

1=response,
0=no response N Mean
Std.
Deviation
Std. Error
Mean t
Sig. (2-
tailed)
0
147 4.544 1.356 .112
Innovation
1
120 4.492 1.484 .135
.299 .765

As noted previously, 67 usable responses to the follow-up survey were
collected. Thus of the 106 respondents to the first round survey concerning the
synchronous interaction mode whose responses were used in the analysis, 39 did not
provide information regarding the asynchronous interaction mode. To ensure no
systematic bias was present concerning non-response to the second round survey, I
compared responders to non-responders in a similar fashion to the first survey. Since
more data were available, responders/non-responders were compared on a variety of
measures, including innovation, learning, sharing diverse knowledge, reflection on
others’ shared knowledge, and CT support variables. ANOVA results are shown in
Table 6-19. The table shows there are no significant differences between the two
sub-samples.

146
Table 6-19: Comparison of Respondents to Non-Respondents for Second Round Survey

1=1st round
only, 0=both
rounds N Mean
Std.
Deviation
Std.
Error
Mean t
Sig (2-
tailed)
1
39 4.436 1.334 .214
Innovation
0
67 4.627 1.594 .195
-.661 .510
1
39 4.759 1.182 .189
Learning

0
67 4.74 1.224 .15
.077 .938
1
39 3.712 1.244 .199
Sharing
Diverse
Knowledge

0
67 3.914 1.149 .140
-.831 .408
1
39 5.06 1.017 .163
Reflection on
others’ shared
knowledge

0
67 5.284 .886 .108
-1.144 .256
1
39 3.08 2.082 .333
CT Support for
a Testing and
adjusting
Strategy

0
67 3.00 2.25 .275
.178 .859
1
39 4.513 1.2166 .195
CT Support for
a Contextual-
ization
Strategy

0
67 4.192 1.4025 .171
1.236 .220

The non-significant results from the analyses of variance do not positively
rule out non-response bias. However, in the absence of any clear evidence of bias
there is reason to believe that any non-response bias, if present, does not significantly
influence the relationships among variables in the study.

6.2 Analysis of the Structural Model
This section reports results from PLS analysis of the structural model.
Analysis proceeded with the full panel of 67 respondents. Since PLS does not allow
for missing values, only the 67 cases in which full data were available were included
in this subsequent analysis. After generating 500 samples using the bootstrap

147
technique, path coefficients were re-estimated along with means and standard
deviations allowing for significance testing (Chin et al. 2003). Figure 6.1 shows the
results for the combined model including the synchronous and asynchronous
interaction mode. As the figure shows, all of the paths were significant except for
the path from sharing diverse knowledge (asynchronous interaction mode) to
learning. Thus, hypothesis 1 was only partially supported; all other hypotheses were
fully supported. For the control variables, number of participants and number of
participant locations had no impact on either sharing diverse knowledge or reflection
on others in the synchronous interaction mode. In the asynchronous interaction
mode, number of participants had a significant positive impact on sharing diverse
knowledge and a significant negative impact on reflection on others’ shared
knowledge. There was no significant impact of the number of participant locations.
Overall the combined model explained 38.9% of the variance in build new mental
models and 31.5% of the variance in build innovative work products.

148
Figure 6-1: Results of the Structural Model
Input Process Outcomes
CT Support for
Testing and
Adjusting Strategy
Sharing Diverse
Knowledge
Innovation
Sharing Diverse
Knowledge
Reflection on Others’
Shared Knowledge
CT Support for
Contextualization
Strategy
CT Support for
Attention Focusing
Strategy
CT Support for
Contextualization
Strategy
CT Support for
Perspective Taking
Strategy
0.524***
0.325**
Reflection on Others’
Shared Knowledge
0.410***
0.458***
0.274**
Synchronous Interaction Mode
Asynchronous Interaction Mode
0.392***
0.321*
0.363**
0.314*
Learning
0.429***
0.244*
0.235
0.389
0.315

* p< .05, ** p<.01, *** p<.001

Since only 67 responses from the first survey could be included in the
combined model analysis, breakout analyses were conducted with each interaction
mode treated separately. Figures 6-2 and 6-3 show these results. In both the
synchronous and the asynchronous interaction mode there was no change in
significant for any model constructs although there were slight changes in magnitude
of the path coefficients. Neither interaction mode by itself accounted for greater
variance in either learning or innovation.

149
Figure 6-2: Breakout of Results for Synchronous Interaction Mode Only, n=106
Input Process Outcomes
CT Support for
Testing and
Adjusting Strategy
Sharing Diverse
Knowledge
Innovation
Reflection on Others’
Shared Knowledge
CT Support for
Contextualization
Strategy
0.331***
0.493**
Synchronous Interaction Mode
0.545***
0.253***
Learning
0.429***
0.297
0.095

* p< .05, ** p<.01, *** p<.001
Figure 6-3: Breakout of Results for Asynchronous Interaction Mode Only, n=67
Input Process Outcomes
Innovation
Sharing Diverse
Knowledge
CT Support for
Attention Focusing
Strategy
CT Support for
Contextualization
Strategy
CT Support for
Perspective Taking
Strategy
Reflection on Others’
Shared Knowledge
0.397***
0.458***
0.274**
Asynchronous Interaction Mode
0.504***
0.494***
Learning
0.244*
0.207
0.254
0.244

* p< .05, ** p<.01, *** p<.001

Chin (1998a) notes that model fit indices that normally accompany structural
equation models are not calculated with PLS Graph. However, noting over-reliance
on fit indices such as NNFI and GFI, Chin recommends assessing model fit based on
R-squared values and magnitude and significance of path coefficients. To verify
goodness of fit of the hypothesized model, the model was compared to the saturated

150
model. All possible paths were included and estimated. Then, paths were removed
one at a time in an iterative fashion and the model re-estimated to verify that the
hypothesized model was the best explanation for the relationships between the
constructs.
Table 6-20: Model Checking - Investigation of Alternative Paths for Significance
From Construct To Construct Path Coefficient Significance
Learning Innovation .004 n/s
Synchronous Interaction Mode Paths
Reflection on others’ shared
knowledge
Learning .196 n/s
Sharing Diverse Knowledge Innovation -0.093 n/s
CT Support for an Attention
Focusing Strategy
Sharing Diverse Knowledge .037 n/s
CT Support for an Attention
Focusing Strategy
Reflection on others’ shared
knowledge
-.089 n/s
CT Support for a Perspective
Taking Strategy
Sharing Diverse Knowledge .034 n/s
CT Support for a Perspective
Taking Strategy
Reflection on others’ shared
knowledge
.004 n/s
CT Support for a Testing and
Adjusting Strategy
Reflection on others’ shared
knowledge
-.054 n/s
CT Support for a
Contextualization Strategy
Sharing Diverse Knowledge .036 n/s
Asynchronous Interaction Mode Paths
CT Support for a Testing and
Adjusting Strategy
Sharing Diverse Knowledge .08 n/s
CT Support for a Testing and
Adjusting Strategy
Reflection on others’ shared
knowledge
.06 n/s
CT Support for a Perspective
Taking Strategy
Sharing Diverse Knowledge .013 n/s
CT Support for a
Contextualization Strategy
Sharing Diverse Knowledge -.003 n/s
CT Support for an Attention
Focusing Strategy
Reflection on others’ shared
knowledge
.111 n/s

One additional significant path was found. Sharing diverse knowledge in the
asynchronous interaction mode was found to have a significant impact on the extent
individuals’ build innovative work products, when, and only when the path from

151
reflection on others’ shared knowledge (in the asynchronous interaction mode) to
build innovative work products was excluded. To investigate the possibility of a
mediated relationship, tests of mediation were run in accordance with the procedures
outlined in Baron and Kenny (1986). Successive PLS models were compared to
conduct the analysis. First, it was shown that reflection on others’ shared knowledge
had a significant impact innovation, as is shown in Figure 6-1. Second, it was
verified again that in the absence of this path, sharing diverse knowledge had a
significant impact on innovation. Third, it was verified that sharing diverse
knowledge had a significant impact on reflection on others’ shared knowledge.
Finally, all three paths were added to the model to verify that the path from sharing
diverse knowledge to innovation became insignificant with the others paths included
in the model. Results indicated a completely mediated relationship. None of the
other alternate paths tested were significant, nor did any other configuration explain
more of the variance in the outcome variables.
In addition un-hypothesized CT support variables were added to the model to
verify that the other CT-supported strategies had no impact on sharing diverse
knowledge or reflection on others’ shared knowledge when not hypothesized.
Results of adding un-hypothesized effects are also listed in Table 6-20. There were
no significant effects.

152
6.3 Summary of Analysis
This chapter presented analysis of the structural model. Analysis of the data
collected (described in chapter 5) demonstrated the data showed acceptable
reliability, convergent validity, and discriminant validity. Reliability is indicated
when composite reliability reported by PLS is greater than 0.7. All constructs
indicated acceptable reliability. Convergent validity is demonstrated in two ways.
First, item loadings were greater than 0.7, except for trust (which was later excluded
from the analysis). Second, the average variance extracted (AVE) for all constructs
was greater than 0.5, indicating convergence (Chin 1998a). Finally, discriminant
validity was assessed using a two step process. First, AVE was compared to inter-
construct correlations for each construct. Discriminant validity was acceptable for
each construct, as the AVE was greater than any of the inter-item correlations.
Second, item cross-loadings were calculated and shown to be high for the indicated
construct and low for all other constructs, in all cases. Therefore, model constructs
demonstrated acceptable discriminant validity.
Two additional checks on the data were conducted. First, interrater
agreement was assessed between individuals interacting together on the same
project. Results indicated that within group agreement was low enough to justify
analysis of the data at an individual level. Second, the data were checked for
common method variance. Analysis showed that common method variance, if
present, had no effect on the significance of the relationships in the model.

153
Results indicate individuals realized successful collaboration outcomes of
learning and innovation through sharing diverse knowledge (sharing) and reflection
on others’ shared knowledge (reflection). For learning, sharing in the synchronous
interaction mode and reflection in the asynchronous mode has a significant impact,
and reflection was found to mediate the impact of sharing asynchronously on
learning. For innovation, reflection from both modes was found to have a significant
impact.
Analysis of the structural model found support for each of the hypothesized
impacts of CT Support. For sharing, in the synchronous interaction mode CT
support for a testing and adjusting strategy supported sharing; however in the
asynchronous interaction mode CT Support for an attention focusing strategy
supported sharing. For reflection, CT Support for a contextualization strategy
supported reflection in both interaction modes; however, CT Support for a
perspective taking strategy supported reflection in the asynchronous interaction
mode only.
Finally, analysis of alternate pathways and un-hypothesized effects found no
additional significant paths and no alternative models that explained a greater
amount of variance in the collaboration outcomes. Figure 6-4 shows the final model
with all non-significant paths removed.

154
Figure 6-4: Final Model, CT Support for PM/PT in Virtual Collaboration for Innovation in the
Synchronous and Asynchronous Interaction Mode

155
CHAPTER 7
DISCUSSION AND CONCLUSIONS
The goal of this dissertation was to present a new theory of CT support for
virtual collaboration for innovation that explained the differential effects CT support
has on learning and innovation in synchronous vs. asynchronous interaction modes.
Virtual collaboration for innovation was defined as the interaction of individuals
from distinct knowledge domains, separated by time and space, and in which they
share and combine their knowledge to develop and implement creative ideas.
Synchronous and asynchronous are the two virtual interaction modes in which
individuals collaborate, and are defined by the ability to share knowledge and give
and receive feedback in real time. Learning was defined as an individual building a
new mental model based on her understanding of others’ knowledge, and innovation
was defined as an individual building innovative work products by creatively
combining her own knowledge with others’ knowledge.
I focused on the cognitive processes rather than the motivational processes of
virtual collaboration for innovation because past literature has suggested that an
individual’s ability to successfully learn and innovate during virtual collaboration for
innovation with others depends on two cognitions performed by the individual
simultaneously: reflecting on others’ shared knowledge (hereafter called
“reflection”) and sharing one’s own diverse knowledge (hereafter called “sharing”).
In addition, I focused on the differences between synchronous vs. asynchronous

156
interaction modes because, when collaborating virtually, each interaction mode
generates different cognitive demands on individuals, suggesting that CT support
may be needed to resolve these cognitive demands differently in order for learning
and innovation to successfully occur.
Given the different interaction capabilities afforded by asynchronous vs.
synchronous interaction modes when collaborating virtually, I asked whether these
interaction modes affected the two cognitions of sharing and reflection, and the role
of CT support. In particular, I asked the following two research questions: 1) Does
an individual’s sharing and reflection effect learning and innovation differently
depending on whether the individual is interacting synchronously vs.
asynchronously? and 2) Are there different kinds of CT Support that facilitate an
individual’s sharing and reflection differently when the individual is interacting with
others synchronously vs. asynchronously?
In addressing the first research question, I found that outcomes were
dependent on the need for attending to self and/or others’ knowledge coupled with
the opportunity to do so in each interaction mode: thus learning is the result of
sharing in both modes and also reflection in the asynchronous interaction mode (but
not the synchronous) because individuals need to explain their own knowledge as
well as understand others’ knowledge which they have less opportunity to do
synchronously. In contrast, innovation is the result of reflection during both
interaction modes, but not directly from sharing in either interaction mode, because

157
individuals must focus on others’ knowledge whenever it is shared rather than their
own knowledge. Thus reflection consistently contributes to innovation in either
interaction mode.
In addressing the second research question, I found that CT support affects
both sharing and reflection in different ways depending on the different cognitive
demands of the interaction modes. I found differential value of four different types
of CT support, based on a framework from Te’eni: CT support for testing and
adjusting, attention-focusing, contextualization, and perspective-taking. To facilitate
sharing, individuals need to quickly understand others’ feedback in the synchronous
mode and they need a means for obtaining feedback in the asynchronous mode –
which CT support for testing and adjusting and attention-focusing, respectively,
provides. For reflection, individuals need contextual information in both modes and
access to others’ knowledge in the asynchronous mode which CT support for
contextualization and perspective-taking provides. Thus different features of CT
support reduce cognitive demands in synchronous vs. asynchronous modes and
facilitate sharing and reflection, which ultimately leads to learning and innovation.
These findings are discussed below in the next section, followed by a discussion of
the limitations and implications of the research.

158
7.1 Discussion of Findings
The findings are organized into two sections, each corresponding to one of
the two research questions. The first section presents the findings for how sharing
and reflection differentially impact learning and innovation depending on the
interaction mode. The second section presents the findings for how CT support
affects sharing and reflection.

7.1.1 Impact of Sharing and Reflection on Learning and Innovation
The first research question asked how sharing and reflection differentially
impacts learning and innovation depending on the interaction mode. To answer this
question, I first reviewed the theory of perspective-making and perspective-taking
(PM/PT) by Boland and Tenkasi (1995). The theory suggests that when individuals
have diverse knowledge, an innovative task, and a forum for dialogue such as in
virtual collaboration for innovation, they engage in two simultaneous cognitive
processes - sharing and reflection – which contribute to learning and innovation.
The cognition associated with sharing is deciding how to explain one’s own
knowledge in a way that others can understand it. The cognition associated with
reflection is the cognitive process of first understanding what others are sharing,
comparing it to what the recipient already knows (the reflection process), and then
phrasing constructive feedback that reaches the sender. Figure 7-1 depicts the
general PM/PT process.

159
Figure 7-1: General PM/PT Process
Outcomes PM/PT Process Inputs
Sharing Diverse
Knowledge
Learning
Diverse Knowledge
from a Number of
Experts
Forum for Dialogue
Innovation
Reflection on Others’
Shared Knowledge
Innovative Task

Although the theory identified both sharing and reflection as necessary for
learning and innovation, it did not specify how these cognitions contributed
differently to learning and innovation nor did it account for interaction mode
differences. I argue that for learning and innovation, individuals may experience
different opportunities for sharing and reflection in the synchronous vs.
asynchronous interaction mode. Therefore sharing and reflection should impact
learning and innovation differently depending on the interaction mode unless the
same opportunities are afforded in each mode. Below I first discuss my extensions
to the PM/PT theory for explaining learning and then discuss my extensions for
explaining innovation.
Learning: Based on the theory of collaborative learning (e.g. Webb and
Palincsar 1996), for a person to build new mental models, they need to share their
diverse knowledge, through which they elaborate and rationalize what they know.
An individual’s self-elaboration of what she knows forms associations between her
existing mental models and new knowledge which is used to build a new mental

160
model. Thus, I predicted that sharing in either interaction mode leads to greater
learning when relying on CT support, since in both modes individuals have the
capability to decide what diverse knowledge to share and the means to share it with
others.
However, learning requires more than sharing one’s diverse knowledge;
individuals also need to reflect on others’ shared knowledge for learning because it is
others’ knowledge that leads to building new mental models rather than reinforcing
existing models. In the synchronous mode, individuals attending to sharing diverse
knowledge and receiving feedback will find it difficult to have the cognitive capacity
to simultaneously reflect on others’ knowledge. Thus reflection in the synchronous
interaction mode does not contribute to learning. However, reflection in the
asynchronous mode does contribute to learning because individuals have the time
and cognitive resources available to attend and reflect on others’ knowledge. Thus, I
predicted that reflection would only affect learning in the asynchronous mode.
I found that, as hypothesized, learning was directly affected by sharing in the
synchronous and asynchronous interaction modes, and from reflection in the
asynchronous interaction mode. However, unexpectedly, I found that the effect of
sharing in the asynchronous interaction mode was completely mediated through
reflection, suggesting that the theory of a direct effect of sharing asynchronously on
learning was not entirely accurate. This finding may be the result of the type of
knowledge individuals share in the asynchronous mode; that is, asynchronously

161
shared knowledge may be more focused on giving feedback to others and resolving
differences and inconsistencies and less focused on their own diverse knowledge
(Kerr and Murthy 2004; Robey et al. 2000). Another possible explanation for
mediation is that individual’s sharing in the asynchronous mode focuses others’
attention and prompts them to share, initiating back and forth dialogue that otherwise
might not have occurred and requiring greater reflection on that additional shared
knowledge for learning to result (Boland et al. 1994; Te’eni 2001).
In sum, my research extends the PM/PT theory by suggesting that, taken at a
single point in time with a single interaction mode, reflection and sharing would not
affect learning when done virtually. Instead, the cycles of interaction modes are
needed to explain learning: reflection is primarily useful when done asynchronously
and sharing is directly useful when done synchronously.
Innovation: For an individual to be able to create innovative work products,
according to Boland and Tenkasi (1995) requires the individual to combine others’
knowledge with her own diverse knowledge, involving, as with learning, both
sharing and reflection. The cognition of reflection is a significant factor in
innovation since it generates recognition of the differences and dependencies
between others’ knowledge and one’s own, which clarify how diverse knowledge
should be combined when building innovative work products. Reflection in both
modes was predicted to be helpful for innovation as long as others’ knowledge is

162
accessible and the individual can devote her cognitive resources to make sense of
that knowledge in each interaction mode.
Sharing in either interaction mode, however, was predicted to not directly
impact innovation but rather to provide the knowledge to combine with others’
knowledge during the reflection process. Thus, while reflection was predicted to
directly impact innovation, sharing was predicted to impact reflection only. The
findings supported this hypothesis. Therefore, there appears to be a tradeoff: sharing
may reduce focus on others’ knowledge and does not directly contribute to one’s
ability to innovate, but it increases the ability to reflect since one understands his or
her own knowledge better in relation to others’ knowledge. It is this reflection that
creates the ability for individual innovation within the group. Clearly then a major
task of groups is creating a balance between individuals sharing and reflection, a
balance that may be more possibly done through the judicial use of synchronous vs.
asynchronous modes. That is, group leaders could limit sharing in some cases to one
interaction mode or the other, using remaining synchronous and/or asynchronous
time for reflection.
In summary, I found that learning was differently impacted by sharing and
reflection in synchronous vs. asynchronous interaction modes, while innovation was
impacted by reflection in both synchronous and asynchronous interaction modes.
Figure 7-2 depicts these results. Significant differences were found between
synchronous and asynchronous interaction modes regarding the contribution of

163
sharing and reflection to learning. In the asynchronous interaction mode, sharing
does not affect learning unless it is mediated by the ability to reflect on that
knowledge. Reflection from both interaction modes contributed to innovation but
sharing did not. And finally, sharing and reflection are significantly correlated, such
that greater sharing leads to greater reflection despite tradeoffs between the two that
impact individual’s capacity for innovation.

Figure 7-2: Findings - Effects of Sharing/Reflection on Learning and Innovation

164
7.1.2 Effect of CT Support on Sharing and Reflection
While the first research question focused on how the differential impact of
sharing and reflection on learning and innovation depends on the interaction mode,
in the second research question I asked what differential effect CT support has in the
synchronous vs. asynchronous mode on individual’s sharing and reflection. No prior
theory of virtual collaboration for innovation had yet been developed that could
explain the differential effect of CT support on critical cognitions relevant to
learning and innovation in synchronous vs. asynchronous interaction modes,
although Boland and Tenkasi (1995) suggested the need for CT support to facilitate
these cognitions.
While there is no comprehensive theory, Te’eni offers a framework for
reducing cognitive complexity during communication with others through the
application of four different communications strategies which can be supported by
CT. These strategies, listed in Table 7-1, are contextualization, testing and adjusting
through feedback, attention-focusing, and perspective-taking. I hypothesized that
sharing and reflection create different cognitive demands in synchronous vs.
asynchronous interaction modes, and that these strategies offer four different ways in
which CT support could reduce the cognitive demands.

165
Table 7-1: Communication Strategies For Reducing Cognitive Complexity From Te'eni (2001),
With Explanation of How CT Supports Each Strategy
Communication
Strategy
Definition How CT supports communication strategy
Testing and
Adjusting
Spontaneously changing
what knowledge is
shared based on
feedback from self and
others
Displaying and editing shared knowledge in real-
time in a visible, tractable format, such as through
document sharing features or whiteboards; instant
feedback available and recorded
Attention
Focusing
Directing knowledge
specifically to others to
initiate dialogue and
obtain feedback
Text-based communication such as email, instant
messaging, discussion forums that permit two-way
asynchronous dialogue
Contextual-
ization
Provision of explicit
context in a message
helps individual
understand why others
made a contribution
Contextualization mechanisms such as displaying
names/profiles of collaborators, attributing verbal
and written contributions, file and document
sharing tools that permit back and forth
comparison of different versions
Perspective
Taking
Cognitively focused on
receiver’s view because
individual can access
their knowledge in their
own words
Access to stored knowledge in repositories or
virtual workspaces when needed for reflection or
use

Figure 7-3 depicts these CT supported strategies and their effect on
individuals cognitive needs for sharing and reflection, in both interaction modes,
which are explained as follows:
Sharing: In particular, in the synchronous interaction mode, the cognition of
sharing requires rapid and meaningful feedback so an individual knows what diverse
knowledge to share, creating a cognitive demand to attend to the ongoing dialogue
sufficiently to easily incorporate feedback regarding what diverse knowledge to
share. When working asynchronously, the cognition of sharing requires directing
more focused diverse knowledge to specific others so they may provide meaningful

166
feedback. Otherwise, individuals experience uncertainty which makes it more
difficult to decide what to share.
I found that for the cognition of sharing, in the synchronous interaction mode
individuals have the means to get direct feedback from others (i.e. synchronous
dialogue) but lack a method for quickly using that feedback to change what diverse
knowledge they are sharing. Thus these individuals rely on CT support for a testing
and adjusting strategy. Conversely, in the asynchronous mode individuals need to
direct their diverse knowledge to others in a format that is understandable so others
can attend to it; yet they don’t have difficulty using others feedback since they can
draft and revise diverse knowledge they are sharing at their own pace. Thus these
individuals require CT support for an attention focusing strategy.

167
Figure 7-3: CT Support Effects on Individuals Cognitive Needs for Sharing and Reflection
Input Process
CT Support for
Testing & Adjusting
Strategy
Sharing Diverse
Knowledge
Sharing Diverse
Knowledge
Reflection on Others
Shared Knowledge
CT Support for
Contextualization
Strategy
CT Support for
Attention Focusing
Strategy
CT Support for
Contextualization
Strategy
CT Support for
Perspective Taking
Strategy
Reflection on Others
Shared Knowledge
Synchronous Interaction Mode
Asynchronous Interaction Mode

Reflection: In contrast, in the synchronous interaction mode, the cognition of
reflection requires contextual cues relevant to the ongoing dialogue so the individual
understands the evolution of the knowledge that has been shared by others (e.g. who
shared what and for what reason), and can stay attentive to the ongoing dialogue
while offering meaningful feedback. When working asynchronously, the cognition
of reflection requires access to more expansive sources of knowledge for comparison
and reference, as well as contextual cues to help interpret the knowledge others have
shared. Without this access and context, individuals experience difficulty making
productive use of their time which increases cognitive complexity and frustrates
attempts at reflection

168
Thus for the cognition of reflection, I found that in the synchronous
interaction mode individuals have access to others’ knowledge as it is shared, but
lack contextual cues to help interpret that knowledge and offer constructive
feedback. Thus these individuals require CT support for a contextualization strategy.
Conversely, in the asynchronous interaction mode, individuals have neither direct
access to others’ shared knowledge nor the cues needed to help them interpret it and
offer constructive feedback. Thus these individuals require CT support for both a
contextualization strategy and a perspective taking strategy.
In sum, my research extends PM/PT theory by showing how different CT
support is needed to diffuse cognitive complexity depending on the interaction mode,
for both sharing and reflection. The final model, which depicts the impact of CT
support through sharing and reflection to realize learning and innovation outcomes,
in both synchronous and asynchronous interaction modes, is shown in Figure 7-4.
Without this CT support, I conclude, an individual will have difficulty with the
cognitions of sharing and reflecting in both interaction modes, and subsequently will
have less success learning and innovating.

169
Figure 7-4: Final Model: Virtual Collaboration for Innovation in Synchronous and
Asynchronous Interaction Modes
Input Process Outcomes
CT Support for
Testing and
Adjusting Strategy
Sharing Diverse
Knowledge
Innovation
Sharing Diverse
Knowledge
Reflection on Others’
Shared Knowledge
CT Support for
Contextualization
Strategy
CT Support for
Attention Focusing
Strategy
CT Support for
Contextualization
Strategy
CT Support for
Perspective Taking
Strategy
Reflection on Others’
Shared Knowledge
Synchronous Interaction Mode
Asynchronous Interaction Mode
Learning

7.2 Limitations
As with any research there are a number of potential limitations, due to
assumptions made during theory development or special conditions that were
encountered during data collection and analysis. Four limitations to validity are
identified which potentially call in to question the quality of the results. For
example, it may be that CT support used in only one interaction mode is actually
needed in both but the model failed to capture this effect. Thus results reported might
not accurately reflect what the effects of CT support in virtual collaboration for
innovation. In addition, two limitations are identified which may impact

170
generalizability of the findings, calling in to question the usefulness of the model in
other contexts or settings.

7.2.1 Limitations Concerning Validity
First, a number of new measures were created specifically for this study
which makes construct validity difficult to assess. New measures were created for
sharing diverse knowledge, reflection on others’ shared knowledge, innovation, CT
support for testing and adjusting, CT support for attention focusing, and CT support
for perspective taking. A systematic process was followed for creating and testing
these measures (i.e. Straub et al. 2004), but repeated use is recommended to assess
validity in a variety of contexts. Without external validation therefore concurrent
validity, or the extent the measures agree with similar instruments in other research,
cannot be verified. While comparison of the constructs in two different interaction
modes provides an initial indication of concurrent validity, it can not be taken as
reliable evidence since both surveys came from the same individuals in the same
organization. Further use of these constructs is necessary to demonstrate acceptable
concurrent validity.
In addition, the innovation measure was a single item measure, which
methodologists advise against because there is no way to assess internal consistency
(Campbell and Fiske 1959; Nunnally 1978). Single items have no less predictive
power than multi-item scales but it is more difficult to assess if they are reliable

171
measures of the construct (Bergkvist and Rossiter 2007). Rossiter (2002) argues that
single items may actually be preferable in situations where the object of a construct
is singular and the attribute evaluated is ‘easily and uniformly imagined,’ by
reducing common method variance, cognitive load during repeated evaluation, and
to reduce non-response. The object of the innovation measure was a distinct
individual and the attribute her contribution of innovative work products, which is
consistent with Rossiter’s definition of a singular and uniformly imagined construct.
Further, the raters (project sponsors) of innovation rated several individuals at the
same time and may not have responded if multiple items were required to rate each
individual. Thus while multiple-item measure continues to be the accepted practice
there may be situations where single items are acceptable and possibly preferable.
Third, although I have demonstrated that common method variance (CMV)
had little to no impact on the relationships between variables, CMV is always a
danger with cross-sectional methods since relationships may appear stronger than
they actually are. Future work might rely less on self-report by coding CT support
from observable use or log data, although as Majchrzak et al. (2005b) noted actual
use does not always correspond with cognitive impact. A network approach might
also be applied to assess sharing/reflection whereby others rated the individuals in
their network based on how much diverse knowledge they shared and how much
they appeared to understand of others’ knowledge, with separate ratings correlated to
obtain scores for each individual.

172
Finally, although the results claim to differentiate between two different
interaction modes there was only a three day lag between survey measurement of the
synchronous and asynchronous interaction modes which may not correspond to how
individuals think about their pattern of virtual collaboration. Further, only one cycle
of a synchronous and asynchronous interaction mode was measured which may not
accurately reflect repeated cycles, although results purport to apply to sustained
cycles of virtual collaboration for innovation. Future work should observe a repeated
set of interaction modes to assess stability of the measures over repeated cycles.

7.2.2 Limitations Concerning Generalizability
Both the case study and the survey were conducted with individuals from a
single organization, which makes it difficult to tell if organizational factors are
influential in virtual collaboration for innovation. While holding organization
constant helps control for a number of macro-level influences on individual
participation in virtual collaboration for innovation, it may limits generalizability
(Wasko and Faraj 2005). The organization may encourage special practices not
typical of other organizations or individuals may have unique characteristics that are
atypical. For example, results indicated that CentCo employees are for the most part
highly educated and used to working virtually. Thus care should be taken when
applying the results in other organizational contexts.

173
Second, the sample was one of convenience rather than one randomly
selected from individuals at CentCo and thus a systematic bias may be present based
on selection criteria. Individuals were chosen specifically because of their
assignment to innovative tasks. Thus individuals in the study may not be
representative of all individuals in work environments. Further, project sponsors
passed on the request to complete the surveys to the individuals on each project (per
CentCo’s request) and their relationships with the individuals may have biased
results or led to greater non-response. In addition, since individuals were required to
complete two rounds of surveys, it is possible that only more productive individuals
participated fully in the survey. Those less likely to complete both rounds of the
survey may be under-represented (Jarvenpaa and Leidner 1999). Analysis of
respondents/non-respondents provided evidence that non-response bias was minimal;
however, the possibility exists that these effects influenced the results.

7.3 Research Implications
Despite these limitations, there are a number of theoretical implications from
this research. Since virtual collaboration for innovation synthesizes research on
innovation, virtual collaboration, and CT design, implications for each of these
streams of research are discussed separately.

174
7.3.1 Implications for Theory Development on Innovation
Past theories of innovation suggest that both sharing (of diverse rather than
commonly-held knowledge) and reflection are necessary for realizing successful
outcomes (Amabile 1988; Carlile 2004; Fiol 1994; Leonard and Swap 1999), but
have not made it clear precisely how they work together, and when. Some have
indicated that primarily sharing in the synchronous interaction mode, and reflection
in the asynchronous mode is optimal for accomplishing divergent (idea generation)
and convergent (decision making) tasks (Dennis et al. 1997; Kerr and Murthy 2004;
Massey et al. 2002). These studies suggest individuals’ have difficulty managing
their diversity in real time which may by symptomatic of difficulties establishing
common ground (Carlile 2002).
Prior research has noted that common ground (also called mutual knowledge)
can make it easier for individuals to exchange knowledge across boundaries
(Cramton 2001; Krauss and Fussell 1990). However, Carlile (2004) notes that
common ground can be a negative influence on innovation since it may constrain
sharing as individuals reuse existing mutual knowledge. My research helps to
resolve this apparent inconsistency in the literature by explaining that both sharing
and reflection are needed to realize learning and innovation, although there may not
be direct effects from each. Instead, through reflection individuals develop common
ground that helps them decide what to share with others; through sharing individuals

175
interject new diverse knowledge into dialogue so reflection does not denigrate into
repeated re-hashing of commonly held knowledge (Stasser and Titus 2003).
In their review of different dimensions of virtuality which impact innovation,
Gibson and Gibbs (2006) note that CT support is negatively associated with
innovation because it makes it difficult for individuals to iterate ideas frequently,
experiment with ideas, and improvise, while at the same time monitor and compare
existing or baseline knowledge. However, my research shows that by breaking out
which of these activities occur in the synchronous and asynchronous interaction
modes we can show how individuals accomplish these activities – for example
frequent iteration may be easier in a synchronous interaction mode since knowledge
is exchanged more quickly, while monitoring and comparison may be easier in an
asynchronous interaction mode since knowledge can be codified and maintained.
Thus in this research I show that CT support may have a positive effect in both
interaction modes if individual’s cognitive complexity associated with these
activities in each interaction mode can be resolved.
In addition, the present research shows that different CT support features
have different effects, at different times. Email, which can support an attention
focusing strategy, may have no bearing on sharing diverse knowledge in
synchronous interaction modes. Conversely, email use might be extensive but if
little sharing of diverse knowledge occurs it is not sustaining any innovation.
However Gibson and Gibbs (2006) coded technology dependence as high-medium-

176
low based on interview transcripts where individuals reported more or less face to
face contact or email use. Thus finding that more technology dependent teams are
less innovative offers no insight into why they are less innovative – i.e. what they are
using the technology for; and offers no guidance for individuals who are engaged in
virtual collaboration for innovation.
Finally, even though the new theory was developed in context of innovative tasks
only, it may have implications for theory that concerns non-innovative tasks. Other
theories may be more suitable for explaining non-innovative virtual collaboration,
where it is not essential that diverse knowledge be shared and understood (and in
fact, it may even impede success if too much is shared). Examples of such theories
are overcoming production blocking (Gallupe et al. 1994; Nijstad et al. 2003) or least
collaborative effort (Clark and Brennan 1991). Production blocking argues that in
synchronous dialogue individuals contribute less because they forget what they want
to say, have no opportunity to share, or withhold knowledge if they think their ideas
are redundant. The principle of least collaborative effort asserts that individuals in
collaboration collectively minimize effort needed to get work done. Needless
sharing and reflection might only complicate a simple task (Hollingshead and
McGrath 1995) Together these theories suggest that using CT support, which incurs
process costs (Hollingshead and McGrath 1995), may foster greater cognitive
complexity by surfacing extraneous diverse knowledge that has to be processed.
Alternatively, the present research suggests that if these theories are applied to CT

177
design then other factors more indicative of success for non-innovative tasks may be
more suitable outcomes for design, such as speed of reaching a decision, equal
contribution from all participants, or minimizing participation or workload.

7.3.2 Implications for Theory Development on Virtual Collaboration
Taking into account two complementary interaction modes in virtual
collaboration for innovation helps resolve prior confusion and inconsistencies in the
virtual team literature regarding what CT support helps individuals collaborate and
what hinders their collaboration (Hollingshead and McGrath 1995). For example,
GSS research has repeatedly held that CT support can aid in sharing diverse
knowledge by allowing anonymous contributions (Carte and Chidambaram 2004;
Dennis et al. 1997; Pinsonneault et al. 1999). Results of the importance of
anonymity have been mixed however (Pinsonneault and Heppel 1998). Theories of
social identity have been used to explain that anonymous contributions may be made
more openly without fear of evaluation or reprisal, and that anonymous evaluation of
ideas promotes more honest and useful feedback (Bhappu et al. 1997; Dennis 1996).
However Pinsonneault and Heppel (1998) and Wittenbaum et al. (2004) argue that
individuals desire to have diverse knowledge identified as their own and that others
need that information for reflection. They argue that actual anonymity is difficult to
achieve, and even when it does the deindividuation and freedom from social
evaluation has positive and negative results, and that in virtual collaboration for

178
innovation the need for contextual information is greater than the need for freedom
of expression.
My research follows in this line of reasoning and shows specifically that
features which support non-anonymity are needed, particularly in collaborative
situations (i.e. participating in active dialogue vs. just making knowledge available to
others), since when attempting reflection individuals require contextualization cues.
These cues provide information about the sender that can help identify not only who
contributed diverse knowledge, but why, what it relates to, and how it should be
evaluated. If individuals are working jointly with others in virtual collaboration for
innovation they need this information to be able to understand others knowledge and
offer feedback. Feedback to others an integral part of the PM/PT process and it is
more meaningful when directed to a specific contributor rather than a general
audience (Te’eni 2001). If lower social evaluation increases sharing but not
reflection than individuals are no better off – they will have additional diverse
knowledge to process and not enough information to reflect on that knowledge.
Thus, if an individual innovates in a non-collaborative context it is possible that
being identified may decrease her sharing of diverse knowledge, but when an
individual needs to interact with others to generate a jointly discussed and agreed
upon outcome, identity is a positive, not negative feature that CT can support.
Another inconsistency this research resolves is the need for face to face
interaction in virtual collaboration for innovation. Griffith et al. (2003) explain that

179
‘purely’ virtual teams are those in which individuals never meet face to face, while
‘hybrid’ teams meet face to face some times and otherwise interact via CT. Prior
research has suggested that face to face interaction is instrumental for learning and
innovation (Kiesler and Cummings 2002; Maznevski and Chudoba 2000; Schunn et
al. 2002). However, it is not strictly necessary – reports of successful virtual teams
have been provided where individuals rely on CT support for all of their interaction
with others (Malhotra et al. 2001; Majchrzak et al. 2005b). The current research
suggests that what is important about face to face communication is not the richness
of cues but the possibility of driving a sustained rhythm of interaction over time,
leveraging the richer environment of face to face communication to accomplish
sharing and reflection to a greater extent. When face to face modes are available
individuals do not have to share or reflect in ‘less-rich’ virtual interaction modes -
this cycle of intensity was depicted in Figure 4-1 as the ‘peaks and valleys’ of
collaboration. What the new model shows is that individuals use CT support to
facilitate sharing and reflection across a sustained cycle of synchronous and
asynchronous virtual interaction modes. Thus the same type of cycle as found by
Maznevski and Chudoba (2000) for hybrid teams is important for purely virtual
collaboration for innovation, but the interaction modes differ and CT support is
needed.

180
7.3.3 Implications for Theory Development for CT Design
What we now know about how individuals require different CT support for
different collaborative needs in both synchronous and asynchronous interaction
modes should be applied to how CT are designed. Malhotra and Majchrzak (2005)
describe how multi-function CT are finally being utilized as an integrated
collaboration tool and explain how different CT support features, such as search,
document storage, and instant messaging are being used. However they note that
several unanswered questions remain regarding why certain combinations of features
are effective, or why some features facilitate sharing diverse knowledge more than
others. The current research validates their assertion that CT support should be
investigated as a bundle of integrated features rather than as separate functionalities
(Bordetsky and Mark 2000). It provides a rationale for the impact of each one of
those features which would be useful in answering the above questions. For
example, Malhotra and Majchrzak (2005) found that some individuals use instant
messaging during synchronous teleconferencing, which individuals in the case study
also used on several occasions. By showing how instant messaging helped
individuals proactively offer feedback during a synchronous presentation where only
one individual was speaking, thus supporting a control of feedback strategy, it
explains why such CT support is valuable – even though those using instant
messaging incur coordination costs. As another example, CT support for a
contextualization strategy was found to be useful for reflection in both the

181
synchronous and asynchronous interaction mode. However, the model explains use
of this strategy in both modes. This particular example shows that CT support
assumed to be useful in both modes, or if the distinction has not been made clear,
should be re-evaluated to verify that a rationale exists for each interaction mode.

7.4 Practical Implications for Managers
As previously explained, the model shows that in virtual collaboration for
innovation individuals are engaged in sharing and reflection consistently over a
sustained cycle of synchronous and asynchronous interaction modes. The ‘peak’ and
‘valley’ model found by Maznevski and Chudoba (2000) for hybrid teams (Figure 7-
5, solid blue line) does not work for virtual collaboration for innovation (dotted pink
line depicts purely virtual interaction of millennial project), because sharing diverse
knowledge and reflection on others’ shared knowledge are important in both modes
for realizing collaboration outcomes. Managers should realize that successful
collaboration outcomes will require support for more constant and continuous
interaction over time, and work processes and incentive structures should mirror this
schedule. One manager at CentCo said “if I don’t fly everyone in here for a week-
long all out session, nothing gets done.” I asked him what everyone did when they
weren’t here (in the face to face session). “Nothing” he replied, “that’s the
problem.” Evidence from the combined body of research on virtual collaboration
suggests that reliance on face to face meetings is a self-fulfilling prophesy and that

182
managers will not realize successful outcomes from virtual collaboration for
innovation until they break that cycle.
Figure 7-5: Nominal Extent of Sharing and Reflection in Virtual Collaboration for Innovation
With Regular Face to Face Interaction Compared to Completely Virtual
Sync
Asynch
Sync
Asynch
Sync
Asynch
Sync
Asynch
Regular Face to Face
Interaction
Completely Virtual

In addition, managers who believe that greater sharing occurs in synchronous
interaction modes (either face to face or virtual) should note that as the individuals in
the case study became more comfortable using CT support, more sharing occurred in
the asynchronous mode. The nature of the individuals work changed from the
‘normal’ model where most sharing occurs in synchronous interaction modes. The
individuals on the project completely restructured their work to leverage the
advantages of each interaction mode, which CT support helped them to do.
Individuals reported they relied less and less on synchronous teleconferences to share
diverse knowledge with others, and accesses to the project virtual workspace
increased. Individuals were also sharing more frequently through emails and instant
messages. This change was only possible because CT supported sharing and
reflection in the asynchronous interaction mode.

183
Finally, it is difficult for managers to assess the value of CT support to their
organization (Barua and Mukhopadhyay 2000; Tallon et al. 2000) because
technology rarely contributes directly to the bottom line. However, this research
presents a fine-grained model of how individuals’ use CT support that can help
managers explain its effectiveness in terms of two collaboration outcomes - learning
and innovation. These are, in the case of innovation, demonstrable and tangible
evidence of success; and in the case of learning, long-lasting benefit that transcend
immediate project goals as individuals become more effective knowledge workers
(Robey et al. 2000). In addition, managers also can use this research to justify if a
CT is right for their project. For example, individuals distributed globally might
have language and time zone barriers that make sharing in the synchronous
interaction mode less effective. If so, they should look for CT support features such
as embedded text messaging or discussion forums or other attention focusing
mechanisms such as alerts for new contributions. These features would make
sharing in the asynchronous interaction mode more effective.

7.5 Future Research
Future research should emphasize contributions in one of three different areas
of research: innovation literature, virtual collaboration literature, and CT design
literature. Suggestions for future work for each of these areas is presented here.

184
7.5.1 Future Research in the Area of Innovation
Many of the implications for how individuals use CT support over a sustained
cycle of virtual collaboration for innovation come from the case study results. The
survey portion however only measures CT support over a single cycle of
synchronous and asynchronous interaction modes. More work is needed to see if the
results from the case study generalize to other projects and if the changes over time
regarding sharing and reflection in the asynchronous interaction mode can be linked
with more successful collaboration outcomes. Implications of greater sharing in the
asynchronous mode might include longer asynchronous interaction modes and
shorter, less frequent synchronous ones. This type of cycle may have unforeseen
influences on innovation that have yet to be explored.
Second, more work is needed to address how CT support helps individuals
sharing across boundaries when those boundaries shift or emerge (Carlile 2004).
Carlile argues that CT support for sharing and reflection can be a dynamic capability
(Teece et al. 1997) for an organization since it helps individuals manage novelty
when converting old (diverse) knowledge into new. Carlile acknowledges that
merely offering up knowledge as a boundary object in this type of environment is not
likely to promote innovation; thus further strategies for CT support may be needed
for not only sharing and reflection, but also dynamic reuse of previously shared
knowledge for novel situations.

185
7.5.2 Future Research in the Area of Virtual Collaboration
Because this research is among the first to distinguish between synchronous
and asynchronous interaction modes in virtual collaboration for innovation (see also
Burke and Chidambaram 1999 and Nowak et al. 2005), it was able to resolve some
conflicts in virtual collaboration research concerning successful collaboration
outcomes in virtual interaction modes. Extending this line of reasoning, this
paradigm should also be applied to help explain other conflicting findings. For
example, some research suggests use of CT support engenders greater
misunderstanding and negative conflict (Cramton 2001; Hinds and Bailey 2003),
while other research suggests use of CT support promotes understanding and positive
task conflict (Carlile 2004; Majchrzak et al. 2004). Affective complexity may also
depend on interaction mode characteristics (Nowak et al. 2005; Te’eni 2001) and it is
not currently understood how this complexity may be resolved differently in
synchronous vs. asynchronous interaction modes nor is it clear what differential
impact that may have on successful collaboration outcomes.

7.5.3 Future Research in the Area of CT Design
Evidence from the case study indicates that there are synergies in how CT
support is used, which may have had un-accounted for influence on the results. For
example, in some cases email was used to direct others to knowledge in the virtual
workspace for subsequent reflection. Thus elements of an attention focusing and

186
perspective taking strategy were mixed together. In this case it’s not clear if
individual’s reflection was a result of access to knowledge in the workspace or the
fact that their attention was directed there, or some combination of both. No
provision was made for assessing the impact of these synergies since the goal of this
research was to identify individual features of CT support and how they were useful
whereas previous models treated technology as a ‘black box.’ Future work is needed
to assess the presence and impact of these synergies.

7.6 Conclusion
This research has broken new ground for research on innovation and virtual
collaboration by explaining how and why virtual collaboration for innovation leads
to successful collaboration outcomes. By using a single theory that explains how
individuals learn and innovate, the theory the model provides a robust framework for
explaining how features of CT support are used in virtual collaboration for
innovation. It will help future designers of CT understand how new features of CT
support will be used and what value they will add.
Boland and Tenkasi (1995), Te’eni (2001) and Carlile (2004) have all
advocated for CT support for virtual collaboration for innovation, yet none were able
to specify exactly how that CT support should be used. Integrating across this legacy
of research however offers new insight into CT support that not only explains the
effectiveness of CT support now, but creates a framework for future research. As

187
features of CT support evolve, this framework will ensure its important role in
supporting virtual collaboration for innovation is acknowledged and expanded.

188
REFERENCES

Agarwal, R., & Karahanna, E. (2000) Time flies when you're having fun: Cognitive
absorption and beliefs about information technology usage. MIS Quarterly
24(4), 665-694.

Agarwal, R., Sambamurthy, V., & Stair, R.M. (2000) Research report: The evolving
relationship between general and specific computer self-efficacy - an
empirical assessment. Information Systems Research 11(4), 418-430.

Alavi, M., & Leidner, D.E. (2001) Review: Knowledge management and knowledge
management systems: Conceptual foundations and research issues," MIS
Quarterly 25(1), 107-136.

Amabile, T.M. (1982) Social psychology of creativity: A consensual assessment
technique. Journal of Personality and Social Psychology 43(5), 997-1013.

Amabile, T.M. (1988) A model of creativity and innovation in organizations. In
B.M. Staw & L.L. Cummings (eds.), Research in Organizational Behavior
(pp. 123-167), JAI Press, Greenwich, CT.

Amabile, T.M., Conti, R., Coon, H., Lazenby, J., & Herron, M. (1996) Assessing the
Work Environment for Creativity. The Academy of Management Journal,
39(5), 1154-1184.

Baron, R. S., Kerr, N. L., & Miller, N. (1992) Social influence and conformity,
Group process, group decision, group action. Pacific Grove, CA:
Brooks/Cole Publishing Company.

Barki, H., & Pinsonneault, A. (2001) Small group brainstorming and idea quality: Is
electronic brainstorming the most effective approach? Small Group Research
32(2), 158-205.

Baron, R. M. & Kenny, D. A. (1986) The Moderator-Mediator Variable Distinction
in Social Psychological Research: Conceptual, Strategic, and Statistical
Considerations. Journal of Personality and Social Psychology, 51(6), 1173-
1182.

Barua, A. & Mukhopadhyay, T. (2000) Information technology and business
performance: Past, present, and future. In Zmud, R.W. (Ed.), Framing the
Domains of IT Management (pp. 65-84), Pinnaflex, Cincinnati, OH.

189

Bergkvist, L., & Rossiter, J. (2007) The predictive validity of multiple-item versus
single-item measures of the same constructs. Journal of Marketing Research,
44, 175-184.

Benbunan-Fich, R., Hiltz, S.R., & Turoff, M. (2002) A comparative content analysis
of face-to-face vs. asynchronous group decision making. Decision Support
Systems, 34, 457-469.

Bhappu, A., Griffith, T. L., & Northcraft, G. B. (1997). Media Effects and
Communication Bias in Diverse Groups. Organizational Behavior and
Human Decision Processes, 70(3), 199-205.

Bock, G.-W., Zmud, R.W., Kim, Y.-G., & Lee, J.-N. (2005) Behavioral intention
formation in knowledge sharing: Examining the roles of extrinsic motivators,
social-psychological forces, and organizational climate. MIS Quarterly 29(1),
87.

Boland, R.J., & Tenkasi, R.V. (1995) Perspective Making and Perspective Taking in
Communities of Knowing. Organization Science 6(4), 350-372.

Boland, R.J., Tenkasi, R.V., & Te'eni, D. (1994) Designing information technology
to support distributed cognition. Organization Science 5(3), 456-475.

Bordetsky, A., & Mark, G. (2000) Memory-Based Feedback Controls to Support
Groupware Coordination. Information Systems Research, 11(4), 366-385.

Bradshaw, G. F. (1992) The airplane and the logic of invention. In R.N. Giere (Ed),
Cognitive Models of Science: Minnesota Studies in the Philosophy of Science,
XV, Minneapolis, LN: University of Minnesota Press.

Brandon, D.P. & Hollingshead, A.B. (2004) Transactive Memory Systems in
Organizations: Matching Tasks, Expertise, and People. Organization Science,
15(6), 633-644.

Brown, S., & Eisenhardt, K. (1995) Product development: Past research, present
findings, and future directions. Academy of Management Review, 20(2), 343-
378.

Bruner, J. S. (1990) Acts of Meaning. Cambridge, MA: Harvard University Press.

190
Burke, K., & Chidambaram, L. (1999) How Much Bandwidth Is Enough? A
Longitudinal Examination of Media Characteristics and Group Outcomes.
MIS Quarterly, 23(4), 557-579.

Campbell, D.T., & Fiske, D.W. (1959) Convergent and Discriminant Validation by
the MultiTrait-MultiMethod Matrix. Psychological Bulletin 56(2), 81-105.

Carlile, P.R. (2002) A pragmatic view of knowledge and boundaries: Boundary
objects in new product development. Organization Science, 13(4), 442-455.

Carlile, P.R. (2004) Transferring, translating, and transforming: An integrative
framework for managing knowledge across boundaries. Organization
Science, 15(5), 555-568.

Carlile, P.R., & Rebentisch, E.S. (2003) Into the black box: The knowledge
transformation cycle. Management Science, 49(9), 1180-1195.

Carte, T., & Chidambaram, L. (2004) A capabilities-based theory of technology
deployment in diverse teams: Leapfrogging the pitfalls of diversity and
leveraging its potential with collaborative technology. Journal of the
Association for Information Systems, 5(11-12), 448-471.

Chi, M. T. H., Bassok, M., Lewis, M. W., Reimann, P., & Glaser, R. (1989) Self-
explanations: how students study and use examples in learning to solve
problems. Cognitive Science, 13, 259-294.

Chidambaram, L., & Tung, L.L. (2005) Is Out of Sight, Out of Mind? An Empirical
Study of Social Loafing in Technology-Supported Groups. Information
Systems Research, 16(:2), 149.

Chin, W.W. (1998a) (1) Issues and Opinion on Structural Equation Modeling. MIS
Quarterly, 22(1), vii-xvi.

Chin, W.W. (1998b) (2) Issues and Opinion on Structural Equation Modeling. MIS
Quarterly, 22(1), vii-xvi.

Chin, W.W., & Gopal, A. (1995) Adoption Intention in GSS: Relative Importance of
Beliefs. DATA BASE for Advances in Information Systems, 26(2&3), 42-64.

191
Chin, W.W., Marcolin, B.L., & Newsted, P.R. (2003) A partial least squares latent
variable modeling approach for measuring interaction effects: Results from a
Monte Carlo simulation study and an electronic-mail emotion / adoption
study. Information Systems Research, 14(2), 189.

Chudoba, K.M., Wynn, E., Lu, M., & Watson-Manheim, M.B. (2005) How virtual
are we? Measuring virtuality and understanding its impact in a global
organization. Information Systems Journal, 15(4), 279.

Churchill, G.A., Jr. (1979) A Paradigm for Developing Better Measures of
Marketing Constructs. Journal of Marketing Research 16(1), 64-73.

Clark, H.H., & Brennan, S.E. (1991) Grounding in Communication. In L.B. Resnick,
J.M. Levine & S.D. Teasley (eds.), Perspectives on Socially Shared
Cognition, American Psychological Association, Washington, DC, 1991.

Cohen, S.G., & Gibson, C.B. (2003) In The Beginning: Introduction and Framework.
In C.B. Gibson & S.G. Cohen (eds.) Virtual Teams That Work, John Wiley &
Sons, Inc., San Francisco, CA.

Cowan, N. (1988) Evolving Conceptions of Memory Storage, Selective Attention,
and Their Mutual Constraints Within the Human Information-Processing
System. Psychological Bulletin, 104(2), 163-191.

Cramton, C.D. (2001) The Mutual Knowledge Problem and Its Consequences for
Dispersed Collaboration. Organization Science, 12(3), 346-371.

Cramton, C.D. (2002) Attribution in distributed work groups. In P. Hinds & S.
Kiesler, (Eds.), Distributed Work, MIT Press, Cambridge, MA.

Cramton, C.D., Orvis, K.L., & Wilson, J.M. (in press) Situation invisibility and
attribution in distributed collaboration. Journal of Management.

Cross, R., Parker, A., Prusak, L & Borgatti, S. (2001). Knowing What We Know:
Supporting Knowledge Creation and Sharing in Social Networks.
Organizational Dynamics, 3(2), 100-120.

Cross, R., & Sproull, L. (2004) More Than an Answer: Information Relationships for
Actionable Knowledge. Organization Science, 15(4), 446–462.

Cummings, J.N. (2004) Work groups, structural diversity, and knowledge sharing in
a global organization. Management Science, 50(3), 352-364.

192
Dennis, A. R. (1996) Information exchange and use in group decision making: You
can lead a group to information, but you can't make it think. MIS Quarterly,
20(4), 433.

Dennis, A. R., Hilmer, K. M., & Taylor, N. J. (1997) Information exchange and use
in GSS and verbal group decision making: effects of minority influence.
Journal of Management Information Systems, 14(3), 61-88.

Dennis, A.R., & Kinney, S.T. (1998) Testing media richness theory in the new
media: The effects of cues, feedback, and task equivocality. Information
Systems Research, 9(3), 256.

DeSanctis, G., & Monge, P. (1999) Introduction to the special issue: Communication
processes for virtual organizations. Organization Science, 10(6), 693.

DeSanctis, G., & Poole, M.S. (1994) Capturing the complexity in advanced
technology use: Adaptive structuration theory. Organization Science, 5(2),
121.

Dillenbourg P. (1999) What do you mean by collaborative learning? In P.
Dillenbourg (Ed), Collaborative-learning: Cognitive and Computational
Approaches (pp.1-19). Elsevier, Oxford.

Dillman, D.A. (1999) Mail and Internet Surveys: The Tailored Design Method, (2nd
ed.) Wiley Interscience, New York.

Dougherty, D. (1992) Interpretive barriers to successful product innovation in large
firms. Organization Science, 3(2), 179-202.

Edmondson, A.C. (1999) Psychological safety and learning behavior in work teams.
Administrative Science Quarterly, 44, 350–383.

Eisenhardt, K.M. (1989) Building Theories from Case Study Research. Academy of
Management Review, 14(4), 532-550.

Finholt, T., Sproull, L., & Kiesler, S. (1990) Communication and Performance in Ad
Hoc Task Groups. In J. Galagher, R. Kraut and C. Egido (eds.), Intellectual
Teamwork, Erlbaum, Hillsdale, NJ, 1990.

Fiol, C.M. (1994) Consensus, diversity, and learning in organizations. Organization
Science, 5(3), 403-420.

193
Fornell, C., and Larcker, D. (1981) Evaluating Structural Equation Models with
Unobservable Variables and Measurement Error. Journal of Marketing
Research, 18, 39-50.

Fulk, J., Flanagin, A., Kalman, M., Monge, P.R., & Ryan, T. (1996) Connective and
communal public goods in interactive communication systems.
Communications Theory, 6, 60-87.

Gallupe, R. B., Cooper, W. H., Grise, M.-L., & Bastianutti, L. M. (1994) Blocking
electronic brainstorms. Journal of Applied Psychology, 79(1), 77.
Gefen, D. (2003) Unidimensional validity: An explanation and example.
Communications of the AIS, 12(2), 23-47.

Gefen, D., & Straub, D. (2005) A practical guide to factorial validity using PLS-
Graph: Tutorial and annotated example. Communications of the AIS, 16, 91-
109.

George, J.M. (1990) Personality, affect, and behavior in groups. Journal of Applied
Psychology, 75, 107-116.

Gersick, C. (1988) Time and transition in work teams: Toward a new model of group
development. Academy of Management Journal, 31(1), 9-31.

Gibson, C.B., & Gibbs, J.L. (2006) Unpacking the concept of virtuality: The effects
of geographic dispersion, electronic dependence, dynamic structure, and
national diversity on team innovation. Administrative Science Quarterly,
51(3), 451-495.

Goodhue, D.L., & Thompson, R.L. (1995) Task-technology fit and individual
performance. MIS Quarterly, 19(2), 213.

Goodman, P.S., & Darr, E.D. (1998) Computer-aided systems and communities:
mechanisms for organizational learning in distributed environments. MIS
Quarterly, 22(4), 417-440.

Grant, R.M. (1996) Prospering in dynamically-competitive environments:
Organizational capability as knowledge integration. Organization Science
7(4), 375-387.

Gray, P.H. (2000) The Effects of Knowledge Management Systems on Emergent
Teams: Towards a Research Model. Journal of Strategic Information
Systems, 9(2&3), 175-191.

194
Gray, P.H., & Meister, D.B. (2004) Knowledge sourcing effectiveness. Management
Science, 50(6), 821-834.

Greitemeyer, T. & Schulz-Hardt, S. (2003) Preference-consistent evaluation of
information in the hidden profile paradigm: Beyond group-level explanations
for the dominance of shared information in group decisions. Journal of
Personality and Social Psychology, 84(2), 322-339.

Griffith, T.L., Sawyer, J.E., & Neale, M.A. (2003) Virtualness and knowledge in
teams: Managing the love triangle of organizations, individuals, and
information technology. MIS Quarterly, 27(2), 265-287.

Hair, J.F., Jr., Anderson, R.E., Tatham, R.L., & Black, W.C. (1988) Multivariate
Data Analysis with Readings, (Fifth ed.) Prentice Hall, Englewood Cliffs, NJ.

Hart, R.F., & Hart, M.K. (2004) Testing for "Near-Normality": The Probability Plot.
Statit Bulletin, Retrieved June 26, 2006
www.statit.com/support/quality_practice_tips/testingfornearnormality.htm.

Hightower, R., & Sayeed, L. (1996) Effects of communication mode and
prediscussion information distribution characteristics on information
exchange in groups. Information Systems Research, 7(4), 451-465.

Hinds, P.J., & Bailey, D.E. (2003) Out of sight, out of sync: Understanding conflict
in distributed teams. Organization Science, 14(6), 615-632.

Hinds, P.J., & Mortensen, M. (2005) Understanding Conflict in Geographically
Distributed Teams: The Moderating Effects of Shared Identity, Shared
Context, and Spontaneous Communication. Organization Science, 16(3),
290.

Hinkle, D.E., Wiersma, W., & Jurs, S.G. (2003) Applied Statistics for the Behavioral
Sciences, 5th Edition Houghton Mifflin, Boston.

Hollingshead, A.B., Fulk, J., & Monge, P.R. (2002) Fostering Intranet Knowledge
Sharing: An Integration of Transactive Memory and Public Goods
Approaches. In P.J. Hinds & S. Kiesler (eds.), Distributed Work, The MIT
Press, Cambridge, MA.

195
Hollingshead, A. B. & McGrath, J. E. (1995) Computer assisted groups: A critical
review of the empirical research. In R. A. Guzzo & E. Salas (Eds.), Team
Effectiveness and Decision making in Organizations. San Francisco: Jossey-
Bass.

Hulland, J.S. (1999) Use of Partial Least Squares (PLS) in Strategic Management
Research: A Review of Four Recent Studies. Strategic Management Journal,
20, 195-204.

James, L.R., Demaree, R.G., & Wold, G. (1984) Estimating within-group interrater
reliability with and without response bias. Journal of Applied Psychology,
69(1), 85-98.

Janz, B.D., Colquitt, J.A., & Noe, R.A. (1997) Knowledge worker team
effectiveness: The role of autonomy, interdependence, team development,
and contextual support variables. Personnel Psychology, 50(4), 877-904.

Jarvenpaa, S., & Leidner, D.E. (1999) Communication and trust in global virtual
teams. Organization Science, 10(6), 791-815.

Jehn, K.A., Northcraft, G.B., & Neale, M.A. (1999) Why difference makes a
difference: a field study of diversity, conflict, and performance in
workgroups. Administrative Science Quarterly, 44(4), 741-763.

John, O.P., & Benet-Martinez, V. (2000) Measurement: Reliability, Construct
Validation, and Scale Construction. In H.T. Reis & C.M. Judd (eds.),
Handbook of Research Methods in Social and Personality Psychology (pp.
339-369), Cambridge University Press, Cambridge.

Kalman, M. E., Monge, P., Fulk, J., & Heino, R. (2002) Motivations to resolve
communication dilemmas in database-mediated collaboration.
Communication Research, 29(2), 125-154.

Kashy, D.A., & Kenny, D.A. (2000) The Analysis of Data from Dyads and Groups.
In H.T. Reis & C.M. Judd (eds.), Handbook of Research Methods in Social
and Personality Psychology, Cambridge University Press, Cambridge.

Kerr, D.S., & Murthy, U.S. (2004) Divergent and Convergent Idea Generation in
Teams: A Comparison of Computer-Mediated and Face-to-Face
Communication. Group Decision and Negotiation, 13, 381-399.

196
Kidwell, R.E., & Bennett, N. (1993) Employee propensity to withhold effort: A
conceptual model to intersect three avenues of research. Academy of
Management Review, 18, 429-456.

Kiesler, S., & Cummings, J.N. (2002) What Do We Know About Proximity and
Distance in Work Groups? A Legacy of Research. In P.J. Hinds & S. Kiesler
(eds.), Distributed Work (pp. 57-80), The MIT Press, Cambridge, MA, 2002.

Kirkman, B.L., Rosen, B., Tesluk, P.E., & Gibson, C.B. (2004) The impact of team
empowerment on virtual team performance: The moderating role of degree of
face to face interaction. Academy of Management Journal, 47, 198-208.

Krauss, R.M., & Fussell, S. (1991) Perspective-taking in communication
representation of others' knowledge in reference. Social Cognition, 9(1), 2-
24.

Latane, B. (1981) The psychology of social impact. American Psychologist, 36(4),
343-356.

Leahey, T. H., & Harris, R. J. (1989) Human Learning. Prentice-Hall, Englewood
Cliffs, NJ.

Leonard, D. & Sensiper, S. (1998) The role of tacit knowledge in group innovation.
California Management Review, 40(3), 112-132.

Leonard, D. & Swap, W. (1999) When Sparks Fly: Igniting Creativity in Groups.
Boston, MA: Harvard Business School Press.

Levitt, B., & March, J.G. (1988) Organizational Learning. Annual Review of
Sociology, 14, 319-340.

Lindell, M.K., Brandt, C.J., & Whitney, D.J. (1999) A Revised Index of Interrater
Agreement for Multi-Item Ratings of a Single Target. Applied Psychological
Measurement, 23(2), 127-135.

Lindell, M.K., & Whitney, D.J. (2001) Accounting for common method variance in
cross-sectional research designs. Journal of Applied Psychology, 86(1), 114-
121.

Lindsay, P. & Norman, D.A. (1977) Human Information Processing. Academic
Press, New York.

197
Majchrzak, A., & Beath, C.M. (2006) A Framework For Studying The Process Of
Implementing Web-Based Learning Systems. In H. O’Neil and R. Perez
(Eds) A State of the Art Assessment for Web-Based Learning (pp.71-86),
Lawrence Erlbaum, Mahwah, NJ.

Majchrzak, A., Beath, C.M., Lim, R.A., & Chin, W.W. (2005a) Managing client
dialogues during information systems design to facilitate client learning. MIS
Quarterly, 29(4), 653.

Majchrzak, A., Malhotra, A., & John, R. (2005b) Perceived Individual Collaboration
Know-How Development Through Information Technology-Enabled
Contextualization: Evidence from Distributed Teams. Information Systems
Research, 16(1), 9.

Majchrzak, A., Malhotra, A., Lipnack, J., &Stamps, J. (2004) Can Absence Make A
Team Grow Stronger? Harvard Business Review, May issue, 131-137.

Malhotra, A., & Majchrzak, A. (2004) Enabling knowledge creation in far-flung
teams: best practices for IT support and knowledge sharing. Journal of
Knowledge Management, 8(4), 75.

Malhotra, A., & Majchrzak, A. (2005) Virtual Workspace Technologies. MIT Sloan
Management Review, 46(2), 11.

Malhotra, A., Majchrzak, A., Carman, R., & Lott, V. (2001) Radical innovation
without collocation: A case study at Boeing-Rocketdyne. MIS Quarterly,
25(2), 229.

Marks, M.A., Mathieu, J.E., & Zaccaro, S.J. (2001) A temporally based framework
and taxonomy of team processes. Academy of Management Review, 26(3),
356.

Markus, M.L. (2001) Towards a Theory of Knowledge Reuse: types of Knowledge
Reuse Situations and Factors in Reuse Success. Journal of Management
Information Systems, 18(1), 57-94.

Massey, A.P., Montoya-Weiss, M., & O’Driscoll, T. (2002) Knowledge management
in pursuit of performance: Insights from Nortel Networks. MIS Quarterly,
26(3), 269-289.

Maznevski, M.L., & Chudoba, K.M. (2000) Bridging space over time: Global virtual
team dynamics and effectiveness. Organization Science, 11(5), 473-492.

198
McGrath, J.E. (1991) Time, interaction, and performance (TIP) A theory of groups.
Small Group Research, 22(2), 147-174.

Millen, D.R., Fontaine, M.A., & Muller, M.J. (2002) Understanding the benefit and
costs of communities of practice. Association for Computing Machinery,
Communications of the ACM, 45(4), 69.

Montoya-Weiss, M.M., Massey, A.P., & Song, M. (2001) Getting it together:
Temporal coordination and conflict management in global virtual teams.
Academy of Management Journal, 44(6), 1251-1262.

Nijstad, B. A., Stroebe, W., & Lodewijkx, HFM. (2003) Production blocking and
idea generation: Does blocking interfere with cognitive processes? Journal of
Experimental Social Psychology, 39(6), 531-548.

Nonaka, I. (1994) A Dynamic Theory of Organizational Knowledge Creation.
Organization Science, 5(1), 14-37.

Nonaka, I., & Takeuchi, I. (1995) The Knowledge-Creating Organization. Oxford
Press, Oxford, UK.

Norman, D. A. (1982) Learning and Memory. San Francisco, CA: W. H. Freeman.

Nowak, K.L., Watt, J., & Walther, J.B. (2005) The influence of synchrony and
sensory modality on the person perception process in computer-mediated
groups. Journal of Computer-Mediated Communication, 10(3).

Nunnally, J.C. (1978) Psychometric Theory, (2nd ed.) McGraw-Hill, New York.

Olson, J.S., Teasley, S., Covi, L., & Olson, G. (2002) The (currently) unique
advantages of collocated work. In P.J. Hinds & S. Kiesler (eds.), Distributed
Work (pp. 113-136), The MIT Press, Cambridge, MA, 2002.

Petty, R. E. & Cacioppo, J.T. (1984) The effects of involvement on response to
argument quantity and quality: Central and peripheral routes to persuasion.
Journal of Personality and Social Psychology, 46, 69-81.

Pinsonneault, A., Barki, H., Gallupe, B., & Hoppen, N. (1999) Electronic
brainstorming: The illusion of productivity. Information Systems Research,
10(2), 110-133.

199
Pinsonneault, A., & Heppel, N. (1998) Anonymity in group support systems
research: A new conceptualization, measure, and contingency framework.
Journal of Management Information Systems, 14(3), 89-108.

Podsakoff, P.M., MacKenzie, S.B., Lee, J.-Y., & Podsakoff, N.P. (2003) Common
method biases in behavioral research: A critical review of the literature and
recommended remedies. Journal of Applied Psychology, 88(5), 879-903.

Postmes, T., Spears, R., & Lea, M. (2000) The formation of group norms in
computer-mediated communication. Human Communication Research,
26(3), 341–371.

Qureshi, S., & Zigurs, I. (2001) Paradoxes and prerogatives in global virtual
collaboration. Association for Computing Machinery. Communications of the
ACM, 44(12), 85.

Ramirez, A., Walther, J.B., Burgoon, J.K., & Sunnafrank, M. (2002) Information
seeking strategies, uncertainty, and computer mediated communication:
Toward a conceptual model. Human Communication Research, 28(2), 213-
228.

Reis, H.T., & Gable, S.L. (2000) Event-sampling and other methods for studying
everyday experience. In H.T. Reis & C.M. Judd (eds.), Handbook of
Research Methods in Social and Personality Psychology, Cambridge
University Press, Cambridge.

Robey, D., Khoo, H.M., & Powers, C. (2000) Situated learning in cross-functional
virtual teams. IEEE Transactions on Professional Communication, 51-66.

Rossiter, J. (2002) The C-OAR-SE procedure for scale development in marketing.
International Journal of Research in Marketing, 19, 305-335.

Saraf, N., & Langdon, C. (2006) The Moderating Role of IS Integration in
Augmenting Absorptive Capacity of Multi-Unit Firms. Presented at the
Annual Meeting of the Academy of Management, Atlanta, GA.

Schaefer, D.R., & Dillman, D.A. (1998) Development of a Standard E-mail
Methodology: Results of an Experiment. Public Opinion Quarterly, 62(3),
378-397.

200
Schunn, C., Crowley, K., & Okada, T. (2002) What makes collaborations across a
distance succeed? The case of the cognitive science community. In P.J. Hinds
& S. Kiesler (eds.), Distributed Work (pp. 113-136), The MIT Press,
Cambridge, MA, 2002.

Sheehan, H.E., Brenton, B.P., & Salkind, N.J. (2004) Statistics for People Who
(Think They) Hate Statistics, 2nd Edition, Sage.

Sivo, S.A., Saunders, C., Chang, Q., & Jiang, J.J. (2006) How low should you go?
Low response rates and the validity of inference in IS questionnaire research.
Journal of the Association for Information Systems, 7(6), 351-414.

Sole, D., & Edmondson, A. (2002) Situated knowledge and learning in dispersed
teams. British Journal of Management, 13, 17-34.

Stasser, G., & Titus, W. (2003) Hidden profiles: A brief history. Psychological
Inquiry, 14(3&4), 304-313.

Steiner, I. D. (1972) Group process and productivity. New York: Academic Press.
Straub, D., Boudreau, M., & Gefen, D. (2004) Validation Guidelines for IS Positivist
Research. Communications of the Association for Information Systems, 13.

Straub, D.W. (1989) Validating Instruments in MIS Research. MIS Quarterly, 13(2),
147-169.

Street, R.L., & Capella, J.H. (eds.) (1985) Sequence and Pattern in Communication
Behaviour. Edward Arnold, London.

Suedfeld, P., Tetlock, P.E., & Streufert, S. (1992) Conceptual/integrative
Complexity. In C.P. Smith (ed.), Motivation and Personality: Handbook of
Thematic Content Analysis, Cambridge University Press, Cambridge.

Swan, K., & Shea, P. (2005) The Development of Virtual Learning Communities. In
S.R. Hiltz & R. Goldman (eds.), Learning Together Online: Research on
Asynchronous Learning Networks (pp. 239-260), Lawrence Erlbaum
Associates, Mahwah, NJ.

Szulanski, G. (1996) Exploring external stickiness: Impediments to the transfer of
best practice within the firm. Strategic Management Journal, 17, 27-43.

201
Tallon, P., Kraemer, K., & Gurbaxani, V. (2000). Executives’ Perceptions of the
Business Value of Information Technology: A Process-oriented Approach.
Journal of Management Information Systems, 16(4), 137-165.

Teece, D.J., Pisano, G., & Scuen, A. (1997) Dynamic capability and strategic
management. Strategic Management Journal, 18, 509-533.

Te'eni, D. (2001) Review: A cognitive-affective model of organizational
communication for designing IT. MIS Quarterly, 25(2), 251-312.

Thomas-Hunt, M.C., Ogden, T.Y., & Neale, M.A. (2003) Who’s Really Sharing?
Effects of Social and Expert Status on Knowledge Exchange Within Groups.
Management Science, 49(4), 464-477.

Valacich, J.S., Wheeler, B.C., Mennecke, B.E., & Wachter, R. (1995) The effects of
numerical and logical group size on computer-mediated idea generation.
Organizational Behavior And Human Decision Processes, 62(3), 318-329.

Vandenbosch, B., & Higgins, C. (1996) Information acquisition and mental models:
An investigation into the relationship between behaviour and learning.
Information Systems Research, 7(2), 198-214.

Visser, P.S., Krosnick, J.A., & Lavrakas, P.J. (2000) Survey Research. In H.T. Reis
& C.M. Judd (eds.), Handbook of Research Methods in Social and
Personality Psychology, Cambridge University Press, Cambridge.

Vygotsky, L.S. (1978) Mind in Society. Harvard University Press, Cambridge, MA.

Walther, J.B. (1995) Relational aspects of computer mediated communication:
Experimental observations over time. Organization Science, 6(2), 186-203.

Walther, J.B. (1996) Computer-mediated communication: Impersonal, interpersonal,
and hyperpersonal interaction. Communication Research, 23(1), 3.

Wasko, M., & Faraj, S. (2005) Why should I share? Examining social capital and
knowledge contribution in electronic networks of practice. MIS Quarterly,
29(1), 1-23.

Webb, N.M., & Palincsar, A.S. (1996) Group processes in the classroom. In D.
Berliner & R. Calfee (eds.), Handbook of Research in Educational
Psychology (pp. 841-873), Prentice Hall, London.

202
Weick, K.E. (1995) Sensemaking in Organizations. Sage, Thousand Oaks, CA.

Weick, K. E. & Meader, D. K. (1993) Sensemaking and group support systems. In L.
Jessup & J. S. Valacich (Eds.), Group Support Systems: New Perspectives.
New York: Macmillan.

Weisband, S.P. (2002) Maintaining Awareness in Distributed Team Collaboration:
Implications for Leadership and Performance. In P.J. Hinds & S. Kiesler
(eds.), Distributed Work, The MIT Press, Cambridge, MA, 2002.

Wenger, E. (1998) Communities of Practice: Learning, Meaning, and Identity.
Cambridge University Press, Cambridge, UK.

Wittenbaum, G.M., Hollingshead, A.B., & Botero, I.C. (2004) From cooperative to
motivated information sharing in groups: Moving beyond the hidden profile
paradigm. Communication Monographs, 7(3), 59-83.

Wold, H. (1980) Model Construction and Evaluation When Theoretical Knowledge
is Scarce: Theory and Application of Partial Least Squares. In J. Kmenta &
J.B. Ramsey (eds.), Evaluation of Econometric Models (pp. 47-75),
Academic Press, NY.

Yuan, Y., Fulk, J., Shumate, M., Monge, P.R., Bryant, J.A., & Matsaganis, M.
(2005) Individual participation in organizational information commons: The
impact of team level social influence and technology-specific competence.
Human Communication Research, 31(2), 212-240.

Yin, R.K. (2002) Case Study Research: Design and Methods, 3rd Edition Sage,
Newbury Park.

Yoo, Y., & Alavi, M. (2001) Media and group cohesion: Relative influences on
social presence, task participation, and group consensus, MIS Quarterly,
25(3), 371.

Zack, M. H. (1993) Interactivity and communication mode choice in ongoing
management groups. Information Systems Research, 4(3), 207.

Zigurs, I. & Buckland, B. K. (1998) A theory of task/technology fit and group
support systems effectiveness. MIS Quarterly, 22(3), 313.

Abstract (if available)

Abstract This dissertation presents a new theory of collaborative technology (CT) support for virtual collaboration for innovation, that explains the differential effects CT support has on learning and innovation in synchronous vs. asynchronous interaction modes. Virtual collaboration for innovation is defined as the interaction of individuals from distinct knowledge domains, separated by time and space, and in which they share and combine their knowledge to develop and implement creative ideas. Synchronous and asynchronous are the two virtual interaction modes in which individuals collaborate, and are differentiated by the immediate or delayed ability to share knowledge and give and receive feedback. Each interaction mode, I argue, has different characteristics which increase the cognitive complexity associated with sharing diverse knowledge (sharing) and reflecting on others' shared knowledge (reflection), and in each mode individuals have different opportunities for realizing learning and innovation outcomes from sharing and reflection.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Building a unicorn: management of innovation, collaboration, and change in a Silicon Valley start-up

PDF

Carte blanche: experiments in the art of interactive political satire and contemplative virtual space

PDF

The role of the chief diversity officer in virtual exchange

PDF

Transforming teaching and learning with technology: a case study of a California public school

PDF

Training qualified rural area teachers in Malawi using synchronous virtual learning (SVL) information and communication technologies (ICTs) platforms

PDF

Exploring three outcomes of online teacher preparation: teaching for social justice, critical reflection, and voluntary collaboration

PDF

Leadership of ports and terminals in the global south and Oceania and the use of technological innovation during the crisis

PDF

Bounded technological rationality: the intersection between artificial intelligence, cognition, and environment and its effects on decision-making

PDF

The informationalization of race: communication technologies and genomics in the information age

PDF

Advanced machine learning techniques for video, social and biomedical data analytics

PDF

Digital fluency and critically conscious computing: a curriculum for undergraduates

PDF

Trust in the 100% remote workplace in high growth technology consulting firms

PDF

Navigating cultural integration: challenges and strategies for Taiwanese technology companies expanding into the United States

PDF

Motivation for participating in virtual religious communities: developing social capital

PDF

Adoption and implementation of innovative diagnostic tools for Alzheimer's Disease: challenges and barriers in primary care

PDF

Modeling dyadic synchrony with heterogeneous data: validation in infant-mother and infant-robot interactions

PDF

Essays on improving human interactions with humans, algorithms, and technologies for better healthcare outcomes

PDF

Knowledge, motivation, and organizational influences within leadership development: a study of a business unit in a prominent technology company

PDF

Increasing graduation and retention for at-risk youth in alternative education

PDF

Preparing English language learners to be college and career ready for the 21st century: the leadership role of secondary school principals in the support of English language learners

Asset Metadata

Creator Yates, Nathan David (author)

Core Title Technology support for virtual collaboration for innovation in synchronous and asynchronous interaction modes

School Marshall School of Business

Degree Doctor of Philosophy

Degree Program Information Systems

Publication Date 06/15/2007

Defense Date 06/07/2007

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag innovation,OAI-PMH Harvest,virtual collaboration

Language English

Advisor Majchrzak, Ann (committee chair), El Sawy, Omar (committee member), Hollingshead, Andrea (committee member), Scott, Steven (committee member)

Creator Email nyates@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m531

Unique identifier UC174785

Identifier etd-Yates-20070615 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-511432 (legacy record id),usctheses-m531 (legacy record id)

Legacy Identifier etd-Yates-20070615.pdf

Dmrecord 511432

Document Type Dissertation

Rights Yates, Nathan David

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu