knowledge sharing: an empirical study of the role of …choo.ischool.utoronto.ca/evans_thesis.pdf!...
TRANSCRIPT
Knowledge Sharing: An empirical study of the role of trust and other social-cognitive factors in an organizational setting
by
Michael Max Evans
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy
Faculty of Information University of Toronto
© Copyright by M. Max Evans 2012
ii
Knowledge Sharing: An empirical study of the role of trust and other social-cognitive factors in an organizational setting
M. Max Evans
Doctor of Philosophy Faculty of Information University of Toronto
2012
Abstract Effective knowledge sharing within project teams is critical to knowledge-intensive
professional service firms. Prior research studies indicate a positive association between
trust, social-cognitive factors, and effective knowledge sharing among co-workers. The
conceptual framework proposed here builds on these studies, and draws from theoretical
foundations from the organizational behavior, psychology, information studies,
sociology, and management literature on organizational trust and knowledge sharing, and
identifies the most significant factors found to influence organizational knowledge
sharing directly and indirectly through trust. The study makes methodological
contributions in the form of conceptualizations for knowledge sharing behavior, trust, and
tie strength. Also, it provides a more nuanced and focused analysis, by factoring for
knowledge type and co-worker working relationship.
Data were collected from 275 knowledge workers (‘legal professionals’ and paralegals)
engaged in shared legal project work, at one of Canada’s largest multijurisdictional law
firms. The nature of their work required a significant reliance on co-workers, for both
explicit and tacit knowledge. Multiple regression analysis, among other statistical
techniques, was used to test the hypotheses and determine significant relationships.
iii
Of the factors examined in the study, the three found to have the strongest effect on
respondents’ trust in their co-workers were shared vision, shared language, and tie
strength. Furthermore, the two factors found to have the strongest effect on organizational
knowledge sharing behavior were trust and shared vision. Overall trust was also found to
have a mediating effect between shared vision and knowledge sharing behavior, and
between shared language and knowledge sharing behavior.
A significant implication for practitioners is that effective knowledge sharing among co-
workers requires a nurturing manager to work on developing co-worker trust and shared
vision. Furthermore, a manager wanting to promote trust between co-workers must
nurture shared language and shared vision.
iv
Table of Contents Abstract ...........................................................................................................................................ii
List of Figures ................................................................................................................................vi
List of Tables.................................................................................................................................vii
Acknowledgments...........................................................................................................................x
Chapter 1: Introduction.................................................................................................................1 1.0 Chapter Overview .........................................................................................................................1 1.1 Problem Statement and Motivation .............................................................................................1 1.2 Purpose and Scope .......................................................................................................................2 1.3 Research Approach ......................................................................................................................3 1.4 Significance of the Study..............................................................................................................4 1.5 Structure of the Dissertation ........................................................................................................5
Chapter 2: Literature Review and Theoretical Framework ......................................................7 2.0 Chapter Overview .........................................................................................................................7 2.1 Knowledge in an Organizational Setting .....................................................................................7
2.1.1 Data, Information, and Knowledge .......................................................................................8 2.1.2 Tacit and Explicit Forms of Knowledge ..............................................................................11 2.1.3 Knowledge Sharing and Knowledge Sharing Behavior Defined .........................................18
2.2 Organizational Trust and Knowledge Sharing .........................................................................20 2.2.1 Understanding Organizational Trust...................................................................................22 2.2.2 Trust and Knowledge Sharing Behavior..............................................................................28
2.3 Social-Cognitive Factors and Trust ...........................................................................................35 2.3.1 Homophily and Trust ...........................................................................................................35 2.3.2 Shared Perspective and Trust ..............................................................................................41 2.3.3 Tie Strength and Trust .........................................................................................................42 2.3.4 Relationship Length and Trust.............................................................................................43
2.4 Social-Cognitive Factors and Knowledge Sharing ...................................................................45 2.4.1 Homophily and Knowledge Sharing Behavior ....................................................................45 2.4.2 Shared Perspective and Knowledge Sharing Behavior .......................................................49 2.4.3 Tie Strength and Knowledge Sharing Behavior ..................................................................51 2.4.4 Relationship Length and Knowledge Sharing Behavior ......................................................52
2.5 Conceptual Framework ..............................................................................................................52
Chapter 3: Research Design and Methodology .........................................................................57 3.0 Chapter Overview .......................................................................................................................57 3.1 Research Questions and Hypotheses .........................................................................................57 3.2 Operationalization of Variables .................................................................................................58
3.2.1 Social-Cognitive Variables (Independent Variables) ..........................................................58 3.2.2 Trust (Mediating Variable) ..................................................................................................65 3.2.3 Knowledge Sharing Behaviors (Dependent Variables) .......................................................74
3.3 Nature of Co-worker Relationship.............................................................................................82 3.4 Reliability and Validity ...............................................................................................................82 3.5 Data Analysis Strategy ...............................................................................................................83 3.6 Selection of Study Population ....................................................................................................84 3.7 Data Collection ...........................................................................................................................85
Chapter 4: Results ........................................................................................................................87 4.0 Chapter Overview .......................................................................................................................87 4.1 Survey Sample Size .....................................................................................................................87 4.2 Survey Respondents ....................................................................................................................87
4.2.1 Respondent Profile by Role and Department ......................................................................88
v
4.2.2 Respondent Profile by Gender .............................................................................................89 4.2.3 Respondent Profile by Birth, Country, Citizenship, and Ethnicity ......................................90 4.2.4 Respondent Profile by Education and Age ..........................................................................90
4.3 Analyses of Measures .................................................................................................................91 4.3.1 Knowledge Sharing Behavior (KSB) ...................................................................................91 4.3.2 Overall Trust ........................................................................................................................95 4.3.3 Shared Language and Shared Vision...................................................................................98 4.3.4 Relationship Length and Tie Strength ...............................................................................101 4.3.5 Homophily..........................................................................................................................104
4.4 Summary of Descriptive Statistics of Study Variables ............................................................107 4.5 Hypothesis Testing....................................................................................................................109
4.5.1 Age, Gender, and Educational Homophily ........................................................................109 4.5.2 Independent Variable 4 – Shared Language .....................................................................134 4.5.3 Independent Variable 5 – Shared Vision ...........................................................................140 4.5.4 Independent Variable 6 – Relationship Length .................................................................146 4.5.5 Independent Variable 7 – Tie Strength ..............................................................................150 4.5.6 Independent Variable 8 – Overall Trust ............................................................................158 4.5.7 Collective Effect of Social-Cognitive Factors and Trust on Knowledge Sharing Behavior....................................................................................................................................................162 4.5.8 Mediating Effect of Overall Trust ......................................................................................168
4.6 Summary of Hypothesis Tests ..................................................................................................180 4.7 Summary of Results by Research Question.............................................................................182
Chapter 5: Discussion and Summary .......................................................................................185 5.0 Chapter Overview .....................................................................................................................185 5.1 Discussion of the Research Findings ......................................................................................191
5.1.1 Relationships Between Social-Cognitive Variables and Trust ..........................................191 5.1.2 Research Question 1 Summary ..........................................................................................204 5.1.3 Relationships Between Social-Cognitive Variables and Knowledge Sharing Behavior ...204 5.1.4 Research Question 2 Summary ..........................................................................................219 5.1.5 Relationships Between Trust and Knowledge Sharing Behavior ......................................219 5.1.6 Research Question 3 Summary ..........................................................................................223 5.1.7 Collective Effect of Social-Cognitive Factors and Trust on Knowledge Sharing Behavior....................................................................................................................................................223 5.1.8 Research Question 4 Summary ..........................................................................................230 5.1.9 Mediating Effects of Trust Between Social-Cognitive Factors and Knowledge Sharing Behavior ......................................................................................................................................231 5.1.10 Research Question 5 Summary ........................................................................................237
5.2 Discussion of the Main Findings and Research Contributions .............................................238 5.3 Limitations of the Study ...........................................................................................................243 5.4 Implications for Practice ..........................................................................................................243 5.5 Conclusion and Future Work ..................................................................................................247
References ...................................................................................................................................250
Appendix .....................................................................................................................................269 A.1 Knowledge Definition Components.........................................................................................269 A.2 Survey Instrument....................................................................................................................279 A.3 Firm-wide Invitation Email .....................................................................................................291 A.4 University of Toronto Ethics Approval Letter ........................................................................293
vi
List of Figures Figure 2.1 Boisot’s Data, Information, and Knowledge Relationship ...........................................10 Figure 2.2 Tsoukas’ Representation of Polanyi’s Personal (Tacit) Knowledge ............................14 Figure 2.3 Original KSB Literature Map .......................................................................................21 Figure 2.4 Mayer, Davis, & Schoorman’s Proposed Model of Trust ............................................25 Figure 2.5 Conceptual Framework Based on Literature Review ...................................................56 Figure 2.6 Conceptual Framework of Decomposed Homophily Factors.......................................56 Figure 4.1 The Focal Relationship and Trust as a Mediating Variable........................................168 Figure 4.2 Significant Relationships and Corresponding Hypotheses for Positive Referents .....171 Figure 4.3 Significant Relationships and Corresponding Hypotheses for Negative Referents....176 Figure 5.1 Conceptual Framework...............................................................................................186 Figure 5.2 Collective Effect of Factors on Knowledge Sharing Behavior for Positive Referents
.............................................................................................................................................225 Figure 5.3 Collective Effect of Factors on Knowledge Sharing Behavior for Negative Referents
.............................................................................................................................................226 Figure 5.4 Summary of the Overall Mediating Effects of Trust for Positive Referents and
Negative Referents...............................................................................................................233
vii
List of Tables Table 2.1 Synonyms of Explicit Knowledge..................................................................................15 Table 3.1 Research Questions and Associated Hypotheses ...........................................................58 Table 3.2 Operationalization and Measurement of the Shared Perspective Research Variables...61 Table 3.3 Operationalization and Measurement of the Tie Strength Research Variable ...............64 Table 3.4 Operationalization and Measurement of the Ability-Based/Competence-Based Trust .69 Table 3.5 Operationalization and Measurement of the Benevolence-Based Trust ........................72 Table 3.6 Operationalization and Measurement of the Integrity-Based Trust ...............................73 Table 3.7 Operationalization and Measurement of the Propensity to Trust...................................74 Table 3.8 Operationalization and Measurement of Willingness to Share and Use Knowledge ....78 Table 3.9 Reliability Results for Measures of Holste’s Knowledge Variables..............................79 Table 3.10 Operationalization and Measurement of Levin and Cross’ Perceived Receipt of Useful
Knowledge Research Variable ..............................................................................................82 Table 4.1 Respondents by Role ......................................................................................................88 Table 4.2 Respondents by Role within the Firm by Department ...................................................89 Table 4.3 Respondents by Age.......................................................................................................90 Table 4.4 Factor Analysis Results for Knowledge Sharing Behavior............................................92 Table 4.5 Reliability Results for Combined Dependent Variables ................................................93 Table 4.6 Descriptive Statistics for Knowledge Sharing Behavior Variables ...............................95 Table 4.7 Factor Analysis Results for Trust...................................................................................96 Table 4.8 Reliability Results for Trust Variables...........................................................................97 Table 4.9 Descriptive Statistics for Trust Variables.......................................................................98 Table 4.10 Factor Analysis Results for Shared Language and Shared Vision...............................99 Table 4.11 Reliability Results for Combined Shared Language and Shared Vision Variables ...100 Table 4.12 Descriptive Statistics for Shared Language and Shared Vision Variables.................100 Table 4.13 Factor Analysis Results for Tie Strength Variables ...................................................102 Table 4.14 Reliability Results for Combined Tie Strength Variables..........................................103 Table 4.15 Descriptive Statistics for Tie Strength and Relationship Length Variables...............103 Table 4.16 Descriptive Statistics for Age and Age Difference ....................................................105 Table 4.17 Descriptive Statistics for Gender and Gender Difference ..........................................106 Table 4.18 Descriptive Statistics for Education and Educational Gap.........................................107 Table 4.19 Descriptive Statistics for Study Variables..................................................................109 Table 4.20 Correlations Between Age and Trust Variables .........................................................110 Table 4.21 Regression of Trust on Age and Other Independent Variables..................................111 Table 4.22 Correlations Between Age and Knowledge Sharing Behavior Variables..................113 Table 4.23 Regression of Knowledge Sharing Behavior on Age and Other Independent Variables
.............................................................................................................................................114 Table 4.24 Correlations Between Age Difference and Trust Variables.......................................116 Table 4.25 Regression of Trust on Age Difference and Other Independent Variables................117 Table 4.26 Correlations Between Age Difference and Knowledge Sharing Behavior Variables 117 Table 4.27 Regression of Knowledge Sharing Behavior on Age Difference and Other
Independent Variables .........................................................................................................118 Table 4.28 Correlations Between Education and Trust Variables ...............................................120 Table 4.29 Regression of Trust on Education and Other Independent Variables ........................121 Table 4.30 Correlations Between Education and Knowledge Sharing Behavior Variables .......121 Table 4.31 Regression of Knowledge Sharing Behavior on Education and Other Independent
Variables ..............................................................................................................................122 Table 4.32 Correlations Between Educational Gap and Trust Variables.....................................123
viii
Table 4.33 Regression of Trust on Educational Gap and Other Independent Variables..............124 Table 4.34 Correlations Between Educational Gap and Knowledge Sharing Behavior Variables
.............................................................................................................................................125 Table 4.35 Regression of Knowledge Sharing Behavior on Educational Gap and Other
Independent Variables .........................................................................................................126 Table 4.36 Correlations Between Gender and Trust Variables....................................................127 Table 4.37 Regression of Trust on Gender and Other Independent Variables.............................127 Table 4.38 Correlations Between Gender and Knowledge Sharing Behavior Variables.............128 Table 4.39 Regression of Knowledge Sharing Behavior on Gender and Other Independent
Variables ..............................................................................................................................129 Table 4.40 Correlations Between Gender Gap and Trust Variables ............................................131 Table 4.41 Regression of Trust on Gender Gap and Other Independent Variables .....................132 Table 4.42 Correlations Between Gender Gap and Knowledge Sharing Behavior Variables .....133 Table 4.43 Regression of Knowledge Sharing Behavior on Gender Gap and Other Independent
Variables ..............................................................................................................................134 Table 4.44 Correlations Between Shared Language and Trust Variables....................................135 Table 4.45 Regression of Trust on Shared Language and Other Independent Variables ............136 Table 4.46 Correlations Between Shared Language and Knowledge Sharing Behavior Variables
.............................................................................................................................................138 Table 4.47 Regression of Knowledge Sharing Behavior on Shared Language and Other
Independent Variables .........................................................................................................139 Table 4.48 Correlations Between Shared Vision and Trust Variables.........................................140 Table 4.49 Regression of Trust on Shared Vision and Other Independent Variables..................142 Table 4.50 Correlations Between Shared Language and Knowledge Sharing Behavior Variables
.............................................................................................................................................143 Table 4.51 Regression of Knowledge Sharing Behavior on Shared Vision and Other Independent
Variables ..............................................................................................................................145 Table 4.52 Correlations Between Relationship Length and Trust Variables ...............................146 Table 4.53 Regression of Trust on Relationship Length and Other Independent Variables........148 Table 4.54 Correlations Between Relationship Length and Knowledge Sharing Behavior
Variables ..............................................................................................................................149 Table 4.55 Regression of Knowledge Sharing Behavior on Relationship Length and Other
Independent Variables .........................................................................................................150 Table 4.56 Correlations Between Tie Strength and Trust Variables............................................152 Table 4.57 Regression of Trust on Tie Strength (Prior to)/Tie Strength (While on) and Other
Independent Variables .........................................................................................................153 Table 4.58 Correlations Between Tie Strength (Prior to) / Tie Strength (While on) and
Knowledge Sharing Behavior Variables .............................................................................155 Table 4.59 Regression of Knowledge Sharing Behavior on Tie Strength (Prior to) / Tie Strength
(While on) and Other Independent Variables ......................................................................158 Table 4.60 Correlations Between Overall Trust and Knowledge Sharing Behavior Variables ...159 Table 4.61 Regression of Knowledge Sharing Behavior on Overall Trust and Other Independent
Variables ..............................................................................................................................161 Table 4.62 Group 1 MRA Significant IV βs and Collective Model Results................................162 Table 4.63 Group 2 MRA Significant IV βs and Collective Model Results ...............................163 Table 4.64 Significant IV βs and Collective Model Results on Overall Willingness to Share
Knowledge in Groups 1 and 2 .............................................................................................164 Table 4.65 Significant IV βs and Collective Model Results on Willingness to Share Explicit and
Tacit Forms of Knowledge in Group 1................................................................................164
ix
Table 4.66 Significant IV βs and Collective Model Results on Willingness to Share Explicit and Tacit Forms of Knowledge in Group 2................................................................................165
Table 4.67 Significant IV βs and Collective Model Results on Overall Willingness to Use Knowledge in Groups 1 and 2 .............................................................................................165
Table 4.68 Significant IV βs and Collective Model Results on Willingness to Use Explicit and Tacit Forms of Knowledge in Group 1................................................................................166
Table 4.69 Significant IV βs and Collective Model Results on Willingness to Use Explicit and Tacit Forms of Knowledge in Group 2................................................................................167
Table 4.70 Significant IV βs and Collective Model Results on Perceived Receipt of Useful Knowledge in Groups 1 and 2 .............................................................................................167
Table 4.71 Results of the HMRA for Positive Referents (Group 1) ............................................172 Table 4.72 HMRA Model Change Summary for Positive Referents (Group 1) ..........................172 Table 4.73 Group 1 HMRA Results for Shared Language on Knowledge Sharing Behavior.....173 Table 4.74 Group 1 HMRA Results for Shared Vision ...............................................................174 Table 4.75 Results of the HMRA for Negative Referents (Group 2)...........................................177 Table 4.76 HMRA Model Change Summary for Negative Referents (Group 2) ........................177 Table 4.77 Group 2 HMRA Results for Shared Vision ...............................................................178 Table 4.78 Summary of the Mediating Effects of Overall Trust..................................................180 Table 4.79 Summary of Data Analysis and whether the Hypotheses were Supported ................181 Table 5.1 Significant Relationships Unique to the Nature of the Working Relationship ............242 Table 5.2 Summary of Abrams, Cross, Lesser and Levin’s Trust Builders and Associated
Managerial Actions..............................................................................................................245
x
Acknowledgments I am immensely proud of the work in this dissertation and I would never change the path
I took to get to this point. Especially because along the way, I have met many wonderful
people to whom I owe my gratitude for all their support, guidance, and encouragement
over the years. Without them, I could not have reached this incredible milestone in my
journey.
In moving to the United States, my parents made the decision to give up everything they
had and start a new life in a country they could not even speak the language of. Without
their incredibly brave decision, my opportunities and directions in life would have
certainly been quite different. I love you very much mom and dad and I understand and
truly appreciate the sacrifice you made on my behalf! I would also like to acknowledge
and dedicate this work to my grandmothers, who instilled in me the value of a good
education and the importance of being a well-rounded, cultured person. Most importantly
they inspired me to be patient, kind, and benevolent and to have strong values and ethics.
This may have taken me some time to learn, but better late than never! I would like to
thank Beata, who bravely endured hearing and reading more about my work than any
human alive. More importantly, she had to deal with my thesis-related anxiety over the
years, which she did very calmly and lovingly. You were right - everything turned out
great. Love you for being so supportive! Next, I would like to thank my Canadian
expatriate family, who were the driving force behind my coming to The University of
Toronto to complete my Masters and subsequent PhD (thanks for the idea, Len). Their
support and love provided much needed strength throughout this process. Finally, I would
like to thank Mark and Lora who have treated me like a member of their own family,
since the first day we met, and always had confidence in my abilities and have shown me
tremendous support and love. I am proud to have you as family.
My co-supervisors and committee members were exceptional at guiding me through the
scholarly process, yet still allowing me to learn, grow, and make my own decisions. I can
never choose the right words or write enough of them to thank Anthony and Chun Wei
for the amount of time and effort they invested in me, from my Masters onward. I truly
xi
look up to both these wonderful people and consider them to be exemplar academics,
personal mentors, heroes, and friends. The same may be said about Kelly, who, while
joining later in the process, was vital in shaping my research and helping me through the
most significant phases. I will never forget the belief they all had in my work and in me,
or the commitment and sacrifice they have made. It is true that we stand on the shoulders
of those that came before us.
There are several other people who were instrumental to the completion of the study and
who I owe an incredible debt of gratitude and acknowledgment. First, I would like to
thank my great friend Catalin, for his tireless comments, edits, support (both mental and
professional), encouragement, consultation, and of course English lessons. Everyone
should be so lucky as to have a friend like Cat in their life. Next, a big Thank You to
Joel, for making an invaluable introduction and for being a wonderful friend and
colleague for many years. Your encouragement meant a lot. Next, I would like to thank
Norm and Andrew for believing in the value of my research, for convincing senior
management of this value, and most importantly for helping me execute an incredibly
successful yet very elaborate survey. I would never have dreamt that we would get so
many professionals to take the time to complete such an intricate survey. Finally, thank
you to Melissa for generously and patiently teaching me how to clean up my data and for
showing me how to use a statistical tool I was unfamiliar with.
I would also like to acknowledge a number of individuals who were instrumental in
shaping and supporting my education over the years. Each of these people took from their
time to answer my questions, always listened and addressed my concerns, were always
up for a spirited debate, and most importantly provided excellent mentorship, guidance,
and support. First, thank you to all of the memorable and influential professors I had in
my undergraduate (especially Dr. T), my Masters, and my PhD (especially Lynne and
Bill). Thank you Dr. Scott, Dr. Herman, Dr. Joseph, and especially Dr. Steve who, as
senior doctoral students helped me understand and prepare for the challenges I would
face in my PhD. Thank you to my colleagues and visiting scholars from the KMRC
(especially Dr. Riva), from KMDI (the main reason I elected UofT), and from my cohort.
xii
Thank you to the Faculty of Information, KMDI, and UofT administrative staff, who
never tired of my questions. A special thank you to all my co-workers at UTM, who I am
proud to call friends. I have tremendously enjoyed teaching there over the years, but of
course it is easy to love your job when you are surrounded by wonderful co-workers.
Finally, thank you to all my students who have helped me shape and refine my teaching
skills and who keep me feeling young.
Lastly, I would like to acknowledge my friends, who are such an important part of my
life. First, to my friends in Chicago, who have continued to make an effort to stay close,
despite the hundreds of miles separating us (especially Andrew, Stanley and Joshua). I
know how easy it is to lose touch, and I appreciate it. A special thank you to Michael and
Shurik for their support throughout the years. They really define how good friends should
treat each other. I would also like to thank my friends in Toronto for welcoming me to
your city, including me in your circles of friends, and of course getting me addicted to
golf. I have had a great time studying in Toronto over the years. I would also like to
express my gratitude to Anthony and Karen who have always made me feel a part of their
family and their home. Last but most certainly not least, I would like to acknowledge
Sammy, who bravely made the trek to Toronto with me and has faithfully been by my
side ever since.
1
Chapter 1: Introduction 1.0 Chapter Overview
The first chapter introduces the present study, which examines factors that influence the
knowledge sharing behavior of knowledge workers in a professional service firm, directly
and indirectly through trust. The chapter begins with a brief statement of the research
problem and discussion of research motivation, purpose, and scope. Next, an overview of
the research approach and the significance of the study are discussed. The chapter
concludes with a summary of the structure of the dissertation.
1.1 Problem Statement and Motivation
A fundamental assumption of the knowledge-based view of the firm (Grant, 2002) is that
knowledge has market value and is one of the most productive resources for
organizations. In addition, knowledge is subject to economies of scale (i.e. initial creation
costs are higher than replication costs) and is a necessary resource for the production of
goods and services for the marketplace (Grant, 2002). The central importance of
knowledge to the production and the creation of value is an important area of study for
researchers in management disciplines, such as strategy, organizational science, and
organizational behavior. A better understanding of the nature of knowledge, its creation,
use, sharing, combination, and value, facilitates an improved understanding of the
functioning of firms in particular, and contemporary developed economies in general.
Such an understanding is both valuable in itself and also a potential source of strategies
and methodologies leading to the improved performance of firms and the economies
within which they are embedded.
It is difficult to argue with the fact that the effective sharing of knowledge has numerous
benefits for the organization and the individuals involved. Cyr and Choo (2010) give a
few examples of the organizational benefits, which include building on past experience
and knowledge, responding more quickly and efficiently to problems, developing new
ideas and insights, and avoiding reinventing the wheel or repeating past mistakes. These
benefits, however, are hard earned for the organization, since finding ways to promote
effective knowledge sharing behavior between employees can be a significant challenge.
2
One reason for this is because “knowledge-sharing behaviors in organizations are highly
complex social interactions” (Dalkir & Wiseman, 2004, p. 64), which are influenced by
trust and other social or cognitive human factors. The study of these human factors is a
relatively new focus area for knowledge management (KM) theorists, which means that
there is not much empirical evidence, or organizational protocol, that assists researchers
in understanding (or practitioners in promoting) effective knowledge sharing behaviors in
an organizational setting (Ruggles, 1998). “While KM is currently experiencing
enormous growth in popularity, little empirical research and development has been done
to better understand the factors influencing knowledge sharing” (Dalkir & Wiseman,
2004, p. 59).
1.2 Purpose and Scope
The phenomenon of interest in the present study is the knowledge sharing behavior
(KSB) of employees in organizations. Specifically, the research examines knowledge
workers on projects in a professional service firm. Of specific interest are trust and other
social factors that have been found to influence or inhibit knowledge sharing behavior in
this setting.
Since a majority of the existing empirical and theoretical research in this area point to
trust as an influencing factor for knowledge sharing behavior, trust was included as a
central factor to the study. For this research, the most important role of trust in the
organization is its ability to support or facilitate knowledge sharing behavior. “A clear
understanding of trust and its causes can facilitate cohesion and collaboration between
people” (Mayer et al., 1995, p. 710-711). Past research has shown trust to have this effect
through an increase in willingness to share information and ideas with others (Davenport
& Prusak, 1998; Empson, 2001; McDermott & O’Dell, 2001; Husted & Michailova,
2002; Hendricks, 1999; Hinds & Pfeffer, 2003; Szulanski 1995; 1996) and decrease in
the time and effort associated with information search and processing (Zaheer, McEvily,
3
& Perrone, 1998; Limerick & Cunnington, 1993; Roberts & O’Reily, 1974; Husted &
Michailova, 2002; Orlikowski, 1993; Mayer et al., 1995; Uzzi, 1997; Burt, 1992).1
This study seeks to expand the understanding of the relationship between trust and
knowledge sharing behavior by answering the overarching research question: what are
the factors that influence knowledge sharing behavior directly and indirectly through
trust? Additional influencing factors were identified through a review of organizational
literature and empirical studies that focused on social or cognitive factors, which
influence organizational trust, especially in the context of knowledge sharing. Numerous
factors were identified and an additional five constructs were added to the study. These
include shared language, shared vision, homophily, tie strength, and relationship
duration. The study sought to test the direct influence of each of these social and
cognitive factors (constructs) on both trust and knowledge sharing behavior. In addition,
the mediating effect of trust was tested between each of these factors and knowledge
sharing behavior. Finally, the collective effect of these factors and trust on knowledge
sharing behavior was tested.
1.3 Research Approach
In the present study, an exploratory approach was used to conduct empirical research on
knowledge workers, at a large legal professional service firm. The study collected
quantitative data from 275 participants, using a web-based survey as a primary research
instrument. The goal of the survey was to understand which of the social and/or cognitive
factors influenced knowledge sharing behavior, directly and indirectly through trust.
Statistical analyses were used to test the hypotheses, which assisted in answering the
overarching and sub-research questions. Statistical analyses of the data included factor
and reliability analysis, correlation analysis, t-tests, and multiple regression analysis. The
mediating (i.e. indirect) effect of trust was tested using hierarchical multiple regression
analysis and steps outlined by Baron and Kenny (1986). Additional data, gathered from
1 A complete discussion of the relationships between trust and knowledge sharing behavior is discussed in Chapter 2.
4
site visits and un-structured interviews with senior executives at the firm, were used to
interpret the findings of the study.
1.4 Significance of the Study
In theoretical terms, the present study contributes to earlier research on knowledge
sharing behavior and trust by building a conceptual framework that includes several of
the most significant factors found to influence or inhibit knowledge sharing behavior
and/or trust in previous empirical studies. This study extends previous research by
investigating the direct influence of these factors on knowledge sharing behavior and
trust. The study also tests whether trust exerts a mediating influence between these
factors and knowledge sharing behavior. This is important, since few research studies
deem trust as a mediating variable; and no known study examines the mediating
influence of this number of factors through trust, onto knowledge sharing behavior. In
addition, no known study considers this number of factors, or tests their collective
influence on knowledge sharing behavior.
The use of a larger set of social and cognitive factors in this study is significant for at
least three reasons. First, measuring the influence of each of the factors on trust and
knowledge sharing behavior can provide a comprehensive analysis of the motivators of
both (as well as their relative levels of influence). Second, this larger set of factors can
form the basis for an empirical model that better predicts, or explains, the existence and
the level of trust/knowledge sharing behavior between parties, in a work relationship.
Finally, the examination of the inter-relatedness of trust, social and cognitive factors, and
knowledge sharing behavior provides a better understanding of the true influencing and
inhibiting factors on effective knowledge sharing behavior in an organizational setting.
The present study also extends the understanding of knowledge sharing behavior found in
previous studies. This perspective takes into account three separate knowledge sharing
behaviors, including: the willingness of the knowledge owner to share their knowledge;
the willingness of the knowledge receiver to use the knowledge shared; and the perceived
benefit or utility of the shared knowledge. Each of these knowledge sharing behaviors has
5
been operationalized and used in previous organizational studies, but this is the first study
where all three elements are included, for a better understanding of organizational
knowledge sharing behavior. An in-depth comprehension of these behaviors was also the
reasoning behind asking participants to answer survey items about two co-workers they
mentally selected based on the nature of their working relationship. For example, the
respondent is asked to comment on one co-worker with whom he or she had a positive
working relationship, and another with whom he or she had a negative working
relationship with. Other important distinctions the study makes include between the form
of knowledge transferred (overall, explicit, and tacit) and the type of trust (overall,
ability-based, integrity-based, and benevolence-based).
In practical terms, the research study helps practitioners in professional service firms
understand the factors that lead to effective knowledge sharing behavior and, more
importantly, suggests methods to foster, promote, and improve them. The reality is that it
can be expensive and time consuming to implement strategies that promote trust and
every social or cognitive factor found to positively influence knowledge sharing behavior
in the firm. In addition, many of the previously identified influencing factors may not
appear as significant, when others are considered at the same time. The results of the
research study provide practitioners guidance and direction with respect to which factors
are most important, or most significant, for the firm to focus its resources on. The study
also offers some suggestions, for practitioners, on the ways they may promote and
nurture the most important factors found.
1.5 Structure of the Dissertation
The rest of this dissertation is organized as follows. Chapter 2 introduces the significant
research questions and the supporting literature review, which is presented in four parts:
Section 2.1 introduces the main phenomenon of interest by defining organizational
knowledge, distinguishing between explicit and tacit knowledge, and defining knowledge
sharing behavior; Section 2.2 examines the literature and theories on relationships
between trust and knowledge sharing behavior; Section 2.3 defines each of the identified
social-cognitive factors and explores the literature on relationships between these factors
6
and trust; and Section 2.4 examines the literature on relationships between the identified
social-cognitive factors and knowledge sharing behavior. Chapter 2 concludes by
presenting a conceptual framework for the research study. Chapter 3 details the research
design and methodology used in the study, beginning with a discussion of the research
questions and associated hypotheses. Next, data measurement and data analysis strategies
are discussed. The chapter concludes with a discussion of the study population, site, and
data collection methods. Chapter 4 presents the results of the statistical analyses and the
testing of the hypotheses using correlation, regression, and mediation analysis. The
chapter closes with a summary of the results by research question and a summary of
whether the hypotheses were upheld. Chapter 5 discusses each of the research findings,
compares them to previous research, and suggests possible reasoning behind the
relationships. The chapter concludes with a discussion of the main findings and research
contributions, the limitations of the study, and the implications for future research as well
as practice.
7
Chapter 2: Literature Review and Theoretical Framework 2.0 Chapter Overview
The following is a review of the literature addressing the main research question: what
are the factors that influence knowledge sharing behavior directly and indirectly
through trust? To further understand this research question, three additional questions
were posed. What are the significant relationships between trust and knowledge sharing
behavior? What are the significant relationships between social-cognitive variables and
trust? And, what are the significant relationships between social-cognitive variables and
knowledge sharing behavior?
Based on these questions, the literature review is presented in four parts: Section 2.1
Knowledge in an Organizational Setting introduces the main phenomenon of interest by
defining organizational knowledge, distinguishing between explicit and tacit knowledge,
and defining knowledge sharing behavior; Section 2.2 Trust and Knowledge Sharing
defines trust and examines the literature and theories on relationships between trust and
knowledge sharing behavior; Section 2.3 Social-Cognitive Factors and Trust defines
each of the identified social-cognitive factors and explores the literature on relationships
between these factors and trust; and finally, Section 2.4 Social-Cognitive Factors and
Knowledge Sharing examines the literature on relationships between the identified social-
cognitive factors and knowledge sharing behavior. The chapter concludes by presenting a
conceptual framework for the research study.
2.1 Knowledge in an Organizational Setting
A great deal of research in knowledge management (KM) is concerned with
epistemological issues relating to how knowledge is acquired and how it might be
differentiated from opinion, belief, and other related concepts. Although these issues are
fundamentally important to explore and explicate for any researcher in KM, they are not
the focus of this research. This research is concerned with examining organizational
knowledge sharing behavior, the lack of which has been identified by Hendricks (1999)
as a significant barrier to effective knowledge management. To fully understand, define,
and frame knowledge sharing behavior, it must first be differentiated from other similar
8
constructs in the field of knowledge management, including data, information,
knowledge, and knowledge sharing.
2.1.1 Data, Information, and Knowledge
Some authors (Alavi & Leidner, 2001; Earl, 2001) feel comfortable using the terms
information and knowledge interchangeably, since they argue for little practicality in
making a distinction. This approach, perhaps, tends to arise from a systems or computer
science background and although it might prove to be easier for scientific inquiry (i.e.
being easily quantifiable and measurable), it does not reflect significant distinctions,
which are discussed below (Tsoukas 2005a, Boisot 1998, 2002; Choo 1998; Davenport &
Prusak 1998; Nonaka, 2002; Nonaka & Takeuchi 1995; Leonard & Sensiper, 2002;
Thompson & Walsham, 2004; Huber, 1991).
Data, information, and knowledge are three independent constructs that can be
considered as constituent elements of a continuum (Tsoukas, 2005a; Nonaka &
Takeuchi, 1995; Nonaka, 2002; Boisot, 1998, 2002; Leonard & Sensiper, 2002). Boisot
(1998) explains the relationship between the three by simply stating, “knowledge builds
on information that is extracted from data” (p. 12). Leonard and Sensiper (2002) argue
that “knowledge is a subset of information” (p. 485). Nonaka and his colleagues (Nonaka
& Takeuchi, 1995; Nonaka, 2002) view data, information, and knowledge as active
rearrangements of each other, where “information is a flow of messages [or meanings],
while knowledge is created by that very flow of information, anchored in the beliefs and
commitment of its holder” (Nonaka & Takeuchi, 1995, pp.58-9). Nonaka (2002) also
asserts that “information is a necessary medium or material for initiating and formalizing
knowledge” (p. 439). Huber (1991) and Boisot (2002) present a similar intellectual
framework, referring to knowledge as interpreted information. For example, Boisot
(2002) notes that:
“[…] it is never knowledge as such that flows between agents, but rather data from which information has to be extracted and internalized. Only when information has been successfully internalized and forms part of an agent's repertoire of expectations and behaviors can it properly be called knowledge” (p. 72).
9
Tsoukas (2005a) locates the meaning of the terms data, information, and knowledge
along a continuum, “depending on the extent to which they reflect human involvement
with, and processing of, the reality at hand […] put simply, data require minimal human
judgment, [and] knowledge maximum judgment” (p. 120).
Data
In information systems terminology, one may think of data as inputs and outputs from a
system (e.g. numbers, characters, images). According to Boisot, data is ‘a discrimination
between physical states’ (1998, p. 12) which is ‘located in the world’ (2002, p. 67) and
‘can be characterized as a property of things’ (1998, p. 12). It is not necessary for data to
convey information to agents, and two separate agents could interpret the same piece of
data as two distinct pieces of information. Extracting the patterns within the data is a
creative task of the agent and can be unique for each agent (Boisot, 2002). Data is often
spoken of as being captured, processed, stored, or disseminated. As mentioned earlier, it
is data, as opposed to knowledge, that flow between agents and systems.
Information
Information can be thought of as a ‘flow of messages’ (Nonaka, 2002, p. 438) that
establishes a relationship between things and agents (Boisot, 1998). This relationship is
best described in Boisot’s (1998) diagram (Figure 2.1), which argues that data are
inherent to objects and events (things) and that agents use their ‘perceptual or conceptual
filters’ (p. 12) to create a subset of these data from the objects and events. Once this
subset of data is created within the agent (i.e. interpretation has taken place) an
established relationship between the two (i.e. agent and object) is formed. This
established relationship between data source and agent is called information.
10
Figure 2.1 Boisot’s Data, Information, and Knowledge Relationship (1998, p. 12)
Knowledge
Using an activity theoretical approach, this research defines knowledge as:
The potential of an activity, situated within a socially constructed domain and bounded by the developmental capacity of the individual.
Although this definition is not presented in precisely this form by any other authors, its
components encompass the core ideas of many noted philosophers, epistemologists, and
organizational theorists. This research does not attempt to resolve a two-thousand-year-
old debate on the definition of knowledge, rather to provide a definition suitable for and
consistent with the rest for the research. It covers what this researcher feels are the most
notable components of knowledge.
Elaborating on these components is important but outside the immediate scope of this
research. Therefore, a thorough explanation and justification for this definition is
provided in the Appendix, along with a relevant peer-reviewed literature review. The five
components of the definition were carefully considered as follows:
1. Knowledge is created, interpreted, disseminated, and displayed through activity;
2. Knowledge is situated within a particular domain;
3. Knowledge is socially constructed and interpreted;
4. Knowledge is personal and bounded by developmental capacity;
5. There is a potential to knowledge (e.g. domain, social construction, and
developmental capacity are partial determinants of the potential value of
knowledge).
11
2.1.2 Tacit and Explicit Forms of Knowledge
Tacit Knowledge
Tacit knowledge is personal and embodied (Nonaka, 2002; Spender 1996b; Polanyi,
1962, 1966), which makes its codification (formalization) and dissemination very
difficult (Nonaka, 2002). Polanyi (1966) famously explained this by stating “I shall
reconsider human knowledge by starting from the fact that we can know more than we
can tell” (p. 4). Most tacit knowledge will find “involvement in a specific context”
(Nonaka, 2002, p. 439) or closely associate with a particular “social or collective
identity” (Spender, 1996b, p.53). If the agent is not associated with the specific domain of
action, then attaining the tacit knowledge becomes next to impossible.
Tacit knowledge, which is somewhat similar to Aristotle’s concept of phronesis
(Thomson, 1955) or practical wisdom, is rooted in the actions of an agent, and is revealed
to the world through practice, action, or activity (Brown & Duguid, 1998; Spender,
1996b; Nonaka, 2002; Leonard & Sensiper, 2002). Ryle (1949) argued this point by
saying that ‘knowing how’ puts ‘knowing that’ into practice. Possessing the relevant
‘know-how’2 makes the knowledge “actionable and operational” (Brown & Duguid,
1998, p. 95).
According to Nonaka (2002), tacit knowledge involves both technical and cognitive
elements. The cognitive elements are working/mental models created by the agent, to
form an understanding of the world around them. These cognitive elements consist of
“analogies, schemata, paradigms, beliefs, and viewpoints” (p. 439). These elements are
also difficult to formalize and disseminate since they are developed within particular
contexts (domains) and may vary across different domains. The technical elements of
tacit knowledge refer to “concrete know-how, crafts, and skills that apply to specific
contexts” (Nonaka, 2002, p. 439) and might best be described using Polanyi’s/Tsoukas’
concept of subsidiary awareness.
2 Particulars need not be apparent
12
Building on the work of Polanyi3, Tsoukas (2005b) discusses the concept of subsidiary
awareness with respect to the ‘know more than we can tell’ element of tacit knowledge.
Agreeing with Polanyi, Tsoukas (2005b) asserts that tacit skills retain opacity and un-
specificity in terms of their particulars. Since these particulars are unknown, the skill
itself cannot be fully accounted or explicated. In other words, the practitioner is able to
perform the skill, without having theoretical knowledge of the particulars involved. For
Tsoukas (2005b),
“Tacit knowledge consists of a set of particulars of which we are subsidiarily aware as we focus on something else. Tacit knowing is vectorial: we know the particulars by relying on our awareness of them for attending to something else” (p. 158).
Perhaps this is best explained through example. Suppose one is driving in the rain and the
car in front suddenly brakes, causing them to skid-stop. In a classroom setting, the driver
may have been told to slightly depress the brakes and steer in the opposite direction of the
way the car skids. In the heat of the moment, however, it is unlikely that the driver will
recall the theoretical lesson. Instead, the driver would rely on their tacit knowledge of
driving under these and other conditions, accumulated over the course of months and
years. They will also not consciously attend to moving their hands on the steering wheel
and their foot on the brake. Thus, based on prior experience, the driver will form a
reaction, which will vary according to the accumulated knowledge. The less experience,
the less likely for the correct or prudent action to occur.
According to Polanyi (1962), tacit knowing occurs through a process of unconscious trial
and error. This trial and error is a feeling-out process, where the agent is improving in
success, over time, without specifically knowing (in a theoretical sense) how. For Nonaka
(2002) tacit knowing is a continuous activity, which develops through action and
reaction. For Choo (2000), tacit knowing is achieved through “extended periods of
experiencing and doing a task, during which the individual develops a feel for and a
capacity to make intuitive judgments about the successful execution of the activity” (p.
395).
3 Polanyi’s subsidiary awareness is discussed in the Instrumentalization section
13
The ability to display the instrumentalization or skill with a tool, what Aristotle called
techne (Thomson, 1955), is one of the few undeniable ways an agent can put their tacit
knowledge on display for the world to see. According to Polanyi (1966), “we can,
accordingly, interpret the use of tools, of probes, and of pointers as further instances of
the art of knowing” (p. 7). In its most basic form, the instrumentalization of an object/tool
simply means to be able to use it properly and un-problematically. Some may call this the
embodiment of a tool; Yuasa (1987, p.25) called it ‘learning with the body’ (tainin/
taitoku) and Boisot (2002, p.73) ‘learning-by-doing’. To achieve mastery with a tool, one
must first learn to use the tool in its intended, domain specific, manner. Polanyi (1962,
1966; Polanyi & Prosch, 1975) and Tsoukas (2005b) suggested that this is realized
through the agent assimilating and dwelling in the tool (i.e. making it feel as if it is an
extension of the body). In order to ‘dwell’ in the tool the agent must be able to focus their
attention through the tool onto the target. The tool must not be in the agent’s focal
awareness4. Instead it “needs to become an instrument through which [they] act [and] of
which [they] are subsidiarily aware” (Tsoukas, 2005b, p. 149). According to Polanyi
(1966), “in an act of tacit knowing we attend from something for attending to something
else; namely, from the first term [proximal] to the second term [distal] of the tacit
relation” (p. 10) and “we are attending from these internal processes to the qualities of
things outside, [t]he transposition of bodily experiences into the perception of things
outside” (p. 14).
Tsoukas (2005b) provides a couple of examples to illustrate this point:
“I have a subsidiary awareness of my holding the hammer in the act of focusing on hitting the nail. In being subsidiarily aware of holding a hammer I see it as having a meaning that is wiped out if I focus my attention on how I hold the hammer. If a pianist shifts her attention from the piece she is playing to how she moves her fingers; if a speaker focuses his attention on the grammar he is using instead of the act of speaking” (p. 146).
4 To place an object into focal awareness is to make it the object of attention
14
According to Tsoukas (2005b), tacit knowledge forms a triangular relationship between
subsidiary particulars, the knower and the focal target (Figure 2.2). By acting (i.e. taking
action), the knower links the subsidiary particulars to the focal target. “Subsidiary, or
instrumental knowledge, is not known in itself, but is known in terms of something
focally known” (Tsoukas, 2005b, p. 156). Without the knower, there is no way to
integrate subsidiaries to the focal target. Therefore, as previously mentioned, tacit
knowing is personal and rooted in action5.
The result of ongoing usage of the tool is two-fold. As the agent gains experience using
the tool, they become less aware of how to use the tool to achieve optimal results
(Tsoukas, 2005a). In addition, the instrumentalization of actions6 enables the agent to
increase their awareness of the situation and refine and improve their skill with the tool
(Tsoukas, 2005a). In Tsoukas’ (2005a) words, “the ongoing process of transforming
experience into subsidiary awareness […] allows one to reach ever higher levels of
skilful achievement” (p. 128). Choo (2000) proposed that tacit knowledge may be,
“revealed through rich modes of discourse that include the use of analogies, metaphors or
models, and through the communal sharing of stories” (p. 396).
Explicit Knowledge
Explicit knowledge is routinely defined as knowledge that can be expressed formally
using some system of symbols or formal systematic language (Choo, 1998, 2000; 5 Tacit knowing is personal because it requires the knower. It is action oriented, because it requires subsidiaries to be linked to focal targets by the knower (requires the knower to act). 6 Actions must be purposeful and justified
Figure 2.2 Tsoukas’ (2005b) Representation of Polanyi’s Personal (Tacit) Knowledge
15
Nonaka, 2002). Explicit knowledge is understood to exist independently from the human
agent who is the knower (De Long & Fahey, 2000). Choo (1998, 2000) further divided
explicit knowledge into object-based or rule-based. Object-based explicit knowledge is
embedded into artifacts and is usually represented using a string of symbols, or is
embodied in the entity itself (Choo, 2000). Some examples given by Choo (2000)
include: products, patents, software code, computer databases, technical drawings, tools,
prototypes, photographs, voice recordings, and films (p. 396). Explicit knowledge is rule-
based when it takes the form of rules, routines, or operating procedures (Choo, 2000).
Other examples of explicit knowledge include: documents, pictures, stories, diagrams and
narratives (Brown & Duguid, 2000, p. 76).
Synonym Author
Codified Nonaka and Takeuchi (1995); Nonaka (2002); Boisot (1998)
Structured De Long and Fahey (2000) Encoded (Similar to Object-Based)
Blackler (2002); Thompson and Walsham (2004)
Embedded (Similar to Rule-Based)
Blackler (2002)
Articulated Knowledge Choo and Bontis (2002) Table 2.1 Synonyms of Explicit Knowledge
The formal expression of knowledge into a system of categories and symbols is called
codification. According to Boisot (2002), codification “refines the categories that the
agent invokes or creates so that it can use them efficiently and in discriminating ways.
The fewer data an agent has to process to distinguish between categories, the more
codified the categories that it has to draw upon” (p. 68). Following a resource-based view
of knowledge7, it is commonly accepted that the more codified something becomes, the
easier it is to disseminate without loss of integrity (Boisot, 2002; Nonaka & Takeuchi,
1995; Choo, 1998, 2000; Choo & Bontis, 2002; De Carolis, 2002; Spender, 1994). This is
not always true, as there might be issues from the receiving party in decoding,
comprehending, and assessing the value of the new explicit, codified knowledge.
7 A resource-based view of knowledge views knowledge as an objective transferable commodity
16
Making Tacit Knowledge Explicit
It has been argued that the level to which something can be codified determines its
placement on the explicit/tacit knowledge continuum. The problem with this viewpoint is
that it assumes all tacit forms of knowledge can become codified. This school of thought
is rather consistent with an information systems perspective that reduces all knowledge to
that which has been codified and that which has yet to be codified. The belief is that
knowledge can be dragged along a supposititious continuum, from tacit to explicit and
back. In other words, all tacit knowledge may be reduced or articulated to an explicit or
codified form.
Polanyi (1962) insisted that tacit knowledge could not be reduced to explicit. Tsoukas
(2005ab) added that tacit knowledge may also not be converted, captured, or translated,
but merely put on display. One reason is that some dimensions of knowledge can never
become fully explicated (Leonard & Sensiper, 2002; Boisot, 2002). This is what Polanyi
(1966) referred to, when he said that “we know more than we can tell” (p. 4). De Carolis
(2002) explained this by arguing that difficulties arise when the knower makes an attempt
to communicate their tacit knowledge, because they realize that they cannot fully
articulate it (i.e. part of it remains tacit). The reality is that agents often know more than
they even realize, making it impossible for the knower to fully articulate what is known.
Any attempt to codify the tacit knowledge will ultimately result in an incomplete
representation, since some of the knowledge stays with the knower (Boisot, 2002).
Hence, one should consider explicit and tacit as two interdependent forms of knowledge,
necessary in all acts of knowing. As Duguid (2005) asserted, “codification is remarkably
powerful, but its power is only released through the corresponding knowing how, which
explains how we get to know and learn to do” (p. 114). Polanyi (1962) also argued that
all codified knowledge requires the skill of a knower, before it can be put into practice.
Tsoukas (2005b) explained this by stating that “even the most theoretical form of
knowledge, such as pure mathematics, cannot be a completely formalized system, since it
is based for its application and development on the skills of mathematicians and how
such skills are used in practice” (p. 142). Duguid (2005) made a similar case arguing that
17
“while knowledge may include codified content, to be used it requires the disposition to
apply it” (p. 111). In other words, to know requires both knowing how something is done
and how to do it (which only resides in the agent). In Blackler’s (2002) words, codified
knowledge “require[s] operators to interpret the selective, de-contextualized and abstract
symbols that machines [and other codified sources] present to them” (p. 53).
Polanyi (1962, p. 50) and Tsoukas (2005b, p. 155) further suggested that the rules of art
do not determine the practice of an art. Rather, rules are only maxims that serve as a
guide to an art, only if they can be integrated into practical knowledge. In other words, it
is necessary for the agent to be able to understand and ‘re-attach’ the information found
in databases or books, in order to put the knowledge into practice. If the agent does not
understand the domain or context of the explicated knowledge, they can make no
valuable use of it. In the words of De Long and Fahey (2000),
“[Codified] resources, no matter how highly analyzed, only become practical knowledge when individuals can apply their own experience and contextual understanding to interpret the details and implications for action” (p. 115).
Choo and Bontis (2002) rephrased:
“The application of explicit knowledge often requires individuals who can interpret, elaborate, demonstrate, or instantiate the formal knowledge with respect to a particular problem setting” (p. 12).
The two forms of knowledge are not in opposition (Tsoukas, 2005b), nor are they
alternative forms of knowing (Spender, 1996b), rather they complement each other in the
knowing process. As Tsoukas (2005b) stated, “two sides of the same coin” (p. 158).
Neither can exist without the other. Explicated artifacts act as guiding lights in providing
meaning and interpretation to a tacit activity. “Uncodified knowledge provides
background context and warrants for assessing the codified. Background no longer works
as background when it is foregrounded” (Duguid, 2005, p. 112). Using Ryle’s (1949)
vernacular, one may say that knowing how helps to make knowing that actionable.
However, getting more ‘know that’ (i.e. explicit, codified information) does not
18
necessarily lead to ‘know how’, which is traditionally generated through practice
(Duguid, 2005).
2.1.3 Knowledge Sharing and Knowledge Sharing Behavior Defined
In a strict and literal sense, knowledge cannot be shared; in that it is not like a
commodity, which can be freely passed around (Hendricks, 1999). Instead, knowledge
sharing is a process involving at least two actors, which has no identifiable starting or
ending point (Boer, van Baalen, & Kumar, 2002). The knowledge owner begins the
knowledge sharing process through an act of externalization, which might or not be a
conscious act. The knowledge receiver then conducts an act of internalization, to absorb
the new stimulus (Hendricks, 1999). Davenport and Prusak (1998) reflected this
perspective in their formula for transfer, which states: Transfer = Transmission +
Absorption.
Perhaps the best way to understand the externalization and internalization processes is to
refer to Boisot’s (2002) concept of ‘resonance’ or Fiol’s (1994) idea of ‘common
understanding’. Boisot (2002) argued that knowledge sharing is no more than “some
degree of resonance being achieved between the knowledge states of two or more agents
following some sharing of data among them” (p. 68). Boisot (2002) further noted that in
addition to resonance, an act of reconstruction is needed for knowledge to be successfully
shared. Knowledge may also be reconstructed through practice and observation
(Hendricks, 1999). The knowledge receiver may then display the reconstruction of
knowledge through action (Duguid, 2005). Without the receiver engaging in some
behavior, it is difficult to determine whether knowledge has been shared, though the
specification of the required behaviors may also be difficult to specify (Wittgenstein,
1953).
The level of success in the reconstruction of knowledge can be quantified, from the
organization’s perspective, in two ways: first, through the actual use of knowledge
shared; and second, through the performance outcomes that result from the knowledge
shared. However, accurately measuring either can be very challenging. Some researchers
19
have approached this problem by measuring the data/information flow or information
awareness (Cross & Cummings, 2004). These approaches could also be conceptually
problematic, as the flow of data or information, alone, does not guarantee that knowledge
was being shared. Additionally, knowing who possesses certain domain knowledge,
alone, does not guarantee that this person is accessible or willing to help. In reality,
neither of these approaches accurately measures knowledge sharing, only remnants of the
knowledge sharing process. These approaches may also raise privacy concerns by mining
data flow (e.g. email) or by asking respondents to identify, name, and rate people in their
professional network, a rather impossible task when trying to keep anonymity.
To better understand knowledge sharing, in this research, an alternative approach was
used; one that considered three important conditions necessary for effective knowledge
sharing behavior to take place. First, the knowledge source must be willing to share the
knowledge they posses. Second, the knowledge receiver must be willing to receive and
use the knowledge that is shared. Finally, the knowledge receiver must perceive the
knowledge shared as being useful to their individual work, the project, or the
organization as a whole. These are important conditions, since all knowledge sharing
requires a willingness to participate in the knowledge sharing process from both ends. For
example, when a person is approached to share what they know, they are asked to make
an investment of their valuable time, often without any likelihood of reward or
recognition. This investment of time may be significant, as extensive interaction may
have to take place to ensure that the knowledge seeker understands. Clearly this
requirement for the investment of time may reduce an individual’s willingness to share
knowledge or use the knowledge shared.
In this research, the three knowledge sharing conditions will be collectively considered as
representing effective knowledge sharing behavior. However, in a pure sense these
conditions do not actually represent behaviors, rather behavioral precursors (i.e.
willingness or intentions) and post-behavioral outcomes (i.e. perceived usefulness). A
more complete discussion of this distinction can be found in Section 3.2.3.
20
2.2 Organizational Trust and Knowledge Sharing
To further understand knowledge sharing and knowledge sharing behavior, a
comprehensive literature review of knowledge sharing factors was conducted (Evans,
2008). Knowledge sharing factors were identified from three different bodies of literature
including studies where authors took a philosophical or theoretical approach; behavioral
approach, and management science or information science (organizational behavior)
approach. Two important assumptions were made: an explicit definition of knowledge
need not be stated in the work, and if a definition for knowledge is provided it need not
be universal.
This initial literature review identified numerous factors as motivators or inhibitors to
knowledge sharing and led to the development of an overview model, or literature map,
representing the main motivators and inhibitors to knowledge sharing (Figure 2.3). The
factors have been organized using a distinction made by Duguid (2005) between factors
that can or cannot be shared (i.e. ability factors) and those that will or will not be shared
(i.e. willingness factors). Given the complexity of the model, creating a study that
investigates all the factors summarized in the figure is not feasible, thus a decision had to
be made on which ones to focus. Of all these factors, trust kept emerging in the literature
and was predominantly the most recurring factor across all three bodies of literature.
Much of this theoretical and empirical work discussed trust as both directly influencing
knowledge sharing and as an important antecedent or precondition, reaffirming that it
should be a central construct in understanding knowledge sharing behavior.
21
Figure 2.3 Original KSB Literature Map8
A second round of literature review of trust studies identified 14 research variables that
were found to have a major effect on trust, especially in the context of knowledge
sharing. Special attention was given to those trust studies that related to knowledge
sharing within an organizational setting. These 14 variables included: group norms,
shared language, shared vision, value homophily, status homophily, tie strength,
reciprocity, formal sanctions, informal sanctions, intrinsic rewards, extrinsic rewards,
relationship duration, tertius gaudens orientation, and tertius iungens orientation. For the
present study, certain variables were combined and others were excluded because of the
concern they posed on the ability to record, measure, and analyze them without affecting
participant anonymity. Therefore, the original list of 14 variables was reduced to five that
jointly will be referred to in this study as social-cognitive factors. These social-cognitive
factors are: shared language, shared vision, homophily, tie strength, and relationship
duration.
8 Figure 2.3 depicts factors that have been found to be motivators or inhibitors to knowledge sharing behavior. The arrows represent significant relationships found in the literature.
22
The following sections explore the existing literature on the interrelatedness of trust, the
identified social-cognitive factors, and knowledge sharing behavior. The sections will
define each of these variables and discuss direct relationships found in previous research
between the social-cognitive factors, trust, and knowledge sharing behavior. The review
will begin with organizational trust, as it is a central construct to this research.
2.2.1 Understanding Organizational Trust
Trust is a construct that has been examined by numerous social science fields of study,
including history, anthropology, psychology, political science, economics, sociology,
information studies, and knowledge management with each of these disciplines applying
their own perspectives and approaches (lenses). Lewicki and Bunker (1996) point out that
“little effort has been made to integrate these different [trust] perspectives or articulate
the key role trust plays in critical social processes (e.g. cooperation, coordination,
performance)” (p. 115). Even though several studies have been conducted since, the role
trust plays in social processes remains an important area of research needing exploration.
Worchel (1979) argued that all these different perspectives on trust may be categorized
into three broad research approaches, which Lewicki and Bunker (1995; 1996) expanded
on. The first research approach, proposed by Worchel (1979), is consistent with the view
of personality theorists. This viewpoint is rooted in early psychological development and
focuses on ‘developmental and social contextual factors’ that shape trust (Lewicki &
Bunker, 1996, p. 115). The second perspective focuses on trust as an institutional
phenomenon and is consistent with research approaches in sociology and economics
(Lewicki & Bunker, 1995; 1996). In this perspective, trust is studied within institutions,
across institutions, or as an individual’s trust in the institution. The final category,
proposed by Worchel (1979), is consistent with the approach of social psychologists and
examines interpersonal relationships and transactions. This is the focus adopted in the
present study. Of specific interest is the “contextual factors that serve to either enhance or
inhibit the development and maintenance of trust” (Lewicki & Bunker, 1996, p.116).
23
According to Lewicki and Bunker (1996), recent work on trust has attempted to make a
distinction between personal and professional relationships. The former suggests that the
research focuses on the development of trust in close personal contexts (e.g. romantic
relationships, friendships, acquaintances). The latter suggests a focus on the development
of trust in working or professional relationships. The authors (Lewicki & Bunker, 1996)
explain that a distinction should be made, because the way these relationships form and
develop is radically different. “In professional relationships [as opposed to personal
ones], trust does not begin with the development of intense emotionality” (Lewicki &
Bunker, 1996, p. 118). The present study focuses only on professional or working
relationships and Mayer et al.’s (1995) model (described in detail below; see Figure 2.4)
is appropriate for use with this type of research, as it is specifically formulated for use
within the organizational setting.
As there are numerous approaches and disciplines to the study of trust, there are also
several definitions. Any definition used must be consistent with, and appropriate to, the
perspective of trust the research intends on selecting. Since the present study emphasizes
a social-psychological perspective, an appropriate definition for trust is one that perceives
it in an interpersonal organizational context.
For most social psychologists, trust is based on “expectations set within particular
contextual parameters and constraints” (Lewicki & Bunker, 1996, p.116). Deutsch (1960)
suggested that an individual decides to trust another when three situational parameters
exist: an uncertain future course of action; an outcome depending on the behavior of
others; and the strength of the detrimental event that is greater than the strength of the
beneficial event. Using similar parameters, Schlenker, Helm, and Tedeschi (1973)
defined trust as the “reliance upon information received from another person about
uncertain environmental states and their accompanying outcomes in a risky situation” (p.
419). Johnson-George and Swap (1982) noted that a “willingness to take risks may be
one of the few characteristics common to all trust situations” (p. 1306). Boon and
Holmes’ (1991) interpretation of trust also focused on risk, defining trust as “a state
involving confident positive expectations about another’s motives with respect to oneself
24
in situations entailing risk” (p. 194). Schoorman, Mayer, and Davis’ (1996b) definition
interprets risk, which the authors describe as “an important component in a model of
trust”, through ‘vulnerability’ (1996, p. 340). According to the authors, “making oneself
vulnerable is taking risk. Trust is not taking risk per se, but rather it is a willingness to
take risk” (Mayer, Davis, & Schoorman, 1995, p. 712) or a “willingness to engage in
risk-taking with the focal party” (Mayer & Davis, 1999, p. 124). According to the
authors, their definition and corresponding model “complements the risk literature by
clarifying the role of interpersonal trust in risk taking” (Mayer, Davis, & Schoorman,
1995, p. 711).
Mayer et al. (1995) define trust as
“[…] the willingness of a party to be vulnerable9 to the actions of another party based on the expectation that the other will perform a particular action important to the trustor, irrespective of the ability to monitor or control that other party” (p. 712).
In Mayer et al.’s (1995) opinion, previous research and models on trust have not been
clear in differentiating trust, factors that lead to trust, and outcomes of trust. The authors’
model conceptualizes trust in a fashion that distinguishes it from its outcomes and
antecedents (Mayer & Davis, 1999). The model (definition) also considers trust factors
from both the characteristics of the trustor (i.e. propensity to trust) and the collective
perceived characteristics of the trustee (i.e. ability, benevolence, integrity), something
that, the authors argue, other models have neglected (Mayer et al., 1995; Schoorman,
Mayer, & Davis, 1996b). “The failure to clearly specify the trustor [the trusting party]
and the trustee [the party to be trusted] encourages the tendency to change referents and
even level of analysis, which obfuscates the nature of the trust relationship” (Mayer et al.,
1995, p. 711).
9 Making oneself vulnerable, implies that something important may be lost. Trust is the willingness to take a risk. The level of trust directly relates to the level of perceived risk (Mayer, Davis, & Schoorman, 1995; Zaheer, McEvily, & Perrone, 1998).
25
Mayer et al.’s (1995) proposed model of organizational trust (Figure 2.4) separates the
relationship between trustor and trustee, in an effort to understand the factors underlying
why a trustor would trust a trustee.
Figure 2.4 Mayer, Davis, & Schoorman’s (1995) Proposed Model of Trust (p. 715)
Mayer et al. (1995) argue that it is the individual traits or characteristics of the trusting
parties that determine the level of trust that may be achieved between them. For instance,
for a trustor to exhibit trust towards a trustee, the trustor must first have the ‘propensity to
trust’10 (Mayer et al., 1995, p.715) that particular trustee, or the propensity to trust in
general (especially when the relationship is new). In return, the trustee must be perceived
as having ability11, benevolence12, and integrity13, which, together, help the trustor
determine the trustee’s ‘trustworthiness’ (Mayer et al., 1995). Any ‘measure’ of a
trustee’s ‘trustworthiness’ is only based on a perception of ‘trustworthiness’ by the
trustor, and not on the actual characteristics or traits of the trustee. In a subsequent paper,
the authors (Schoorman, Mayer, & Davis, 1996b) justify this perspective by claiming that
10 Propensity to trust is defined as “the general willingness to trust others” (Mayer, Davis, & Schoorman, 1995, p. 715) 11 Ability is defined as the skills, competencies, and characteristics necessary to have influence in a specific domain. (Mayer, Davis, & Schoorman, 1995, p. 717) 12 Benevolence is defined as the extent to which a trustor believes the trustee wants to do good to the trustor. Act in a way that is not egocentric. (Mayer, Davis, & Schoorman, 1995, p. 718) 13 Integrity is determined by the trustor, by making an assessment as to whether or not the trustee will adhere to an acceptable (to the trustor) set of principles. (Mayer, Davis, & Schoorman, 1995, p. 719)
26
it is necessary “to account for why a particular trustee might be highly trusted by one
trustor but not by another” (p. 338).
“A trustor will be willing to be vulnerable to another party based both on the trustor’s propensity to trust other people in general and on the trustor’s perception that the particular trustee is trustworthy” (Mayer & Davis, 1999, p. 124).
To better explain Mayer et al.’s (1995) proposed model of trust, it is best to separate the
trustor’s trust characteristics from those of the trustee. The former considers the trustor’s
“general willingness to trust others” (Mayer et al., 1995, p. 714), which affects the
likelihood that the trustor will trust in general, and presumably carries with the person, as
they interact in different situations (Mayer et al., 1995). “In this approach trust is viewed
as a trait that leads to a generalized expectation about the trustworthiness of others”
(Mayer et al., 1995, p.715). The authors refer to this trait as the ‘propensity to trust’.
“Propensity might be thought of as the general willingness to trust others. Propensity will
influence how much trust one has for a trustee prior to data on that particular party
becoming available” (Mayer et al., 1995, p. 715).
Mayer et al. (1995) also suggest that one should consider the trustee’s characteristics that
convey trustworthiness to the trustor. “Characteristics and actions of the trustee will lead
that person to be more or less trusted” (Mayer et al., 1995, p. 717). To identify these
characteristics Mayer et al. (1995) conducted a review of antecedent factors14 and found
three characteristics to appear most often. Mayer et al. (1995) refer to these three
characteristics (i.e. ability, benevolence, and integrity) as The Factors of Trustworthiness.
“Our decision to treat all three as contributors to trust was based on our view that they
have an additive quality in determining the level of trust. [Nevertheless,] all three
concepts are theoretically distinct” (Schoorman, Mayer, & Davis, 1996b, p. 339).
According to Mayer et al. (1995), “ability is that group of skills, competencies, and
characteristics that enable a party to have influence within some specific domain” (p.
717). If a trustee is perceived as having high domain specific knowledge, then that person
14 For a complete list of the Mayer et al.’s literature review and antecedent factors see Table 1 Trust Antecedent (1995, p. 718)
27
is afforded trust, by the trustor, on tasks relating to that domain (Mayer et al., 1995). In
justifying the inclusion of ability in their model, Mayer et al. (1995) cite a number of
theorists who discuss either the same (i.e. ability15) or similar constructs in their work.
The authors note that similar constructs include competence, expertise, business sense,
and judgment (Mayer et al., 1995). Competence has been a popular synonym in studies
measuring trust and will be discussed in more detail in the next chapter. Replacing one
term for another is not a concern, since “competence and ability are clearly similar”
(Mayer et al., 1995, p. 722). In later work, the authors even use the terms interchangeably
(Schoorman, Mayer, & Davis, 1996b).
Mayer et al. (1995) define benevolence as “the extent to which a trustee is believed to
want to do good to the trustor, aside from an egocentric profit motive” (p. 718).
Benevolence suggests that there is “some specific attachment” (p. 718) of the trustee to
the trustor (e.g. the trust between a mentor and his/her protégé). Unlike ability, which is
domain specific, benevolence implies a personal orientation. Mayer et al. (1995) justify
the inclusion of benevolence, by citing several theorists16 who have also used the exact
term in their interpretations of trust. The authors (Mayer et al,, 1995) also review other
theorists who use different terminology, but essentially have similar perceptions (i.e.
consider trust to be tied to a persons’ altruistic or loyalty-based intentions and/or
motives).
Integrity “involves the trustor’s perception that the trustee adheres to a set of principles
that the trustor finds acceptable” (Mayer et al., 1995, p. 719). According to the authors
(Mayer et al., 1995), the trustor makes a judgment about the trustee’s integrity, in four
ways: through the consistency of the trustee’s past actions; through communication with
others (i.e. others say the trustee is credible); through an assessment of the trustee’s sense
of justice17; and through an assessment of the extent to which the trustee’s actions match
their words. Like in the case of the previous two constructs, the authors (Mayer et al.,
15 “Cook and Wall (1980), Deutsch (1960), Jones, James, and Bruni (1975), and Sitkin and Roth (1993) all considered ability an essential element of trust.” (Mayer et al., 1995, p. 717) 16 Larzelere & Huston, 1980; Solomon, 1960; Stickland, 1958, as cited in Mayer et al., 1995 17 The trustee is thought as having high integrity when their perceived sense of justice is high
28
1995) justify the inclusion of integrity, by citing numerous theorists18 who use it (or a
very similar construct) as an antecedent to trust.
Ability, benevolence, and integrity are all related to one another, but each can be
separated and varies independently of the others (Mayer et al., 1995). All three factors are
important to trust, yet each, by itself, is insufficient for trust. “Each element contributes a
unique perceptual perspective from which the trustor considers the trustee” (Mayer et al.,
1995, p. 722). If all three are perceived as high by the trustor, then the trustee is deemed
trustworthy. Trustworthiness should be understood as a continuous variable, rather than
mutually exclusive (i.e. either present or not). High trust normally presumes a high level
of all three variables, but it is possible for lesser degrees of the three variables to still
yield ‘meaningful’ amounts of trust (Mayer et al., 1995, p. 721).
The extent to which one person is willing to trust another is a function of both the
trustor’s perceived judgment of the trustee (with respect to ability, benevolence, and
integrity) and the trustor’s propensity to trust.
2.2.2 Trust and Knowledge Sharing Behavior
The majority of research examining the impact of trust on knowledge sharing builds upon
previous work in the fields of psychology and behavioral science, that have been
examining trust as an antecedent for social behavior. The consensus of this research
seems to suggest that higher trust among individuals leads to higher and more productive
knowledge sharing behaviors and activities between them. It is also generally accepted
that trust is a prerequisite for knowledge sharing (Nonaka, 1991; Adler, 2002; De Long &
Fahey, 2000; McAllister, 1995). As Mayer et al. (1995) stated, “a clear understanding of
trust and its causes can facilitate cohesion and collaboration between people” (pp. 710-
11).
Before discussing the literature in more detail, it is important to acknowledge a few
limitations in attempting to synthesize a literature review on trust and knowledge sharing. 18 Leiberman, 1981; Sitkin & Roth, 1993; Butler & Cantrell, 1964; Butler, 1991; Gabarro, 1978; Hart, Capps, Cangemi, & Caillouet, 1986, as cited in Mayer et al., 1995
29
First, the reviewed studies did not all explore similar settings, respondents, or relationship
dyads; and in no study were these characteristics identical to those proposed in this
research. Second, in most cases, the knowledge sharing constructs in the literature were
conceptually related to the three behavioral conditions proposed in this study for effective
knowledge sharing behavior, yet rarely identical to them. Finally, trust greatly varied in
the way it was defined and measured in the literature, which has been recognized by
McEvily and Tortoriello (2011) as a problem for this area of research – the authors’
review of 171 empirical papers measuring trust in organizational contexts (published
from 1962-2010) found a staggering 207 different psychometric trust measures.
Looking past these limitations, the literature revealed that trust had a positive influence
on effective knowledge transfer, behaviors, and activities. The same can be said for the
positive effect of trust on outcomes associated with effective knowledge transfer (e.g.
decision-making and problem solving). The following sections present findings from the
literature showing the relationship between trust and 1) specific knowledge sharing
behaviors; 2) other motivating behaviors; and 3) resource and information exchange /
decision making and problem solving.
Trust and Specific Knowledge Sharing Behaviors
A number of research studies examined a direct relationship between trust and knowledge
sharing behavior specifically. For example, Renzl’s (2008) study found a direct positive
relationship between trust and knowledge sharing behavior within and between project
groups in two large organizations (15 interviews and 201 survey respondents). Andrews
and Delahaye (2000) study with scientific staff representing 5 partner organizations at a
Bio-Medical Consortium (15 semi-structured interviews) found trustworthiness among
co-workers to positively affect knowledge sharing activities. Trust was also found to
have a significant positive effect on knowledge sharing in two research studies (437
survey respondents) done at three large technology companies implementing KM in
Taiwan (Ho, Kuo, Lin, & Lin, 2010; Ho, Kuo, & Lin, 2011).
30
Using a social capital perspective, Nahapiet and Ghoshal (1998) developed a model that
viewed trust as a necessary facet of the relational dimension of social capital.
The authors argued that this relational dimension (with trust as a facet) had a positive
effect and influence on the transfer of knowledge. Tsai and Ghoshal (1998) conducted
research to test this model on 45 managers (3 managers in each of the 15 business units)
in a multinational electronics company. The results found a positive relationship between
trust and the willingness to share knowledge. Other researchers suggested similar positive
relationships between trust and the willingness to share knowledge, including Hinds and
Pfeffer (2003), Van den Hooff and Van Weenen (2004), Lin (2007), and Hislop (2003).
In his doctoral research, Holste (2003) examined the effect of trust on the willingness to
share knowledge and willingness to use knowledge. To gain a comprehensive
understanding of the effect of trust, Holste (2003) also factored for the type of knowledge
(explicit vs. tacit), as well as the nature of relationship with referent (positive vs. negative
referent19). Holste’s (2003) results showed that both affect-based trust and cognition-
based20 trust had positive effects on the willingness to share knowledge and willingness
to use knowledge with positive referents. With negative referents only, affect-based trust
(not cognition-based) was found to have a positive effect on the willingness to share
knowledge (Holste, 2003). However, both affect-based trust and cognition-based trust
had positive effects on the willingness to use knowledge from negative referents (Holste,
2003).
In their study of three divisions at an American pharmaceutical company, a British bank,
and a Canadian oil and gas company, Levin and Cross (2004) found trust (specifically
competence-based trust) to have a positive impact on knowledge transfers involving
19 Positive referents were those individuals the respondents felt they worked best with and negative referents were those individuals the respondents felt they did not work well with. 20 In his study, McAllister (1995) created new measures to assess affect- and cognition-based trust. These measures were based on the work of Lewis and Weigert (1985) who described social trust as having cognitive and affective foundations. According to the authors (Lewis & Weigert, 1985) “trust is based on a cognitive process which discriminates among persons and institutions that are trustworthy, distrusted, and unknown. In this sense, we cognitively choose whom we will trust” (p. 970). In addition, “trust is also constructed on an emotional base that is complementary to its cognitive base. This affective component of trust consists in an emotional bond among all those who participate in the relationship” (Lewis & Weigert, 1985, p. 971).
31
highly tacit knowledge. More interestingly, for transfers involving codified knowledge,
competence-based trust was found not to provide any benefit. Yang and Farn’s (2009)
research suggested similar findings, where affect-based trust (rather than competence-
based) was found to motivate the intention to share tacit knowledge. Parallel findings
were made in a survey of 212 students in the Business Administration department at the
University of Taiwan that found trust in co-workers to be positively related to tacit
knowledge sharing (Lin, 2007). Holste’s (2003) work showed a more complex
relationship between trust and explicit and tacit knowledge sharing behavior, one that
depended on how the respondent perceived the relationship between them and the
referent. For example, with positive referents the effect of trust on the willingness to
share knowledge and willingness to use knowledge consisted more of tacit knowledge
sharing and use, rather than explicit (Holste, 2003). For negative referents, the opposite
was found to be true (Holste, 2003).
Finally, Levin and Cross (2004) found significant positive relationships to exist between
benevolence/competence-based trust and the perceived receipt of useful knowledge.
These results were also consistent with Szulanski, Cappetta, and Jensen (2004), who
examined 122 knowledge transfers in 38 practices and found perceptions of
trustworthiness to have direct effect on the accuracy of knowledge transfer. Other
research also found evidence of the positive effect of trust on knowledge importing
(Andrews & Delahaye, 2000) and knowledge combination (Tsai & Ghoshal, 1998).
Trust and Other Motivating Behaviors
Research examining trust in an organizational setting was not specifically concerned with
knowledge sharing constructs or the knowledge transfer process. Instead, in these studies
the focus was on the effect of trust on other organizational behaviors and actions. It
should be noted that the behaviors below are not specifically knowledge sharing
behaviors; yet these studies were included in the review because the behaviors can be
argued to have a positive relationship on effective knowledge sharing behavior.
32
The discussion of these other organizational behaviors will start by examining the role
trust played in minimizing behaviors that might be argued as being detrimental to
effective knowledge sharing behavior. First, trust and trustworthiness between employees
have been positively associated with a decrease in efforts needed for information search
and processing (Zaheer, McEvily, & Perrone, 1998). When trust exists, efforts needed for
information search and processing are minimized, since the receiving party does not have
to scrutinize the quality or veracity of the information (Zaheer, McEvily, & Perrone,
1998). "High levels of trust help reduce transaction costs” (Limerick & Cunnington,
1993, p. 95). Trust and trustworthiness have also been associated with a decrease in
information monitoring and safeguarding behaviors (Zaheer, McEvily, & Perrone, 1998;
Roberts & O’Reily, 1974; Husted & Michailova, 2002; and Orlikowski, 1993). Control
mechanisms are reduced as interaction increases and trust is developed (Mayer et al,
1995). The existence of a trusting relationship reassures the sender that the receiver will
not misappropriate the information entrusted to them, reducing their monitoring,
safeguarding behaviors, and conserving cognitive resources (Uzzi, 1997). Ultimately this
leads to more ‘openness’ in the exchange (Zaheer, McEvily & Perrone, 1998). According
to Limerick and Cunnington (1993) “trust lubricates the smooth, harmonious functioning
of the organization by eliminating friction and minimizing the need for bureaucratic
structures that specify the behavior of participants who do not trust each other” (pp. 95-
6).
Other organizational studies have made similar conclusions. For example, Zand’s (1972)
study examining problem-solving groups in management development programs found
trust to minimize control-based monitoring behavior. McAllister’s (1995) study of 194
managers and professionals reporting on cross-functional dyadic relationships with co-
workers found trust to minimize defensive cooperative behavior; also an empirical
finding of Tsai and Ghoshal (1998) and Hislop (2003). Zaheer, McEvily and Perrone’s
(1998) study of 205 purchasing managers found trust to minimize conflict, as well as
reduce costs associated with negotiation. Finally, Bromiley and Cummings (1995) argued
that trust reduced transaction costs.
33
On the contrary, there are a number of organizational trust studies that examined the role
trust plays in motivating behaviors that may be argued as leading to effective knowledge
sharing behavior. For example, trust has been found to influence an individual’s desire to
share information and ideas (Davenport & Prusak, 1998; Empson, 2001; McDermott &
O’Dell, 2001; Husted & Michailova, 2002; Hendricks, 1999; and Hinds & Pfeffer, 2003)
or what Szulanski (1995; 1996) calls a desire to ‘initiate a transfer’. Zand (1972) found
trust to increase confidence in the reliability of others, and the work of Tsai and Ghoshal
(1998) found it to influence the lending of support to co-workers in the achievement of
their goals. McAllister (1995) found that managers who had higher trust in their workers
paid more attention to their needs. Looking at relationships with management in a large
US federal government training organization, McCauley and Kuhnert (1992) found a
positive relationship between trust and willingness to listen. Similarly, Penley and
Hawkins’ (1985) research with employees in the personnel and support services areas of
a large southern US insurance company and from the logistics and support division of a
large military base, found a positive relationship between trust and communication
responsiveness, which they defined as a willingness to listen and respond to issues raised.
The authors’ (1985) findings also suggested a direct effect of trust on task
communication, which they defined as willingness to explain work tasks and policies.
Finally, numerous researchers (Ring & Van de Ven, 1994; Tsai & Ghoshal, 1998; Dirks
& Ferrin, 2002; Kramer, 1999; and Tyler & Degoey, 1996) have found direct positive
connections between trust and cooperation or cooperative behavior.
Trust and Resource/Information Exchange & Trust and Decision Making/Problem
Solving
The final section of trust literature explores the effect of trust on resource/information
exchange, decision-making, and problem-solving within the firm. Even though these
outcomes do not specifically address or relate to knowledge sharing behavior, it could be
argued that they represent the desired residual effects of the organizational knowledge
sharing process. In other words, in knowledge intensive organizations, the desired or
expected outcomes of knowledge sharing would be measured in increased exchange,
better decision-making, and better problem-solving.
34
The literature suggests that trust has a positive effect on resource exchange (Tsai &
Ghoshal, 1998). In addition, several other research studies found trust to have a positive
effect on the exchange of accurate, comprehensive, and timely information, including
ideas, goals, opinions, and problems (Zand, 1972; Tsai & Ghoshal, 1998; Mital, Isreal, &
Agarwal, 2010; McCauley & Kuhnert, 1992; Burt, (1992). McCauley and Kuhnert (1992)
explained this relationship as behaviorally driven, where trust had a positive effect on a
feeling of openness in the information exchange. Other authors have made similar
connections to behavioral and cognitive drivers for this relationship with information
exchange, finding trust to have a positive influence on information available (DeLong &
Fahey, 2000; Husted & Michailova, 2002), information commitment (Dirks & Ferrin,
2002; Zand, 1972; Ouchi, 1980), and information disclosure. Zand’s (1972) research also
found trust to have a positive effect on the information search process.
Trust has also been positively linked to effective decision-making and problem solving.
For example, McCauley and Kuhnert’s (1992) study discovered a positive relationship
between trust and willingness to participate in decision-making. Zand (1972) found trust
to have a positive influence on the framing of problem alternatives and effective problem
solving within groups. This may be because there is less scrutiny in the exchange, and the
group members are able to draw better distinctions on the information they have
(Tsoukas 2005a, 2005b), which gives them the ability to reach quicker and better
decisions (Roberts & O’Reily, 1974). “A group within which there is extensive
trustworthiness and extensive trust is able to accomplish more than a comparable group
without the trustworthiness and trust” (Coleman, 1988, p. S101).
Summary of Trust and Knowledge Sharing
Despite a lack of consistency in the definition of trust and knowledge sharing behavior,
the overall literature findings suggest a positive relationship between trust and knowledge
sharing behavior in organizations. This can be witnessed in the sections above, that
summarize the literature showing a positive relationship between trust and 1) specific
35
knowledge sharing behaviors; 2) other related motivating behaviors; and 3) resource and
information exchange / decision-making and problem-solving.
2.3 Social-Cognitive Factors and Trust
The following section defines each of the social-cognitive factors identified in the study,
as well as examines the existing literature for relationships between these variables and
trust. The section begins with homophily factors and then discusses shared perspective,
tie strength, and relationship length.
2.3.1 Homophily and Trust
In network terms, homophily implies the existence of a positive relationship between the
degree of similarity between two actors and the strength of the relationship between
them. In other words, greater similarity between two individuals leads to a stronger
relationship between them. McPherson, Smith-Lovin, and Cook, (2001) define
homophily as “the principle that contact between similar people occurs at a higher rate
than among dissimilar people” (p. 416). According to Coleman (1988), a higher rate of
contact (caused by homophily) allows for the development of reputation through the
proliferation of obligations and expectations. Further, the development of reputation leads
to higher trust.
According to Toh and Srinivas (2011) “apparent physical cues […] form salient bases of
an individual’s assessment of how much to trust a given target” (p. 3). In his work, Byrne
(1971) discovered that individuals with characteristics similar to those of the respondents
being studied were rated as more attractive and respondents were more willing to trust
those individuals. Several other researchers including Burt (1992), Levin, Whitener, and
Cross (2006), and Chatman and Flynn (2001) have also established direct connections
between rate of contact and homophily, and between trusting behavior and homophily.
In his work, Burt (1992) established a direct connection between homophily and trust,
arguing that similar agents are more likely to trust each other than those that are
dissimilar. Burt (1992) argued that two individuals with high homophily are more likely
36
to trust each other because “the operational guide to the formation of close, trusting
relations seems to be that a person more like me is less likely to betray me” (p. 16).
Levin, Whitener, and Cross (2006) suggested a similar relationship stating “trust may be
built on perceived demographic similarities” (p. 1164). Chatman and Flynn (2001)
argued that “people often use immediately apparent physical features, such as race, sex,
and national origin, to categorize others and predict their behavior” (p. 957). Other
researchers have also found that people believe those with demographic similarities to
themselves as being more honest, trustworthy, and cooperative (Brewer, 1979;
McAllister, 1995; Shore, Cleveland, & Goldberg, 2003; Tsui & O’Reilly, 1989). Brewer
(1981) called this a “depersonalized trust based on category membership” (p. 356).
Noted causes of homophily include geography21, family ties22, organizational foci23,
isomorphic sources24, and cognitive processes25 (McPherson, Smith-Lovin, & Cook,
2001). According to the authors (2001), there are two distinct types of homophily: status
homophily26 and value homophily27. However, this research study will only explore status
homophily, as value homophily may be too conceptually similar to shared vision, which
is discussed below. Nevertheless, it is important to note that there is research suggesting
that trust is influenced by homophily on values and goals28 (e.g. Hogg & Terry, 2000;
McAllister, 1995; Sitkin & Roth, 1993; and Tsai & Ghoshal, 1998). The selected
characteristics used to explore the relationship of status homophily with trust include both
ascribed characteristics (age, gender, race, ethnicity and nationality, and immigration)
21 Geography relates to geographic distance. More likely to have contact with those that are closer 22 Family Ties refers to a family relation (biological tie). Likely to be the same race, ethnicity, and religion 23 Organizational Foci relates to a focused activity which fosters the relationship (e.g. school, work, or voluntary organizations) 24 Isomorphic Sources relates to occupied positions or roles (e.g. workplace roles (status, seniority, functional division), family roles (wives), or political roles (senators)) 25 Cognitive Processes refers to perceived similarity (e.g. people who share similar knowledge domains) 26 Status Homophily is based on informal, formal, and ascribed status. Includes ascribed characteristics (race, ethnicity, gender, age) and acquired characteristics (religion, education, occupation, behavior patterns) 27 Value Homophily is based on values, attitudes, and beliefs 28 These connections are discussed in more detail in the Shared Vision section.
37
and acquired characteristics (education, experience, and marital status)29. The review will
begin with the ascribed characteristics.
Age Homophily and Trust
Status homophily motivated by a similarity in age has been prevalent in psychological
studies over the years, especially when homophily was measured through rate of contact.
Existing research on age homophily also seems especially prevalent in studies of
community, friendship networks, and school-related acquaintances. Fischer (1977) and
Verbrugge (1977) found age homophily to be a higher influencer of rate of contact than
any other dimensions among close friends. Feld (1982) and Shrum, Cheek Jr., and Hunter
(1988) found similar results in their studies of superficial friendships and school-related
acquaintances. In his research of personal friend networks, Fisher (1982) expanded on the
influence of age homophily as more than an increase in the rate of contact between
individuals, suggesting that age homophilous ties were closer, longer-lived, and more
personal (in addition to having a higher number of exchanges). Although, once again,
trust was not specifically mentioned, Marsden’s (1988) study, using data from the 1985
General Social Survey (GSS)30, discovered that respondents had a tendency to confide in
someone of the same age. The study (1998) also found that people were less likely to
discuss ’important matters’ with someone farther away from them in age. Both actions
could suggest a clear sign of trusting behavior.
Gender Homophily and Trust
Gender homophily begins when children enter school (Smith-Lovin & McPherson, 1993;
McPherson, Smith-Lovin, & Cook, 2001). As students move from early grades to their
adolescence, boys are more likely to form larger heterogeneous groups, while girls form
smaller, more homogeneous, ones (Shrum et al., 1988). By adulthood, most people have
29 Another notable network effect of homophily is evidenced by what is termed ‘selective tie dissolution’. This phenomenon arises as a result of low homophily within a group leading to weakened ties and hence a higher probability of subsequent dissolution (McPherson, Smith-Lovin, & Cook, 2001, p. 436). 30 The GSS gathered data on interpersonal environments for a sample of 1534 persons drawn from the non-institutionalized population of the United States.
38
gender-integrated networks (37% are perfectly mixed), with only 22% not having cross-
gender confidants (Marsden, 1987).
In their research, Huckfeldt and Sprague (1995) found content-based, less intimate
relationships, like the ones that might traditionally be found in a corporate environment,
to be gender homophilous (e.g. 84% of men discussed politics only with other men).
Research specifically taking place in corporate environments has confirmed the existence
of gender homophily (Bielby & Baron, 1986; Kalleberg, Knoke, Marsden, & Spaeth,
1996), which most frequently occurred among upper-level managers and entrepreneurs
(Ibrarra, 1992, 1997; Brass, 1985). Also, men were found to have more gender
homophilous networks, especially in institutions where men were the majority (Ibrarra,
1992, 1997; Brass, 1985). According to McPherson, Smith-Lovin, and Cook, (2001) “this
pattern [of male homophilous networks] is especially strong when we consider
instrumental or status-loaded ties of advice, respect, and mentoring; socio-emotional ties
of friendship and support are much more gender homophilous, in spite of skewed
environments” (p. 424). In these types of situations, one gender is clearly provided an
unfair and unequal access to advice, respect, and mentoring (McPherson, Smith-Lovin, &
Cook, 2001). Without access to these opportunities, trust has little chance to develop.
Race / Ethnicity Homophily and Trust
According to McPherson, Smith-Lovin, and Cook (2001) race and ethnicity create the
largest divide among social networks within the United States. Strong homophily based
on race or ethnicity has been discovered in several different types of relationships
including married couples (Kalmijn, 1998), individuals that confide in one another
(Marsden, 1987; 1988), school-related relationships (Shrum et al., 1988; Maccoby, 1998),
and work-related relationships (Lincoln & Miller, 1979; Ibarra, 1995). Interestingly, in
his research, Marsden (1987) found that only 8% of adults sampled discussed important
matters with someone of opposite race. Similar results were found by Blum, Blau, and
their colleagues (Blum, 1984; Blau, Beeker, & Fitzpatrick, 1984; Blau, Blum, &
Schwartz, 1982; Blau, Ruan, & Monika, 1991), when they examined different measures
of ethnicity (e.g. national origin, native tongue, ethnic group, and birth location). With
39
respect to the organizational settings, Ibarra (1995) found that ethnic minorities were
much more likely to seek advice and support from their heterogeneous counterparts, than
the ethnic majority.
Even though no direct relationship between ethnicity-based homophily and trust is
specifically mentioned, it is reasonable to assume that discussing important matters,
seeking advice, and seeking support from another individual are all clear signs of trusting
behavior, which appear to be occurring more frequently within ethic homogeneous
relationships.
Citizenship / Immigration Homophily and Trust
Earley and Mosakowski (2000) suggested that “nationality is a superordinate determinant
of a person's self-identity, derived through a meaning system shared with others” (p. 26).
Geringer (1988), Oetzel (1995), and Bornhorst et al. (2010) further explained this
relationship by suggesting that nationality determines communication patterns and styles
of interaction. In addition, shared “nationality can become a key driver of social
categorization” (Mäkelä, Andersson, and Seppälä, 2012, p. 442) through in-group
association, which in turn influences trust among members. However, empirical evidence
of this relationship was not very consistent.
In their research involving almost 900 students from a Dutch university, Curseu and
Schruijer (2010) found nationality diversity to be negatively related to trust. In other
words, respondents trusted those individuals of similar nationalities more than those with
different nationalities. Research by Uslaner (2011) found that similar nationality
influences trust, but only for certain backgrounds. Banai and Reisel’s (1999) research
found host-country nationals to report high levels of trust for their superiors with similar
nationalities in nationally homogeneous settings. However in nationally heterogeneous
settings, trust between employees and their superiors was not affected by nationality.
Finally, Mäkelä, Andersson, and Seppälä’s (2012) study found no relationship between
similar nationality and trust.
40
Educational Homophily and Trust
Unlike the homophily characteristics previously discussed, which are all ascribed,
education, occupation, experience, and marital status are acquired or achieved by the
individual. However, similar to the presence of homophily based on ascribed
characteristics, research also shows evidence of homophily based on acquired
characteristics. This would seem only logical, since learning institutions locate
educationally like-minded students in similar physical settings and create opportunities
for networking and social exchange through group discussions and projects. For example,
as students begin their university education, they are separated into particular degrees. As
they progress through their degrees, they are further narrowed into different faculties,
schools (e.g. a school of management), and specialty programs (e.g. strategy
management). This occurs again at the graduate level, when individuals seek a focused
advanced education (e.g. a law degree). With each step, the students’ network becomes
more educationally homogeneous. This segregating behavior seems to continue into the
work environment, as organizations tend to group employees with similar professional
experience or education together (e.g. accountants, sales persons). As with specific
faculties, these professional groups are also traditionally physically grouped together, in a
department or industry. Routine work then affords these employees opportunities for
social interaction and exchange, through projects or shared work. Overtime, close ties
and friendships are established in such environments, starting at young ages. One
outcome of the new close ties and friendships is an increased trust between the
individuals (Coleman, 1988; Burt, 1992).
Existing research found similar connections between close personal ties and educational
homophily. For example, Marsden’s (1987) results showed about 30% of the close
personal networks to be homophilous on education (with a standard deviation of less than
1 year of education) (p. 126) with extremely high and low educational levels showing the
greatest tendency for homophily (Marsden, 1988). Louch (2000) found similar results,
suggesting that individuals were more likely to form close network ties with others who
share a similar education. Several other authors found a positive relationship between
cooperative ties and educational homophily among individuals. Galaskiewicz and Shatin
41
(1981) found this relationship in turbulent communities, Laumann, Gagnon, Michael, and
Michaels (1994) in romantic relationships, Verbrugge (1977) in friendship networks, and
Reagans (2005) in a small corporate setting. Marsden’s (1987) study found that people
were more likely to confide in those with similar educational levels. Finally, Wright’s
(1997) study, which was not directly related to education or educational level, similarly
found the difference in skill level to be a significant barrier to friendship31.
Marital Homophily and Trust
Although no specific study examines marital status as a homophily factor in an
organizational setting, a number of researchers have noted marital status to be an
important homophily characteristic. For example, McPherson, Smith-Lovin, and Cook
(2001) mention marital status several times in their review of homophily in social
networks, but provide no empirical contribution. In their study, van Duijn, van
Busschbach, and Snijders (1999) found friendships whose members were similar in
marital status to be more stable. Finally, in her study, Popielarz (1999) found that in all
female groups relationships between individuals were more likely to be homophilous on
marital status than in all-male or mixed groups. One might argue that there is a
relationship between marital status and trust, yet further specific research is needed to
bring forth empirical evidence.
2.3.2 Shared Perspective and Trust
Organizational shared perspective between employees is best conceptualized by Levin,
Whitener, and Cross (2006) who defined it as an amalgamation of shared language and
shared vision between co-workers. The authors (2006) defined shared language as the
extent to which the “knowledge receiver and source seem on the same wavelength” (p.
1166). ‘Same wavelength’ was an idiom used to describe a situation in which the sender
and receiver were able to easily understand, communicate, and agree. According to the
authors (Levin, Whitener, & Cross, 2006), “as a manifestation of shared perspective,
parties develop a sense of the extent to which they share the same language or jargon” (p.
1166). Argyres (1999) referred to shared language as “a ‘technical grammar’ for 31 Other research studies (Marsden, 1987; Campbell, Marsden, & Hurlbert, 1986; Campbell, 1988; Fischer, 1982) have shown evidence of higher educated males as having more diverse networks.
42
communication” (p. 162). Shared vision is defined as the extent to which a source and
receiver (in the eyes of the receiver) share goals, concerns, and purpose (Levin, Whitener,
& Cross, 2006; Tsai & Ghoshal, 1998). Tsai and Ghoshal (1998) explained shared vision
as embodying “the collective goals and aspirations of the members of an organization;
[…] we can view a shared vision as a bonding mechanism that helps different parts of an
organization to integrate or to combine resources” (p. 467).
According to Levin, Cross, and Abrams (2003), “both of these elements […] are among
the most important factors in relation to who trust[s] whom in the knowledge transfer
context” (p. 24). Bracken and Oughton (2006) argued for this relationship by stating that
“if there is to be any chance of success in developing common understandings, the first
step is the development of trust” (p. 380). One possible explanation for this connection
was suggested by Lee (2009), who argued that shared language “overlaps with informal
interaction during daily business activity [which] helps to develop general levels of
solidarity32 [and motivate] benevolence of individual actors” (p. 258). The positive
relationship between shared perspective and trust was empirically supported by Mäkelä
and Brewster (2009), and Levin, Whitener, and Cross (2006). In addition, Tsai and
Ghoshal (1998) found shared vision to have a significant positive effect on perceived
trustworthiness.
2.3.3 Tie Strength and Trust
Tie strength was defined using Granovetter (1973), who conceptualized it as “a
combination of the amount of time, the emotional intensity, the intimacy, and the
reciprocal services which characterize the tie” (p. 1361). Granovetter (1973) further
added that each of the three determinants act independently, although the set, as a whole,
are highly intra-correlated. Several other researchers (Hansen, 1999; Levin & Cross,
2004; McFadyen & Cannella, 2004) followed similar conceptualizations, defining tie
strength as a mix of interaction frequency and closeness. Interaction frequency can
practically be thought of as synonymous with communication frequency and, in a
corporate setting, could presumably include interactions that are face-to-face, by phone,
32 Solidarity is the basis for the formation of trust
43
through email, etc. Closeness relates to an emotional intensity felt by the individuals in
the relationship towards one another, which Marsden and Campbell (1984) argued is “on
balance the best indicator of the concept of tie strength” (p. 498). In fact, the authors
(1984) warned that measures of tie strength based on interaction frequency alone do not
accurately capture tie strength among co-workers. In a more recent paper, Levin, Walter,
and Murnighan (2011) supported Marsden and Campbell’s (1984) claim by arguing that
interaction frequency is not a perfect indicator of tie strength, because not all interactions
are “the same length, do not occur at a constant rate, and have unequal emotional impact”
(2011, p. 924) The variation in these ultimately causes different lasting outcomes.
With respect to trust, Levin and Cross (2004) found that tie strength had a positive impact
on benevolence-based and competence-based (i.e. ability-based) trust in organizational
settings. The authors also found benevolence and competence based trust to act as
mediator between strong ties and the perception that the knowledge received from the
coworker was useful. Krackhardt’s (1992) research suggested similar conclusions,
finding strong ties to be more accessible and more willing to help than weak ones.
2.3.4 Relationship Length and Trust
Relationship length was defined by Levin, Whitener, and Cross (2004) as how long one
co-worker has known another. Knowing someone for a longer period of time should
theoretically lead to more trust, since it provides the individuals with ample opportunities
to develop trust between them. However, according to the authors (2004) the “construct
of relationship length has been largely neglected in the trust literature” (p. I1). For
example, Dirks and Ferrin’s (2002) meta-analysis of studies and papers written over the
last four decades on trust and its implications on leadership, only found 5 of the 106
studies to contain a measure for relationship length.
Dirks and Ferrin (2002) suggested that, “the level of trust may be greater in a relationship
of long duration than in a relationship of short duration owing to the level of knowledge
and familiarity acquired” (p. 615). Lewicki and Bunker (1996) made similar claims,
arguing that levels of trust in a relationship increase and develop over time. The authors
44
(Lewicki & Bunker, 1996) proposed that trust (specifically in professional relationships)
becomes deeper as it moves through identified stages, which they call “The Stagewise
Evolution of Trust” (p. 124). The movement through these stages occurs over time, as
individuals develop a history of interaction. "In professional relationships, trust develops
gradually as the parties move from one stage to another" (Lewicki and Bunker, 1996, p.
124). The authors (1996) argued that trust is "a dynamic phenomenon which takes on a
different character in the early, developing, and ‘mature’ stages of a relationship" (p.
118).
Similar connections between trust and relationship length can be found in Coleman’s
(1988) work on network closure. Coleman (1988) reasoned that a reputation was required
to build trust between individuals, which developed over time through the proliferation of
obligations and expectations, a process similar to Dirks and Ferrin’s (2002) ‘knowledge
and familiarity acquired’ or Lewicki and Bunker’s (1996) history of interaction. Levin,
Whitener, and Cross (2004) explained these interactions as a “chance to gather
information about each other’s idiosyncrasies and perspectives, expectations can be
rooted in knowing if they share the same goals, perspective, and self-identity” (p. I3).
Much like Coleman (1988), Levin, Whitener, and Cross (2004) argued that, early in a
relationship, the trustor does not have accurate or reliable information about the trustee to
gauge benevolence. The authors propose that, over time and with direct social interaction,
the trust between the individuals becomes rooted in expectations that are based on actual
observations of behavior. Their study (Levin, Whitener, & Cross, 2004) found that “[the]
relationship length significantly moderated the bases of benevolence trust [and it] did not
have a direct association with a person’s trust in another party’s benevolence, but rather a
complex and curvilinear one” (pp. I4-5).
“We found that, in new/early relationships, the bases of trust in another party’s benevolence are rooted primarily in expectations associated with demographic prototypes33; in medium-length relationships, they are rooted primarily in behavioral expectations gathered from moderate social interaction; and in old/long
33 E.g. gender similarity was significantly associated
45
relationships, they are rooted primarily in personal knowledge of shared perspectives” (Levin, Whitener, & Cross, 2004 p. I5).
Interestingly, the authors’ (Levin, Whitener, & Cross, 2004) research findings showed
trustworthy behaviors to be the greatest in relationships that were neither new nor old, but
in-between.
2.4 Social-Cognitive Factors and Knowledge Sharing
The following section explores existing literature for relationships between the previously
discussed social-cognitive factors and knowledge sharing behavior. As with the previous
section, the review will begin with homophily factors and then discuss relationships with
shared perspective, tie strength, and relationship length. Each social-cognitive factor was
defined in the previous section.
2.4.1 Homophily and Knowledge Sharing Behavior
Below is a brief summary of the literature exploring the relationships between status
homophily and knowledge sharing or knowledge sharing behavior. Both ascribed and
acquired characteristics are explored, with ascribed characteristics including age, gender,
race, ethnicity, and citizen and immigration data, and acquired characteristics including
education, experience, and marital status.
Age Homophily and Knowledge Sharing Behavior
Research that found a positive relationship between age homophily and rate of contact
was previously discussed (Fisher, 1977; Verbrugge, 1977; Feld, 1982; Shrum, Cheek Jr.,
& Hunter, 1988). Building on this research, one may assume that an increase in the rate
of contact between co-workers gives them ample opportunities for knowledge sharing,
use, and explanation. Research studies conducted by Ojha (2005) and Riege (2005) may
suggest evidence for this theory, with findings showing that the more age compatible a
team was, the more likely that team would be to engage in effective knowledge sharing.
Furthermore, both authors provided evidence that age differences were likely to suppress
knowledge sharing. Ojha (2005) explained this relationship by suggesting that persons of
similar age were more likely to group together and interact freely.
46
Gender Homophily and Knowledge Sharing Behavior
As previously discussed, the work of Huckfeldt and Sprague (1995) found that less
intimate and more content-based relationships (like the ones one may find in a
professional organization) tend to be significantly more gender segregated. Other studies
set in professional environments also found evidence of gender segregation (e.g. Bielby
& Baron, 1986; Kalleberg, Knoke, Marsden, & Spaeth, 1996). Ibarra’s (1992) work
found that gender homophily played a role in inequality for access to professional
interaction networks and their knowledge sources and contents. Connelly and Kelloway
(2003) offer a nice illustration of this relationship by stating “if knowledge sharing is
most likely to occur among friends, and employees are most likely to become friends
with similar others (e.g. of the same gender), then employees of a minority gender may
be less likely to share knowledge freely” (p. 300). In his work on knowledge sharing in
software project teams, Ojha (2005) discovered that the more compatible a person was
with their group in terms of gender, the more likely they were to practice knowledge
sharing.
Race / Ethnicity Homophily and Knowledge Sharing Behavior
Mehra, Kilduff, and Brass (1998) suggest that “research on the patterning of social
relations in organizations has shown the importance of visible categories such as race as
basis for identification and network formation” (p. 441) and “[as] basis for shared identity
and social interaction” (p. 450). Along similar lines, it was previously discussed that
strong homophily on race and ethnicity has been discovered in marriages (Kalmijn,
1998), individuals that confide in each other (Marsden, 1987; 1988), school-related
relationships (Shrum et al., 1988; Maccoby, 1998), and work-related relationships
(Lincoln & Miller, 1979; Ibarra, 1995). Further, Marsden (1987), Blum (1984), Blau and
his colleagues (Blau, Beeker, & Fitzpatrick, 1984; Blau, Blum, & Schwartz, 1982; Blau,
Ruan, & Monika, 1991) found race to predict with whom individuals shared important
matters.
Within the organizational setting, Ibarra (1993) argues that “cross-race developmental
relationships [e.g. mentor/apprentice] appear to be harder to develop, and they provide a
47
narrower range of benefits for minorities” (p. 70). Ironically, Ibarra (1993) insists that it
is not that these cross-race contacts are not available to their counterparts; it is just that,
statistically, these relationships seem more difficult and less likely to form in the
organizational setting (1993). Ibarra (1993) further argues that this difficultly in forming
relationships limits the knowledge sharing opportunities provided to these employees
(e.g. often visible minorities) in the form of mentoring. In subsequent research, Ibarra
(1995) found that ethnic minorities were much more likely (than the ethnic majority) to
seek advice and support from their heterogeneous counterparts. Other organizational
studies (Brass, Galaskiewicz, Greve, & Tsai, 2004; Mehra, Kilduff, & Brass, 1998) have
suggested similar findings, even citing race homophily as a driver for effective
organizational knowledge sharing (Makela, Kalla, & Piekkari, 2007).
Citizenship / Immigration Homophily and Knowledge Sharing Behavior
Geringer (1988), Oetzel (1995), and Bornhorst et al. (2010) found that similarity in
nationality influenced communication patterns and styles of interaction, which arguably
play a role in the sharing and use of knowledge. In their study, Manev and Stevenson
(2001) found shared nationality to be a foundation on which managers create and sustain
strong network ties. These strong network ties create opportunities for interaction, which
further open opportunities for knowledge sharing to take place. According to Mäkelä,
Andersson, and Seppälä (2012) “knowledge sharing is a bi-product of interaction: the
more extensive interaction there is between two people, the more opportunities they have
for both intentional and serendipitous knowledge sharing” (p. 442). The authors also
argued that network ties tend to be stronger between employees coming from similar
national backgrounds, because nationality can become a key driver for social
categorization. Their findings supported their reasoning, showing evidence that
individuals of the same nationality shared more knowledge across unit boundaries.
Educational Homophily and Knowledge Sharing Behavior
As previously mentioned, individuals of similar education and experience are more likely
to form network ties (Louch, 2000) and cooperative ties (Galaskiewicz & Shatin, 1981;
Laumann, Gagnon, Michael & Michaels, 1994; Verbrugge, 1977). Ojha’s (2005) research
48
found that individuals with greater differences in their levels of education were less likely
to share. Ojha (2005) also found differences in the type of education to be an inhibitor for
participating in knowledge sharing. Finally, Riege’s (2005) work established a
contributory relationship between educational level and the likelihood to share
knowledge.
Marital Homophily and Knowledge Sharing Behavior
Fischer (1982) found significant evidence of marital status homophily in his study of
personal networks, where married respondents more frequently named married referents,
never married named the never married referents, and the divorced named divorced.
Kalmijn and Vermunt (2007) offered an explanation for this, by arguing that “people may
have a preference for interacting with others in the same marital status category [because
those] in the same marital status position may better understand, they may have more
relevant information for each other, and they may share a certain lifestyle which increases
possibilities for joint activities” (p. 27). The authors (2007) also suggested that this may
be because individuals want to interact with people who went through similar life
transitions, or they are also going through these life transitions. “If a person’s friends start
getting married, for example, this may speed up this person’s decision to get married as
well, thereby increasing the degree of marital status homogeneity. Friends who marry
will become disconnected from friends who remain single, and friends who divorce will
become disconnected from friends who remain married” (Kalmijn & Vermunt, 2007, p.
27).
There was no research found to discuss the direct effect of marital status on knowledge
sharing or knowledge sharing behavior. However, there are some studies that suggest
planning groups, such as Parent Teacher Associations (PTAs), to be homogeneous with
respect to marital status (McPherson & Smith-Lovin, 1987). Also, in their review of
homophily, Rogers and Bhowmik (1970) mentioned a 1967 study at the University of
Chicago, conducted by Palmore that found individuals living in the Chicago ghettos to
share family planning ideas with others of like marital status.
49
2.4.2 Shared Perspective and Knowledge Sharing Behavior
As previously discussed, knowledge is highly contextual and circumstantial (Goman,
2002); it is always developed in a specific context and is rarely interpreted by the receiver
in the exact way as it was intended by the transmitter (Husted & Michailova, 2002). A
key problem is representing the context in which knowledge is created and is relevant
(Choo, 2000), which makes transferring knowledge problematic (Brown & Duguid,
1991; Kogut & Zander, 1992; Empson, 2001). Reasons for contextual mismatches
include differences in mental/conceptual frameworks or culture and language (Hendricks,
1999). For this reason, several researchers have identified a shared perspective as a driver
for knowledge sharing (e.g. Nahapiet & Ghoshal, 1998; Nonaka & Takeuchi, 1995;
Brown & Duguid, 2000; Mäkelä & Brewster, 2009).
Knowledge is easier to transfer when it is rooted in the domain or practice of the
individuals participating (Brown & Duguid, 1998). Nonaka (2002) makes a similar
argument with information: “the mere transfer of information will often make little sense
if it is abstracted from embedded emotions and nuanced contexts that are associated with
shared experiences” (p. 442). For knowledge to be shared, the receiver and the transmitter
must share a contextual base (Mäkelä & Brewster, 2009). Brown and Duguid (2000)
called this “a shared framework for interpretation” (p. 107). The receiver must possess
what Swap et al. (2001) called a hook or receptor, which assimilates the information
provided by the transmitter. Argyres (1999) called this “a ‘technical grammar’ for
communication” (p. 162).
Shared Language and Knowledge Sharing Behavior
Triandis (1960) found connections between similarity in language and effective
knowledge understanding and use. Tsoukas and Vladimirou (2001) proclaimed a
common language as the most important cultural tool needed in assisting an individual in
‘drawing distinctions’ within a collectively generated ‘domain of action’. Nonaka (1994)
argued that shared language was paramount to the transfer and integration of tacit
knowledge rooted in the sharing of common schemata and frameworks, such as stories,
analogies, and metaphors.
50
Research also found that shared language facilitated knowledge sharing behavior
through: a common understanding of how to act (Tsai & Ghoshal, 1998); an ability to
gain access to the right people for information (Nahapiet & Ghoshal, 1998); a “common
conceptual apparatus for evaluating the likely benefits of exchange” (Chiu, Hsu, &
Wang, 2006, p. 1878); and a common framework for the combination of knowledge
(Nahapiet & Ghoshal, 1998). Several of these benefits were represented in Zenger and
Lawrence’s (1989) research, which found that shared language determined the efficiency
of communication, by acting as guide for how information was interpreted and responded
to. Henderson’s (2005) study of multinational corporation international project teams also
concluded that language diversity affected knowledge sharing through interpretation,
ultimately influencing overall team performance. Ojha’s (2005) research, which collected
data on 83 software project teams in 20 organizations, found a significant relationship
between language compatibility and the likelihood of participating in knowledge sharing
behavior. Chiu, Hsu, and Wang (2006) added a distinction between the differential
effects shared language had on the quality and quantity of knowledge, suggesting shared
language to have a significant positive effect on the quality of knowledge shared, and no
effect on quantity.
Shared Vision and Knowledge Sharing Behavior
Research also found that shared vision helps workers see the potential value of their
knowledge exchange (Tsai & Ghoshal, 1998) and provides a guideline for understanding
which knowledge was worth acquiring and disseminating (Hoe & McShane, 2002). Also,
if shared vision was high among co-workers, then the knowledge sharing process could
tolerate a certain degree of ‘inefficiency’, as long as the bulk of employee actions were
pointed in a unified direction (Hoe & McShane, 2002).
Tsai and Ghoshal (1998) found that management teams who shared a vision were more
likely to participate in knowledge sharing and resource exchange activities. Chang et al.
(2011) showed that shared vision was a necessary precondition for knowledge sharing,
and that it had a positive effect on the overall willingness to share knowledge, ideas, and
opinions, and to answer colleague questions. Similarly, Hoe and McShane’s (2002) study
51
of IT sales, customer service, and technical consulting employees in 11 business units of
a large ICT firm, found shared vision to be a strong predictor for knowledge sharing.
Finally, Chiu, Hsu, and Wang (2006) found shared vision to have a significant positive
effect on the quality of knowledge shared and, more interestingly, a significant negative
effect on quantity. This finding suggests that having a higher shared vision encouraged
more succinct meaningful exchanges between individuals.
2.4.3 Tie Strength and Knowledge Sharing Behavior
In his seminal paper, Granovetter (1973) concluded that weak ties were more likely to act
as a source for unique and useful information. Granovetter (1973) reasoned that an
individual’s strong ties likely had the same or similar information and network access to
those already in the network. On the other hand, weak ties provided the opportunity to
access new network ties, as well as useful novel information. “Weaker ties reflect a path
along which new information or novel insights are more likely to travel in comparison to
stronger ones” (Levin & Cross, 2004 p. 1480). Subsequent research by Granovetter
(1982), as well as Rogers (1995), found that weak ties were also instrumental in the
diffusion of ideas. Research on weak ties has also demonstrated they are beneficial in the
dissemination of information (Uzzi & Lancaster, 2003; Cross & Cummings, 2004) and
technical advice (Constant, Sproull, & Kiesler, 1996). Hansen (1999) found weak ties
useful, because they were, for the individual, less costly to maintain than strong ties.
Other research suggested the exact opposite relationship to exist. For instance,
Krackhardt (1992) suggested that strong ties were more important than weak ones to the
individual, because these were the ties that made themselves accessible and, more
importantly, were willing to help. Numerous other studies have confirmed this claim, by
suggesting that strong ties are channels through which useful knowledge travels, and that
these paths have greater knowledge exchange occurring within them (Hansen, 1999;
Szulanski, 1996; Ghoshal, Korine, & Szulanski, 1994; Uzzi, 1996, 1997). Levin and
Cross’s (2004) research even found a positive statistical relationship between strong ties
and the perception that the knowledge received was useful. Much like Krackhardt (1992),
Levin and Cross (2004) reasoned that this occurred because “strong ties [were] more
52
likely to expend effort to ensure that a knowledge seeker sufficiently understands and can
put into use newly acquired knowledge” (p. 1479).
2.4.4 Relationship Length and Knowledge Sharing Behavior
Very few research studies were found that tested the length of relationship, and a direct
relationship between relationship length and knowledge sharing behavior. In their
examination of interpersonal relationships and knowledge sharing at multinational
corporations, Mäkelä and Brewster (2009) found that relationship length influenced the
extent of knowledge sharing. In subsequent research Mäkelä, Andersson, and Seppälä
(2012) found knowledge sharing to be “through enhanced interaction extensiveness
[which is] the product of ‘frequency of interaction’ and ‘length of the relationship’” (p.
447).
Other researchers expanded their theoretical models to include knowledge sharing
behaviors, by proposing a more complex relationship between knowledge sharing
behavior and relationship length, mediated through trust (Renzl, 2008; Levin, Whitener,
& Cross, 2006; Coleman 1988).
2.5 Conceptual Framework
This section presents the conceptual framework for examining the effect social-cognitive
factors and trust have on organizational knowledge sharing behavior. The conceptual
framework is based on research studies, presented in the previous sections, which
explored various relationships between trust and knowledge sharing (or knowledge
sharing behavior) in organizations. It is also based on research studies, also presented in
the previous sections, which identified social or cognitive factors that have been found to
influence trust, knowledge sharing behavior or both. Research studies conducted within
organizational settings were given priority in the review, but were not always available.
Knowledge sharing behavior was identified in the conceptual framework using three
important behavioral conditions (presented in Section 2.1.3) necessary for effective
knowledge sharing behavior to take place. First, the knowledge source must be willing to
53
share the knowledge they posses (i.e. willingness to share knowledge). Second, the
knowledge receiver must be willing to receive and use the knowledge that is shared (i.e.
willingness to use knowledge). Finally, the knowledge receiver must perceive the
knowledge shared as being useful to their individual work, the project, or the
organization as a whole (i.e. perceived receipt of useful knowledge). In addition,
willingness to share and use knowledge are further separated to consider overall
knowledge, tacit knowledge, and explicit knowledge.
The first social-cognitive factor identified in the literature and appearing in the
conceptual framework (Figure 2.5) is homophily. According to existing research,
homophily (or similarity) between two individuals based on certain ascribed and acquired
characteristics (i.e. age, gender, race/ethnicity, citizenship/ immigration, education, and
marital status) has been found to have a positive effect on the level of trust the two
individuals perceive to have for one another (Section 2.3.1). Homophily based on these
characteristics has also been found to positively influence effective knowledge sharing
activities and behaviors (Section 2.4.1). Figure 2.6 shows a decomposition of the
homophily factor, noting each of the ascribed and acquired characteristics identified in
the study, and their proposed relationship with trust and knowledge sharing behavior.
Arrows in Figures 2.5 and 2.6 indicate these relationships. The influence of each type of
homophily factor on trust and knowledge sharing behavior will be explored in the study.
The second and third social-cognitive factors identified in the conceptual framework are
shared language and shared vision between co-workers (Figure 2.5). These factors were
based on Levin, Whitener, and Cross’ (2006) concept of organizational shared
perspective, which is an amalgamation of the two. Several previous organizational
studies (Section 2.3.2) have found positive relationships between shared language and
trust, as well as between shared vision and trust. Specifically, co-workers who shared a
higher degree of language or vision were found to have higher trust for one another.
Previous studies (Section 2.4.2) have also found shared language and shared vision to
have a positive influence on effective knowledge sharing activities and behaviors. Arrows
in Figure 2.5 indicate these relationships.
54
The fourth social-cognitive factor identified in the literature review is tie strength
between co-workers. Tie strength was primarily based on the work of Granovetter (1973),
who conceptualized it as “a combination of the amount of time, the emotional intensity,
the intimacy, and the reciprocal services which characterize the tie” (p. 1361). Research
by Levin and Cross (2004) and Krackhardt (1992) suggested that a stronger tie strength
between individuals led to higher perceived trust among them. However, the relationship
between tie strength and knowledge sharing behaviors has not been entirely
straightforward. For example, Granovetter (1973), Levin and Cross (2004), Rogers
(1995) and other authors (Uzzi & Lancaster, 2003; Cross & Cummings, 2004; Constant,
Sproull, & Kiesler, 1996) found weak ties to provide access to more information and
channels for knowledge, as well as new knowledge. These ties are also more efficient,
since they are less costly (e.g. time) to maintain (Hansen, 1999). On the other hand,
several other researchers (Krackhardt, 1992; Hansen, 1999; Szulanski, 1996; Ghoshal,
Korine, & Szulanski, 1994; Uzzi, 1996, 1997) found strong ties to promote more
effective knowledge sharing activities and behaviors because these are the ties that are
more accessible and willing to help. The conceptual framework in this study takes into
consideration both possibilities by simply identifying some type of relationship as
existing between tie strength and knowledge sharing behavior and between tie strength
and trust.
The final social-cognitive factor identified in the literature review, and included in the
conceptual framework (Figure 2.5), is relationship length. Relationship length was
defined using Levin, Whitener, and Cross (2004) as how long one co-worker has known
another. Numerous studies (Levin, Whitener, & Cross 2004; Dirks & Ferrin, 2002;
Lewicki & Bunker, 1996; Coleman, 1988; Levin, Whitener, & Cross, 2004) have
suggested that the longer an individual knows another, the more they should theoretically
trust them, since, over time, the individual becomes more familiar with their co-worker
through interaction. This interaction allows for both individuals to learn about each other
(Levin, Whitener, & Cross, 2004) and build reputations (Coleman, 1988). As previously
mentioned, few studies (Mäkelä & Brewster, 2009; Mäkelä, Andersson, & Seppälä,
2012) were found that discovered a direct influence of relationship length on knowledge
55
sharing behavior but several studies suggested a relationship existed (and was mediated)
through trust (Renzl, 2008; Levin, Whitener, & Cross, 2006; Coleman 1988). The
conceptual framework in this study considers both types of relationships by simply
indicating some type of relationship as existing between relationship length and
knowledge sharing behavior and between relationship length and trust.
The final factor included in the conceptual framework found to influence knowledge
sharing behavior is trust (Figure 2.5). Trust was defined using Mayer et al.’s (1995)
proposed model of organizational trust, which includes benevolence-based trust, ability-
based trust, integrity-based trust, and propensity to trust. The literature review (Section
2.2.2) identified an overwhelming positive relationship between trust and specific
knowledge sharing behaviors, other organizational behaviors (which might be deemed as
beneficial for knowledge sharing), resource/information exchange, and decision-making
and problem solving (which are outcomes of knowledge sharing behavior). Arrows in
Figure 2.5 indicate this relationship.
Since each of the social-cognitive factors has been found in the literature to have some
direct effect on both knowledge sharing behavior and trust, in addition to depicting these
direct relationships, the conceptual framework also considers the indirect or mediating
effect through trust onto knowledge sharing behavior.
Each of the concepts explored in the present study, and possible relationships between
them, are depicted in the conceptual framework in Figure 2.5. Factors examined, along
with trust, include homophily, shared language, shared vision, tie strength, and
relationship length. Figure 2.6 depicts a decomposition of the homophily factor,
separating ascribed and acquired characteristics.
56
Figure 2.5 Conceptual Framework Based on Literature Review
Figure 2.6 Conceptual Framework of Decomposed Homophily Factors
57
Chapter 3: Research Design and Methodology 3.0 Chapter Overview
This chapter details the research design and methodology used in this study. The chapter
begins with a discussion of the research questions and associated hypotheses. Next, the
data measurement strategies are discussed, with a clarification of the variables and
operationalization. The chapter continues with data analysis strategies, elaborating the
validity and reliability of measures, as well as how hypotheses were tested. The chapter
concludes with a discussion of population, site, and data collection methods.
3.1 Research Questions and Hypotheses
As previously mentioned, the overarching research question was what are the factors
that influence knowledge sharing behavior directly and indirectly through trust? To
elaborate this research question, three additional questions were posed. What are the
significant relationships between trust and knowledge sharing behavior? What are the
significant relationships between social-cognitive variables and trust? And, what are the
significant relationships between social-cognitive variables and knowledge sharing
behavior?
To better understand the direct and indirect effects of social-cognitive variables and trust
on knowledge sharing behavior, two additional research questions were posed. The first
examined the collective effect social-cognitive factors and trust had on knowledge
sharing behavior (i.e. what is the collective effect of the identified social-cognitive
variables and trust on knowledge sharing behavior?). The second additional research
question was concerned with the indirect or intervening effect trust had between social-
cognitive factors and knowledge sharing behavior (i.e. does trust act as a mediating
variable between social-cognitive variables and knowledge sharing behavior?).
Based on the findings from the literature review, the relationships between the
independent and dependent variables were hypothesized. Testing these hypothesized
relationships formed the basis through which the research questions were answered. A
summary of the five research questions and associated hypotheses is shown in Table 3.1.
58
Research Question Associated Hypotheses
RQ1: What are the significant relationships between social-cognitive variables and trust?
H1: Age homophily will be positively related to Trust H3: Educational homophily will be positively related to Trust H5: Gender homophily will be positively related to Trust H18: Race/Ethnic Homophily will be positively related to Trust H20: Citizen/Immigration homophily will be positively related to Trust H22: Marital Homophily will be positively related to Trust H7: Shared Language will be positively related to Trust H9: Shared Vision will be positively related to Trust H11: Relationship Length will be positively related to Trust H13: Tie Strength will be positively related to Trust
RQ2: What are the significant relationships between social-cognitive variables and knowledge sharing behavior?
H2: Age Homophily will be positively related to Knowledge Sharing Behavior H4: Educational Homophily will be positively related to Knowledge Sharing Behavior H6: Gender Homophily will be positively related to Knowledge Sharing Behavior H19: Race/Ethnic Homophily will be positively related to Knowledge Sharing Behavior H21: Citizen/Immigration homophily will be positively related to Knowledge Sharing Behavior H23: Marital Homophily will be positively related to Knowledge Sharing Behavior H8: Shared Language will be positively related to Knowledge Sharing Behavior H10: Shared Vision will be positively related to Knowledge Sharing Behavior H12: Relationship Length will be positively related to Knowledge Sharing Behavior H14: Tie Strength will be positively related to Knowledge Sharing Behavior
RQ3: What are the significant relationships between trust and knowledge sharing behavior?
H15: Trust will be positively related to Knowledge Sharing Behavior
RQ4: What is the collective effect of the identified social-cognitive variables and trust on knowledge sharing behavior?
H16: Overall Trust and Social-Cognitive Factors explain Knowledge Sharing Behavior
RQ5: Does trust act as a mediating variable between social-cognitive variables and knowledge sharing behavior?
H17:Overall Trust will be a mediating variable between Social-Cognitive Factors and Knowledge Sharing Behavior
Table 3.1 Research Questions and Associated Hypotheses
3.2 Operationalization of Variables
3.2.1 Social-Cognitive Variables (Independent Variables)
Homophily Measures
To measure status homophily between individuals, this research used variables to capture
both ascribed and acquired characteristics. Each respondent was asked to answer
questions that provided information related to characteristics about themselves; and then,
as best as they could, answer questions describing characteristics about two fellow co-
workers they mentally selected. Ascribed characteristic variables of interest included
measures of age, gender, race/ethnicity, as well as citizen/immigration data. Acquired
characteristic variables of interest included educational history, experience, and marital
59
status. Since questions regarding both acquired and ascribed characteristics may be
sensitive in nature, appropriate wording became imperative. To avoid such concerns, this
study used Statistics Canada’s 2006 Census questions (see Survey Instrument in the
Appendix), to capture all homophily related survey items (Statistics Canada, 2006). This
research did not introduce or measure any variables that attempt to capture value
homophily, as these measures may be conceptually too similar to shared vision.
Shared Perspective (Shared Language and Shared Vision)
Based on the work of Levin, Whitener, and Cross (2006), two variables were introduced
to measure the extent to which a shared perspective (shared language and shared vision)
existed between participants in the study. To measure shared perspective, Levin,
Whitener, and Cross (2006) conducted a cross-sectional survey of employees working in
a knowledge intensive division of a U.S. pharmaceutical company, a British bank, and a
Canadian oil and gas company. A total of six items were used: three items to measure
shared language (Cronbach’s alpha of .67) and three items for shared vision (Cronbach’s
alpha of .78). Table 3.2 summarizes the items used by Levin, Whitener, and Cross (2006)
for shared language, the authors’ study population, and the proposed adapted items that
were used in the research presented here. The proposed adapted measure for shared
language is also indicated using a three-item measure. Items included: “I could
understand completely what this person meant when he or she was talking”; “I was
familiar with the jargon/terminology that he or she used”; and, “it felt like we could
communicate on the same wavelength”. All shared language items were measured using
a five-point scale (1 – Strongly Disagree, 5 – Strongly Agree).
Levin, Whitener, and Cross (2006) also developed three items to measure shared vision,
which, they claim, were motivated by, and are similar to, the two-item measure for
shared vision used by Tsai and Ghoshal (1998), in their research of management teams in
a multinational electronics company. Table 3.2 summarizes the items used by Levin,
Whitener, and Cross (2006) and Tsai and Ghoshal (1998) for shared vision, the authors’
study population, and the proposed adapted items that were used in the research presented
here.
60
Based on the work of Levin, Whitener, and Cross (2006) and Tsai and Ghoshal (1998), in
this study, shared vision was measured using a 5-item scale. The items were: “I felt like
this person and I were working toward completely different goals”; “I assumed that this
person and I cared about the same issues”; “I believed that this person and I shared a
commitment to a common purpose”; “I believed that this person and I shared the same
ambitions and vision”; and, “I believed that this person and I shared enthusiasm about
pursuing the collective goals and missions of the whole organization”. All shared vision
items were measured on a five-point scale (1 – Strongly Disagree, 5 – Strongly Agree).
61
Author(s) Factor Study Population
Authors’ Item Adapted Item (final question items in
bold)
Levin, Whitener, and Cross, 2006
Shared Language
Knowledge workers on a project
Prior to seeking information/advice from this person on this project, I could understand completely what this person meant when he or she was talking.
I could understand completely what this person meant when he or she was talking.
Levin, Whitener, and Cross, 2006
Shared Language
Knowledge workers on a project
Prior to seeking information/advice from this person on this project, I was familiar with the jargon/terminology that he or she used.
I was familiar with the jargon/terminology that he or she used.
Levin, Whitener, and Cross, 2006
Shared Language
Knowledge workers on a project
Prior to my seeking information/advice from this person on this project, it felt like we could communicate on the same "wavelength”.
It felt like we could communicate on the same "wavelength”.
Levin, Whitener, and Cross, 2006
Shared Vision
Knowledge workers on a project
Prior to seeking information/advice from this person on this project, I felt like this person and I were working toward completely different goals. [reverse coded]
I felt like this person and I were working toward completely different goals.
Levin, Whitener, and Cross, 2006
Shared Vision
Knowledge workers on a project
Prior to seeking information/advice from this person on this project, I assumed that this person and I cared about the same issues.
I assumed that this person and I cared about the same issues.
Levin, Whitener, and Cross, 2006
Shared Vision
Knowledge workers on a project
Prior to seeking information/advice from this person on this project, I believed that this person and I shared a commitment to a common purpose.
I believed that this person and I shared a commitment to a common purpose.
Tsai and Ghoshal, 1998
Shared Vision
Management Teams
Our unit shares the same ambitions and vision with other units at work
I believed that this person and I shared the same ambitions and vision.
Tsai and Ghoshal, 1998
Shared Vision
Management Teams
People in our unit are enthusiastic about pursuing the collective goals and missions of the whole organization.
I believed that this person and I shared enthusiasm about pursuing the collective goals and missions of the whole organization.
Table 3.2 Operationalization and Measurement of the Shared Perspective Research Variables
Tie Strength
Based on the work of Granovetter (1973), and two items used by Hansen (1999) in his
study for inter-unit tie weakness (i.e. one for interaction frequency and another for
closeness; see Table 3.3), Levin and Cross (2004) developed their measure for
organizational tie strength. Using pre-test feedback, the authors (Levin & Cross, 2004)
62
clarified Hansen’s (1999) scales, and introduced new ties for those participants with no
prior contact. Levin and Cross (2004) also included a third item for communication
frequency, to increase reliability. Since the three measures (Table 3.3) used different
scales, Levin and Cross (2004) normalized each of them, before creating the overall
variable. Cronbach’s alpha for Levin and Cross’ (2004) three-item operationalization of
tie strength was .9. For the present study, the authors’ (Levin & Cross, 2004) items
(Table 3.3) were further adapted, as a result of conversations with Daniel Levin (Evans,
personal communication, October, 2010), who, with his colleagues, suggested a
rethinking of Marsden and Campbell’s (1984) understanding on tie strength. Levin,
Walter, and Murnighan (2011) suggested that emotion-based (or closeness) measures
were, at the very least, as important as the commonly used interaction and
communication frequency measures. With this in mind, a fourth item relating to closeness
was introduced in this study, to increase reliability (Table 3.3) (i.e. two of the four tie
strength items adapted for use in the present study related to closeness and the other two
items to interaction/communication frequency).
In addition, it was reasoned that tie strength prior to working on a project could be
conceptually different from tie strength developed while working on a project. The
perceived difference between tie strength prior to and while on could be justified by
considering that the frequency with which two co-workers interact can easily vary from
project to project, based on the specifics of the project and its requirements. Also, the
closeness two co-workers feel for each other can change, as they continue to work
together and have opportunities to interact. It is possible that high levels of interaction
frequency exist between co-workers, prior to working on the project, and still have no
relation to the tie strength felt while on a new project. Making this distinction is
important, since one practical application for understanding tie strength may be choosing
to allocate workers to projects based on their prior communication or interaction
frequency. However, past interaction or communication may not guarantee, or be a
predictor for, tie strength between the same co-workers while they are working on a new
project together. Since this conceptual difference may have a different effect on the
dependent variables, a distinction between the two was made. Therefore, each of the tie
63
strength items appeared twice on the survey; once asking the respondent about their
relationship with their co-workers prior to the project and once again to measure the
same relationship while on the project.
In this study, tie strength was measured using an eight-item scale. Items were: “Prior to
(While) working with each of the co-workers on the project you shared, how close was
your working relationship?”; “My relationship with each of the co-workers I mentally
selected was a very intense, strong relationship prior to (while) working on the matter or
firm-related project we shared”; “Prior to (While) working with each of the co-workers
on the project you shared, how often did you communicate?”; “Prior to (While) working
with each of the co-workers on the project you shared, to what extent did you typically
interact with each of them?”. The response categories for the four tie strength items
varied (see Survey Instrument in the Appendix for specific scales).
64
Author(s) Factor Study Population
Authors’ Item Adapted Item (final question items in
bold)
Levin and Cross, 2004
Tie Strength
Knowledge workers (Pharmaceutical)
Prior to seeking information/advice from this person on this project how close was your working relationship with each person? If you had no prior contact at all with this person before you sought information /advice from him or her on this project, please choose 7. (1 = very close; 4 = somewhat close; 7 = distant)
Prior to (While) working with each of the co-workers on the project you shared, how close was your working relationship?
Levin and Cross, 2004
Tie Strength
Knowledge workers (Pharmaceutical)
Prior to seeking information/advice from this person on this project how often did you communicate with each person? If you had no prior contact at all with this person before you sought information /advice from him or her on this project, please choose 7. (1 = daily; 2 = twice a week; 3 = once a week; 4 = twice a month; 5 = once a month; 6 = once every 2nd month; 7 = once every 3 months or less (or never))
Prior to (While) working with each of the co-workers on the project you shared, how often did you communicate?
Levin and Cross, 2004
Tie Strength
Knowledge workers (Pharmaceutical)
Prior to seeking information/advice from this person on this project to what extent did you typically interact with each person? (1 = to no extent; 2 = to little extent; 3 = to some extent; 4 = to a great extent; 5 = to a very great extent)
Prior to (While) working with each of the co-workers on the project you shared, to what extent did you typically interact with each of them?”
Hansen, 1999
Interunit Tie Weakness
Divisions in a large electronics company
How frequently do (did) people in your division interact with this division (on average over the past two years)? (0 = once a day, 1 = twice a week, 2 = once a week, 3 = twice a month, 4 = once a month, 5 = once every 2nd month, 6 = once every 3 months.)
Prior to (While) working with each of the co-workers on the project you shared, to what extent did you typically interact with each of them?” *Repeated item
Hansen, 1999
Interunit Tie Weakness
Divisions in a large electronics company
How close is (was) the working relationship between your division and this division? (0 = "Very close, practically like being in the same work group," 3 = "Somewhat close, like discussing and solving issues together," 6 = "Distant, like an arm's-length delivery of the input".)
Prior to (While) working with each of the co-workers on the project you shared, how close was your working relationship? *Repeated item
New Item Tie Strength
N/A N/A My relationship with each of the co-workers I mentally selected was a very intense, strong relationship prior to (while) working on the matter or firm-related project we shared.
Table 3.3 Operationalization and Measurement of the Tie Strength Research Variable
65
Relationship Duration (Length)
Based on the work of Levin, Whitener, and Cross (2004; 2006), a one-item variable was
used to measure relationship duration/length between co-workers. The authors (2006)
simply asked “How long had you known this person prior to your seeking
information/advice from him or her on this project (in months and years)?” For the
present study, a similar item was used: “Approximately, how long have you known each
of the co-workers?” This study recognizes that relationship duration may not be a
predictor for interaction pattern, although, as Levin, Whitener, and Cross (2004; 2006)
argue, it is one important proxy for the extent of interaction.
3.2.2 Trust (Mediating Variable)
In Mayer et al.’s (1995) opinion, previous research, models, and measures for trust have
not been clear in differentiating trust, factors that led to trust, and the outcomes of trust.
However, the authors’ model conceptualizes trust in a fashion that distinguishes it from
its outcomes and from its antecedents. Their model and definition also attempts to
measure trust factors with respect to the characteristics of the trustor (i.e. propensity) and
the perceived characteristics of the trustee (i.e. ability, benevolence, and integrity),
something the authors argue other models have neglected (Mayer et al., 1995;
Schoorman, Mayer, & Davis, 1996b). “The failure to clearly specify the trustor [the
trusting party] and the trustee [the party to be trusted] encourages the tendency to change
referents and even level of analysis, which obfuscates the nature of the trust relationship”
(Mayer et al, 1995, p. 711). As discussed in the literature review chapter (Section 2.2.1),
the model suggests a separate measure for ‘propensity to trust’ (p. 715), which captures a
personal characteristic of the trustor and perceived trustworthiness, a set of characteristics
specific to the trustee (i.e. ability, benevolence, and integrity) (Mayer et al., 1995).
Using Mayer et al.’s (1995) distinction, the present research proposed to measure trust by
first asking respondents to answer items relating to their ‘propensity to trust’, and then
asking the respondents to make an assessment of their co-worker’s factors of perceived
trustworthiness (i.e. ability, benevolence, and integrity). This allowed for the study to
explore the relationship of collective or overall trust (i.e. the sum of the three perceived
66
trustworthiness factors) and each of the individual trustworthiness factors, against other
independent variables in the study, as well as dependent variables. The operationalization
of each type of trust factor is presented below.
Ability-based Trust (Competence-based trust)
Based on the work of several researchers (Mayer & Davis, 1999; Levin & Cross, 2004;
McAllister, 1995; Chattopadhyay, 1999), in this study, nine items (see Table 3.4) were
introduced to measure perceived ability-based trust in a co-worker. To increase
reliability, items for ability-based trust were adapted and combined from similar
constructs, including Mayer and Davis’ (1999) ability-based trust, Levin and Cross’
(2004) competence-based trust, McAllister’s (1995) cognition-based trust, and
Chattopadhyay (1999) trust.
In their study of employee trust for top management, Mayer and Davis (1999) introduced
a measure to reflect ability-based trust in top management’s domain specific skills and
competencies. Ability-based trust was rated as high when management’s decisions
showed competence, or when management demonstrated skills in understanding
problems and resolving employee work-related issues. To measure ability-based trust,
Mayer and Davis (1999) used six items, which are summarized in Table 3.4. Cronbach’s
alpha for these six measures was .85 for the first wave of the study and .88 for the second
wave (Mayer & Davis, 1999).
Mayer, Davis, and Schoorman (1995; 1996ab) argue that ability-based trust and
competence-based trust are conceptually similar constructs, which suggests that
competence-based trust items may be included alongside ability-based ones. Similarly,
competence-based trust may be considered synonymous with McAllister’s (1995)
cognition-based trust. For example, Levin and Cross (2004) measured competence-based
trust using a measure adapted from two top loading items for cognition-based trust,
developed and used by McAllister (1995) in his study. Cronbach’s alpha for competence-
based trust measures was .80 (Levin & Cross, 2004). Like Levin and Cross (2004),
Chattopadhyay (1999) also adapted McAllister’s (1995) instrument, in his measurement
67
of trust as a mediating factor between demographic characteristics and organizational
citizenship behavior. Chattopadhyay (1999) used four of McAllister’s (1995) items,
including an adaptation of the three suggested for this research (Table 3.4).
McAllister’s (1995) work on trust suggested that there are two principal forms of trust:
cognition-based and affect based. To develop measures for these two forms of trust,
McAllister (1995) conducted a literature review on available measures of trust. Eleven
behavioral scholars were then asked to classify the initial pool of 48 items, into the
suggested two forms of trust. McAllister (1995) used these evaluations to create a subset
of twenty items, ten for each form of trust. Through exploratory factor analysis of pre-test
data on M.B.A. and business students, McAllister (1995) reduced the number of items to
the eleven ‘strongest-loaded items’ (p. 36). Cronbach’s alphas were .91 for cognition-
based trust items (six in total) and .89 for affect-based items (five in total).
According to McAllister (1995), cognition-based trust relates to the cognitive foundations
of trust. This form of trust is circumstantial and is based on a choice made for “good
reasons” (p. 25). These good reasons are usually based on past evidence of behavior, of
which ability and/or competence is a part. “Measures of trust in organizational settings
suggest that competence and responsibility are central elements” (McAllister, 1995, p.
26). Although not all of McAllister’s (1995) items fit within the current study, three items
of the cognition-based trust (Table 3.4) did relate to ability/competence-based trust.
68
Author(s) Factor Study Population
Authors’ Item Adapted Item (final question items in bold)
Mayer and Davis, 1999
Ability Top Management & Employees
Top management is very capable of performing its job.
This person is very capable of performing his/her job.
Mayer and Davis, 1999
Ability Top Management & Employees
Top management is known to be successful at the things it tries to do.
This person is known to be successful at the things he/she tries to do.
Mayer and Davis, 1999
Ability Top Management & Employees
Top management has much knowledge about the work that needs done.
This person has much knowledge about the work that needs done.
Mayer and Davis, 1999
Ability Top Management & Employees
I feel confident about top management’s skills.
I feel confident about this person’s skills.
Mayer and Davis, 1999
Ability Top Management & Employees
Top management has specialized capabilities that can increase performance.
This person has specialized capabilities that can increase performance.
Mayer and Davis, 1999
Ability Top Management & Employees
Top management is well qualified.
This person is well qualified.
Levin and Cross, 2004
Competence Knowledge workers
Prior to seeking information / advice from this person, I believed that this person approached his or her job with professionalism and dedication.
I believed that this person approached his or her job with professionalism and dedication.
Levin and Cross, 2004
Competence Knowledge workers
Prior to seeking information / advice from this person, given his or her track record, I saw no reason to doubt this person’s competence and preparation.
Given his or her track record, I saw no reason to doubt this person’s competence and preparation.
McAllister, 1995
Cognition Manager and knowledge workers
This person approaches his or her job with professionalism and dedication.
I believed that this person approached his or her job with professionalism and dedication. *Repeated item
McAllister, 1995
Cognition Manager and knowledge workers
Given this person’s track record, I see no reason to doubt his/her competence and preparation.
Given his or her track record, I saw no reason to doubt this person’s competence and preparation. *Repeated item
McAllister, 1995
Cognition Manager and knowledge workers
I can rely on this person not to make my job more difficult by careless work.
I can rely on this person not to make my job more difficult by careless work.
69
Chattopadhyay, 1999
Trust Workgroups (employees and peers)
The members of my work group approach their jobs with professionalism and dedication.
I believed that this person approached his or her job with professionalism and dedication. *Repeated item
Chattopadhyay, 1999
Trust Workgroups (employees and peers)
Given my work group’s track record, I see no reason to doubt his/her competence and preparation.
Given his or her track record, I saw no reason to doubt this person’s competence and preparation. *Repeated item
Chattopadhyay, 1999
Trust Workgroups (employees and peers)
I can rely on my group members not to make my job more difficult by careless work.
I can rely on this person not to make my job more difficult by careless work. *Repeated item
Table 3.4 Operationalization and Measurement of the Ability-Based / Competence-Based Trust
In this study, ability-based trust was measured with a nine-item scale (Table 3.4). Items
were: “This person is very capable of performing his/her job”; “This person is known to
be successful at the things he/she tries to do”; “This person has much knowledge about
the work that needs done”; “I feel confident about this person’s skills”; “This person has
specialized capabilities that can increase performance”; “This person is well qualified”;
“I believed that this person approached his or her job with professionalism and
dedication”; “Given his or her track record, I saw no reason to doubt this person’s
competence and preparation”; and, “I can rely on this person not to make my job more
difficult by careless work”. All ability-based trust items were measured on a five-point
scale (1 – Strongly Disagree, 5 – Strongly Agree).
Benevolence-Based Trust
Based on the work of several researchers (Mayer & Davis, 1999; Levin, Whitener, &
Cross, 2006; Johnson, Cullen, Sakano, & Takenouchi, 1996), in this study, nine items
(Table 3.5) were adapted to measure perceived benevolence-based trust in a co-worker.
To increase reliability for this measure, items for benevolence-based trust were adapted
and combined from similar items used by Mayer and Davis (1999); Levin, Whitener, and
Cross (2006) and Johnson, Cullen, Sakano, and Takenouchi (1996).
In their study, Mayer and Davis (1999) explained benevolence as the extent to which an
employee believed their manager cared about their interests and acted in an altruistic
70
manner toward them (i.e. acted in a good way towards the employee). The authors
(Mayer & Davis, 1999) operationalized benevolence into five items (Table 3.5) that
collectively had Cronbach’s alpha of .87 for the second wave of the study and .89 for the
third wave. Other researchers have also used similar constructs, in their work on trust in
organizations. For example, in their research on relationship length and trust, Levin,
Whitener, and Cross (2006) measured perceived trustworthiness as a “perception of
trustworthiness in terms of benevolence” (p. 1166). The authors (Levin, Whitener, &
Cross, 2006) used three items (Table 3.5) to measure perceived trustworthiness between
employees. Collectively, the three items had a Cronbach’s alpha of .84 (Levin, Whitener,
& Cross, 2006).
Johnson, Cullen, Sakano, and Takenouchi’s (1996) work on trust, in strategic alliances
between US and Japanese firms, also recognized benevolence-based trust, by dividing
trust into two dimensions: credibility and benevolence. The authors (Johnson et al., 1996)
operationalized the benevolence dimension by using four items (Table 3.5) adapted from
the work of Ganesan (1994). Factor and reliability analyses were not presented for this
specific dimension, but the authors did get Cronbach’s alphas of .94 for the overall
measure of trust in US firms, and .92 in their Japanese partners (Johnson et al., 1996).
In this study, benevolence-based trust was measured using a nine-item scale (Table 3.5).
Items were: “This person is very concerned about my welfare”; “My needs and desires
are very important to this person”; “This person would not knowingly do anything to hurt
me”; “This person really looks out for what is important to me”; “This person will go out
of his or her way to help me”; “This person would always look out for my interests”;
“This person would go out of his or her way to make sure I am not damaged or harmed in
this relationship”; “I feel like this person cares what happens to me”; and, “I feel like this
person is on my side”. All benevolence-based trust items were measured on a five-point
scale (1 – Strongly Disagree, 5 – Strongly Agree).
71
Author(s) Factor Study Population
Authors’ Item Adapted Item (final question items in bold)
Mayer and Davis, 1999
Benevolence Top Management & Employees
Top management is very concerned about my welfare.
This person is very concerned about my welfare.
Mayer and Davis, 1999
Benevolence Top Management & Employees
My needs and desires are very important to top management.
My needs and desires are very important to this person.
Mayer and Davis, 1999
Benevolence Top Management & Employees
Top management would not knowingly do anything to hurt me.
This person would not knowingly do anything to hurt me.
Mayer and Davis, 1999
Benevolence Top Management & Employees
Top management really looks out for what is important to me.
This person really looks out for what is important to me.
Mayer and Davis, 1999
Benevolence Top Management & Employees
Top management will go out of its way to help me.
This person will go out of his or her way to help me.
Levin, Whitener, and Cross, 2006
Perceived Trustworthiness
Knowledge workers on a project
Prior to seeking information / advice from this person on this project, I assumed that he or she would always look out for my interests.
This person would always look out for my interests.
Levin, Whitener, and Cross, 2006
Perceived Trustworthiness
Knowledge workers on a project
Prior to seeking information / advice from this person on this project, I assumed that he or she would go out of his or her way to make sure I was not damaged or harmed.
This person would go out of his or her way to make sure I am not damaged or harmed in this relationship.
Levin, Whitener, and Cross, 2006
Perceived Trustworthiness
Knowledge workers on a project
Prior to seeking information / advice from this person on this project, I felt like he or she cared what happened to me.
I feel like this person cares what happens to me.
Johnson, Cullen, Sakano, and Takenouchi, 1996
Benevolence Dimension of Trust
US and Japanese strategic business partners
Our Japanese/US partner would go out of its way to make sure our firm is not damaged or harmed in this relationship.
This person would go out of his or her way to make sure I am not damaged or harmed in this relationship. *Repeated item
Johnson, Cullen, Sakano, and Takenouchi, 1996
Benevolence Dimension of Trust
US and Japanese strategic business partners
In this relationship, we feel like our Japanese/US partner cares what happens to us.
I feel like this person cares what happens to me. *Repeated item
Johnson, Cullen, Sakano, and Takenouchi, 1996
Benevolence Dimension of Trust
US and Japanese strategic business partners
Our Japanese/US partner always looks out for our interests in this alliance.
This person would always look out for my interests. *Repeated item
72
Johnson, Cullen, Sakano, and Takenouchi, 1996
Benevolence Dimension of Trust
US and Japanese strategic business partners
We feel like our Japanese/US partner is on our side.
I feel like this person is on my side.
Table 3.5 Operationalization and Measurement of the Benevolence-Based Trust Integrity-Based Trust
Based on the work of Mayer and Davis (1999) and Schoorman, Mayer, and Davis
(1996a) six items (Table 3.6) were adapted to measure perceived integrity-based trust in
a co-worker. Mayer and Davis (1999) operationalized integrity-based trust into six items
(Table 3.6), using slightly altered measures, originally developed by Schoorman, Mayer,
and Davis (1996a). The authors’ (Mayer & Davis, 1999) items intend to measure justice,
honesty, fairness, and consistency. The items are also used to gauge adherence to the
general perceived values or principles of the trustee. The six items used by Mayer and
Davis (1999) were found to have Cronbach’s alphas of .82 for the first wave of the study
and .88 for the second wave.
In this study, integrity-based trust was measured using a similar six-item scale to that of
the one used by Mayer and Davis (1999) (Table 3.5). Items were: “This person has a
strong sense of justice”; “I never have to wonder whether this person will stick to his/her
word”; “This person tries hard to be fair in dealings with others”; “This person’s actions
and behaviors are not very consistent” (reverse coded); “I like this person’s values”; and
“Sound principles seem to guide this person’s behavior”. All integrity-based trust items
were measured on a five-point scale (1 – Strongly Disagree, 5 – Strongly Agree).
73
Author(s) Factor Study Population
Authors’ Item Adapted Item (final question items in bold)
Mayer and Davis, 1999
Integrity Top Management & Employees
Top management has a strong sense of justice.
This person has a strong sense of justice.
Mayer and Davis, 1999
Integrity Top Management & Employees
I never have to wonder whether top management will stick to its word.
I never have to wonder whether this person will stick to his/her word.
Mayer and Davis, 1999
Integrity Top Management & Employees
Top management tries hard to be fair in dealings with others.
This person tries hard to be fair in dealings with others.
Mayer and Davis, 1999
Integrity Top Management & Employees
Top management’s actions and behaviors are not very consistent. (reverse coded)
This person’s actions and behaviors are not very consistent. (reverse coded)
Mayer and Davis, 1999
Integrity Top Management & Employees
I like top management’s values. I like this person’s values.
Mayer and Davis, 1999
Integrity Top Management & Employees
Sound principles seem to guide top management’s behavior.
Sound principles seem to guide this person’s behavior.
Table 3.6 Operationalization and Measurement of the Integrity-Based Trust
Propensity to Trust
Based on the work of Mayer and Davis (1999) and Schoorman, Mayer, and Davis
(1996a) eight items (Table 3.7) were adapted to measure the respondent’s propensity to
trust. Mayer and Davis (1999) operationalized propensity to trust, using the same eight
items used by Schoorman, Mayer, and Davis (1996a) in their study of trust and the
delegation of risky tasks by veterinarians to their staff (Table 3.7). The authors’ (Mayer
& Davis, 1999) items were found to have Cronbach’s alphas of .55 for the first wave of
the study and .66 for the second wave. Schoorman, Mayer, and Davis’s (1996a) earlier
work yielded a slightly higher Cronbach’s alpha of .71.
In this study, propensity to trust was measured using a similar eight-item scale (Table
3.7) to the one used by Mayer and Davis (1999) and Schoorman, Mayer, and Davis
(1996a). Items were: “One should be very cautious with strangers”; “Most experts tell the
truth about the limits of their knowledge”; “Most people can be counted on to do what
they say they will do”; “These days, you must be alert or someone is likely to take
advantage of you”; “Most salespeople are honest in describing their products”; “Most
74
repair people will not overcharge people who are ignorant of their specialty”; “Most
people answer public opinion polls honestly”; and “Most adults are competent at their
jobs”. All propensity to trust items were measured on a five-point scale (1 – Strongly
Disagree, 5 – Strongly Agree).
Author(s) Factor Study Population Authors’ Item Adapted Item (final question items in
bold)
Mayer and Davis, 1999 Schoorman, Mayer and Davis, 1996a
Propensity to Trust
Top Management & Employees Veterinarians & hospital staff
One should be very cautious with strangers.
One should be very cautious with strangers.
Mayer and Davis, 1999 Schoorman, Mayer and Davis, 1996a
Propensity to Trust
Top Management & Employees Veterinarians & hospital staff
Most experts tell the truth about the limits of their knowledge.
Most experts tell the truth about the limits of their knowledge.
Mayer and Davis, 1999 Schoorman, Mayer and Davis, 1996a
Propensity to Trust
Top Management & Employees Veterinarians & hospital staff
Most people can be counted on to do what they say they will do.
Most people can be counted on to do what they say they will do.
Mayer and Davis, 1999 Schoorman, Mayer and Davis, 1996a
Propensity to Trust
Top Management & Employees Veterinarians & hospital staff
These days, you must be alert or someone is likely to take advantage of you.
These days, you must be alert or someone is likely to take advantage of you.
Mayer and Davis, 1999 Schoorman, Mayer and Davis, 1996a
Propensity to Trust
Top Management & Employees Veterinarians & hospital staff
Most salespeople are honest in describing their products.
Most salespeople are honest in describing their products.
Mayer and Davis, 1999 Schoorman, Mayer and Davis, 1996a
Propensity to Trust
Top Management & Employees Veterinarians & hospital staff
Most repair people will not overcharge people who are ignorant of their specialty.
Most repair people will not overcharge people who are ignorant of their specialty.
Mayer and Davis, 1999 Schoorman, Mayer and Davis, 1996a
Propensity to Trust
Top Management & Employees Veterinarians & hospital staff
Most people answer public opinion polls honestly.
Most people answer public opinion polls honestly.
Mayer and Davis, 1999 Schoorman, Mayer and Davis, 1996a
Propensity to Trust
Top Management & Employees Veterinarians & hospital staff
Most adults are competent at their jobs.
Most adults are competent at their jobs.
Table 3.7 Operationalization and Measurement of the Propensity to Trust
3.2.3 Knowledge Sharing Behaviors (Dependent Variables)
As discussed in the previous chapter, for the present research, three knowledge sharing
conditions were adopted to collectively understand knowledge sharing behavior. These
75
conditions were: the knowledge source must be willing to share the knowledge they
posses; the knowledge receiver must be willing to receive and use the knowledge that is
shared; and the knowledge receiver must perceive the knowledge shared as being useful
to their individual work, the project, or the firm. Even though the present study refers to
these conditions as knowledge sharing behaviors, as previously mentioned, in a pure
sense these conditions do not represent actual employee behaviors, rather self-reported
behavioral precursors (i.e. willingness or intentions) and post-behavioral outcomes (i.e.
perceived usefulness). It is believed that this approach would be appropriate in
understanding knowledge sharing behaviors, as these types of measures have often been
used as close surrogates to indicate or measure behavior (e.g. the technology acceptance
model (TAM), interpersonal trust measures, etc.). In addition, the actual knowledge
sharing behavior could not be captured, as there was no practical way of measuring actual
employee behavior.
Based on this perspective, three variables were introduced to measure knowledge sharing
behavior in this study: willingness to share knowledge; willingness to use knowledge; and
perceived receipt of useful knowledge. The operationalization of each of these three
knowledge sharing behavior variables is discussed below.
Willingness to Share Knowledge and Willingness to Use Knowledge
As part of his doctoral dissertation on knowledge sharing and trust, Holste (2003)
operationalized knowledge sharing into four categories of measures, based on direction
and knowledge type (i.e. sharing explicit knowledge, sharing tacit knowledge, using
explicit knowledge, and using tacit knowledge). Holste (2003) created sixteen items (four
for each category; see Table 3.8), using examples of explicit and tacit knowledge34
identified in his literature review. “Each statement simply asks the respondent to indicate
his [or her] willingness to share or use a specific example of explicit or tacit knowledge
identified by experts in the literature” (Holste, 2003, p. 75). Holste (2003) calculated the
score for each of his four knowledge-sharing dependent variables by “determining the
34 A complete list of Holte’s descriptions and examples of explicit and tacit knowledge can be found in Table 3 of his work (Holste, 2003, pp. 23-25).
76
mean of his/her responses to the items associated with each dependent variable” (p. 77).
Since the sixteen items were unique to Holste’s (2003) study, he also conducted factor
analysis, to determine if he was measuring four distinct variables. Each of the factors had
clean loadings, and reliability analysis of the four variables and two combined knowledge
measures yielded high Cronbach’s alphas (Table 3.9).
In this study, willingness to share knowledge and willingness to use knowledge were each
measured using five-item scales, similar to those used in Holste’s (2003) study (Table
3.8). The one notable difference is that the eight items Holste (2003) used to measure
willingness to share and use explicit knowledge were reduced to two items, because the
remaining examples of explicit knowledge were not relevant to this study (e.g. lectures,
databases, spreadsheets, etc.). Items for willingness to share knowledge were: “I would
take the initiative to provide this individual with useful tools I have developed (e.g.
precedents, memos, client information, industry information)”; “I would allow this
individual to spend significant time observing me in order for them to better understand
and learn from my work”; “I would willingly share with this person rules of thumb, tricks
of the trade, and other insights into the work of my office and that of the organization I
have learned”; “I would willingly share my new ideas with this individual”; and “I would
willingly share with this individual the latest organizational rumors, if significant”. Items
for willingness to use knowledge were: “I would eagerly receive and use tools developed
by this person including precedents, memos, client information, and industry
information”; “I would welcome the opportunity to spend significant time observing and
collaborating with this individual in order for me to better understand and learn from
their work”; “I would welcome and use any rules of thumb, tricks of the trade, and other
insights they have learned”; “I would eagerly receive and consider any new ideas this
individual might have”; and “I would tend to believe organizational rumors shared by
this individual and would use such knowledge as appropriate”. All willingness to share
knowledge and willingness to use knowledge items were measured on a five-point scale
(1 – Strongly Disagree, 5 – Strongly Agree).
77
Author(s) Factor Authors’ Item Adapted Item (final question items in bold)
Holste (2003); adapted from Choo (2000)
Willingness to share explicit organizational knowledge
I would take the initiative to provide this individual with tools I have developed in connection with my work that I believe would be useful to him/her.
I would take the initiative to provide this individual with useful tools I have developed (e.g. precedents, memos, client information, industry information).
Holste (2003); adapted from Haldin-Herrgard, 2000
Willingness to share explicit organizational knowledge
I would take the initiative to provide this individual with lectures/presentations I have prepared that I believe would be useful to him/her.
*Item not used in the present study
Holste (2003); adapted from Choo, 2000; Clarke & Rollo, 2001; Epstein, 2000
Willingness to share explicit organizational knowledge
Assuming I had permission to do so, I would take the initiative to provide this individual with data/databases/spreadsheets I am maintaining that I believe would be useful to him/her.
*Item not used in the present study
Holste (2003); adapted from Smith, 2001; Wong & Radcliffe, 2000
Willingness to share explicit organizational knowledge
Assuming I had permission to do so, I would take the initiative to provide this individual with printed or electronic copies of documents and/or manuals I have produced that I believe would be useful to him/her.
*Item not used in the present study
Holste (2003); adapted from Choo, 2000 Haldin-Herrgard, 2000
Willingness to use explicit organizational knowledge
I would eagerly receive and use tools developed by this person, if relevant to my work.
I would eagerly receive and use tools developed by this person including precedents, memos, client information, and industry information.
Holste (2003); adapted from Haldin-Herrgard, 2000
Willingness to use explicit organizational knowledge
I would eagerly receive and use lectures/presentations prepared by this person, if relevant to my work.
*Item not used in the present study
Holste (2003); adapted from Choo, 2000; Clarke & Rollo, 2001; Epstein, 2000
Willingness to use explicit organizational knowledge
I would eagerly receive and use data/databases/spreadsheets developed by this person, if relevant to my work.
*Item not used in the present study
Holste (2003); adapted from Smith, 2001; Wong & Radcliffe, 2000
Willingness to use explicit organizational knowledge
I would eagerly receive and use printed or electronic copies of documents and/or manuals produced by this person, if relevant to my work.
*Item not used in the present study
78
Holste (2003); adapted from Choo, 2000; Clarke & Rollo, 2001; Davenport & Grover, 2001; Scott, 2000
Willingness to share tacit organizational knowledge
If requested to do so, I would allow this individual to spend significant time observing and collaborating with me in order for him/her to better understand and learn from my work.
I would allow this individual to spend significant time observing me in order for them to better understand and learn from my work.
Holste (2003); adapted from Haldin-Herrgard, 2000; Wong & Radcliffe, 2000
Willingness to share tacit organizational knowledge
I would willingly share with this person rules of thumb, tricks of the trade, and other insights into the work of my office and that of the organization I have learned.
I would willingly share with this person rules of thumb, tricks of the trade, and other insights into the work of my office and that of the organization I have learned.
Holste (2003); adapted from Epstein, 2000
Willingness to share tacit organizational knowledge
I would willingly share my new ideas with this individual.
I would willingly share my new ideas with this individual.
Holste (2003); adapted from Epstein, 2000
Willingness to share tacit organizational knowledge
I would willingly share with this individual the latest organizational rumors, if significant.
I would willingly share with this individual the latest organizational rumors, if significant.
Holste (2003); adapted from Choo, 2000; Clarke & Rollo, 2001; Davenport & Grover, 2001; Scott, 2000
Willingness to use tacit organizational knowledge
If relevant to my work, I would welcome the opportunity to spend significant time observing and collaborating with this individual in order for me to better understand and learn from his/her work.
I would welcome the opportunity to spend significant time observing and collaborating with this individual in order for me to better understand and learn from their work.
Holste (2003); adapted from Haldin-Herrgard, 2000; Wong & Radcliffe, 2000
Willingness to use tacit organizational knowledge
If relevant to my work, I would welcome and use any rules of thumb, tricks of the trade, and other insights he/she has learned.
I would welcome and use any rules of thumb, tricks of the trade, and other insights they have learned.
Holste (2003); adapted from Epstein, 2000
Willingness to use tacit organizational knowledge
I would eagerly receive and consider any new ideas this individual might have.
I would eagerly receive and consider any new ideas this individual might have.
Holste (2003); adapted from Epstein, 2000
Willingness to use tacit organizational knowledge
I would tend to believe organizational rumors shared by this individual and would use such knowledge as appropriate.
I would tend to believe organizational rumors shared by this individual and would use such knowledge as appropriate.
Table 3.8 Operationalization and Measurement of Willingness to Share and Use Knowledge
79
Holste’s Knowledge Measure Cronbach’s Alpha
Explicit Knowledge Sharing .90 Explicit Knowledge Use .94 Tacit Knowledge Sharing .85 Tacit Knowledge Use .84 Combined Knowledge Sharing .95 Combined Knowledge Use .94
Table 3.9 Reliability Results for Measures of Holste’s (2003) Knowledge Variables Perceived Receipt of Useful Knowledge
The previous measures for willingness to share knowledge and willingness to use
knowledge only consider characteristics of the exchange itself, and do not address the
effectiveness of the shared knowledge. To be considered successful, the result of the
interaction must have a positive impact on the individuals involved in the exchange, the
project, or the organization. In other words, the knowledge shared must be useful or have
utility.
In their work on weak ties, trust, and knowledge transfer, Levin and Cross (2004) adapted
a measure to capture this type of utility for project-related work, which they called
perceived receipt of useful knowledge. They operationalized this variable by creating
eight unique items (Table 3.10) adapted from four organizational knowledge sharing
research studies35. “These eight items asked to what extent the knowledge received from
each person hurt or helped key aspects of the project’s outcomes” (Levin & Cross, 2004,
p. 1482). Reliability analysis for these eight measures (receipt of useful knowledge)
produced a Cronbach’s alpha of .93 (Levin & Cross, 2004).
In this study, perceived receipt of useful knowledge was measured using a five-item scale
similar to the one used by Levin and Cross (2004) (Table 3.10). The one notable
difference is the exclusion of three items used by Levin and Cross (2004), which
specifically related to the overall project budget, overall project time, and overall project
cost. For most respondents, this type of information may be unknown. Therefore, these
three items used by Levin and Cross (2004) were reduced and adapted into one item,
35 Item 1 is adapted from Szulanski (1996); 2-6, Keller (1994); 7, Haas and Hansen (2007); 8, Hansen (1999).
80
asking respondents to reflect on the cost and time it took them to complete the part of the
project they were responsible for.
Items for perceived receipt of useful knowledge were: “The information I received from
each of the co-workers made (or is likely to make) the following contribution to: client
satisfaction with the project; this projects quality; the project teams overall performance;
the overall success of FIRM NAME; the cost and/or time it took to complete the portion
of the project I am responsible for; [and] my individual performance on the project.” All
perceived receipt of useful knowledge items were measured on a five-point scale (1 –
Very negative, 5 – Very positive).
81
Author(s) Factor Authors’ Item Adapted Item (final question items in bold)
Levin and Cross, 2004; adapted from Szulanski, 1996
Perceived Receipt of Useful Knowledge
The information/advice I received from this person made (or is likely to make) the following contribution to client satisfaction with this project.
The information I received from each of the co-workers made (or is likely to make) the following contribution to client satisfaction with the project.
Levin and Cross, 2004; adapted from Keller, 1994
Perceived Receipt of Useful Knowledge
The information/advice I received from this person made (or is likely to make) the following contribution to this project's quality.
The information I received from each of the co-workers made (or is likely to make) the following contribution to this project’s quality.
Levin and Cross, 2004; adapted from Keller, 1994
Perceived Receipt of Useful Knowledge
The information/advice I received from this person made (or is likely to make) the following contribution to this project team's overall performance.
The information I received from each of the co-workers made (or is likely to make) the following contribution to the project team’s overall performance.
Levin and Cross, 2004; adapted from Keller, 1994
Perceived Receipt of Useful Knowledge
The information/advice I received from this person made (or is likely to make) the following contribution to my organization.
The information I received from each of the co-workers made (or is likely to make) the following contribution to the overall success of FIRM NAME.
Levin and Cross, 2004; adapted from Keller, 1994
Perceived Receipt of Useful Knowledge
The information/advice I received from this person made (or is likely to make) the following contribution to this project's coming in on budget or closer to coming in on budget.
The information I received from each of the co-workers made (or is likely to make) the following contribution to the cost and/or time it took to complete the portion of the project I am responsible for.
Levin and Cross, 2004; adapted from Keller, 1994
Perceived Receipt of Useful Knowledge
The information/advice I received from this person made (or is likely to make) the following contribution to reducing costs on this project.
The information I received from each of the co-workers made (or is likely to make) the following contribution to the cost and/or time it took to complete the portion of the project I am responsible for. *Repeated item
Levin and Cross, 2004; adapted from Haas and Hansen, 2007
Perceived Receipt of Useful Knowledge
The information/advice I received from this person made (or is likely to make) the following contribution to my being able to spend less time on this project.
The information I received from each of the co-workers made (or is likely to make) the following contribution to my individual performance on the project.
Levin and Cross, 2004; adapted from Hansen, 1999
Perceived Receipt of Useful Knowledge
The information/advice I received from this person made (or is likely to make) the following contribution to shortening the time this project took.
The information I received from each of the co-workers made (or is likely to make) the following contribution to the cost and/or time it took to complete the portion of the project I am responsible for. *Repeated item
82
(1=Contributed very negatively; 2=Contributed negatively; 3= Contributed somewhat negatively; 4=Contributed neither positively nor negatively; 5=Contributed somewhat positively; 6=Contributed positively; 7=Contributed very positively)
Table 3.10 Operationalization and Measurement of Levin and Cross’ (2004) Perceived Receipt of Useful Knowledge Research Variable
3.3 Nature of Co-worker Relationship
In this study, the complete survey was divided into three sections: an individual section,
asking the respondents to answer questions about themselves and their background; a co-
worker section, asking the respondents to answer questions about a positive referent (i.e.
someone they worked best with, on a project they worked on recently); and finally
another co-worker section, asking respondents to answer questions about a negative
referent (i.e. someone they did not work well with, on a project they worked together on
recently). This approach of getting respondents to comment on both a positive and
negative referent was based on similar distinctions made in previous research studies
(McAllister, 1995; Tsui, 1984, 1986; Holste, 2003). This approach was also motivated by
a conceptual distinction in the types of relationships that occur within these settings. For
example, one of the interesting features of knowledge-intensive organizations is that their
employees rarely have free choice in deciding whom they work with, and whom they are
required to share knowledge with. To achieve project objectives, in most instances,
employees are required to share knowledge with both individuals they work well with
and those who they do not work well with.
3.4 Reliability and Validity
The validity of the study primarily relied on construct validity, since all adapted scales
and items had been previously used in empirical organizational studies that confirmed
their operationalizations through factor and/or reliability analysis. An indicator of
reliability (i.e. Cronbach’s alpha) was used for each of the original measures above and it
exceeded the minimal acceptable range of 0.65–0.70 (DeVellis, 1991, p. 85). The present
study aimed to confirm the operationalization of each of the measures, using exploratory
factor analysis and reliability analysis (i.e. Cronbach’s alpha). Any weak-loading, or
incorrect loading, item discovered using factor analysis was excluded from further
analysis. In addition, any measure that did not exceed the minimal acceptable Cronbach’s
alpha range (DeVellis, 1991) was also excluded from further analysis. This method
83
assured the reliability and validity of measures and provided construct validity for future
studies.
3.5 Data Analysis Strategy
Each of the hypotheses identified above was tested using statistical techniques. Initially,
correlation analysis was used to measure bivariate relationships between independent and
dependent variables, and then t-tests were used to test differences between the means of
variables (Tabachnick & Fidell, 2007). To measure correlation between variables, the
Pearson product-moment correlation coefficient (r) was used. Multiple regression
analysis was performed to examine relationships between dependent variables and the
focal independent variable, while controlling for all the other variables in the model. The
standardized regression coefficients, or βetas (β), represented the amount of net change
that occurred in the dependent variable(s) for an independent variable change of one
standard deviation (Bohrnstedt & Knoke, 1994, p. 519). A t-test was conducted to assess
whether the βeta coefficients were significantly different than zero. Significance was
tested for the regression coefficient, as well as for the overall equation (p < .05). Finally,
an F-test was done to “assess whether the proportion of variance of the dependent
variable that [was] accounted for by the regression on the independent variables [was]
statistically significant” (Chan, 2002, p. 117). If the results of the regression analysis
contradicted the results of the correlation analysis, or t-tests, the results of the regression
analysis were given priority and accepted, as multiple-regression analysis controlled for
all the variables in the conceptual framework of the study, and therefore represented the
more stringent analytical framework.
To test for the mediating effect of trust, hierarchical multiple regression analysis was
used, in accordance with a testing method suggested by Baron and Kenny (1986),
whereby four conditions were met. As first step, the direct effect of each social-cognitive
factor was tested against the mediating variable (overall trust) and each of the dependent
variables (knowledge sharing behaviors). To qualify for inclusion, the social-cognitive
factor needed to have a significant relationship with overall trust and at least one
knowledge sharing behavior. Next, hierarchical multiple regression equations were run,
84
using a two-step model (i.e. one with and one without the mediating variable), to
determine whether the relationship between the social-cognitive factor and knowledge
sharing behavior was reduced, when overall trust was introduced. If, after introducing
overall trust, the original path between the social-cognitive factor and knowledge sharing
behavior became insignificant (i.e. complete mediating effect), or was reduced (i.e.
partial mediating effect), it could be concluded that overall trust had a mediating effect.
3.6 Selection of Study Population
Professional service firms (PSFs) rely greatly on their employees’ knowledge for
developing solutions to difficult and complex problems (Hitt, Bierman, Shimizu, &
Kochhar, 2001; Criscuolo, Salter, & Ter Wal, 2010). Developing solutions, however,
depends on individual knowledge and on successfully sharing knowledge with co-
workers and applying it effectively (Cross & Sproull, 2004).
The legal profession is one of several situated in PSFs. Other situated professions include
accountancy, medicine, architecture, and engineering (Clegg & Bailey, 2008). The legal
profession is knowledge-based, as “in its core, [it] is about providing specialized
knowledge and services in a variety of ways to a variety of clients” (du Plessis & du Toit,
2006, p. 360). Effective knowledge sharing practices are perhaps the most important
assets of PSFs.
The firm selected for this study is one of Canada's preeminent and largest law firms, with
offices in six cities across Canada, making it an excellent example of a large PSF. The
firm has implemented an “Excellent Knowledge Management” initiative, aimed at
leveraging and sharing the intellectual assets of the firm, to better serve their clients. As
explained by one of the senior partners, the philosophy of the firm is “thought leadership,
knowledge, and collaboration [to achieve] service excellence. […] our firm aims to build,
manage, and share our specialized knowledge and broad expertise” (Evans, personal
communication, October, 2010).
85
The firm manages a large number of project teams, whose composition consists of
knowledge workers working on projects (i.e. legal matters). The nature of these projects
allow survey respondents to objectively evaluate the project’s outcomes, giving a better
sense of the resulting effects of knowledge shared, as well as a better understanding of
knowledge sharing behavior
The selection of legal professionals to participate in the study flowed naturally from the
selection of the firm, as they are the heart of the firm and are the key individuals who
provide knowledge-based services to their clients. Another important reason was because
of the nature of their work, which is primarily knowledge-intensive and, more
importantly, relies heavily on the co-workers’ knowledge sharing and use behaviors. As
such, the firm provides an ideal environment for studying knowledge sharing behavior
between knowledge workers.
3.7 Data Collection
A senior partner at the national multijurisdictional law firm was contacted to discuss the
opportunity of conducting this study. The objectives of the study were explained to senior
partners and the Director of Knowledge Management at the firm, after which the firm
agreed to participate. Before launching the survey at the study site, a pre-test was
conducted with twenty knowledge workers, who were not affiliated with the firm. The
purpose of the pre-test was to analyze the survey instrument and ensure that respondents
understood and could answer the questions. The knowledge workers made several
valuable suggestions, with respect to delivery method, survey instructions, and wording
of questions. The updated survey was then published on the web, using a private web-
based academic survey tool called Qualtrics. Once the survey was online, it was pre-
tested again with ten additional knowledge workers, who were not affiliated with the
firm, and with three senior executives working at the firm. The purpose of the second
pre-test was to evaluate the delivery mechanism, time the survey, and re-evaluate the
wording of the questions. The second pre-test group reported no issues with the delivery
mechanism and an approximate time of 15-20 minutes per survey, which was deemed
appropriate. However, based on subsequent interviews with the three senior executives
86
involved in the pre-test, some questions were further re-worded, to match the vernacular
of the firm.
After all the changes were in place, a senior partner at the firm sent a firm-wide email
(Appendix A.3), asking all ‘legal professionals’ and paralegals/law clerks in six
nationally distributed offices to participate in the survey (approximately 900 employees
in total). All the employees contacted were knowledge workers engaged in knowledge-
intensive legal project work, the nature of which required a significant reliance on others,
for both explicit and tacit forms of knowledge. The survey was active from January 26th,
2011 to February 25th, 2011 (31 days,) and respondents could participate at anytime
between those dates. In addition, the senior partner sent out two reminder emails during
the active survey period. The first email was sent two weeks after the survey was
launched and the second email was sent three days before the closing date. Respondents
who completed the survey were rewarded for participation with a $5 gift card to a
popular coffee shop.
Once the data were analyzed, the site (firm) was visited to conduct 90-120 minute un-
structured interviews with senior partners and the Director of Knowledge Management,
to gauge whether the survey captured a representative sample of the firm, and to further
understand and interpret the unique significant relationships found during analysis. In
several cases, unique relationships found between variables were due to the specific
industry or firm characteristics, which these interviews helped understand and shape. The
results of the analysis are presented in the next chapter.
87
Chapter 4: Results 4.0 Chapter Overview
This chapter begins by presenting information about the size of the study and its
respondents, describing the organization as a whole. Next, it will discuss how measures
were analyzed using factor and reliability analysis. Descriptive statistics are presented for
each variable. The chapter then focuses on the testing of the hypotheses using correlation
and regression analysis. The chapter closes with summaries of the research questions and
whether the hypotheses were upheld.
4.1 Survey Sample Size
As discussed in the previous chapter, adequate sample size (N) was determined using
power analysis. Based on Cohen and Cohen’s (1983) power of significance test analysis
table (Table F, p. 528), to attain a statistical power of .80 with a population effect size of r
= .30, the study needed to have at least 84 respondents. Tabachnick and Fidell’s (2007, p.
123) more stringent sample size heuristic (N ≥ 50 + 8(number of independent variables))
required at least 146 respondents36. The study sample size of 275 well exceeded both
heuristics.
4.2 Survey Respondents
Approximately 900 invitations were sent to all ‘legal professionals’ and paralegals/law
clerks working at one of Canada’s largest multijurisdictional law firms. Of the 900
invitations, 775 were distributed to ‘legal professionals’, of whom 735 were lawyers, 30
trademark or patent agents, 5 or 6 accountants, and 5 or 6 (GR) governmental
professional. Approximately 37% of all the legal professionals were associates and 63%
were partners. In addition, 120 invitations were sent to all ‘paralegals’ and ‘law clerks’.
No administrative staff were distributed invitations to participate.
36 146 respondents were based on the 12 independent variables used in the proceeding hypothesis testing, including demographic variables. Variables included are: Age, Age Difference, Gender, Gender Gap, Education, Educational Gap, Shared Language, Shared Vision, Tie Strength (Prior to), Tie Strength (While on), Relationship Length, and Overall Trust.
88
A total of 275 completed questionnaires were received from the population of 900
eligible knowledge professionals who had worked on a project or matter, for a response
rate of 30.6%. A breakdown of respondent by role appears in Table 4.1.
Role N % of Sample
Clerk / Paralegal 89 32.36% Articling Student 4 1.45% Associate 91 33.09% Partner 88 32.00% No Answer 3 1.09%
Table 4.1 Respondents by Role
Four respondents identified themselves as ’students‘, because at the time of the survey
they were officially designated as such by the firm. However, according to an interview
with one of the senior partners, these respondents should be considered associate lawyers,
since this was their role on projects.
4.2.1 Respondent Profile by Role and Department
Approximately 24% of all ‘legal professionals’ completed the questionnaire (i.e. 183 of
the 775 invited to participate). Compared to the actual firm distribution, partners were
underrepresented (i.e. 88 of 183 or 48.1%) and associates overrepresented (i.e. 95 of 183
or 51.9%). This should not be surprising, since partners tended to be busier and less
motivated to complete this type of questionnaire. One senior partner explained this, in an
interview, by saying that “an associate who sees an email from a partner saying fill this
out is more likely to do it than a partner who sees an email from a partner saying please
fill this out” (Evans, personal communication, August, 2011). For similar reasons, legal
clerks and paralegals were highly overrepresented in the survey, with a response
percentage of 74.2% (i.e. 89 of 120 invited to participate completed the survey). There
were no anticipated implications of the overrepresentation of law clerks/paralegals and
the underrepresentation of partners, as the intention of this study was to capture the
general knowledge sharing behavior of knowledge workers on projects, regardless of
their role within the firm. However, this distinction may be made in future research.
The professional law services of the firm were separated into two departments: business
and litigation. The actual firm distribution, at the time of the survey, was about 55% in
89
the business services department and 45% in the litigation services department. This was
similar to the distribution found in the survey respondents (i.e. business department was
57% and litigation department was 42%).
Role Business Litigation
Clerk / Paralegal 49 40 Articling Student 4 0 Associate 50 40 Partner 54 34
Table 4.2 Respondents by Role within the Firm by Department Table 4.2 shows the role within the firm by department. According to a senior partner at
the firm, clerks/paralegals and associates were accurately represented in the survey
respondents, for both departments. On the other hand, litigation partners were
underrepresented in the survey respondents. One senior partner explained this by
reasoning that litigators spend more time out of the office, on depositions and motions,
mediations, and trials. Corporate (i.e. business) lawyers tend to spend more time in the
office, which gives them more opportunities to complete the questionnaire.
4.2.2 Respondent Profile by Gender
According to one senior manager, when compared to the actual distribution in the firm,
women were overrepresented in the survey respondents (i.e. women accounted for 62%
of all respondents). In actuality, the distribution of associates was approximately 55%
men and 45% women and the distribution of partners was approximately 70% men and
30% women.
Based on the actual firm distribution, associate women were slightly overrepresented in
the survey respondents (i.e. 53% as opposed to 45%). Women partners were also
overrepresented, as 41% of all partners who completed the survey were women (as
opposed to 30%). However, the higher number of women respondents who were
paralegals or clerks could explain the generally larger percentage of women in the survey
respondents. According to a senior partner at the firm, the position of clerk/paralegal was
predominantly held by women, who outnumbered men in the same role by 10-1, both in
the firm and in the survey respondents (i.e. men ~3% / women ~30%).
90
4.2.3 Respondent Profile by Birth, Country, Citizenship, and Ethnicity
Almost 85% of those who completed the survey were born in Canada (233 of 275) with
3% from the US (8 of 275) and 3% from the UK (7 of 275). Approximately 9% (25 of
275) were born in other countries and less than 1% (2 of 275) did not answer. More than
98% (270 of 275) of respondents had Canadian citizenship (1.5% other; >.4% no
answer). Almost 87% (239 of 275) of respondents identified themselves as, at least
partially, Caucasian (Chinese = ~4%; South Asian = ~3%; Arab = ~2%; Prefer not to say
= ~2%)37
According to senior executives at the firm, the survey respondents accurately represented
the birth country, citizenship, and ethnic distributions present in those invited to
participate.
4.2.4 Respondent Profile by Education and Age
Of all the respondents that completed the survey, more than half (139 of 275) achieved
some sort of professional degree (e.g. JD/LLB, MD). In addition, 3% earned a doctoral
degree (8 of 275), 13% a Masters degree (36 of 275), 13% a 4-year degree (36 of 275)
and 17% of respondents (47 of 275) finished some college. Less than 3% only finished
high school and only one respondent did not provide their education. The age breakdown
of respondents is presented in Table 4.3. According to senior executives at the firm, the
survey respondents accurately represented the education and age distributions of those
invited to participate.
Age N %
21-30 54 19.64% 31-40 102 37.09% 41-50 68 24.73% 51-60 37 13.45% Over 61 9 3.27% No Answer 5 1.82%
Table 4.3 Respondents by Age
37 The percentages are approximate, as a person may have identified themselves with more than one ethnicity/race
91
4.3 Analyses of Measures
4.3.1 Knowledge Sharing Behavior (KSB)
Factor Analysis
As discussed in the previous chapter, this study used three sets of measures for
knowledge sharing behavior, which included willingness to share knowledge, willingness
to use knowledge, and perceived receipt of useful knowledge.
Overall willingness to share knowledge and overall willingness to use knowledge were
each measured using five survey items. Four of the five items, in each group, were
intended to measure tacit knowledge and one item measured explicit knowledge. Overall
willingness to share knowledge (WSO) was separated into willingness to share explicit
knowledge (WSE) and willingness to share tacit knowledge (WST). Similarly, overall
willingness to use knowledge (WUO) was separated into willingness to use explicit
knowledge (WUE) and willingness to use tacit knowledge (WST). With perceived receipt
of useful knowledge (PRUK), a total of seven variables were used to measure knowledge
sharing behavior in the study. To analyze the variance in these knowledge sharing
behavior items, SPSS was used to perform an exploratory factor analysis (i.e. principal
components analysis with varimax rotation). Table 4.4 presents the results of the factor
analysis for Knowledge Sharing Behavior.
Positive Referents Negative Referents
Component Component Willingness to Share Knowledge 1 2 3 1 2 3 1 I would take the initiative to provide
this individual with useful tools I have developed (e.g. precedents, memos, client information, industry information).
0.83 0.842
2 I would allow this individual to spend significant time observing me in order for them to better understand and learn from my work.
0.783 0.87
3 I would willingly share with this person rules of thumb, tricks of the trade, and other insights into the work of my office and that of the organization I have learned.
0.842 0.902
92
4 I would willingly share my new ideas with this individual.
0.66 0.844
5 I would willingly share with this individual the latest organizational rumors, if significant.
0.218 0.485
Willingness to Use Knowledge 1 2 3 1 2 3 1 I would eagerly receive and use tools
developed by this person including precedents, memos, client information and industry information.
0.702 0.75
2 I would welcome the opportunity to spend significant time observing this individual in order for me to better understand and learn from their work.
0.836 0.784
3 I would welcome and use any rules of thumb, tricks of the trade, and other insights they have learned.
0.859 0.873
4 I would eagerly receive and consider any new ideas this individual might have.
0.754 0.827
5 I would tend to believe organizational rumors shared by this individual and would use such knowledge as appropriate.
0.704 0.673
Perceived Receipt of Useful Knowledge 1 2 3 1 2 3 The information I received from each of the co-workers made (or is likely to make) the following contributions to:
1 Client satisfaction with the matter/project
0.818 0.856
2 The matter's/project's quality 0.838 0.882 3 The project team's overall
performance 0.868 0.876
4 The overall success of BLG 0.853 0.871 5 The cost and/or the time it took to
complete the portion of the matter/project I am responsible for
0.716 0.715
6 My individual performance on the matter/project
0.715 0.744
Table 4.4 Factor Analysis Results for Knowledge Sharing Behavior Based on Comrey and Lee (1992), a factor loading above 0.55 (30% overlapping
variance) was set as cutoff point for inclusion of an item in a factor. The results of the
varimax rotation (Table 4.4) suggested that, with both positive (Group 1) and negative
(Group 2) referents, factor analysis distinguished principal factors comprising measures
93
for overall willingness to share knowledge, overall willingness to use knowledge, and
perceived receipt of useful knowledge. However, within overall willingness to share
knowledge, the fifth survey item (i.e. “I would willingly share with this individual the
latest organizational rumors, if significant”) had ‘poor’ factor loadings in Group 1 and
‘fair’ factor loadings in Group 238. Since these loadings were lower than the set cutoff
point, the fifth item was excluded from all subsequent data analyses involving overall
willingness to share knowledge or willingness to share tacit knowledge.
Reliability Analysis
Reliability analysis was performed on the three combined dependent measures for
knowledge sharing behavior (i.e. overall willingness to share knowledge39, overall
willingness to use knowledge, and perceived receipt of useful knowledge). Additionally,
reliability analysis was performed for willingness to share tacit knowledge and
willingness to use tacit knowledge. As noted in Table 4.5, the values of Cronbach’s α
well exceeded the minimally acceptable range of 0.65–0.70 (DeVellis, 1991, p. 85), in
both study groups.
Group 1 Group 2 Reliability Statistics α N α N
Overall Willingness to Share (Excluding Item 5) 0.859 4 0.908 4 Willingness to Share (Tacit Only) 0.799 3 0.889 3 Overall Willingness to Use 0.877 5 0.878 5 Willingness to Use (Tacit Only) 0.857 4 0.869 4 Perceived Receipt of Useful Knowledge 0.917 6 0.917 6
Table 4.5 Reliability Results for Combined Dependent Variables Descriptive Statistics
As per the factor and reliability analysis, knowledge sharing behaviors were grouped into
three constructs: willingness to share knowledge, willingness to use knowledge, and
perceived receipt of useful knowledge. Willingness to share/use knowledge was further
divided by type of knowledge (i.e. explicit vs. tacit). Measures for overall willingness to
38 One possible explanation as to why item five did not load with the others in WSO is because the idea of sharing “organizational rumors” might have been frowned upon within the organizational culture, and thus respondents did not want to claim they would willingly share any rumors, regardless of whether they were appropriate. 39 Excluding item five
94
share knowledge and overall willingness to use knowledge were calculated by adding
explicit and tacit knowledge together. Through interviews with senior partners in the
firm, it was discovered that, in the legal setting, explicit knowledge referred to
precedents, memos, client information, and industry information. Tacit knowledge
included rules of thumb, tricks of the trade, insights, new ideas, and, in some cases,
rumors. Perceived receipt of useful knowledge was a respondent’s assessment of the
usefulness of the knowledge received from their co-workers, on the individual, group,
and firm performance outcomes. Performance outcomes related to the individual, as well
as to client satisfaction, overall project quality, overall project team performance, and the
overall firm success.
Table 4.6 presents the descriptive statistics for the knowledge sharing behavior variables.
The results suggested that individuals within the firm exhibited significantly higher
overall knowledge sharing behavior toward positive referents (i.e. those individuals with
whom respondents felt they worked best). According to these descriptive statistics,
respondents were 21% (4.3854 vs. 3.4644) more willing to share knowledge and 25%
(4.247 vs. 3.2023) more willing to use knowledge with individuals they felt they worked
best with. Interestingly, respondents perceived the knowledge they received from positive
referents to be 28% more useful than knowledge received from negative referents (4.3954
vs. 3.1869).
N Minimum Maximum Mean Std.
Deviation Overall Willingness to Share Knowledge Group 1 264 3 5 4.3854 0.51303 Group 2 260 1 5 3.4644 0.97578 Willingness to Share Explicit Knowledge Group 1 267 2 5 4.44 0.588 Group 2 262 1 5 3.58 1.1 Willingness to Share Tacit Knowledge Group 1 264 3 5 4.3674 0.52654 Group 2 262 1 5 3.43 0.9985 Overall Willingness to Use Knowledge Group 1 264 1 5 4.247 0.62049 Group 2 260 1 5 3.2023 0.90907 Willingness to Use Explicit Knowledge Group 1 268 1 5 4.36 0.653 Group 2 266 1 5 3.41 1.082
95
Willingness to Use Tacit Knowledge Group 1 267 1 5 4.2257 0.64516 Group 2 262 1 5 3.1613 0.93311 Perceived Receipt of Useful Knowledge Group 1 263 3 5 4.3954 0.49739 Group 2 263 1 5 3.1869 0.83245
Table 4.6 Descriptive Statistics for Knowledge Sharing Behavior Variables
4.3.2 Overall Trust
Factor Analysis
Based on Mayer et al.’s (1995; Mayer & Davis, 1999) construct of perceived
trustworthiness, the measure of overall trust (Otrust) was calculated by combining a
series of items designed to capture three distinct types of perceived trust: ability-based
trust (ABtrust), integrity-based trust (IBtrust)40, and benevolence-based trust (BBtrust).
To analyze the variance and test for three distinct measures of trust, SPSS was used to
perform an exploratory factor analysis (i.e. principal components analysis with varimax
rotation). Table 4.7 presents a summary of the factor analysis results for overall trust, by
type.
Positive Referents Negative Referents Component Component
Ability-Based Trust 1 2 3 1 2 3 1 I believed that this person
approached his or her job with professionalism and dedication.
0.7 0.532
2 Given his or her track record, I saw no reason to doubt this person’s competence and preparation.
0.757 0.735
3 This person is very capable of performing his/her job.
0.818 0.861
4 This person is known to be successful at the things he/she tries to do.
0.778 0.8
5 This person has much knowledge about the work that needs to be done.
0.805 0.843
6 I feel confident about this person’s skills.
0.824 0.88
40 The procedure for reverse coding involved wording the first IBtrust item, such that high values of the theoretical construct were reflected by low scores on the item, while other IBtrust items were worded such that high values of the construct were reflected by high scores on the item. This was done to encourage respondents to pay attention to the questions. It should also be noted that all trust items were first transformed, so that they were all oriented in the same direction before they were averaged.
96
7 This person has specialized capabilities that can increase performance.
0.767 0.795
8 This person is well qualified. 0.841 0.872 9 I can rely on this person not to make
my job more difficult by careless work.
0.647 0.721
Benevolence-Based Trust 1 2 3 1 2 3 1 This person is very concerned about
my welfare. 0.821 0.844
2 My needs and desires are very important to this person.
0.819 0.851
3 This person would not knowingly do anything to hurt me.
0.589 0.777
4 This person really looks out for what is important to me.
0.812 0.846
5 This person would go out of his or her way to help me.
0.759 0.848
6 This person would go out of his or her way to make sure I am not damaged or harmed in this relationship.
0.781 0.81
7 I feel like this person cares what happens to me.
0.827 0.834
8 This person would always look out for my interests.
0.85 0.879
9 I feel like this person is on my side. 0.758 0.815 Integrity-Based Trust 1 2 3 1 2 3 1 This person has a strong sense of
justice. 0.516 0.579
2 I never have to wonder whether this person will stick to his/her word.
0.661 0.573
3 This person tries hard to be fair in dealings with others.
0.618 0.643
4 This person’s actions and behaviors are not very consistent. [reverse coded]
0.391 0.635
5 I like this person’s values. 0.63 0.644 6 Sound principles seem to guide this
person’s behavior. 0.656 0.661
Table 4.7 Factor Analysis Results for Trust
The results of the varimax rotation suggested that, in both groups, factor analysis
distinguished principal factors comprising measures for ability-based trust, benevolence-
based trust, and integrity-based trust. Notably, within Group 1, the first item for
integrity-based trust (i.e. This person has a strong sense of justice.) and the first item for
ability-based trust (i.e. I believed that this person approached his or her job with
97
professionalism and dedication.) had slightly lower factor loadings than the set cutoff of
.55 (i.e. .512, .532). Also, the fourth item for integrity-based trust (i.e. This person’s
actions and behaviors are not very consistent.) had a notably lower factor loading than
the set cutoff (i.e. 0.391). However, based on the fact that each of these items had a
strong factor loading in the corresponding group, and on factor loadings found in
previous research, a decision was made not to omit the three trust items from the study.
Reliability Analysis
Reliability analysis was performed on the independent measures used to calculate overall
trust in both groups (i.e. ABtrust, BBtrust, and IBtrust). Reliability analysis was also
performed for the combined measure of Otrust, which added the previous three together.
As noted in Table 4.8, the value of Cronbach’s α well exceeded the acceptable minimum,
for these measures of trust, in both study groups. Cronbach’s α for propensity to trust did
not meet the minimum acceptable range for analysis. Therefore, all propensity to trust
items were excluded from all further analyses. This was not expected to affect the results,
since the remaining measures fully captured trust in a co-worker, regardless of the
respondent’s propensity to trust.
Group 1 Group 2 Reliability Statistics α N α N
All Trust Items (Overall Trust) 0.943 24 0.935 24 Ability-Based Trust 0.932 9 0.927 9 Benevolence-Based Trust 0.943 9 0.956 9 Integrity-Based Trust 0.783 6 0.863 6
Table 4.8 Reliability Results for Trust Variables Descriptive Statistics
Table 4.9 presents descriptive statistics for the trust variables (i.e. Otrust, ABtrust,
BBtrust, and IBtrust). As expected, the results suggested that respondents exhibited
significantly higher trust towards positive referents versus negative ones; averaging 23%
(4.4696 vs. 3.4415) higher in ABtrust, 38.5% higher in BBtrust (4.0868 vs. 2.5115), and
32% (4.1812 vs. 2.841) higher in IBtrust. As a combined measure, Otrust towards
positive referents was 31% (4.2643 vs. 2.9388), higher than Otrust towards negative
ones.
98
N Minimum Maximum Mean Std. Deviation Overall Trust Group 1 253 2.71 5 4.2643 0.47851 Group 2 248 1.17 4.5 2.9388 0.65614 Ability-Based Trust Group 1 269 2.78 5 4.4696 0.5228 Group 2 263 1.22 5 3.4415 0.86473 Benevolence-Based Trust Group 1 265 1.56 5 4.0868 0.65311 Group 2 260 1 4.67 2.5115 0.85928 Integrity-Based Trust Group 1 264 2.83 5 4.1812 0.52393 Group 2 262 1 5 2.841 0.79747
Table 4.9 Descriptive Statistics for Trust Variables
4.3.3 Shared Language and Shared Vision
Factor Analysis
The measure of shared language (SL) was calculated by combining three question items
designed by Levin, Whitener, and Cross (2006), to capture the extent to which the
knowledge sender and receiver were able to easily understand and communicate with
each other. The measure of shared vision (SV) was calculated by combining five items
(three designed by Levin, Whitener, and Cross (2006) and two by Tsai and Ghoshal
(1998)), designed to capture the extent to which the knowledge sender and receiver
shared common goals, concerns, and purpose.
To analyze the variance of the two measures, and to confirm that shared language and
shared vision measured distinctly different concepts, SPSS was used to perform an
exploratory factor analysis on the eight items (i.e. principal components analysis with
varimax rotation). Table 4.10 presents a summary of the factor analysis results for SL and
SV.
99
Positive Referents
Negative Referents
Component Component Shared Language 1 2 1 2 1 I could understand completely what
this person meant when he or she was talking.
0.844 0.845
2 I was familiar with the jargon/terminology that he or she used.
0.882 0.796
3 It felt like we could communicate on the same "wavelength”.
0.822 0.781
Shared Vision 1 2 1 2 1 I felt like this person and I were
working toward completely different goals. [reverse coded]
0.35 0.422
2 I assumed that this person and I cared about the same issues.
0.759 0.798
3 I believed that this person and I shared a commitment to a common purpose.
0.82 0.874
4 I believed that this person and I shared the same ambitions and vision.
0.862 0.861
5 I believed that this person and I shared enthusiasm about pursuing the collective goals and mission of the whole organization.
0.795 0.751
Table 4.10 Factor Analysis Results for Shared Language and Shared Vision
The results of the varimax rotation suggested that, in both groups, factor analysis
distinguished principal factors comprising measures for shared language and shared
vision. However, within shared vision, the first item (i.e. I felt like this person and I were
working toward completely different goals [reverse coded41]) had factor loadings of
lower than the set cutoff point, for both groups (i.e. .350 in Group 1 and .422 in Group
41 The first item for shared vision was reverse coding, which involved wording the item such that high values of the theoretical construct were reflected by low scores on the item, while other Shared Vision items were worded such that high values of the construct were reflected by high scores on the item. This was done to encourage respondents to pay attention to the questions. It should also be noted that all Shared Vision items were first transformed, so that they were all oriented in the same direction before they were averaged.
100
2)42. Therefore, the item was excluded from all subsequent data analysis involving shared
vision.
Reliability Analysis
Reliability analysis was performed on the combined measures for shared language and
shared vision43. As noted in Table 4.11, the value of Cronbach’s α well exceed the
minimally acceptable range of 0.65–0.70 (DeVellis, 1991, p. 85), in both study groups.
Group 1 Group 2 Reliability Statistics α N α N
Shared Language 0.882 3 0.776 3 Shared Vision 0.887 4 0.862 4
Table 4.11 Reliability Results for Combined Shared Language and Shared Vision Variables
Descriptive Statistics
Table 4.12 presents descriptive statistics for the SL and SV variables. The results
suggested that respondents had significantly higher shared language and shared vision
with positive referents. Specifically, respondents averaged 24% higher in SL (i.e. means
of 4.2388 in Group 1 vs. 3.213 in Group 2) and 24% higher in SV (i.e. means of 4.1783 in
Group 1 vs. 3.166 in Group 2).
N Minimum Maximum Mean Std. Deviation Shared Language Group 1 268 1 5 4.2388 0.57715 Group 2 266 1 5 3.213 0.82603 Shared Vision Group 1 265 2.5 5 4.1783 0.53747 Group 2 262 1 5 3.166 0.83712
Table 4.12 Descriptive Statistics for Shared Language and Shared Vision Variables
42 One possible explanation as to why item one did not load with the others for SV, was because respondents could not think of a situation where they would be asked to work towards different goals. Or they might never have been put in a situation to work towards different goals. Item one not loading might have also been a factor of the reverse coding. 43 Excluding item one
101
4.3.4 Relationship Length and Tie Strength
Factor Analysis
Based on the works of Levin, Whitener, and Cross (2004) and Dirks and Ferrin (2002),
the measure of relationship length (RL) was captured using a single survey item:
“Approximately, how long have you known each of the two co-workers you selected?”
Tie strength was measured by combining four question items influenced by the works of
Levin and Cross (2004) and Hansen (1999). These items were further transformed as a
result of conversations with Daniel Levin (Evans, personal communication, October,
2010), who suggested a rethinking of Marsden and Campbell’s (1984) work on tie
strength. Levin, Walter, and Murnighan (2011) suggested that ’emotion-based‘ (or
closeness) measures were, at the very least, an equally important indicator of tie strength
to the commonly used interaction and communication frequency measures. With this in
mind, two of the four tie strength items used in this study related to ’closeness‘, and the
other two items to interaction frequency. These four items44 were asked of both groups.
Finally, the study examined tie strength both prior to and while working on a project
together.
Tie strength items were combined and standardized following a similar method to the one
used by Levin and Cross (2004). Once the data was collected, the questions with six and
eight response choices were averaged and reduced to five, so that the four tie strength
questions had the same number of response choices. This made it possible for them to be
combined and analyzed.
Next, at the suggestion of Daniel Levin (Evans, personal communication, March, 2011),
an exploratory factor analysis of the four standardized items was done using principal
44 Tie Strength Closeness items: 1. Prior to (While) working with each of the co-workers on the matter or firm-related project you shared, how close was your working relationship? 2. My relationship with each of the co-workers I mentally selected was a very intense, strong relationship prior to (while) working on the matter or firm-related project we shared. Tie Strength Interaction / Communication Frequency items: 1. Prior to (While) working with each of the co-workers on the matter or firm-related project you shared, how often did you communicate? 2. Prior to (While) working with each of the co-workers on the matter or firm-related project you shared, to what extent did you typically interact with each of them?
102
axis factoring (in SPSS). Based on previous research (Levin & Cross, 2004; Marsden &
Campbell, 1984), it was expected that factor analysis would extract into two factors
(emotion-based and frequency-based tie strength), but in fact, as can be seen in Table
4.13, it only extracted as one component in both groups (prior to and while on the
project). Based on factor analysis, all subsequent analyses of tie strength used one
combined and standardized variable for tie strength (TSp – prior to the project) and one
variable for tie strength (TSw – while on the project).
Positive
Referents Negative Referents
Component Component Tie Strength - Prior to the project 1 1 1 Prior to working with each of the co-workers on the
matter or firm-related project you shared, how close was your working relationship?
0.808 0.826
2 Prior to working with each of the co-workers on the matter or firm-related project you shared, how often did you communicate?
0.876 0.831
3 Prior to working with each of the co-workers on the matter or firm-related project you shared, to what extent did you typically interact with each of them?
0.889 0.881
4 My relationship with each of the co-workers I mentally selected was a very intense, strong relationship prior to working on the matter or firm-related project we shared.
0.758 0.659
Tie Strength – While on the project 1 1 1 While working on the matter or firm-related project
you shared, how close was your working relationship with each of the co-workers you mentally selected?
0.755 0.804
2 While working on the matter or firm-related project you shared, how often did you communicate with the two co-workers you mentally selected?
0.463 0.628
3 While working on the matter or firm-related project you shared, to what extent did you typically interact with the two co-workers you mentally selected?
0.837 0.853
4 My relationship with each of the co-workers I mentally selected was a very intense, strong relationship while working on the matter or firm-related project we shared.
0.611 0.672
Table 4.13 Factor Analysis Results for Tie Strength (Prior to and While on the Project/Matter)
103
Reliability Analysis
Reliability analysis was performed on the combined and standardized items for TSp and
TSw in both groups. As noted in Table 4.14, the value of Cronbach’s α exceeded the
acceptable range, in both study groups.
Group 1 Group 2 Reliability Statistics α N α N
Tie Strength (Prior to the project) 0.889 4 0.752 4 Tie Strength (While on the project) 0.887 4 0.827 4
Table 4.14 Reliability Results for Combined Tie Strength Variables Descriptive Statistics
Table 4.15 presents descriptive statistics for the tie strength and relationship length
variables.
N Minimum Maximum Mean Std. Deviation Tie Strength (Prior to the project)
Group 1 269 1 5 3.2326 1.19006 Group 2 268 1 5 2.4953 1.04899 Tie strength (While on the project)
Group 1 269 1.5 5 4.0895 0.65364 Group 2 265 1 5 3.3079 0.85965 Relationship Length Group 1 272 2 17 10.17 3.89 Group 2 269 1 17 9.19 3.998
Table 4.15 Descriptive Statistics for Tie Strength and Relationship Length Variables The results suggested that respondents had significantly higher overall tie strength with
positive referents. Specifically, averaging 23% higher TSp (i.e. means of 3.2326 in Group
1 vs. 2.4953 in Group 2) and 19% higher TSw (i.e. means of 4.0895 in Group 1 vs.
3.3079 in Group 2).
Means for relationship length were relatively similar for positive (i.e. 10.17) and negative
referents (i.e. 9.19). The results were expected to be similar, as it is reasonable to assume
that respondents would know co-workers they work well with and those they do not work
well with equal amounts of time.
104
4.3.5 Homophily
Factor Analysis
In order to investigate the significance of homophily, the survey instrument was designed
so that it could capture seven types of ascribed or acquired characteristics: age, gender,
ethnicity, education, marital status, country of birth, and citizenship. Each respondent was
asked to identify their own personal characteristics for these seven categories and the
same characteristics again for both positive and negative referents.
Once the data were collected, further examination was conducted on the characteristics,
by reviewing the overall descriptive statistics for the sample. The review led to the
discovery that some of the characteristic variables were not appropriate for further
homophily analysis. Specifically, birth country, citizenship, ethnicity, and marital status
did not have adequate variability or distribution, across the sample, to warrant inclusion.
Section 4.2.3 mentioned that 85% of respondents who completed the survey were born in
Canada and more than 98% of all respondents had Canadian citizenship. With such an
overwhelming majority of Canadian respondents, there was not enough variability to
effectively analyze homophily based on country of birth or country of citizenship.
Similarly, almost 87% of respondents identified themselves as, at least partially,
Caucasian, causing a similar variability concern with homophily based on ethnicity.
Finally, of all the respondents surveyed, 58% were legally married. Due to the lack of
variability in birth country, citizenship, ethnicity, and marital status responses, a decision
was made not to continue with any further analysis of homophily for these variables.
Descriptive Statistics – Age Homophily
To analyze age homophily, the data set was first reduced to include only those cases
where respondents provided a complete set of answers for their age and the ages of the
positive and negative referents. Next, the “prefer not to say” and “I am not able to
assess” responses were eliminated from the new data set, which also reduced the
response choices of all the age items to 5 (i.e. 1 = 21-30, 2 = 31-40, 3 = 41-50, 4 = 51-60
and 5 = Over 61). Finally, two new variables were created to measure the age difference
105
between the respondent’s age group and the age groups of the two co-workers they were
asked to think about. For example, age difference in Group 1 was calculated by taking the
absolute value of the respondent’s age subtracted from the age of the person they worked
best with. The resulting factor represented the number of age categories that separated the
two. For example, an age difference of 1 would equal a one-step separation between age
categories, or approximately 10 years.
Table 4.16 presents descriptive statistics for the age and age difference variables. The
table suggests that the survey respondents were, on average, very similar in age to
positive referents (i.e. mean of 2.4 vs. 2.45 in Group 1) and only marginally younger than
negative referents (mean of 2.68 in Group 2). Further examination of the means for age
difference suggested no evidence of age homophily, since the average age gap between
respondents and referents was about one age category, or approximately 10 years (i.e.
with positive referents .984 and with negative referents 1.091).
N Minimum Maximum Mean Std. Deviation Respondent's Age 252 1 5 2.4 1.023 Age in Group 1 252 1 5 2.45 0.95 Age in Group 2 252 1 5 2.68 1.047 Age Difference in Group 1 252 0 3 0.9841 0.85605 Age Difference in Group 2 252 0 4 1.0913 0.89909
Table 4.16 Descriptive Statistics for Age and Age Difference Descriptive Statistics – Gender Homophily
Gender homophily was analyzed using a similar technique to the one used for age
homophily. First, the data set was reduced to include only those cases where respondents
provided a complete set of answers for their gender and the genders of the positive and
negative referents. Next, the “prefer not to say” responses were eliminated from the
updated data set. This also reduced the item scales to have two matching gender
categories (i.e. 1 = Male 2 = Female). Finally, two new variables were created to measure
the gender difference between the respondent and referent. Gender difference was
calculated by taking the absolute value of the respondent’s gender subtracted from the
gender of the referent co-workers. The resulting factor was a binary code suggesting
whether the two genders were the same (i.e. 0 = same, 1 = different). For example, if the
106
respondent’s gender was female (coded 2) and the positive referent was also a female
(coded 2), then the resulting gender difference factor would be 0 (ABS(2-2)), or the
same. If, on the other hand, the respondent was male (coded 1) and the positive referent
was female (coded 2), then the resulting gender difference factor would be 1 (ABS(1-2)),
or different.
Table 4.17 presents descriptive statistics for the gender and gender difference variables.
The results suggested that, on average, no one gender was considered better or worse to
work with, as the means were evenly distributed across the survey respondents and the
referents. A gender difference mean of close to zero would have suggested the
possibility of gender homophily. A gender difference mean close to one would have
suggested the opposite effect. The gender difference means in Table 4.15 suggested that
no gender homophily was present in either group.
N Minimum Maximum Mean Std. Deviation Respondent's Gender 266 1 2 1.62 0.485 Gender in Group 1 266 1 2 1.5 0.501 Gender in Group 2 266 1 2 1.51 0.501 Gender Difference in Group 1 266 0 1 0.3459 0.47655 Gender Difference in Group 2 266 0 1 0.4098 0.49272
Table 4.17 Descriptive Statistics for Gender and Gender Difference Descriptive Statistics – Educational Homophily
Educational homophily was analyzed using a similar technique to the one used to identify
age and gender homophily. First, the data set was reduced to include only those cases
where respondents provided a complete set of answers for their education and the
education of the positive and negative referents.
Next, the “I don’t know” responses were eliminated from the updated data set, which also
reduced the response choices of all the education items to six matching education
categories (i.e. 1 = high school/GED, 2 = some college/2-year college degree, 3 = 3 or 4-
year university degree, 4 = professional degree (e.g. JD/LLB, MD), 5 = Masters degree
and 6 = Doctoral degree). Finally, two new variables were created to measure the
educational gap between the respondents and each of the two co-workers they selected.
107
Educational gap was calculated by taking the absolute value of the respondent’s
educational level subtracted from the educational level of the two referents. The resulting
factors represented the educational gap that separated the two co-workers. An
educational gap of 1 would equal a one-step separation between educational levels (e.g.
some college (2) to 4-year degree (3)).
Table 4.18 presents descriptive statistics for the education and educational gap variables.
The table suggests that the survey respondents were, on average, almost identical in
educational level to both positive (i.e. mean of 3.75 vs. 3.77 in Group 1) and negative
referents (i.e. mean of 3.74 in Group 2). These means and the ones related to educational
gap suggest the possible existence of homophily based on education, in both groups. For
example, the average educational gaps found between respondents and referents, both
positive (.532) and negative (.612), was a difference of less than one educational
category.
N Minimum Maximum Mean Std. Deviation Respondent's Education 237 1 6 3.75 1.018 Education in Group 1 237 1 6 3.77 0.823 Education in Group 2 237 1 6 3.74 0.942 Educational Gap in Group 1 237 0 3 0.5316 0.77299 Educational Gap in Group 2 237 0 4 0.6118 0.87886
Table 4.18 Descriptive Statistics for Education and Educational Gap 4.4 Summary of Descriptive Statistics of Study Variables
Table 4.19 provides a summary of the descriptive statistics for the independent and
dependent variables analyzed in the study. The table presents the sample sizes,
minimums, maximums, means, and standard deviations for each of the variables
discussed in the previous sections.
N Minimum Maximum Mean Std.
Deviation Overall Willingness to Share Knowledge Group 1 264 3 5 4.3854 0.51303 Group 2 260 1 5 3.4644 0.97578 Willingness to Share Explicit Knowledge Group 1 267 2 5 4.44 0.588 Group 2 262 1 5 3.58 1.1
108
Willingness to Share Tacit Knowledge Group 1 264 3 5 4.3674 0.52654 Group 2 262 1 5 3.43 0.9985 Overall Willingness to Use Knowledge Group 1 264 1 5 4.247 0.62049 Group 2 260 1 5 3.2023 0.90907 Willingness to Use Explicit Knowledge Group 1 268 1 5 4.36 0.653 Group 2 266 1 5 3.41 1.082 Willingness to Use Tacit Knowledge Group 1 267 1 5 4.2257 0.64516 Group 2 262 1 5 3.1613 0.93311 Perceived Receipt of Useful Knowledge Group 1 263 3 5 4.3954 0.49739 Group 2 263 1 5 3.1869 0.83245 Overall Trust Group 1 253 2.71 5 4.2643 0.47851 Group 2 248 1.17 4.5 2.9388 0.65614 Ability-Based Trust Group 1 269 2.78 5 4.4696 0.5228 Group 2 263 1.22 5 3.4415 0.86473 Benevolence-Based Trust Group 1 265 1.56 5 4.0868 0.65311 Group 2 260 1 4.67 2.5115 0.85928 Integrity-Based Trust Group 1 264 2.83 5 4.1812 0.52393 Group 2 262 1 5 2.841 0.79747 Shared Language Group 1 268 1 5 4.2388 0.57715 Group 2 266 1 5 3.213 0.82603 Shared Vision Group 1 265 2.5 5 4.1783 0.53747 Group 2 262 1 5 3.166 0.83712 Tie Strength (Prior to) Group 1 269 1 5 3.2326 1.19006 Group 2 268 1 5 2.4953 1.04899 Tie strength (While on) Group 1 269 1.5 5 4.0895 0.65364 Group 2 265 1 5 3.3079 0.85965 Relationship Length Group 1 272 2 17 10.17 3.89 Group 2 269 1 17 9.19 3.998 Respondent's Age 252 1 5 2.4 1.023 Age in Group 1 252 1 5 2.45 0.95 Age in Group 2 252 1 5 2.68 1.047 Age Difference in Group 1 252 0 3 0.9841 0.85605 Age Difference in Group 2 252 0 4 1.0913 0.89909 Respondent's Gender 266 1 2 1.62 0.485 Gender in Group 1 266 1 2 1.5 0.501 Gender in Group 2 266 1 2 1.51 0.501
109
Gender Difference in Group 1 266 0 1 0.3459 0.47655 Gender Difference in Group 2 266 0 1 0.4098 0.49272 Respondent's Education 237 1 6 3.75 1.018 Education in Group 1 237 1 6 3.77 0.823 Education in Group 2 237 1 6 3.74 0.942 Educational Gap in Group 1 237 0 3 0.5316 0.77299 Educational Gap in Group 2 237 0 4 0.6118 0.87886
Table 4.19 Descriptive Statistics for Study Variables 4.5 Hypothesis Testing
In the following sections, each of the hypotheses identified in the previous chapter was
tested using statistical techniques. Correlation analysis was used to measure the bivariate
relationships between the independent and dependent variables. Then, t-tests were used to
test differences between the means of the variables. Multiple regression analysis was
used to examine the relationships between the dependent variables and the focal
independent variables, while controlling for all the other variables. And finally,
hierarchical multiple regression analysis was used to test for a mediating effect between
the independent and dependent variables.
If the results of the regression analysis contradicted the results of the correlation analysis,
or t-tests, the results of the regression analysis were given priority and accepted as MRA
controlled for all of the study’s variables in the theoretical framework, and therefore
represented the more stringent analytical approach.
4.5.1 Age, Gender, and Educational Homophily
Hypotheses 1, 3, and 5 were concerned with the direction of the relationship between
trust and age, education, and gender based homophily. Hypotheses 2, 4, and 6 were
concerned with the direction of the relationship between knowledge sharing behaviors
and age, education, and gender based homophily. For all six hypotheses, homophily
variables were expected to have positive direct effects on trust and knowledge sharing
behavior.
4.5.1.1 Independent Variable 1 – Age Homophily
Hypothesis 1 stated that “age homophily will be positively related to trust” and
hypothesis 2 stated that “age homophily will be positively related to knowledge sharing
110
behavior”. Correlation and multiple regression analysis were used to test these
hypotheses.
As an initial step in the analysis of age homophily, the demographic variable of age was
considered with each of the dependent variables for trust and knowledge sharing
behavior. As a second step, the variable of age difference was analyzed with each of the
variables for trust and knowledge sharing behavior.
Age and Trust
Correlation analysis was first used to examine the bivariate relationships between age and
each of the trust variables. As it can be seen in Table 4.20, the correlation between age
and overall trust (Otrust) was not significant, in either group. However, when age was
analyzed with each type of trust, there was a statistically significant positive correlation
between ability-based trust (ABtrust) and age (.232) in Group 2. No other correlations
were found between age and the other forms of trust, in either group. The results of the
correlation analysis suggested that as the age of the negative referent increased, so did the
respondents ABtrust in that co-worker.
Group 1: Positive
Referents Group 2: Negative
Referents Trust Dependent Variable(s) Correlation N Correlation N Overall Trust 0.058 253 0.069 246 Ability-Based Trust 0.068 269 0.232*** 261 Integrity-Based Trust 0.005 264 -0.027 260 Benevolence-Based Trust 0.052 265 -0.074 258
Table 4.20 Correlations Between Age and Trust Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
age and each of the trust variables, while controlling for the effect of the other
independent variables. MRA of trust on age, along with the remaining independent
111
variables, took the following form: Trust DV = β1*Age+ β2*Edu+ β3*Gen+ β4*R.Len+
β5*T.S.p+ β6*T.S.w+ β7*S.Lan+ β8*S.Vis+ β9*AgeDiff+ β10*EduGap+ β11*GenGap45.
The adjusted R2 was the proportion of the variance of the dependent variable(s) (i.e. trust)
that was accounted for, by the set of independent variables in the regression equation,
adjusted for the total number of variables in the equation.
Table 4.21 shows the results of the multiple regression analysis of the trust dependent
variables on age and other independent variables. The MRA was repeated for each of the
trust dependent variables (i.e. overall, ability-based, integrity-based, and benevolence-
based), which allowed the study to examine the effect of age on each type of trust46.
Trust Dependent Variable(s)
β for Age only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Trust 0.059 1.056 195 0.527 20.657 0 Ability-Based Trust 0.091 1.413 205 0.336 10.369 0 Integrity-Based Trust 0.014 0.237 203 0.489 18.560 0 Benevolence-Based Trust 0.049 .854 204 0.471 17.458 0 Group 2: Negative Referents Overall Trust .176* 2.216 194 0.182 4.906 0 Ability-Based Trust .343*** 4.428 208 0.158 4.527 0 Integrity-Based Trust -0.009 -.111 205 0.145 4.141 0 Benevolence-Based Trust -0.033 -.406 206 0.124 3.629 0
Table 4.21 Regression of Trust on Age and Other Independent Variables (*p < .05, **p < .01, ***p < .001)
The adjusted R2 was then calculated for each type of trust. Adjusted R2 for Otrust was
.527 in Group 1 and .182 in Group 2. Adjusted R2 for ABtrust was .336 in Group 1 and
.158 in Group 2, IBtrust was .489 in Group 1 and .145 in Group 2, and BBtrust was .471
in Group 1 and .124 in Group 2. For both groups, adjusted R2 for each model had p <
.001.
45 (Age, Edu = Education, Gen=Gender, R.Len = Relationship Length, T.S.p = Tie Strength (prior to), T.S.w = Tie Strength (while on), S. Lan = Shared Language, S.Vis = Shared Vision, AgeDiff = Age Difference, EduGap = Educational Gap, GenGap = Gender Gap) 46 ABtrust, IBtrust, and BBtrust are each a subset of Otrust which is a combination of the three
112
For positive referents, the standardized regression coefficient of OTrust on age was found
to be not significant. However, for negative referents, the standardized regression
coefficient of Otrust on age was .176 (p < .05). The results also showed that, for negative
referents, the standardized regression coefficient of ABtrust on age was .343 (p < .001).
No other form of trust was found to be significant, with respect to age, in Group 2 and no
significant relationships were found in Group 1, with age.
Overall, the multiple regression analysis indicated that, for the group of negative
referents, respondents had higher Otrust (consisting of ABtrust) in those co-workers who
were older. In other words, as the age of negative referents increased, so did the
respondent’s Otrust and ABtrust in those individuals.
The strength of the relationships between each type of trust and age was determined using
the square of semi-partial coefficient (sr2). In Group 2, the amount of variance in Otrust
that was uniquely explained by age was 3% and the amount of variance in ABtrust that
was uniquely explained by age was 12%. No relationships were found to be significant in
Group 1.
Age and Knowledge Sharing Behavior
Correlation analysis was first used to examine the bivariate relationships between age and
each of the knowledge sharing behavior variables. As it can be seen in Table 4.22, the
correlation between age and overall willingness to share (WSO) was found to be not
significant in Group 1, and had a correlation of -.196 (p < .01) in Group 2. After running
correlations separately on the two types of knowledge shared, statistically significant
negative correlations were found with both willingness to share explicit knowledge (-
.144, p < .01) and willingness to share tacit knowledge (-.192, p < .01), in Group 2.
Neither WSE or WST was found to be significantly correlated to age in Group 1. The
correlations between age and overall willingness to use knowledge (WUO), willingness to
use explicit knowledge (WUE), willingness to use tacit knowledge (WUT), or perceived
receipt of useful knowledge were found to be not significant in either group (Table 4.22).
The results of the correlation analysis indicated that, as the age of the negative referents
113
increased, an overall willingness to share (both explicit and tacit forms of knowledge)
with that co-worker decreased.
Group 1: Positive
Referents Group 2: Negative
Referents Knowledge Sharing Behavior Dependent Variable(s) Correlation N Correlation N Overall Willingness to Share -0.062 264 -0.196** 258 Willingness to Share (Explicit) -0.004 267 -0.144** 260 Willingness to Share (Tacit) -0.08 264 -0.192** 260 Overall Willingness to Use 0.081 264 -0.031 258 Willingness to Use (Explicit) 0.008 268 0.004 264 Willingness to Use (Tacit) 0.099 264 -0.031 260 Perceived Receipt of Useful Knowledge 0.012 263 0.054 261
Table 4.22 Correlations Between Age and Knowledge Sharing Behavior Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
age and each of the knowledge sharing behavior variables, while controlling for the effect
of the the other independent variables. MRA of knowledge sharing behavior on age,
along with the remaining independent variables, took the following form: KSB DV =
β1*Age+ β2*Edu+ β3*Gen+ β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+ β8*S.Vis+
β9*AgeDiff+ β10*EduGap+ β11*GenGap+ β12*O.Trust47.
Table 4.23 shows the results of the multiple regression analysis of the knowledge sharing
behavior dependent variables on age and other independent variables. The MRA was
repeated for each of the KSB dependent variables (i.e. WSO, WSE, WST, WUO, WUE,
WUT, PRUK), which allowed the study to individually examine the effect of age on each
of the knowledge sharing behaviors48.
47 (Age, Edu = Education, Gen=Gender, R.Len = Relationship Length, T.S.p = Tie Strength (prior to), T.S.w = Tie Strength (while on), S. Lan = Shared Language, S.Vis = Shared Vision, AgeDiff = Age Difference, EduGap = Educational Gap, GenGap = Gender Gap, Otrust = Overall Trust) 48 WSE and WST are each a subset of WSO, which is a combination of the two variables. WUE and WUT are each a subset of WUO, which is a combination of the two variables.
114
Knowledge Sharing Behavior Dependent Variable(s)
β for Age only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Willingness to Share -0.065 -1.028 189 0.424 12.540 0 Willingness to Share (Explicit) -0.053 -0.780 192 0.329 8.789 0 Willingness to Share (Tacit) -0.062 -0.971 189 0.399 11.384 0 Overall Willingness to Use .201** 2.924 188 0.316 8.185 0 Willingness to Use (Explicit) 0.019 0.276 192 0.288 7.444 0 Willingness to Use (Tacit) .234** 3.378 190 0.298 7.674 0 Perceived Receipt of Useful Knowledge -0.037 -0.591 189 0.434 13.030 0 Group 2: Negative Referents Overall Willingness to Share -.229** -2.818 189 0.203 4.996 0 Willingness to Share (Explicit) -.212** -2.592 191 0.177 4.416 0 Willingness to Share (Tacit) -.222** -2.698 189 0.186 4.574 0 Overall Willingness to Use 0.090 1.149 186 0.277 6.896 0 Willingness to Use (Explicit) 0.100 1.264 191 0.215 5.328 0 Willingness to Use (Tacit) 0.075 0.953 187 0.266 6.613 0 Perceived Receipt of Useful Knowledge 0.130 1.671 189 0.247 6.151 0 Table 4.23 Regression of Knowledge Sharing Behavior on Age and Other Independent
Variables (*p < .05, **p < .01, ***p < .001) The adjusted R2 was then calculated for each knowledge sharing behavior. Adjusted R2
for WSO was .424 in Group 1 and .203 in Group 2; WSE was .329 in Group 1 and .177 in
Group 2; WST was .399 in Group 1 and .186 in Group 2. The adjusted R2 for WUO was
.316 in Group 1 and .277 in Group 2; WUE was .288 in Group 1 and .215 in Group 2;
WUT was .298 in Group 1 and .266 in Group 2. The adjusted R2 for PRUK was .434 in
Group 1 and .247 in Group 2. For both groups, the adjusted R2 for each model had p <
.001.
For Group 1, the standardized regression coefficient of WSO, WSE, and WST or PRUK on
age was found to be not significant. However, the standardized regression coefficient of
WUO on age was .201 (p < .01). In addition, the standardized regression coefficient of
WUT on age was .234 (p < .01); WUE on age was found to be not significant. For Group
2, the standardized regression coefficient of WUO, WUE, and WUT or PRUK on age was
found to be not significant. However, the standardized regression coefficient of WSO on
age was -.229 (p < .01). In addition, the standardized regression coefficient of WSE on
age was -.212 (p < .01) and WST on age was -.222 (p < .01).
115
The strength of the relationships between each type of knowledge sharing behavior and
age was determined using the square of semi-partial coefficient (sr2). In Group 1, the
amount of variance in WUO that was uniquely explained by age was 4% and the amount
of variance in WUT that was uniquely explained by age was 6%. In Group 2, the amount
of variance in WSO that was uniquely explained by age was 5%. In the same group, the
amount of variance in WSE that was uniquely explained by age was 5% and the amount
of variance in WST that was uniquely explained by age was 5%.
Overall, the multiple regression analysis indicated that, for the group of positive
referents, respondents had a higher WUO and WUT from those co-workers who were
older. In the negative referent group, respondents had a higher WSO, WSE, and WST with
those co-workers who were younger.
Age Difference and Trust
Prior to running correlation and regression analysis on the age difference variable, the
entire data set was sorted and reduced to include only those cases where respondents
provided a complete set of answers for their age and the ages of the two co-workers they
mentally selected. Any case not containing complete age information was excluded from
further analysis.
Correlation analysis was then used to examine the bivariate relationships between age
difference and each of the trust variables. As it can be seen in Table 4.24, the correlation
between age difference and Otrust was not significant in Group 1, and .143 (p < .05) in
Group 2. However, when age difference was analyzed with each type of trust separately,
there was a statistically significant positive correlation found, in Group 2, between
ABtrust and age difference (.172, p < .01). No other correlations were found between age
difference and the other forms of trust in Group 2, and no significant correlations were
found between trust and age difference in Group 1. The results of the correlation analysis
suggested that, as the age gap between the respondent and the negative referent increased,
so did overall trust and ability-based trust in that co-worker.
116
Group 1: Positive
Referents Group 2: Negative
Referents Trust Dependent Variable(s) Correlation N Correlation N Overall Trust 0.062 234 0.143* 232 Ability-Based Trust 0.05 247 0.172** 246 Integrity-Based Trust 0.048 245 0.124 244 Benevolence-Based Trust 0.062 245 0.05 244
Table 4.24 Correlations Between Age Difference and Trust Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
age difference and each of the trust variables, while controlling for the effect of the other
independent variables. MRA of trust on age difference, along with the remaining
independent variables, took the following form: Trust DV = β1*Age+ β2*Edu+ β3*Gen+
β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+ β8*S.Vis+ β9*AgeDiff+ β10*EduGap+
β11*GenGap. The adjusted model R2 for each type of trust is discussed in Section 4.5.1.1.
Table 4.25 shows the results of the multiple regression analysis of the trust dependent
variables on age difference and other independent variables. The MRA was repeated for
each of the trust dependent variables (i.e. overall, ability-based, integrity-based, and
benevolence-based), which allowed the study to examine the effect of age difference on
each type of trust49. For both groups, the standardized regression coefficient of Otrust on
age difference was found to be not significant. In addition, no other form of trust was
found to be significant with respect to age difference in either group. Overall, the
multiple regression analysis indicated that age homophily had no significant effect on
trust, in either group.
Trust Dependent Variable(s) β for Age Diff. only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Trust 0.076 1.443 195 0.527 20.657 0 Ability-Based Trust 0.046 0.763 205 0.336 10.369 0 Integrity-Based Trust 0.096 1.776 203 0.489 18.560 0 Benevolence-Based Trust 0.068 1.246 204 0.471 17.458 0
49 ABtrust, IBtrust, and BBtrust are each a subset of Otrust which is a combination of the three
117
Group 2: Negative Referents Overall Trust 0.076 1.076 194 0.182 4.906 0 Ability-Based Trust 0.073 1.029 208 0.158 4.527 0 Integrity-Based Trust 0.099 1.389 205 0.145 4.141 0 Benevolence-Based Trust 0.014 0.195 206 0.124 3.629 0
Table 4.25 Regression of Trust on Age Difference and Other Independent Variables (*p < .05, **p < .01, ***p < .001)
Therefore, hypothesis 1 was not supported by the results of the analysis in either group.
Correlation analysis suggested that the opposite effect may be true with negative
referents, but the regression analysis showed no significant evidence of age homophily
(as it related to trust) in either group.
Age Difference and Knowledge Sharing Behavior
Correlation analysis was first used to examine the bivariate relationships between age
difference and each of the knowledge sharing behavior variables. As it can be seen in
Table 4.26, the correlation between age difference and WSO, WSE, or WST was not
significant in either group. The correlation between age difference and WUO, WUE, or
WUT was also not significant in either group. Finally, the correlation between age
difference and PRUK was not significant in Group 1, and .141 (p < .05) in Group 2. The
results of the correlation analysis suggested that, as the age gap between the respondents
and negative referents increased, so did a perception that the knowledge received from
that co-worker was useful and had a positive effect on the performance outcomes.
Group 1: Positive
Referents Group 2: Negative
Referents Knowledge Sharing Behavior Dependent Variable(s) Correlation N Correlation N Overall Willingness to Share 0.023 246 0.042 254 Willingness to Share (Explicit) 0.017 249 0.060 256 Willingness to Share (Tacit) 0.023 246 0.032 256 Overall Willingness to Use -0.001 244 0.069 254 Willingness to Use (Explicit) 0.084 248 0.042 260 Willingness to Use (Tacit) -0.030 246 0.078 256 Perceived Receipt of Useful Knowledge 0.070 243 0.141* 257
Table 4.26 Correlations Between Age Difference and Knowledge Sharing Behavior Variables (*p < .05, **p < .01, ***p < .001)
118
Multiple regression analysis (MRA) was then used to examine the relationship between
age difference and each of the knowledge sharing behavior variables, while controlling
for the effect of the other independent variables. MRA of knowledge sharing behavior on
age difference, along with the remaining independent variables, took the following form:
KSB DV = β1*Age+ β2*Edu+ β3*Gen+ β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+
β8*S.Vis+ β9*AgeDiff+ β10*EduGap+ β11*GenGap+ β12*O.Trust. The adjusted model R2
for each type of KSB was discussed in Section 4.5.1.1.
Table 4.27 shows the results of the multiple regression analysis of the knowledge sharing
behavior dependent variables on age difference and other independent variables. The
MRA was repeated for each of the KSB dependent variables (i.e. WSO, WSE, WST,
WUO, WUE, WUT, PRUK), which allowed the study to individually examine the effect
of age difference on each of the knowledge sharing behaviors.
Knowledge Sharing Behavior Dependent Variable(s)
β for Age Diff. only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Willingness to Share 0.086 1.462 189 0.424 12.540 0 Willingness to Share (Explicit) 0.062 0.974 192 0.329 8.789 0 Willingness to Share (Tacit) 0.088 1.460 189 0.399 11.384 0 Overall Willingness to Use -0.109 -1.692 188 0.316 8.185 0 Willingness to Use (Explicit) 0.032 0.494 192 0.288 7.444 0 Willingness to Use (Tacit) -.144* -2.222 190 0.298 7.674 0 Perceived Receipt of Useful Knowledge 0.069 1.174 189 0.434 13.030 0 Group 2: Negative Referents Overall Willingness to Share 0.094 1.319 189 0.203 4.996 0 Willingness to Share (Explicit) 0.101 1.409 191 0.177 4.416 0 Willingness to Share (Tacit) 0.083 1.151 189 0.186 4.574 0 Overall Willingness to Use 0.009 0.137 186 0.277 6.896 0 Willingness to Use (Explicit) 0.022 0.308 191 0.215 5.328 0 Willingness to Use (Tacit) 0.004 0.063 187 0.266 6.613 0 Perceived Receipt of Useful Knowledge 0.069 0.995 189 0.247
6.151 0
Table 4.27 Regression of Knowledge Sharing Behavior on Age Difference and Other Independent Variables (*p < .05, **p < .01, ***p < .001)
For Group 1, the standardized regression coefficients of WSO, WSE, WST, or PRUK on
age difference was found to be not significant. Further, the standardized regression
coefficient of WUO or WUE on age difference was not significant. However, in Group 1,
119
the standardized regression coefficient of WUT on age difference was -.144 (p < .05).
Therefore, in Group 1, there was a 2% variance in WUT that was uniquely explained by
age homophily.
Overall, the multiple regression analysis indicated that respondents were more willing to
use tacit knowledge from positive referents, if they were closer to their age. In other
words, a smaller gap in age difference or higher age homophily was found to be a
predictor of a WUT in Group 1. No significant age homophily and knowledge sharing
behavior relationships were found for Group 2.
Therefore, hypothesis 2 was partially supported by the results of the analysis in Group 1
and not supported in Group 2. The correlation analysis showed no significant
relationships between age homophily and willingness to share knowledge, or willingness
to use knowledge, in either group. However, the analysis did reveal a statistically
significant negative relationship between age homophily and perceived receipt of useful
knowledge, with negative referents. The regression analysis did not result in similar
findings; showing no evidence of age homophily with relation to willingness to use
knowledge, perceived receipt of useful knowledge, or overall willingness to use
knowledge. However, the regression analysis did suggest age homophily to be present
with respect to WUT from positive referents.
4.5.1.2 Independent Variable 2 – Educational Homophily
Hypothesis 3 stated that “educational homophily will be positively related to trust” and
hypothesis 4 stated that “educational homophily will be positively related to knowledge
sharing behavior”. Correlation and multiple regression analysis were used to test these
hypotheses.
As an initial step in the analysis of educational homophily, the demographic variable of
education was considered with each of the dependent variables for trust and knowledge
sharing behavior. As second step, the variable of educational gap was analyzed with each
of the variables for trust and knowledge sharing behavior.
120
Education and Trust
Correlation analysis was first used to examine the bivariate relationships between
education and each of the trust variables. As it can be seen in Table 4.28, the correlation
between education and Otrust, ABtrust, IBtrust, or BBtrust was found to be not
significant for either group. The results of the correlation analysis suggested that no
significant relationships existed between trust and education in either group.
Group 1: Positive
Referents Group 2: Negative
Referents Trust Dependent Variable(s) Correlation N Correlation N Overall Trust -0.013 252 0.046 246 Ability-Based Trust 0.030 268 -0.035 261 Integrity-Based Trust 0.018 263 0.055 260 Benevolence-Based Trust -0.023 264 0.075 258
Table 4.28 Correlations Between Education and Trust Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
education and each of the trust variables, while controlling for the effect of the other
independent variables. MRA of trust on education, along with the remaining independent
variables, took the following form: Trust DV = β1*Age+ β2*Edu+ β3*Gen+ β4*R.Len+
β5*T.S.p+ β6*T.S.w+ β7*S.Lan+ β8*S.Vis+ β9*AgeDiff+ β10*EduGap+ β11*GenGap.
The adjusted model R2 for each type of trust was discussed in Section 4.5.1.1.
Table 4.29 shows the results of the multiple regression analysis of the trust dependent
variables on education and other independent variables. The MRA was repeated for each
of the trust dependent variables (i.e. overall, ability-based, integrity-based, and
benevolence-based), which allowed the study to examine the effects of education on each
type of trust.
Trust Dependent Variable(s)
β for Education only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Trust 0.045 0.861 195 0.527 20.657 0 Ability-Based Trust 0.060 0.999 205 0.336 10.369 0 Integrity-Based Trust .122* 2.295 203 0.489 18.560 0 Benevolence-Based Trust -0.011 -0.208 204 0.471 17.458 0
121
Group 2: Negative Referents Overall Trust 0.025 0.346 194 0.182 4.906 0 Ability-Based Trust 0.018 0.261 208 0.158 4.527 0 Integrity-Based Trust 0.000 0.001 205 0.145 4.141 0 Benevolence-Based Trust 0.029 0.409 206 0.124 3.629 0
Table 4.29 Regression of Trust on Education and Other Independent Variables (*p < .05, **p < .01, ***p < .001)
For Group 1, the standardized regression coefficient of Otrust, ABtrust, or BBtrust on
education was found to be not significant. However, in the same group, the standardized
regression coefficient of IBtrust on education was .122 (p < .05). Therefore, in Group 1,
the amount of variance in IBtrust that was uniquely explained by education was 1%. No
significant education and trust relationships were found in Group 2. Overall, the multiple
regression analysis indicated that respondents had higher integrity-based trust in those
positive referent co-workers who had a higher level of education.
Education and Knowledge Sharing Behavior
Correlation analysis was first used to examine the bivariate relationships between
education and each of the knowledge sharing behavior variables. As it can be seen in
Table 4.30, the correlation between education and WSO, WSE, WST, WUO, WUE, WUT,
or PRUK was not significant in either group. The results of the correlation analysis
suggested that no significant relationships existed between knowledge sharing behavior
and education in either group.
Group 1: Positive
Referents Group 2: Negative
Referents Knowledge Sharing Behavior Dependent Variable(s) Correlation N Correlation N Overall Willingness to Share -0.111 263 0.024 258 Willingness to Share (Explicit) -0.069 266 0.027 260 Willingness to Share (Tacit) -0.118 263 0.021 260 Overall Willingness to Use -0.071 263 -0.118 258 Willingness to Use (Explicit) 0.082 267 -0.136 264 Willingness to Use (Tacit) 0.064 266 -0.102 260 Perceived Receipt of Useful Knowledge -0.029 262 0.014 261
Table 4.30 Correlations Between Education and Knowledge Sharing Behavior Variables (*p < .05, **p < .01, ***p < .001)
122
Multiple regression analysis (MRA) was then used to examine the relationships between
education and each of the knowledge sharing behavior variables, while controlling for the
effect of the other independent variables. MRA of knowledge sharing behavior on
education, along with the remaining independent variables, took the following form:
KSB DV = β1*Age+ β2*Edu+ β3*Gen+ β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+
β8*S.Vis+ β9*AgeDiff+ β10*EduGap+ β11*GenGap+ β12*O.Trust. The adjusted model R2
for each type of KSB was discussed in Section 4.5.1.1.
Table 4.31 shows the results of the multiple regression analysis of the knowledge sharing
behavior variables on education and other independent variables. The MRA was repeated
for each of the KSB dependent variables (i.e. WSO, WSE, WST, WUO, WUE, WUT,
PRUK), which allowed the study to individually examine the effect of education on each
of the knowledge sharing behaviors.
Knowledge Sharing Behavior Dependent Variable(s)
β for Education only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Willingness to Share 0.041 0.712 189 0.424 12.540 0 Willingness to Share (Explicit) 0.077 1.243 192 0.329 8.789 0 Willingness to Share (Tacit) 0.024 0.397 189 0.399 11.384 0 Overall Willingness to Use 0.016 0.246 188 0.316 8.185 0 Willingness to Use (Explicit) 0.052 0.806 192 0.288 7.444 0 Willingness to Use (Tacit) 0.010 0.151 190 0.298 7.674 0 Perceived Receipt of Useful Knowledge -0.039 -0.668 189 0.434
13.030 0
Group 2: Negative Referents Overall Willingness to Share -0.086 -1.188 189 0.203 4.996 0 Willingness to Share (Explicit) -0.064 -0.881 191 0.177 4.416 0 Willingness to Share (Tacit) -0.088 -1.203 189 0.186 4.574 0 Overall Willingness to Use -0.084 -1.224 186 0.277 6.896 0 Willingness to Use (Explicit) -0.045 -0.637 191 0.215 5.328 0 Willingness to Use (Tacit) -0.090 -1.296 187 0.266 6.613 0 Perceived Receipt of Useful Knowledge 0.003 0.048 189 0.247 6.151 0
Table 4.31 Regression of Knowledge Sharing Behavior on Education and Other Independent Variables (*p < .05, **p < .01, ***p < .001)
The standardized regression coefficient of WSO, WSE, WST, WUO, WUE, WUT, or
PRUK on education was found to be not significant, in either group. Overall, the multiple
123
regression analysis confirmed the correlation analysis, indicating that no significant
education and knowledge sharing behavior relationships existed in either group.
Educational Gap and Trust
Prior to running correlation and regression analysis on the educational gap variable, the
entire data set was sorted and reduced to include only those cases where respondents
provided a complete set of answers for their education and the education of the two co-
workers they mentally selected. Any case not containing complete educational
information was excluded from further analysis.
Correlation analysis was used to examine the bivariate relationships between educational
gap and each of the trust variables. As it can be seen in Table 4.32, the correlation
between educational gap and Otrust, ABtrust, IBtrust, or BBtrust was not significant in
either group. The results of the correlation analysis suggested that no significant
relationships existed between trust and educational homophily in either group.
Group 1: Positive
Referent Group 2: Negative
Referent Trust Dependent Variable(s) Correlation N Correlation N Overall Trust -0.031 221 0.009 216 Ability-Based Trust -0.038 233 0.007 231 Integrity-Based Trust 0.019 230 0.012 229 Benevolence-Based Trust -0.028 230 -0.03 228
Table 4.32 Correlations Between Educational Gap and Trust Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
educational gap and each of the trust variables, while controlling for the effect of the
other independent variables. MRA of trust on educational gap, along with the remaining
independent variables, took the following form: Trust DV = β1*Age+ β2*Edu+ β3*Gen+
β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+ β8*S.Vis+ β9*AgeDiff+ β10*EduGap+
β11*GenGap. The adjusted model R2 for each type of trust was discussed in Section
4.5.1.1.
124
Table 4.33 shows the results of the multiple regression analysis of the trust dependent
variables on educational gap and other independent variables. The MRA was repeated
for each of the trust dependent variables (i.e. overall, ability-based, integrity-based, and
benevolence-based), which allowed the study to examine the effect of educational gap on
each type of trust.
The standardized regression coefficient of Otrust, ABtrust, IBtrust, or BBtrust on
educational gap was found to be not significant in either group. Overall, the multiple
regression analysis confirmed the correlation analysis, indicating that no significant
relationships existed between educational homophily and trust, in either group. Since the
correlation analysis and regression analysis showed no significant evidence of
educational homophily (as it relates to trust) in either group, hypothesis 3 was not
supported by the results of the analysis in either group.
Trust Dependent Variable(s)
β for Educational Gap only t N
Model Adj. R2
Model F Model Sig.
Group 1: Positive Referents Overall Trust -0.012 -0.242 195 0.527 20.657 0 Ability-Based Trust -0.040 -0.691 205 0.336 10.369 0 Integrity-Based Trust 0.028 0.556 203 0.489 18.560 0 Benevolence-Based Trust -0.001 -0.017 204 0.471 17.458 0 Group 2: Negative Referents Overall Trust 0.043 0.649 194 0.182 4.906 0 Ability-Based Trust 0.005 0.077 208 0.158 4.527 0 Integrity-Based Trust 0.071 1.082 205 0.145 4.141 0 Benevolence-Based Trust 0.034 0.518 206 0.124 3.629 0
Table 4.33 Regression of Trust on Educational Gap and Other Independent Variables (*p < .05, **p < .01, ***p < .001)
Educational Gap and Knowledge Sharing Behavior
Correlation analysis was first used to examine the bivariate relationships between
educational gap and each of the knowledge sharing behavior variables. As it can be seen
in Table 4.34, the correlation between educational gap and WSO, WSE, WST, WUO,
WUE, WUT, or PRUK was not significant in either group. The results of the correlation
analysis suggested that no significant relationships existed between knowledge sharing
behavior and educational homophily, in either group.
125
Group 1: Positive
Referents Group 2: Negative
Referents Knowledge Sharing Behavior Dependent Variable(s) Correlation N Correlation N Overall Willingness to Share -0.031 230 -0.018 230 Willingness to Share (Explicit) -0.053 233 -0.026 232 Willingness to Share (Tacit) 0.033 230 -0.001 231 Overall Willingness to Use 0.037 228 0.043 227 Willingness to Use (Explicit) -0.031 232 0.047 233 Willingness to Use (Tacit) 0.061 231 0.029 229 Perceived Receipt of Useful Knowledge 0.017 227 0.024 231
Table 4.34 Correlations Between Educational Gap and Knowledge Sharing Behavior Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
educational gap and each of the knowledge sharing behavior variables, while controlling
for the effect of the other independent variables. MRA of knowledge sharing behavior on
educational gap, along with the remaining independent variables, took the following
form: KSB DV = β1*Age+ β2*Edu+ β3*Gen+ β4*R.Len+ β5*T.S.p+ β6*T.S.w+
β7*S.Lan+ β8*S.Vis+ β9*AgeDiff+ β10*EduGap+ β11*GenGap+ β12*O.Trust. The
adjusted model R2 for each type of KSB was discussed in Section 4.5.1.1.
Table 4.35 shows the results of the multiple regression analysis of the knowledge sharing
behavior variables on educational gap and other independent variables. The MRA was
repeated for each of the KSB dependent variables (i.e. WSO, WSE, WST, WUO, WUE,
WUT, PRUK), which allowed the study to individually examine the effect of educational
gap on each of the knowledge sharing behaviors.
The standardized regression coefficient of WSO, WSE, WST, WUO, WUE, WUT, or
PRUK on educational gap was found to be not significant in either group. Overall, the
multiple regression analysis confirmed the correlation analysis, indicating that no
significant educational gap and knowledge sharing behavior relationships existed in
either group. Since the correlation analysis and regression analysis showed no significant
evidence of educational homophily as it related to knowledge sharing behavior, in either
group, hypothesis 4 was not supported by the results of the analysis, in either group.
126
Knowledge Sharing Behavior Dependent Variable(s)
β for Educational Gap only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Willingness to Share 0.016 0.280 189 0.424 12.540 0 Willingness to Share (Explicit) -0.042 -0.706 192 0.329 8.789 0 Willingness to Share (Tacit) 0.035 0.619 189 0.399 11.384 0 Overall Willingness to Use 0.036 0.588 188 0.316 8.185 0 Willingness to Use (Explicit) -0.019 -0.303 192 0.288 7.444 0 Willingness to Use (Tacit) 0.060 0.973 190 0.298 7.674 0 Perceived Receipt of Useful Knowledge 0.011 0.203 189 0.434 13.030 0 Group 2: Negative Referents Overall Willingness to Share -0.049 -0.747 189 0.203 4.996 0 Willingness to Share (Explicit) -0.072 -1.074 191 0.177 4.416 0 Willingness to Share (Tacit) -0.039 -0.580 189 0.186 4.574 0 Overall Willingness to Use 0.055 0.852 186 0.277 6.896 0 Willingness to Use (Explicit) 0.000 -0.007 191 0.215 5.328 0 Willingness to Use (Tacit) 0.063 0.985 187 0.266 6.613 0 Perceived Receipt of Useful Knowledge 0.069 1.070 189 0.247 6.151 0 Table 4.35 Regression of Knowledge Sharing Behavior on Educational Gap and Other
Independent Variables (*p < .05, **p < .01, ***p < .001)
4.5.1.3 Independent Variable 3 – Gender Homophily
Hypothesis 5 stated that “gender homophily will be positively related to trust” and
hypothesis 6 stated that “gender homophily will be positively related to knowledge
sharing behavior”. Correlation and multiple regression analysis were used to test these
hypotheses.
As an initial step in the analysis of gender homophily, the demographic variable of
gender was considered with each of the dependent variables for trust and knowledge
sharing behavior. As second step, the variable of educational gap was analyzed with each
of the variables for trust and knowledge sharing behavior.
Gender and Trust
Correlation analysis was first used to examine the bivariate relationships between gender
and each of the trust variables. As it can be seen in Table 4.36, the correlation between
gender and Otrust, ABtrust, IBtrust, or BBtrust was found to be not significant for either
127
group. The results of the correlation analysis suggested that no significant relationships
existed between trust and education, in either group.
Group 1: Positive
Referents Group 2: Negative
Referents Trust Dependent Variable(s) Correlation N Correlation N Overall Trust 0.012 253 -0.011 247 Ability-Based Trust -0.013 268 -0.067 262 Integrity-Based Trust 0.069 264 -0.004 261 Benevolence-Based Trust -0.008 265 0.011 259
Table 4.36 Correlations Between Gender and Trust Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
gender and each of the trust variables, while controlling for the effect of the other
independent variables. MRA of trust on gender, along with the remaining independent
variables, took the following form: Trust DV = β1*Age+ β2*Edu+ β3*Gen+ β4*R.Len+
β5*T.S.p+ β6*T.S.w+ β7*S.Lan+ β8*S.Vis+ β9*AgeDiff+ β10*EduGap+ β11*GenGap.
The adjusted model R2 for each type of trust was discussed in Section 4.5.1.1.
Table 4.37 shows the results of the multiple regression analysis of the trust dependent
variables on gender and other independent variables. The MRA was repeated for each of
the trust dependent variables (i.e. overall, ability-based, integrity-based, and
benevolence-based), which allowed the study to examine the effect of gender on each
type of trust.
Trust Dependent Variable(s) β for Gender only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Trust 0.075 1.365 195 0.527 20.657 0 Ability-Based Trust 0.017 0.266 205 0.336 10.369 0 Integrity-Based Trust .183** 3.242 203 0.489 18.560 0 Benevolence-Based Trust 0.039 0.674 204 0.471 17.458 0 Group 2: Negative Referents Overall Trust -0.014 -0.187 194 0.182 4.906 0 Ability-Based Trust 0.039 0.565 208 0.158 4.527 0 Integrity-Based Trust -0.028 -0.384 205 0.145 4.141 0 Benevolence-Based Trust -0.112 -1.545 206 0.124 3.629 0
Table 4.37 Regression of Trust on Gender and Other Independent Variables (*p < .05, **p < .01, ***p < .001)
128
For Group 1, the standardized regression coefficient of Otrust, ABtrust, or BBtrust on
gender was not significant. However, in the same group, the standardized regression
coefficient of IBtrust on gender was .183 (p < .01). In Group 1, the amount of variance in
IBtrust that was uniquely explained by gender was 3%. No significant relationships
between gender and trust were found in Group 2. The results of the multiple regression
analysis indicated that respondents had higher integrity-based trust in those positive
referent co-workers that were women.
Gender and Knowledge Sharing Behavior
Correlation analysis was first used to examine the bivariate relationships between gender
and each of the knowledge sharing behavior variables. As it can be seen in Table 4.38,
the correlation between educational gap and WSO, WSE, WST, WUO, WUE, WUT, or
PRUK was not significant in either group. The results of the correlation analysis
suggested that no significant relationships existed between knowledge sharing behavior
and gender, in either group.
Group 1: Positive
Referent Group 2: Negative
Referent Knowledge Sharing Behavior Dependent Variable(s) Correlation N Correlation N Overall Willingness to Share 0.044 264 0.069 259 Willingness to Share (Explicit) 0.017 267 0.080 261 Willingness to Share (Tacit) 0.050 264 0.063 261 Overall Willingness to Use -0.034 264 0.052 259 Willingness to Use (Explicit) 0.017 268 0.061 265 Willingness to Use (Tacit) -0.043 267 0.046 261 Perceived Receipt of Useful Knowledge -0.086 263 0.045 262
Table 4.38 Correlations Between Gender and Knowledge Sharing Behavior Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
gender and each of the knowledge sharing behavior variables, while controlling for the
effect of the other independent variables. MRA of knowledge sharing behavior on
gender, along with the remaining independent variables, took the following form: KSB
DV = β1*Age+ β2*Edu+ β3*Gen+ β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+
129
β8*S.Vis+ β9*AgeDiff+ β10*EduGap+ β11*GenGap+ β12*O.Trust. The adjusted model
R2 for each type of KSB was discussed in Section 4.5.1.1.
Table 4.39 shows the results of the multiple regression analysis of the knowledge sharing
behavior variables on gender and other independent variables. The MRA was repeated
for each of the KSB dependent variables (i.e. WSO, WSE, WST, WUO, WUE, WUT,
PRUK), which allowed the study to individually examine the effect of gender on each of
the knowledge sharing behaviors.
Knowledge Sharing Behavior Dependent Variable(s)
β for Gender only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referent Overall Willingness to Share .131* 2.112 189 0.424 12.540 0 Willingness to Share (Explicit) .146* 2.198 192 0.329 8.789 0 Willingness to Share (Tacit) 0.113 1.775 189 0.399 11.384 0 Overall Willingness to Use 0.051 0.756 188 0.316 8.185 0 Willingness to Use (Explicit) .138* 2.014 192 0.288 7.444 0 Willingness to Use (Tacit) 0.033 0.485 190 0.298 7.674 0 Perceived Receipt of Useful Knowledge -0.049 -0.788 189 0.434 13.030 0 Group 2: Negative Referent Overall Willingness to Share 0.031 0.428 189 0.203 4.996 0 Willingness to Share (Explicit) 0.066 0.894 191 0.177 4.416 0 Willingness to Share (Tacit) 0.013 0.183 189 0.186 4.574 0 Overall Willingness to Use .162* 2.326 186 0.277 6.896 0 Willingness to Use (Explicit) .179* 2.495 191 0.215 5.328 0 Willingness to Use (Tacit) .140* 2.000 187 0.266 6.613 0 Perceived Receipt of Useful Knowledge 0.129 1.824 189 0.247 6.151 0
Table 4.39 Regression of Knowledge Sharing Behavior on Gender and Other Independent Variables (*p < .05, **p < .01, ***p < .001)
For Group 1, the standardized regression coefficient of WSO on gender was .131 (p <
.05) and the standardized regression coefficient of WSE on gender was .146 (p < .05). In
the same group, the standardized regression coefficient of WUE on gender was .138 (p <
.05). Regression analysis found no other significant relationships between the remaining
KSB (i.e. WST, WUO, WUT or PRUK) variables and gender in this group. In Group 1,
the amount of variance in WSO that was uniquely explained by gender was 2%. Further,
130
the amount of variance in WSE that was uniquely explained by gender was 2%. Finally,
the amount of variance in WUE that was uniquely explained by gender was 2%.
For Group 2, the standardized regression coefficient of WSO, WSE, WST, or PRUK on
gender was not significant. However, in the same group, the standardized regression
coefficient of WUO on gender was .162 (p < .05). In addition, the standardized regression
coefficient of WUE on gender was .179 (p < .05) and the standardized regression
coefficient of WUT on gender was .140 (p < .05). In Group 2, the amount of variance in
WUO that was uniquely explained by gender was 3%. Further, the amount of variance in
WUE that was uniquely explained by gender was 3%. Finally, the amount of variance in
WUT that was uniquely explained by gender was 2%.
Overall, the multiple regression analysis indicated that respondents had a higher WSO,
WSE, and WUE from positive referents who were women. In Group 2, the multiple
regression analysis indicated that respondents had a higher WUO (both tacit and explicit)
from those individuals who were women.
Gender Gap and Trust
Prior to running correlation and regression analysis on the gender gap variable, the entire
data set was sorted and reduced to include only those cases where respondents provided a
complete set of answers for their gender and the genders of two co-workers they mentally
selected. Any case not containing complete gender information was excluded from
further analysis.
Correlation analysis was used to examine the bivariate relationships between educational
gap and each of the trust variables. As it can be seen in Table 4.40, the correlation
between gender gap and Otrust was not significant in either group. However, there was a
positive correlation found in Group 1 between ABtrust and gender gap of .148 (p < .05)
and a negative correlation in Group 2 between BBtrust and gender gap of -.136 (p < .05).
No other significant correlations were found between gender gap and the other forms of
trust, in either group. The results of the correlation analysis suggested that respondents
131
had a higher ability-based trust in positive referents that were of opposite gender. With
negative referents however, respondents had a higher benevolence-based trust in the co-
workers who were of the same gender.
Group 1: Positive
Referents Group 2: Negative
Referents Trust Dependent Variable(s) Correlation N Correlation N Overall Trust 0.051 246 -0.11 244 Ability-Based Trust 0.148* 261 -0.029 259 Integrity-Based Trust 0.019 257 -0.114 258 Benevolence-Based Trust -0.008 258 -0.136* 256
Table 4.40 Correlations Between Gender Gap and Trust Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
gender gap and each of the trust variables, while controlling for the effect of the other
independent variables. MRA of trust on gender gap, along with the remaining
independent variables, took the following form: Trust DV = β1*Age+ β2*Edu+ β3*Gen+
β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+ β8*S.Vis+ β9*AgeDiff+ β10*EduGap+
β11*GenGap. The adjusted model R2 for each type of trust was discussed in Section
4.5.1.1.
Table 4.41 shows the results of the multiple regression analysis of the trust dependent
variables on gender gap and other independent variables. The MRA was repeated for
each of the trust dependent variables (i.e. overall, ability-based, integrity-based, and
benevolence-based), which allowed the study to examine the effect of gender gap on
each type of trust. The standardized regression coefficient of Otrust, ABtrust, IBtrust, or
BBtrust on gender gap was found to be not significant, in either group. Overall, the
multiple regression analysis indicated that no significant relationships existed between
gender homophily and trust, in either group.
132
Trust Dependent Variable(s)
β for Gender Gap only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Trust -.043 -0.808 195 0.527 20.657 0 Ability-Based Trust .035 0.565 205 0.336 10.369 0 Integrity-Based Trust -.038 -0.692 203 0.489 18.560 0 Benevolence-Based Trust -.093 -1.694 204 0.471 17.458 0 Group 2: Negative Referents Overall Trust -.079 -1.107 194 0.182 4.906 0 Ability-Based Trust -.003 -0.049 208 0.158 4.527 0 Integrity-Based Trust -.090 -1.257 205 0.145 4.141 0 Benevolence-Based Trust -.128 -1.804 206 0.124 3.629 0
Table 4.41 Regression of Trust on Gender Gap and Other Independent Variables (*p < .05, **p < .01, ***p < .001)
The correlation analysis suggested that gender homophily was positively related to
BBtrust in Group 2, and negatively related to ABtrust in Group 1. In both cases, however,
the regression analysis showed no significant evidence of gender homophily, as it related
to trust, for either group. Therefore, hypothesis 5 was not supported by the results of the
analysis, in either group.
Gender Gap and Knowledge Sharing Behavior
Correlation analysis was first used to examine the bivariate relationships between gender
gap and each of the knowledge sharing behavior variables. As it can be seen in Table
4.42, the correlation between gender gap and WUO was .124 (p < .05) in Group 1, and
not significant in Group 2. No other correlations were found to be significant between
gender gap and the remaining KSBs (i.e. WSO, WSE, WST, WUE, WUT, or PRUK), in
either group. The results of the correlation analysis suggested that respondents were more
willing to use knowledge from positive referents, when they were of opposite gender.
133
Group 1: Positive
Referents Group 2: Negative
Referents Knowledge Sharing Behavior Dependent Variable(s) Correlation N Correlation N Overall Willingness to Share 0.032 257 -0.099 256 Willingness to Share (Explicit) 0.031 260 -0.055 258 Willingness to Share (Tacit) 0.031 257 -0.107 258 Overall Willingness to Use 0.124* 257 -0.026 256 Willingness to Use (Explicit) 0.118 261 -0.012 262 Willingness to Use (Tacit) 0.11 260 -0.018 258 Perceived Receipt of Useful Knowledge 0.103 256 0.008 259
Table 4.42 Correlations Between Gender Gap and Knowledge Sharing Behavior Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
gender gap and each of the knowledge sharing behavior variables, while controlling for
the effect of the other independent variables. MRA of knowledge sharing behavior on
gender gap, along with the remaining independent variables, took the following form:
KSB DV = β1*Age+ β2*Edu+ β3*Gen+ β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+
β8*S.Vis+ β9*AgeDiff+ β10*EduGap+ β11*GenGap+ β12*O.Trust. The adjusted model R2
for each type of KSB was discussed in Section 4.5.1.1.
Table 4.43 shows the results of the multiple regression analysis of the knowledge sharing
behavior dependent variables on gender gap and other independent variables. The MRA
was repeated for each of the KSB dependent variables (i.e. WSO, WSE, WST, WUO,
WUE, WUT, PRUK), which allowed the study to individually examine the effect of
gender gap on each of the knowledge sharing behaviors.
For Group 1, the standardized regression coefficient of WUE on gender gap was .143 (p
< .05). The amount of variance in WUE that was uniquely explained by gender gap was
2%. For Group 2, the standardized regression coefficient of PRUK on gender gap was
.148 (p < .05). The amount of variance in PRUK that was uniquely explained by gender
gap was 2%. No other relationships were found to be significant between gender gap and
the other KSBs, in either group. Overall, the multiple regression analysis indicated that
respondents perceived the knowledge from negative referents as more useful, when they
134
were of different gender. The analyses also indicated that respondents had higher WUE
from those positive referents different from them in gender.
Knowledge Sharing Behavior Dependent Variable(s)
β for Gender Gap only T N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Willingness to Share .062 1.022 189 0.424 12.540 0 Willingness to Share (Explicit) .042 0.658 192 0.329 8.789 0 Willingness to Share (Tacit) .063 1.012 189 0.399 11.384 0 Overall Willingness to Use .084 1.271 188 0.316 8.185 0 Willingness to Use (Explicit) .143* 2.146 192 0.288 7.444 0 Willingness to Use (Tacit) 0.063 0.950 190 0.298 7.674 0 Perceived Receipt of Useful Knowledge 0.075 1.257 189 0.434 13.030 0 Group 2: Negative Referents Overall Willingness to Share .033 0.458 189 0.203 4.996 0 Willingness to Share (Explicit) .100 1.384 191 0.177 4.416 0 Willingness to Share (Tacit) .004 0.056 189 0.186 4.574 0 Overall Willingness to Use .116 1.667 186 0.277 6.896 0 Willingness to Use (Explicit) .067 0.945 191 0.215 5.328 0 Willingness to Use (Tacit) .119 1.711 187 0.266 6.613 0 Perceived Receipt of Useful Knowledge .148* 2.116 189 0.247 6.151 0
Table 4.43 Regression of Knowledge Sharing Behavior on Gender Gap and Other Independent Variables (*p < .05, **p < .01, ***p < .001)
The correlation analysis found no effect of gender homophily. Instead, it suggested a
gender heterogeneous effect might be present in relation to WUO, in Group 1. The
regression analysis also showed no significant evidence of gender homophily, as it
related to knowledge sharing behavior, in either group. In fact, regression analysis
showed similar heterogenous relationships with WUE in Group 1, and PRUK in Group 2.
Based on these results, hypothesis 6 was not supported by the analysis, in either group.
4.5.2 Independent Variable 4 – Shared Language
Hypothesis 7 stated that “shared language will be positively related to trust” and
hypothesis 8 stated that “shared language will be positively related to knowledge sharing
behavior”. Correlation and multiple regression analysis were used to test these
hypotheses.
135
Shared Language and Trust
Correlation analysis was used to examine the bivariate relationships between shared
language and each of the trust variables. As it can be seen in Table 4.44, the correlation
between shared language and Otrust was .498 (p < .001) in Group 1, and .195 (p < .01)
in Group 2. In Group 1, the correlation between shared language and ABtrust was .435;
between shared language and IBtrust was .517; and between shared language and
BBtrust was .395 (all three correlations p < .001). In Group 2, the correlation between
shared language and ABtrust was .165 (p < .01); between shared language and IBtrust
was .150 (p < .05); and between shared language and BBtrust was .156 (p < .05). The
results of the correlation analysis suggested that respondents had higher trust (of all
types) in those co-workers with whom they shared more of a common language. The
results were consistent in both groups, although they showed weaker correlations
between respondents and negative referents.
Group 1: Positive
Referents Group 2: Negative
Referents Trust Dependent Variable(s) Correlation N Correlation N Overall Trust 0.498*** 249 0.195** 258 Ability-Based Trust 0.435*** 263 0.165** 265 Integrity-Based Trust 0.517*** 259 0.150* 264 Benevolence-Based Trust 0.395*** 261 0.156* 261
Table 4.44 Correlations Between Shared Language and Trust Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
shared language and each of the trust variables, while controlling for the effect of the
other independent variables. MRA of trust on shared language, along with the remaining
independent variables, took the following form: Trust DV = β1*Age+ β2*Edu+ β3*Gen+
β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+ β8*S.Vis+ β9*AgeDiff+ β10*EduGap+
β11*GenGap. The adjusted model R2 for each type of trust was discussed in Section
4.5.1.1.
Table 4.45 shows the results of the multiple regression analysis of the trust dependent
variables on shared language and other independent variables. The MRA was repeated
for each of the trust dependent variables (i.e. overall, ability-based, integrity-based, and
136
benevolence-based), which allowed the study to examine the effect of shared language
on each type of trust.
For Group 1, the standardized regression coefficient of Otrust on shared language was
.259 (p < .001). In addition, the standardized regression coefficient of ABtrust on shared
language was .273 (p < .001); IBtrust on shared language was .330 (p < .001); and
BBtrust on shared language was .149 (p < .05). In Group 1, the amount of variance in
Otrust that was uniquely explained by shared language was 7%. The amount of variance
in ABtrust that was uniquely explained by shared language was 7%. The amount of
variance in IBtrust that was uniquely explained by shared language was 11%. Finally,
the amount of variance in BBtrust that was uniquely explained by shared language was
2%. No significant relationships between shared language and any of the trust variables
were found in Group 2.
Overall, the multiple regression analysis indicated that respondents had higher trust (of
all types) in those positive referents with whom they shared more of a common language.
The MRA found no significant relationships between trust and shared language with
negative referents.
Trust Dependent Variable(s)
β for Shared Language only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Trust .259*** 4.070 195 0.527 20.657 0 Ability-Based Trust .273*** 3.719 205 0.336 10.369 0 Integrity-Based Trust .330*** 5.052 203 0.489 18.560 0 Benevolence-Based Trust .149* 2.294 204 0.471 17.458 0 Group 2: Negative Referents Overall Trust 0.040 0.544 194 0.182 4.906 0 Ability-Based Trust 0.012 0.165 208 0.158 4.527 0 Integrity-Based Trust 0.037 0.503 205 0.145 4.141 0 Benevolence-Based Trust 0.066 0.898 206 0.124 3.629 0 Table 4.45 Regression of Trust on Shared Language and Other Independent Variables
(*p < .05, **p < .01, ***p < .001)
137
Even though correlation analysis found statistically significant relationships between trust
and shared language in both groups, the regression analysis showed only significant
effects in Group 1. Therefore, hypothesis 7 was supported by the results of the analysis in
Group 1 and not supported in Group 2.
Shared Language and Knowledge Sharing Behavior
Correlation analysis was first used to examine the bivariate relationships between shared
language and each of the knowledge sharing behavior variables. As it can be seen in
Table 4.46, the correlation between shared language and WSO was .473 in Group 1 (p <
.001) and .152 in Group 2 (p < .05). In addition, the correlation between shared language
and WSE was .434 in Group 1 (p < .001) and .161 in Group 2 (p < .01). The correlation
between shared language and WST was .454 in Group 1 (p < .001) and not significant in
Group 2. Next, the correlation between shared language and WUO was .343 in Group 1
(p < .001) and .210 in Group 2 (p < .01). The correlation between shared language and
WUE was .333 in Group 1 (p < .001) and .130 in Group 2 (p < .05). The correlation
between shared language and WUT was .322 in Group 1 (p < .001) and .191 in Group 2
(p < .01). Finally, the correlation between shared language and PRUK was .431 in Group
1 (p < .001). No significant correlation was found between shared language and PRUK
in Group 2.
The results of the correlation analysis suggested that respondents had a higher WSO,
WSE, and WST with positive referents, if the two shared more of a common language.
This was also the case between respondents and negative referents, where higher shared
language was correlated to higher WSO and WSE. The results of the correlation analysis
also suggested that respondents had a higher WUO, WUE, and WUT from co-workers, in
both groups, if they shared more of a common language. Finally, the results suggested
that respondents had a higher perception that knowledge received from positive referents
was useful, if they shared a common language with them. This same relationship was not
found with negative referents.
138
Group 1: Positive
Referents Group 2: Positive
Referents Knowledge Sharing Behavior Dependent Variable(s) Correlation N Correlation N Overall Willingness to Share 0.473*** 260 0.152* 257 Willingness to Share (Explicit) 0.434*** 263 0.161** 259 Willingness to Share (Tacit) 0.454*** 260 0.130 259 Overall Willingness to Use 0.343*** 259 0.210** 256 Willingness to Use (Explicit) 0.333*** 263 0.130* 262 Willingness to Use (Tacit) 0.322*** 262 0.191** 258 Perceived Receipt of Useful Knowledge 0.431*** 259 0.108 259
Table 4.46 Correlations Between Shared Language and Knowledge Sharing Behavior Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
shared language and each of the knowledge sharing behavior variables, while controlling
for the effect of the other independent variables. MRA of knowledge sharing behavior on
shared language, along with the remaining independent variables, took the following
form: KSB DV = β1*Age+ β2*Edu+ β3*Gen+ β4*R.Len+ β5*T.S.p+ β6*T.S.w+
β7*S.Lan+ β8*S.Vis+ β9*AgeDiff+ β10*EduGap+ β11*GenGap+ β12*O.Trust. The
adjusted model R2 for each type of KSB was discussed in Section 4.5.1.1.
Table 4.47 shows the results of the multiple regression analysis of the knowledge sharing
behavior dependent variables on shared language and other independent variables. The
MRA was repeated for each of the KSB dependent variables (i.e. WSO, WSE, WST,
WUO, WUE, WUT, PRUK), which allowed the study to individually examine the effect
of shared language on each of the knowledge sharing behaviors.
For Group 1, the standardized regression coefficient of WSO on shared language was
.320 (p < .001). In the same group, the standardized regression coefficient of WSE on
shared language was .325 (p < .001) and WST on shared language was .294 (p < .001).
In Group 1, the amount of variance in WSO that was uniquely explained by shared
language was 10%. The amount of variance in WSE that was uniquely explained by
shared language was 11%. The amount of variance in WST that was uniquely explained
by shared language was 9%.
139
For Group 2, the standardized regression coefficient of WSO on shared language was not
significant. However, the standardized regression coefficient of WSE on shared language
was.146 (p < .05) (i.e. WST on shared language was not significant). The standardized
regression coefficient of WUO, WUE, WUT, or PRUK on shared language was not
significant in either group. In Group 2, the amount of variance in WUE that was uniquely
explained by shared language was 2%.
Overall, the multiple regression analysis indicated that respondents had higher WSO,
WSE, and WST with those positive referents they shared more of a common language
with. The MRA also suggested that respondents had a higher WSE with those negative
referents they felt they shared a more of a common language with.
Knowledge Sharing Behavior Dependent Variable(s)
β for Shared Language only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Willingness to Share .320*** 4.330 189 0.424 12.540 0 Willingness to Share (Explicit) .325*** 4.074 192 0.329 8.789 0 Willingness to Share (Tacit) .294*** 3.892 189 0.399 11.384 0 Overall Willingness to Use 0.047 0.584 188 0.316 8.185 0 Willingness to Use (Explicit) 0.095 1.156 192 0.288 7.444 0 Willingness to Use (Tacit) 0.036 0.438 190 0.298 7.674 0 Perceived Receipt of Useful Knowledge 0.123 1.667 189 0.434 13.030 0 Group 2: Negative Referents Overall Willingness to Share 0.099 1.362 189 0.203 4.996 0 Willingness to Share (Explicit) .146* 1.993 191 0.177 4.416 0 Willingness to Share (Tacit) 0.073 0.999 189 0.186 4.574 0 Overall Willingness to Use -0.026 -0.371 186 0.277 6.896 0 Willingness to Use (Explicit) -0.043 -0.595 191 0.215 5.328 0 Willingness to Use (Tacit) -0.026 -0.365 187 0.266 6.613 0 Perceived Receipt of Useful Knowledge -0.039 -0.558 189 0.247 6.151 0 Table 4.47 Regression of Knowledge Sharing Behavior on Shared Language and Other
Independent Variables (*p < .05, **p < .01, ***p < .001) In Group 1, both the correlation and regression analyses suggested statistically significant
relationships between shared language and willingness to share knowledge (overall,
explicit and tacit). However, unlike the correlation analysis, the regression analysis did
140
not find significant relationships between shared language and the other KSBs, for
Group 1. In Group 2, correlation analysis found significant relationships between shared
language and all of the KSB variables. However, the regression analysis found only one
of these relationships to be statistically significant (i.e. between shared language and
WSE). Therefore, hypothesis 8 was partially supported in both groups.
4.5.3 Independent Variable 5 – Shared Vision
Hypothesis 9 stated that “shared vision will be positively related to trust” and hypothesis
10 stated that “shared vision will be positively related to knowledge sharing behavior”.
Correlation and multiple regression analysis were used to test these hypotheses.
Shared Vision and Trust
Correlation analysis was used to examine the bivariate relationships between shared
vision and each of the trust variables. As it can be seen in Table 4.48, in Group 1 the
correlation between shared vision and Otrust was .637; between shared vision and
ABtrust was .531; between shared vision and IBtrust was .635; and between shared
vision and BBtrust was .535. All four relationships in this group had p < .001. In Group 2,
the correlation between shared vision and Otrust was .465; between shared vision and
ABtrust was .292; between shared vision and IBtrust was .436; and between shared
vision and BBtrust was .360. All four relationships in this group had p < .001. The results
of the correlation analysis suggested that respondents had higher trust (of all types) in
positive and negative referents with whom they shared more of a vision.
Group 1: Positive
Referents Group 2: Negative
Referents Trust Dependent Variable(s) Correlation N Correlation N Overall Trust 0.637*** 246 0.465*** 243 Ability-Based Trust 0.531*** 260 0.292*** 257 Integrity-Based Trust 0.635*** 256 0.436*** 254 Benevolence-Based Trust 0.535*** 258 0.360*** 255
Table 4.48 Correlations Between Shared Vision and Trust Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
shared vision and each of the trust variables, while controlling for the effect of the other
141
independent variables. MRA of trust on shared vision, along with the remaining
independent variables, took the following form: Trust DV = β1*Age+ β2*Edu+ β3*Gen+
β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+ β8*S.Vis+ β9*AgeDiff+ β10*EduGap+
β11*GenGap. The adjusted model R2 for each type of trust was discussed in Section
4.5.1.1.
Table 4.49 shows the results of the multiple regression analysis of the trust dependent
variables on shared vision and other independent variables. The MRA was repeated for
each of the trust dependent variables (i.e. overall, ability-based, integrity-based, and
benevolence-based), which allowed the study to examine the effect of shared vision on
each type of trust.
For Group 1, the standardized regression coefficient of Otrust on shared vision was .374.
In the same group, the standardized regression coefficient of ABtrust on shared vision
was .362; the standardized regression coefficient of IBtrust on shared vision was .346;
and the standardized regression coefficient of BBtrust on shared vision was .286. All four
relationships in this group had p < .001. In Group 1, the amount of variance in Otrust that
was uniquely explained by shared vision was 14%. The amount of variance in ABtrust
that was uniquely explained by shared vision was 13%. The amount of variance in
IBtrust that was uniquely explained by shared vision was 12%. Finally, the amount of
variance in BBtrust that was uniquely explained by shared vision was 8%.
For Group 2, the standardized regression coefficient of Otrust on shared vision was .375
(p < .001). In the same group, the standardized regression coefficient of ABtrust on
shared vision was .277 (p < .001); the standardized regression coefficient of IBtrust on
shared vision was .322 (p < .001); and the standardized regression coefficient BBtrust on
shared vision on was .251 (p < .01). In Group 2, the amount of variance in Otrust that
was uniquely explained by shared vision was 14%. The amount of variance in ABtrust
that was uniquely explained by shared vision was 8%. The amount of variance in IBtrust
that was uniquely explained by shared vision was 10%. Finally, the amount of variance in
BBtrust that was uniquely explained by shared vision was 6%.
142
Overall, the multiple regression analysis indicated that respondents had higher trust (of
all types) in those positive and negative referents with whom they had more of a shared
vision.
Trust Dependent Variable(s)
β for Shared Vision only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Trust .374*** 5.939 195 0.527 20.657 0 Ability-Based Trust .362*** 4.871 205 0.336 10.369 0 Integrity-Based Trust .346*** 5.313 203 0.489 18.560 0 Benevolence-Based Trust .286*** 4.384 204 0.471 17.458 0 Group 2: Negative Referents Overall Trust .375*** 5.039 194 0.182 4.906 0 Ability-Based Trust .277*** 3.773 208 0.158 4.527 0 Integrity-Based Trust .322*** 4.320 205 0.145 4.141 0 Benevolence-Based Trust .251** 3.329 206 0.124 3.629 0
Table 4.49 Regression of Trust on Shared Vision and Other Independent Variables (*p < .05, **p < .01, ***p < .001)
Both the correlation and regression analyses showed statistically significant positive
relationships between shared vision and trust (i.e. overall, ability-based, integrity-based,
and benevolence-based). Therefore, hypothesis 9 was supported by the results of the
analysis, in both groups.
Shared Vision and Knowledge Sharing Behavior
Correlation analysis was first used to examine the bivariate relationships between shared
vision and each of the knowledge sharing behaviors. As it can be seen in Table 4.50, the
correlations between shared vision and WSO was .557 in Group 1 and .310 in Group 2;
between shared vision and WSE was .463 in Group 1 and .279 in Group 2; between
shared vision and WST was .553 in Group 1 and .285 in Group 2. Next, the correlations
between shared vision and WUO was .511 in Group 1 and .439 in Group 2; between
shared vision and WUE was .440 in Group 1 and .301 in Group 2; between shared vision
and WUT was .499 in Group 1 and .451 in Group 2. All the correlations between shared
vision and each of the knowledge sharing behavior variables had p < .001. The results of
the correlation analysis suggested that, in both groups, having more of a shared vision
among co-workers had a positive effect on each of the knowledge sharing behaviors.
143
Group 1: Positive
Referents Group 2: Positive
Referents Knowledge Sharing Behavior Dependent Variable(s) Correlation N Correlation N Overall Willingness to Share 0.557*** 257 0.310*** 254 Willingness to Share (Explicit) 0.463*** 260 0.279*** 256 Willingness to Share (Tacit) 0.553*** 257 0.285*** 256 Overall Willingness to Use 0.511*** 256 0.439*** 254 Willingness to Use (Explicit) 0.440*** 260 0.301*** 259 Willingness to Use (Tacit) 0.499*** 259 0.451*** 255 Perceived Receipt of Useful Knowledge 0.631*** 255 0.209*** 256
Table 4.50 Correlations Between Shared Vision and Knowledge Sharing Behavior Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
shared vision and each of the knowledge sharing behavior variables, while controlling for
the effect of the other independent variables. MRA of knowledge sharing behavior on
shared vision, along with the remaining independent variables took the following form:
KSB DV = β1*Age+ β2*Edu+ β3*Gen+ β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+
β8*S.Vis+ β9*AgeDiff+ β10*EduGap+ β11*GenGap+ β12*O.Trust. The adjusted model R2
for each type of KSB was discussed in Section 4.5.1.1.
Table 4.51 shows the results of the multiple regression analysis of the knowledge sharing
behavior dependent variables on shared vision and other independent variables. The
MRA was repeated for each of the KSB dependent variables (i.e. WSO, WSE, WST,
WUO, WUE, WUT, PRUK), which allowed the study to individually examine the effects
of shared vision on each of the knowledge sharing behaviors.
For Group 1, the standardized regression coefficient of WSO on shared vision was .240 (p
< .05); the standardized regression coefficient of WST on shared vision was .260 (p <
.001), and the standardized regression coefficient of WSE on shared vision was not
significant. For the same group, the standardized regression coefficient of WUO on
shared vision was .265 (p < .01); the standardized regression coefficient of WST on
shared vision was .282 (p < .001); and the standardized regression coefficient of WUE on
144
shared vision on was not significant. Finally, in Group 1, the standardized regression
coefficient of PRUK on shared vision was .385 (p < .001).
In Group 1, the amount of variance in WSO that was uniquely explained by shared vision
was 6%. The amount of variance in WST that was uniquely explained by shared vision
was 7%. Next, the amount of variance in WUO that was uniquely explained by shared
vision was 7%, and the amount of variance in WUT that was uniquely explained by
shared language was 8%. Finally, the amount of variance in PRUK that was uniquely
explained by shared vision was 15%.
For Group 2, the standardized regression coefficient of WSO on shared vision was .222 (p
< .01); the standardized regression coefficient of WSE on shared vision was .165 (p <
.05); and the standardized regression coefficient of WST on shared vision was .227 (p <
.01). For the same group, the standardized regression coefficient of WUO on shared
vision was .312 (p < .001); the standardized regression coefficient of WUT on shared
vision was .348 (p < .001); and the standardized regression coefficient of WUE on shared
vision was not significant. Finally, the standardized regression coefficient of PRUK on
shared vision was not significant in Group 2.
In Group 2, the amount of variance in WSO that was uniquely explained by shared vision
was 5%. The amount of variance in WSE that was uniquely explained by shared vision
was 3% and the amount of variance in WST that was uniquely explained by shared vision
was 5%. Finally, the amount of variance in WUO that was uniquely explained by shared
vision was 10%, and the amount of variance in WUT that was uniquely explained by
shared language was 12%.
Overall, the multiple regression analysis indicated that respondents had a higher WSO
and WST with those positive referents they felt they shared more of a vision with.
Respondents also had a higher WUO and WUT from positive referents they shared more
of a vision with. More shared vision between respondents and positive referents also
related to a higher perception that the knowledge received from those co-workers was
145
useful. Similarly, for negative referents, respondents had a higher WSO, WSE, and WST
with those individuals that they felt they shared more of a vision with. Higher shared
vision between respondents and negative referents also related to a higher WUO and
WUT.
Knowledge Sharing Behavior Dependent Variable(s)
β for Shared Vision only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Willingness to Share .240* 3.106 189 0.424 12.540 0 Willingness to Share (Explicit) .153 1.851 192 0.329 8.789 0 Willingness to Share (Tacit) .260*** 3.293 189 0.399 11.384 0 Overall Willingness to Use .265** 3.169 188 0.316 8.185 0 Willingness to Use (Explicit) 0.072 0.846 192 0.288 7.444 0 Willingness to Use (Tacit) .282*** 3.358 190 0.298 7.674 0 Perceived Receipt of Useful Knowledge .385*** 5.049 189 0.434 13.030 0 Group 2: Negative Referents Overall Willingness to Share .222** 2.796 189 0.203 4.996 0 Willingness to Share (Explicit) .165* 2.064 191 0.177 4.416 0 Willingness to Share (Tacit) .227** 2.825 189 0.186 4.574 0 Overall Willingness to Use .312*** 4.031 186 0.277 6.896 0 Willingness to Use (Explicit) 0.128 1.618 191 0.215 5.328 0 Willingness to Use (Tacit) .348*** 4.483 187 0.266 6.613 0 Perceived Receipt of Useful Knowledge -0.082 -1.065 189 0.247 6.151 0
Table 4.51 Regression of Knowledge Sharing Behavior on Shared Vision and Other Independent Variables (*p < .05, **p < .01, ***p < .001)
In both groups, the correlation and regression analyses suggested statistically significant
relationships between shared vision and WSO, as well as between shared vision and WST.
The relationships found using correlation analysis in both groups between shared vision
and WSE were significant, but only remained significant for negative referents after
regression analysis. In both groups, the correlation and regression analyses suggested
statistically significant relationships between shared vision and WUO, as well as between
shared vision and WUT. The relationships between shared vision and WUE were both
significant using correlation analysis, but failed to remain so, in either group, after
regression analysis. Finally, the relationships between shared vision and PRUK were
significant in both groups using correlation analysis, but only remained significant for
positive referents after regression analysis. Therefore, hypothesis 10 was partially
supported in both groups.
146
4.5.4 Independent Variable 6 – Relationship Length
Hypothesis 11 stated that “relationship length will be positively related to trust” and
hypothesis 12 stated that “relationship length will be positively related to knowledge
sharing behavior”. Correlation and multiple regression analysis were used to test these
hypotheses.
Relationship Length and Trust
Correlation analysis was used to examine the bivariate relationships between relationship
length and each of the trust variables. As it can be seen in Table 4.52, the correlation
between relationship length and Otrust was .199 in Group 1 (p < .001) and not significant
in Group 2. The correlation between relationship length and ABtrust was not significant
in Group 1 and .147 in Group 2 (p < .05). The correlation between relationship length
and IBtrust was .195 in Group 1 (p < .001) and -.143 in Group 2 (p < .05). The
correlation between relationship length and BBtrust was .194 in Group 1 (p < .01) and
not significant in Group 2. The results of the correlation analysis suggested that the
longer the respondents knew the positive referents, the more Otrust, IBtrust, and BBtrust
they had in those individuals. The results were mixed in the second group, where the
longer respondents knew the negative referents, the more ABtrust and less IBtrust they
had in them.
Group 1: Positive
Referents Group 2: Negative
Referents Trust Dependent Variable(s) Correlation N Correlation N Overall Trust 0.199*** 252 -0.028 247 Ability-Based Trust 0.115 267 0.147* 262 Integrity-Based Trust 0.195*** 263 -0.143* 261 Benevolence-Based Trust 0.194** 264 -0.116 259
Table 4.52 Correlations Between Relationship Length and Trust Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
relationship length and each of the trust variables, while controlling for the effect of the
other independent variables. MRA of trust on relationship length, along with the
remaining independent variables, took the following form: Trust DV = β1*Age+ β2*Edu+
β3*Gen+ β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+ β8*S.Vis+ β9*AgeDiff+
147
β10*EduGap+ β11*GenGap. The adjusted model R2 for each type of trust was discussed in
Section 4.5.1.1.
Table 4.53 shows the results of the multiple regression analysis of the trust dependent
variables on relationship length and other independent variables. The MRA was repeated
for each of the trust dependent variables (i.e. overall, ability-based, integrity-based, and
benevolence-based), which allowed the study to examine the effect of relationship length
on each type of trust.
In Group 1, no significant trust and relationship length relationships resulted from the
MRA. For Group 2, the standardized regression coefficient of Otrust on relationship
length was -.196 (p < .05). Further, the standardized regression coefficient of IBtrust on
relationship length was -.197 (p < .05) and the standardized regression coefficient of
BBtrust on relationship length was -.190 (p < .05) (ABtrust was not significant). In this
group, the amount of variance in Otrust that was uniquely explained by relationship
length was 4%. The amount of variance in IBtrust that was uniquely explained by
relationship length was 4%, and the amount of variance in BBtrust that was uniquely
explained by relationship length was 4%.
Overall, the multiple regression analysis indicated that respondents had higher Otrust,
IBtrust, and BBtrust in those negative referents they had a shorter relationship length
with. This may also be thought in terms of trust decreasing, as relationship length
increased between the respondent and the person they did not work well with.
148
Trust Dependent Variable(s)
β for Relationship Length only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Trust -0.010 -0.172 195 0.527 20.657 0 Ability-Based Trust 0.015 0.218 205 0.336 10.369 0 Integrity-Based Trust 0.054 0.893 203 0.489 18.560 0 Benevolence-Based Trust -0.077 -1.246 204 0.471 17.458 0 Group 2: Negative Referents Overall Trust -0.196* -2.582 194 0.182 4.906 0 Ability-Based Trust -0.031 -0.391 208 0.158 4.527 0 Integrity-Based Trust -0.197* -2.500 205 0.145 4.141 0 Benevolence-Based Trust -0.190* -2.370 206 0.124 3.629 0 Table 4.53 Regression of Trust on Relationship Length and Other Independent Variables
(*p < .05, **p < .01, ***p < .001) In Group 1, the statistically significant relationships found between relationship length
and trust, using correlation analysis, were no longer found to be significant using
regression analysis. This was also the case for ABtrust in Group 2. On the other hand,
IBtrust was found to have a statistically significant negative relationship with
relationship length in Group 2, using both correlation and regression analysis. Otrust and
BBtrust had significant positive relationships with relationship length using correlation
analysis, but were found to have a significant negative relationship using MRA.
Therefore, hypothesis 11 was not supported by the results of the analysis in either group.
Interestingly, in Group 2 the MRA suggested partial support of the opposite effect.
Relationship Length and Knowledge Sharing Behavior
Correlation analysis was first used to examine the bivariate relationships between
relationship length and each of the knowledge sharing behaviors. As it can be seen in
Table 4.54, the correlation between relationship length and WSO was .164 (p < .01) in
Group 1 and not significant in Group 2. The correlation between relationship length and
WSE was .176 (p < .01) in Group 1 and not significant in Group 2. The correlation
between relationship length and WST was not significant in either group. No significant
correlations were found between relationship length and WUO, WUE, or WUT in either
group. Finally, the correlation between relationship length and PRUK was .140 (p < .05)
in Group 1 and not significant in Group 2.
149
The results of the correlation analysis suggested that the longer the respondents knew the
positive referents, the more willing they were to share knowledge (i.e. overall, tacit and
explicit) with them. The results also suggested that the longer the respondents knew the
positive referents, the more they perceived the knowledge received from them to be
useful. No significant relationships were found between relationship length and
knowledge sharing behaviors, with negative referents.
Group 1: Positive
Referents Group 2: Negative
Referents Knowledge Sharing Behavior Dependent Variable(s) Correlation N Correlation N Overall Willingness to Share 0.164** 264 -0.056 260 Willingness to Share (Explicit) 0.176** 267 -0.004 262 Willingness to Share (Tacit) 0.146 264 -0.060 262 Overall Willingness to Use 0.075 263 -0.013 259 Willingness to Use (Explicit) 0.099 267 0.024 265 Willingness to Use (Tacit) 0.071 266 -0.017 261 Perceived Receipt of Useful Knowledge 0.140* 262 -0.073 262
Table 4.54 Correlations Between Relationship Length and Knowledge Sharing Behavior Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
relationship length and each of the knowledge sharing behavior variables, while
controlling for the effect of the other independent variables. MRA of knowledge sharing
behavior on relationship length, along with the remaining independent variables, took the
following form: KSB DV = β1*Age+ β2*Edu+ β3*Gen+ β4*R.Len+ β5*T.S.p+ β6*T.S.w+
β7*S.Lan+ β8*S.Vis+ β9*AgeDiff+ β10*EduGap+ β11*GenGap+ β12*O.Trust. The
adjusted model R2 for each type of KSB was discussed in Section 4.5.1.1.
Table 4.55 shows the results of the multiple regression analysis of the knowledge sharing
behavior dependent variables on relationship length and other independent variables. The
MRA was repeated for each of the KSB dependent variables (i.e. WSO, WSE, WST,
WUO, WUE, WUT, PRUK), which allowed the study to individually examine the effect
of relationship length on each of the knowledge sharing behaviors.
150
The standardized regression coefficient of each of the KSBs on relationship length was
found to be not significant, in either group. Overall, the multiple regression analysis
indicated that, for both groups, there were no significant relationships between KSB and
relationship length.
Knowledge Sharing Behavior Dependent Variable(s)
β for Relationship Length only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Willingness to Share 0.009 0.131 189 0.424 12.540 0 Willingness to Share (Explicit) 0.054 0.757 192 0.329 8.789 0 Willingness to Share (Tacit) -0.012 -0.170 189 0.399 11.384 0 Overall Willingness to Use -0.034 -0.468 188 0.316 8.185 0 Willingness to Use (Explicit) 0.070 0.938 192 0.288 7.444 0 Willingness to Use (Tacit) -0.053 -0.721 190 0.298 7.674 0 Perceived Receipt of Useful Knowledge 0.025 0.376 189 0.434 13.030 0 Group 2: Negative Referents Overall Willingness to Share 0.046 0.572 189 0.203 4.996 0 Willingness to Share (Explicit) 0.092 1.129 191 0.177 4.416 0 Willingness to Share (Tacit) 0.024 0.296 189 0.186 4.574 0 Overall Willingness to Use -0.066 -0.838 186 0.277 6.896 0 Willingness to Use (Explicit) -0.055 -0.685 191 0.215 5.328 0 Willingness to Use (Tacit) -0.063 -0.798 187 0.266 6.613 0 Perceived Receipt of Useful Knowledge -0.117 -1.484 189 0.247 6.151 0
Table 4.55 Regression of Knowledge Sharing Behavior on Relationship Length and Other Independent Variables (*p < .05, **p < .01, ***p < .001)
In Group 1, the statistically significant relationships, found using correlation analysis,
between KSB and relationship length were no longer significant using regression
analysis (i.e. WSO, WSE, PRUK). Neither correlation nor regression analysis found
significant relationships between relationship length and knowledge sharing behaviors, in
the second group. Therefore, hypothesis 12 was not supported by the results of the
analysis in either group.
4.5.5 Independent Variable 7 – Tie Strength
Hypothesis 13 stated that “tie strength will be positively related to trust” and hypothesis
14 stated that “tie strength will be positively related to knowledge sharing behavior”.
Correlation and multiple regression analysis were used to test these hypotheses. Tie
151
strength was measured twice for each group, once examining the tie strength between the
respondent and the co-worker prior to the project or matter (i.e. TSp), and once more for
the tie strength between the two co-workers while on the project or matter (i.e. TSw).
Tie Strength and Trust
Correlation analysis was used to examine the bivariate relationships between tie strength
and each of the trust variables. As it can be seen in Table 4.56, the correlation between
TSp and Otrust was .283 in Group 1 (p < .001), and not significant in Group 2. The
correlation between TSw and Otrust was .502 in Group 1 (p < .001) and .148 in Group 2
(p < .05). After controlling for the type of trust, the correlation between TSp and ABtrust
was not significant, in either group. The correlation between TSw and ABtrust was .297 in
Group 1 (p < .001) and not significant in Group 2. The correlation between TSp and
IBtrust was .200 in Group 1 (p < .001) and not significant in Group 2. The correlation
between TSw and IBtrust was .423 in Group 1 (p < .001) and not significant in Group 2.
The correlation between TSp and BBtrust was .345 in Group 1 (p < .001) and .138 in
Group 2 (p < .05). Finally, the correlation between TSw and BBtrust was .533 in Group 1
(p < .001) and .165 in Group 2 (p < .01).
The results of the correlation analysis suggested that respondents had higher trust (Otrust,
IBtrust, and BBtrust) in those positive referents with whom they had higher TSp. In
addition, higher TSp between respondents and negative referents resulted in higher
BBtrust in those co-workers. The results of the correlation analysis also suggested that
respondents had higher trust (of all types) in those positive referents that they had more
TSw. For negative referents, higher TSw was found to positively influence Otrust and
BBtrust in that co-worker.
Group 1: Positive
Referents Group 2: Negative
Referents Trust Dependent Variable(s) Correlation N Correlation N Tie Strength – Prior to the project Overall Trust 0.283*** 249 0.108 246 Ability-Based Trust 0.115 264 0.061 261 Integrity-Based Trust 0.200*** 260 0.028 260 Benevolence-Based Trust 0.345*** 261 0.138* 258
152
Tie Strength – While on the project Overall Trust 0.502*** 249 0.148* 243 Ability-Based Trust 0.297*** 264 0.079 258 Integrity-Based Trust 0.423*** 260 0.066 257 Benevolence-Based Trust 0.533*** 261 0.165** 255
Table 4.56 Correlations Between Tie Strength and Trust Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
tie strength and each of the trust variables, while controlling for the effect of the other
independent variables. MRA of trust on tie strength, along with the remaining
independent variables, took the following form: Trust DV = β1*Age+ β2*Edu+ β3*Gen+
β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+ β8*S.Vis+ β9*AgeDiff+ β10*EduGap+
β11*GenGap. The adjusted model R2 for each type of trust was discussed in Section
4.5.1.1.
Table 4.57 shows the results of the multiple regression analysis of the trust dependent
variables on tie strength and other independent variables. The MRA was repeated for
each of the trust dependent variables (i.e. overall, ability-based, integrity-based, and
benevolence-based), which allowed the study to examine the effect of tie strength on each
type of trust.
For Group 1, the standardized regression coefficient of Otrust on TSp was not significant.
However, in the same group, the standardized regression coefficient of BBtrust on TSp
was .207 (p < .01) (ABtrust and IBtrust were not significant in this group). For Group 1,
the standardized regression coefficient of Otrust on TSw was .251 (p < .001). In addition,
for this group, the standardized regression coefficient of IBtrust on TSw was .136 (p <
.05), and the standardized regression coefficient of BBtrust on TSw was .353 (p < .001)
(ABtrust was not significant). In this group, the amount of variance in BBtrust that was
uniquely explained by TSp was 4%. The amount of variance in Otrust that was uniquely
explained by TSw was 6%. The amount of variance in IBtrust that was uniquely
explained by TSw was 2%; and the amount of variance in BBtrust that was uniquely
explained by TSw was 13%. For Group 2, the MRA results suggested no statistically
significant relationships between tie strength and any of the trust variables.
153
Overall, the multiple regression analysis indicated that for both positive and negative
referents, respondents had higher BBtrust in those individuals they felt they had a
stronger tie strength with (prior to and while on the project). Respondents also had higher
Otrust and IBtrust in positive referents they felt they had stronger tie strength with, while
on the project or matter.
Trust Dependent Variable(s)
β for Tie Strength (Prior to the Project) only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Trust 0.071 1.175 195 0.527 20.657 0 Ability-Based Trust -0.080 -1.143 205 0.336 10.369 0 Integrity-Based Trust -0.027 -0.426 203 0.489 18.560 0 Benevolence-Based Trust .207** 3.305 204 0.471 17.458 0 Group 2: Negative Referents Overall Trust 0.044 0.550 194 0.182 4.906 0 Ability-Based Trust 0.005 0.058 208 0.158 4.527 0 Integrity-Based Trust 0.025 0.313 205 0.145 4.141 0 Benevolence-Based Trust 0.052 0.661 206 0.124 3.629 0
Trust Dependent Variable(s)
β for Tie Strength (While on the project) only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Positive Referents Overall Trust 0.251*** 4.267 195 0.527 20.657 0 Ability-Based Trust 0.073 1.070 205 0.336 10.369 0 Integrity-Based Trust 0.136* 2.264 203 0.489 18.560 0 Benevolence-Based Trust 0.353*** 5.909 204 0.471 17.458 0 Group 2: Negative Referents Overall Trust 0.049 0.634 194 0.182 4.906 0 Ability-Based Trust 0.045 0.589 208 0.158 4.527 0 Integrity-Based Trust -0.041 -0.542 205 0.145 4.141 0 Benevolence-Based Trust 0.062 0.803 206 0.124 3.629 0
Table 4.57 Regression of Trust on Tie Strength (Prior to)/Tie Strength (While on) and Other Independent Variables (*p < .05, **p < .01, ***p < .001)
The correlation analysis for Group 1 found significant relationships between TSp and
Otrust, IBtrust, and BBtrust. However, subsequent regression analysis for this group only
confirmed the relationship between TSp and BBtrust. In addition, the correlation analysis
for Group 1 showed statistically significant relationships between TSw and each of the
trust variables. Subsequent regression analysis confirmed the relationships between TSw
and Otrust; between TSw and IBtrust; and between TSw and BBtrust. The correlation
154
analysis for Group 2 showed statistically significant relationships between TSp and
BBtrust, that subsequent regression analysis did not confirm. Finally, the correlation
analysis for Group 2 showed statistically significant relationships between TSw and
Otrust, and between TSw and BBtrust. However, subsequent regression analysis did not
confirm either relationship in this group. Therefore, hypothesis 13 was partially
supported by the results of the analysis in Group 1 (both prior to and while on the
project) and not supported in Group 2.
Tie Strength and Knowledge Sharing Behavior
Correlation analysis was used to examine the bivariate relationships between tie strength
and each of the knowledge sharing behaviors. As it can be seen in Table 4.58, the
correlation between TSp and WSO was .172 in Group 1 (p < .01) and .179 in Group 2 (p
< .01). The correlation between TSp and WSE was .154 in Group 1 (p < .05) and .148 in
Group 2 (p < .05). The correlation between TSp and WST was .164 in Group 1 (p < .01)
and .173 in Group 2 (p < .01). The correlation between TSp and WUO was .141 in Group
2 (p < .05) and not significant in Group 1. The correlation between TSp and WUE was
.197 in Group 2 (p < .05) and not significant in Group 1. The correlation between TSp
and WUT was .127 (p < .05) and not significant in Group 1. Finally, the correlation
between TSp and PRUK was .197 in Group 1 (p < .01) and not significant in Group 2.
The correlation between TSw and WSO was .373 in Group 1 (p < .001) and .148 in Group
2 (p < .05). The correlation between TSw and WSE was .321 in Group 1 (p < .001) and
.144 in Group 2 (p < .05). The correlation between TSw and WST was .359 in Group 1 (p
< .001) and .136 in Group 2 (p < .05). The correlation between TSw and WUO was .372
in Group 1 (p < .001) and .146 in Group 2 (p < .05). The correlation between TSw and
WUE was .301 in Group 1 (p < .001) and .150 in Group 2 (p < .05). The correlation
between TSw and WUT was .364 in Group 1 (p < .001) and .131 in Group 2 (p < .05).
Finally, the correlation between TSw and PRUK was .343 in Group 1 (p < .001) and .211
in Group 2 (p < .01).
155
The results of the correlation analysis suggested that respondents had a higher willingness
to share knowledge (overall, explicit and tacit) with individuals, in both groups, they had
higher tie strength with, prior to and while on the project or matter. The results of the
correlation analysis also suggested that respondents had a higher willingness to use
knowledge (overall, explicit and tacit) with negative referents they had higher tie strength
with, prior to the project or matter; and with co-workers from both groups they had
higher tie strength with, while on the project or matter. Finally, the results of the
correlation analysis suggested that respondents had a higher perception that knowledge
received from positive referents was useful, if they had higher tie strength with that co-
worker, prior to the project or matter. Higher tie strength with both negative and positive
referents, while on the project, also positively influenced the perception that the co-
worker knowledge was useful.
Group 1: Positive
Referents Group 2: Negative
Referents Knowledge Sharing Behavior Dependent Variable(s) Correlation N Correlation N Tie Strength – Prior to the project Overall Willingness to Share 0.172** 260 0.179** 258 Willingness to Share (Explicit) 0.154* 263 0.148* 260 Willingness to Share (Tacit) 0.164** 260 0.173** 260 Overall Willingness to Use 0.082 260 0.141* 258 Willingness to Use (Explicit) 0.047 264 0.197** 264 Willingness to Use (Tacit) 0.088 263 0.127* 260 Perceived Receipt of Useful Knowledge 0.197** 260 0.09 261 Tie Strength – While on the project Overall Willingness to Share 0.373*** 261 0.148* 256 Willingness to Share (Explicit) 0.321*** 264 0.144* 258 Willingness to Share (Tacit) 0.359*** 261 0.136* 258 Overall Willingness to Use 0.372*** 261 0.146* 256 Willingness to Use (Explicit) 0.301*** 265 0.150* 262 Willingness to Use (Tacit) 0.364*** 264 0.131* 257 Perceived Receipt of Useful Knowledge 0.343*** 260 0.211** 258
Table 4.58 Correlations Between Tie Strength (Prior to) / Tie Strength (While on) and Knowledge Sharing Behavior Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
tie strength and each of the knowledge sharing behavior variables, while controlling for
156
the effect of the other independent variables. MRA of knowledge sharing behavior on tie
strength, along with the remaining independent variables, took the following form: KSB
DV = β1*Age+ β2*Edu+ β3*Gen+ β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+
β8*S.Vis+ β9*AgeDiff+ β10*EduGap+ β11*GenGap+ β12*O.Trust. The adjusted model R2
for each type of KSB was discussed in Section 4.5.1.1.
Table 4.59 shows the results of the multiple regression analysis of the knowledge sharing
behavior dependent variables on tie strength and other independent variables. The MRA
was repeated for each of the KSB dependent variables (i.e. WSO, WSE, WST, WUO,
WUE, WUT, PRUK), which allowed the study to individually examine the effect of tie
strength on each of the knowledge sharing behaviors.
For Group 1, the MRA showed no statistically significant relationships between TSp or
TSw and the knowledge sharing behavior variables. For Group 2, the standardized
regression coefficient of WSO on TSp was .179 (p < .05); and the standardized regression
coefficient of WST on TSp was .189 (p < .05) (WSE on TSp was not significant). For the
same group, the standardized regression coefficient of WUO on TSp was .178 (p < .05);
the standardized regression coefficient of WUE on TSp was .230 (p < .01); and the
standardized regression coefficient of WUT on TSp was .159 (p < .05). The MRA showed
no significant relationships between PRUK and TSp, in either group. In addition, for
Group 2, the standardized regression coefficient of PRUK on TSw was .189 (p < .05). The
MRA showed no other significant relationships between TSw and any other KSB
variable, in this group.
In Group 2, the amount of variance in WSO that was uniquely explained by TSp was 3%.
The amount of variance in WST that was uniquely explained by TSp was 4%. Next, the
amount of variance in WUO that was uniquely explained by TSp was 3%. The amount of
variance in WUE that was uniquely explained by TSp was 5%; and the amount of
variance in WUT that was uniquely explained by TSp was 3%. Finally, in this group, the
amount of variance in PRUK that was uniquely explained by TSw was 4%.
157
Overall, the multiple regression analysis results suggested no statistically significant
relationships to exist between TSp or TSw and knowledge sharing behaviors, with
positive referents. However, respondents did have higher WSO and WST with those
negative referents they felt they had a stronger TSp. Respondents also had a higher
willingness to use knowledge (overall, tacit and explicit) from negative referents, when
there were higher levels of TSp between them. Finally, respondents who had higher TSw
with negative referents also had a higher perception that the knowledge received from
those co-workers was useful.
Knowledge Sharing Behavior Dependent Variable(s)
β for Tie Strength (Prior to the project) only t N
Model Adj. R2 Model F
Model Sig.
Group 1: Positive Referents Overall Willingness to Share -0.019 -0.280 189 0.424 12.540 0 Willingness to Share (Explicit) -0.086 -1.181 192 0.329 8.789 0 Willingness to Share (Tacit) 0.008 0.113 189 0.399 11.384 0 Overall Willingness to Use -0.082 -1.080 188 0.316 8.185 0 Willingness to Use (Explicit) -0.135 -1.773 192 0.288 7.444 0 Willingness to Use (Tacit) -0.069 -0.898 190 0.298 7.674 0 Perceived Receipt of Useful Knowledge 0.026 0.383 189 0.434 13.030 0 Group 2: Negative Referents Overall Willingness to Share 0.179* 2.230 189 0.203 4.996 0 Willingness to Share (Explicit) 0.124 1.533 191 0.177 4.416 0 Willingness to Share (Tacit) .189* 2.338 189 0.186 4.574 0 Overall Willingness to Use .178* 2.252 186 0.277 6.896 0 Willingness to Use (Explicit) .230** 2.896 191 0.215 5.328 0 Willingness to Use (Tacit) .159* 2.004 187 0.266 6.613 0 Perceived Receipt of Useful Knowledge 0.005 0.067 189 0.247 6.151 0
Knowledge Sharing Behavior Dependent Variable(s)
β for Tie Strength (While on the project) only T N
Model Adj. R2 Model F
Model Sig.
Group 1: Positive Referents Overall Willingness to Share 0.036 0.520 189 0.424 12.540 0 Willingness to Share (Explicit) 0.078 1.059 192 0.329 8.789 0 Willingness to Share (Tacit) 0.018 0.251 189 0.399 11.384 0 Overall Willingness to Use 0.030 0.401 188 0.316 8.185 0 Willingness to Use (Explicit) -0.065 -0.860 192 0.288 7.444 0 Willingness to Use (Tacit) 0.037 0.492 190 0.298 7.674 0 Perceived Receipt of Useful Knowledge -0.114 -1.653 189 0.434 13.030 0
158
Group 2: Negative Referents Overall Willingness to Share -0.109 -1.407 189 0.203 4.996 0 Willingness to Share (Explicit) -0.066 -0.842 191 0.177 4.416 0 Willingness to Share (Tacit) -0.118 -1.502 189 0.186 4.574 0 Overall Willingness to Use -0.029 -0.380 186 0.277 6.896 0 Willingness to Use (Explicit) -0.003 -0.039 191 0.215 5.328 0 Willingness to Use (Tacit) -0.043 -0.558 187 0.266 6.613 0 Perceived Receipt of Useful Knowledge .189* 2.517 189 0.247 6.151 0 Table 4.59 Regression of Knowledge Sharing Behavior on Tie Strength (Prior to) / Tie
Strength (While on) and Other Independent Variables (*p < .05, **p < .01, ***p < .001) The correlation analysis for both groups showed significant relationships between TSp /
TSw and willingness to share knowledge (overall, explicit and tacit). Subsequent
regression analysis confirmed statistically significant relationships, with negative
referents, between WSO and TSp, and between WST and TSp. The correlation analysis
also found statistically significant relationships with negative referents between TSp /
TSw and willingness to use knowledge (overall, explicit and tacit), and with positive
referents between WUO and TSw. Subsequent regression analysis confirmed relationships
between willingness to use knowledge (overall, explicit and tacit) and TSp with negative
referents. Finally, the correlation analysis found statistically significant relationships
between TSp and PRUK with positive referents, and between TSw and PRUK in both
groups. Subsequent regression analysis confirmed a statistically significant relationship
between PRUK and TSw with negative referents. Based on these results, hypothesis 14
was not supported for Group 1, either for tie strength prior to or tie strength while on the
project or matter. However, the hypothesis was partially supported for Group 2, both for
tie strength prior to and for tie strength while on the project or matter.
4.5.6 Independent Variable 8 – Overall Trust
Overall Trust and Knowledge Sharing Behavior
Hypothesis 15 stated that “trust will be positively related to knowledge sharing
behavior”. Correlation and multiple regression analysis were used to test this hypothesis.
Correlation analysis was used to examine the bivariate relationships between Otrust and
each of the knowledge sharing behaviors. As it can be seen in Table 4.60, the correlation
between Otrust and WSO was .541 in Group 1 and .302 in Group 2. The correlation
159
between Otrust and WSE was .469 in Group 1 and .266 in Group ,2 and between Otrust
and WST was .530 in Group 1 and .298 in Group 2. The correlation between Otrust and
WUO was .507 in Group 1 and .468 in Group 2. The correlation between Otrust and
WUE was .491 in Group 1 and .444 in Group 2, and between Otrust and WUT was .485
in Group 1 and .445 in Group 2. Finally, the correlation between Otrust and PRUK was
.586 in Group 1 and .482 in Group 2. All correlations, in both groups, had p < .001. The
results of the correlation analysis suggested that the higher trust respondents had in co-
workers from both groups, the more effective knowledge sharing behaviors were likely to
take place with those co-workers.
Group 1: Positive
Referents Group 2: Negative
Referents Knowledge Sharing Behavior Dependent Variable(s) Correlation N Correlation N Overall Willingness to Share 0.541*** 244 0.302*** 240 Willingness to Share (Explicit) 0.469*** 247 0.266*** 242 Willingness to Share (Tacit) 0.530*** 244 0.298*** 241 Overall Willingness to Use 0.507*** 244 0.468*** 238 Willingness to Use (Explicit) 0.491*** 248 0.444*** 244 Willingness to Use (Tacit) 0.485*** 247 0.445*** 240 Perceived Receipt of Useful Knowledge 0.586*** 244 0.482*** 242
Table 4.60 Correlations Between Overall Trust and Knowledge Sharing Behavior Variables (*p < .05, **p < .01, ***p < .001)
Multiple regression analysis (MRA) was then used to examine the relationships between
Otrust and each of the knowledge sharing behavior variables, while controlling for the
effect of the other independent variables. MRA of knowledge sharing behavior on Otrust,
along with the remaining independent variables, took the following form: KSB DV =
β1*Age+ β2*Edu+ β3*Gen+ β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+ β8*S.Vis+
β9*AgeDiff+ β10*EduGap+ β11*GenGap+ β12*O.Trust. The adjusted model R2 for each
type of KSB was discussed in Section 4.5.1.1.
Table 4.61 shows the results of the multiple regression analysis of the knowledge sharing
behavior dependent variables on Otrust and other independent variables. The MRA was
repeated for each of the KSB dependent variables (i.e. WSO, WSE, WST, WUO, WUE,
160
WUT, PRUK), which allowed the study to individually examine the effect of Otrust on
each of the knowledge sharing behaviors.
For Group 1, the standardized regression coefficient of WSO on Otrust was .178 (p <
.05). The standardized regression coefficient of WST on Otrust was .172 (p < .05) (WSE
on Otrust was found to be not significant). For the same group, the standardized
regression coefficient of WUO on Otrust was .308 (p < .001). The standardized
regression coefficient of WUE on Otrust was .461 (p < .001); and the standardized
regression coefficient of WST on Otrust was .266 (p < .01). Finally, in Group 1, the
standardized regression coefficient of PRUK on Otrust was .309 (p < .001).
In Group 1, the amount of variance in WSO that was uniquely explained by Otrust was
3%. The amount of variance in WST that was uniquely explained by Otrust was 3%.
Next, the amount of variance in WUO that was uniquely explained by Otrust was 10%.
The amount of variance in WUE that was uniquely explained by Otrust was 21%; and the
amount of variance in WUT that was uniquely explained by Otrust was 7%. Finally, in
this group, the amount of variance in PRUK that was uniquely explained by Otrust was
10%.
For Group 2, the standardized regression coefficient of WSO on Otrust was .211 (p <
.01). The standardized regression coefficient of WSE on Otrust was .226 (p < .01); and
the standardized regression coefficient of WST on Otrust was .194 (p < .05). For the same
group, the standardized regression coefficient of WUO on Otrust was .306 (p < .001).
The standardized regression coefficient of WUE on Otrust was .359 (p < .001); and the
standardized regression coefficient of WST on Otrust was .277 (p < .001). Finally, the
standardized regression coefficient of PRUK on Otrust was .462 (p < .001).
In Group 2, the amount of variance in WSO that was uniquely explained by Otrust was
5%. The amount of variance in WSE that was uniquely explained by Otrust was 5%; and
the amount of variance in WST that was uniquely explained by Otrust was 4%. Next, the
amount of variance in WUO that was uniquely explained by Otrust was 9%. The amount
161
of variance in WUE that was uniquely explained by Otrust was 13%; and the amount of
variance in WUT that was uniquely explained by Otrust was 8%. Finally, in Group 2, the
amount of variance in PRUK that was uniquely explained by Otrust was 21%.
Overall, the multiple regression analysis suggested that, for positive referents,
respondents had a higher WSO and WST with those co-workers they had a higher Otrust
in. A higher Otrust in positive referents also related to a higher willingness to use
knowledge (overall, explicit and tacit) from them. Additionally, higher Otrust in the
positive referent positively related to a higher perception that the knowledge received
from those co-workers was useful. The MRA also showed that, for negative referents,
respondents had a higher willingness to share knowledge (overall, explicit and tacit) with
those co-workers they had a higher Otrust in. A higher Otrust in those same individuals
also related to a higher willingness to use knowledge (overall, explicit and tacit) from
them. Finally, a higher Otrust by respondents in negative referents positively related to a
higher perception that the knowledge received from those individuals was useful.
Knowledge Sharing Behavior Dependent Variable(s)
β for Overall Trust only t N
Model Adj. R2
Model F
Model Sig.
Group 1: Person they worked best with Overall Willingness to Share 0.178* 2.139 189 0.424 12.540 0 Willingness to Share (Explicit) 0.153 1.720 192 0.329 8.789 0 Willingness to Share (Tacit) 0.172* 2.029 189 0.399 11.384 0 Overall Willingness to Use 0.308*** 3.401 188 0.316 8.185 0 Willingness to Use (Explicit) 0.461*** 5.017 192 0.288 7.444 0 Willingness to Use (Tacit) 0.266** 2.921 190 0.298 7.674 0 Perceived Receipt of Useful Knowledge .309*** 3.790 189 0.434 13.030 0 Group 2: Person they did not work well with Overall Willingness to Share 0.211** 2.848 189 0.203 4.996 0 Willingness to Share (Explicit) 0.226** 3.017 191 0.177 4.416 0 Willingness to Share (Tacit) 0.194* 2.586 189 0.186 4.574 0 Overall Willingness to Use .306*** 4.276 186 0.277 6.896 0 Willingness to Use (Explicit) .359*** 4.901 191 0.215 5.328 0 Willingness to Use (Tacit) .277*** 3.845 187 0.266 6.613 0 Perceived Receipt of Useful Knowledge .462*** 6.413 189 0.247 6.151 0
Table 4.61 Regression of Knowledge Sharing Behavior on Overall Trust and Other Independent Variables (*p < .05, **p < .01, ***p < .001)
162
In both groups, the correlation analysis suggested statistically significant relationships
between Otrust and all the knowledge sharing behaviors variables. Subsequent regression
analysis confirmed the correlation analysis in all cases for Group 2, and all cases except
Otrust and WSE in Group 1. Therefore, hypothesis 15 was partially supported in Group 1
and fully supported in Group 2.
4.5.7 Collective Effect of Social-Cognitive Factors and Trust on Knowledge Sharing
Behavior
Hypothesis 16 stated that “overall trust and social-cognitive factors explain knowledge
sharing behavior.” The goal of testing this hypothesis was to assess the collective effect
of the overall model (including Otrust) on knowledge sharing behaviors, in both groups.
The hypothesis was tested for each KSB, in both groups, using regression analysis.
Tables 4.62 and 4.63 summarize the collective model results for each of the knowledge
sharing behaviors in both groups. For each knowledge sharing behavior, the tables
present significant independent variables (based on their β values), β significance,
adjusted R2 for the model, model F value, model significance, and sample size. The semi-
partial coefficient squared (sr2) for each variable is discussed below.
Since KSB was conceptualized in the study as a composite of three distinct knowledge
sharing behaviors, the findings are divided and presented in three sections, consistent
with the three distinct knowledge behaviors: willingness to share knowledge, willingness
to use knowledge, and perceived receipt of useful knowledge.
Group 1 Significant Independent Variables Dependent Variables -
KSB Age β Gen β S.Lan β S.Vis β O.Trust β
Age Diff β
Gender Gap β
Model Adj. R2
F Value Sig. N
KSB - WSO .131* .320*** .240** .178* 0.424 12.54 0 189 KSB - WSE .146* .325*** 0.329 8.789 0 192 KSB - WST .294*** .260** .172* 0.399 11.384 0 189 KSB - WUO .201** .265** .308** 0.316 8.185 0 188 KSB - WUE .138* .461*** .143* 0.288 7.444 0 192 KSB - WUT .234** .282** .266** -.144* 0.298 7.674 0 190 KSB - PRUK .385*** .309*** 0.434 13.03 0 189
Table 4.62 Group 1 MRA Significant IV βs and Collective Model Results (*p < .05, **p < .01, ***p < .001)
163
Group 2 Significant Independent Variables Dependent Variables
- KSB Age β Gen β
T.S.P. β
T.S.P. β
S.Lan β S.Vis β
O.Trust β
Gender Gap β
Model Adj. R2
F Value Sig N
KSB - WSO -.229** .179* .222** .211** 0.203 4.996 0 189 KSB - WSE -.212** .146* .165* .226** 0.177 4.416 0 191 KSB – WST -.222** .189* .227** .194* 0.186 4.574 0 189 KSB - WUO .162* .178* .312*** .306*** 0.277 6.896 0 186 KSB - WUE .179* .230** .359*** 0.215 5.328 0 191 KSB - WUT .140* .159* .348*** .277*** 0.266 6.613 0 187 KSB - PRUK .189* .462*** .148* 0.247 6.151 0 189
Table 4.63 Group 2 MRA Significant IV βs and Collective Model Results (*p < .05, **p < .01, ***p < .001)
4.5.7.1 Collective Effect of Social-Cognitive Factors and Trust on Willingness to Share
Knowledge
The collective effect of all the independent variables (including overall trust) on
willingness to share knowledge, in both groups, is presented in Table 4.64. Using MRA,
the adjusted model R2 for WSO in the first group was found to be .424, with an F value of
12.54. The Group 1 model had p < .001 and a sample size of 189. Significant
independent variables included gender (β of .131 and sr2 of 2%), shared language (β of
.320 and sr2 of 10%), shared vision (β of .240 and sr2 of 6%), and Otrust (β of .178 and
sr2 of 3%). All βs had p < .05. In Group 2, the adjusted model R2 for WSO was .203 with
an F value of 4.996. The Group 2 model had p < .001 and a sample size of 189.
Significant independent variables included age (β of -.229 and sr2 of 5%), TSp (β of .179
and sr2 of 3%), shared vision (β of .222 and sr2 of 5%), and Otrust (β of .211 and sr2 of
5%). All βs had p < .05.
164
Overall Willingness to Share Knowledge (Group 1)
Overall Willingness to Share Knowledge (Group 2)
Model Adj. R2 F Value Model Sig. N Model Adj. R2 F Value Model Sig. N 0.424 12.54 0 189 0.203 4.996 0 189
Sig. Independent Variables β β Sig. sr2
Sig. Independent Variables β β Sig. sr2
Gender .131 .05 2% Age -.229 .01 5% Shared Language .320 .001 10% Tie Strength (Prior) .179 .05 3% Shared Vision .240 .01 6% Shared Vision .222 .01 5% Overall Trust .178 .05 3% Overall Trust .211 .01 5%
Table 4.64 Significant IV βs and Collective Model Results on Overall Willingness to Share Knowledge in Groups 1 and 2
The collective effect of the overall model on WSE and WST, in both groups, is presented
in Tables 4.65 (Group 1) and 4.66 (Group 2). In Group 1, the adjusted model R2 for WSE
was .329 with an F value of 8.789 (Table 4.65). The Group 1 WSE model had p < .001
and a sample size of 192. Significant independent variables included gender (β of .146
and sr2 of 2%) and shared language (β of .325 and sr2 of 11%). All βs had p < .05. For the
same group, the adjusted model R2 for WST was .399 with an F value of 11.384. The
Group 1 WST model had p < .001 and a sample size of 189. Significant independent
variables included shared language (β of .294 and sr2 of 9%), shared vision (β of .260
and sr2 of 7%), and Otrust (β of .172 and sr2 of 3%). All βs had p < .05.
Willingness to Share Explicit Knowledge
(Group 1) Willingness to Share Tacit Knowledge
(Group 1) Model Adj. R2 F Value Model Sig. N Model Adj. R2 F Value Model Sig. N
0.329 8.789 0 192 0.399 11.384 0 189 Sig. Independent Variables β β Sig. sr2
Sig. Independent Variables β β Sig. sr2
Gender .146 .05 2% Shared Language .294 .001 9% Shared Language .325 .001 11% Shared Vision .260 .01 7% Overall Trust .172 .05 3%
Table 4.65 Significant IV βs and Collective Model Results on Willingness to Share Explicit and Tacit Forms of Knowledge in Group 1
In Group 2, the adjusted model R2 for WSE was .177 with an F value of 4.416 (Table
4.66). The Group 2 WSE model had p < .001 and a sample size of 191. Significant
independent variables included age (β of -.212 and sr2 of 5%), shared language (β of .146
and sr2 of 2%), shared vision (β of .165 and sr2 of 3%), and Otrust (β of .226 and sr2 of
5%). All βs had p < .05. For the same group, the adjusted model R2 for WST was .186,
with an F value of 4.574. The Group 2 WST model had p < .001 and a sample size of 189.
Significant independent variables included age (β of -.222 and sr2 of 5%), TSp (β of .189
165
and sr2 of 4%), shared vision (β of .227 and sr2 of 5%), and Otrust (β of .194 and sr2 of
4%). All βs had p <.05.
Willingness to Share Explicit Knowledge
(Group 2) Willingness to Share Tacit Knowledge
(Group 2) Model Adj. R2 F Value Model Sig. N Model Adj. R2 F Value Model Sig. N
0.177 4.416 0 191 0.186 4.574 0 189 Sig. Independent Variables β β Sig. sr2
Sig. Independent Variables β β Sig. sr2
Age -.212 .01 5% Age -.222 .01 5% Shared Language .146 .05 2% Tie Strength (Prior) .189 .05 4% Shared Vision .165 .05 3% Shared Vision .227 .01 5% Overall Trust .226 .01 5% Overall Trust .194 .01 4%
Table 4.66 Significant IV βs and Collective Model Results on Willingness to Share Explicit and Tacit Forms of Knowledge in Group 2
4.5.7.2 Collective Effect of Social-Cognitive Factors and Trust on Willingness to Use
Knowledge
Next, the analysis looked at the collective effect of all the independent variables on
willingness to use knowledge, in both groups. Table 4.67 presents the results of the MRA
for WUO. In Group 1, the adjusted model R2 for WUO was .316, with an F value of
8.185. The Group 1 model had p < .001 and a sample size of 188. Significant
independent variables included age (β of .201 and sr2 of 4%), shared vision (β of .265 and
sr2 of 7%), and Otrust (β of .308 and sr2 of 10%). All βs had p < .05. In Group 2, the
adjusted model R2 for WUO was .277, with an F value of 6.896. The Group 2 model had
p < .001 and a sample size of 186. Significant independent variables included gender (β
of .162 and sr2 of 3%), TSp (β of .178 and sr2 of 3%), shared vision (β of .312 and sr2 of
10%), and Otrust (β of .306 and sr2 of 9%). All βs had p < .05.
Overall Willingness to Use Knowledge
(Group 1) Overall Willingness to Use Knowledge
(Group 2)
Model Adj. R2 F Value Model Sig. N Model Adj. R2 F Value Model
Sig. N 0.316 8.185 0 188 0.277 6.896 0 186
Sig. Independent Variables β β Sig. sr2
Sig. Independent Variables β β Sig. sr2
Gender .201 .01 4% Gender .162 .05 3% Shared Vision .265 .01 7% Tie Strength (Prior) .178 .05 3% Overall Trust .308 .01 10% Shared Vision .312 .001 10% Overall Trust .306 .001 9% Table 4.67 Significant IV βs and Collective Model Results on Overall Willingness to Use
Knowledge in Groups 1 and 2
166
The collective effect of the overall model on WUE and WUT, in both groups is presented
in Tables 4.68 (Group 1) and 4.69 (Group 2). In Group 1, the adjusted model R2 for WUE
was .288, with an F value of 7.444 (Table 4.68). The Group 1 WUE model had p < .001
and a sample size of 192. Significant independent variables included gender (β of .138
and sr2 of 2%), Otrust (β of .461 and sr2 of 21%), and gender gap (β of .143 and sr2 of
2%). All βs had p < .05. For the same group, the adjusted model R2 for WUT was .298,
with an F value of 7.674. The Group 1 WUT model had p < .001 and a sample size of
190. Significant independent variables included age (β of .234 and sr2 of 6%), shared
vision (β of .282 and sr2 of 8%), Otrust (β of .266 and sr2 of 7%), and age difference (β of
-.144 and sr2 of 2%). All βs had p < .05.
Willingness to Use Explicit Knowledge (Group 1)
Willingness to Use Tacit Knowledge (Group 1)
Model Adj. R2 F Value Model Sig. N Model Adj. R2 F Value Model Sig. N 0.288 7.444 0 192 0.298 7.674 0 190
Sig. Independent Variables β β Sig. sr2
Sig. Independent Variables β β Sig. sr2
Gender .138 .05 2% Age .234 .01 6% Overall Trust .461 .001 21% Shared Vision .282 .01 8% Gender Gap .143 .05 2% Overall Trust .266 .01 7% Age Difference -.144 .05 2%
Table 4.68 Significant IV βs and Collective Model Results on Willingness to Use Explicit and Tacit Forms of Knowledge in Group 1
In Group 2, the adjusted model R2 for WUE was .215, with an F value of 5.328 (Table
4.69). The Group 2 WUE model had p < .001 and a sample size of 191. Significant
independent variables included gender (β of .179 and sr2 of 3%), TSp (β of .230 and sr2 of
5%), and Otrust (β of .359 and sr2 of 13%). All βs had p < .05. For the same group, the
adjusted model R2 for WUT was .266, with an F value of 6.613. The Group 2 WUT model
had p < .001 and a sample size of 187. Significant independent variables included gender
(β of .140 and sr2 of 2%), TSp (β of .159 and sr2 of 3%), shared vision (β of .348 and sr2
of 12%), and Otrust (β of .277 and sr2 of 8%). All βs had p < .05.
167
Willingness to Use Explicit Knowledge (Group 2)
Willingness to Use Tacit Knowledge (Group 2)
Model Adj. R2 F Value Model Sig. N Model Adj. R2 F Value Model
Sig. N 0.215 5.328 0 191 0.266 6.613 0 187
Sig. Independent Variables β β Sig. sr2
Sig. Independent Variables β β Sig. sr2
Gender .179 .05 3% Gender .140 .05 2% Tie Strength (Prior) .230 .01 5% Tie Strength (Prior) .159 .05 3% Overall Trust .359 .001 13% Shared Vision .348 .001 12% Overall Trust .277 .001 8%
Table 4.69 Significant IV βs and Collective Model Results on Willingness to Use Explicit and Tacit Forms of Knowledge in Group 2
4.5.7.3 Collective Effect of Social-Cognitive Factors and Trust on Perceived Receipt of
Useful Knowledge
Finally, the analysis looked at the collective effect of all of the independent variables on
the perceived receipt of useful knowledge, in both groups. Table 4.70 presents the results
of the MRA for PRUK. In Group 1, the adjusted model R2 for PRUK was .434, with an F
value of 13.03. The Group 1 model had p < .001 and a sample size of 189. Significant
independent variables included shared vision (β of .385 and sr2 of 15%) and Otrust (β of
.309 and sr2 of 10%). In Group 2, the adjusted model R2 for PRUK was .247, with an F
value of 6.151. The Group 2 model had p < .001 and a sample size of 189. Significant
independent variables included TSw (β of .189 and sr2 of 4%), Otrust (β of .462 and sr2 of
21%), and gender gap (β of .148 and sr2 of 2%). All βs, in both groups, had p < .05.
Perceived Receipt of Useful Knowledge
(Group 1) Perceived Receipt of Useful Knowledge
(Group 2)
Model Adj. R2 F Value Model Sig. N Model Adj. R2 F Value Model
Sig. N 0.434 13.03 0 189 0.247 6.151 0 189
Sig. Independent Variables β β Sig. sr2
Sig. Independent Variables β β Sig. sr2
Shared Vision .385 .001 15% Tie Strength (While) .189 .05 4% Overall Trust .309 .001 10% Overall Trust .462 .001 21% Gender Gap .148 .05 2%
Table 4.70 Significant IV βs and Collective Model Results on Perceived Receipt of Useful Knowledge in Groups 1 and 2
Multiple regression analysis showed that overall trust and social-cognitive factors
explained each of the knowledge sharing behaviors identified in the study, with both
positive and negative referents. However, considerably more of knowledge sharing
168
behaviors could be explained by the identified IVs in relationships with positive
referents. In either case, hypothesis 16 was fully supported in both groups.
4.5.8 Mediating Effect of Overall Trust
Hypothesis 17 stated that “overall trust will be a mediating variable between social-
cognitive factors and knowledge sharing behavior.” The theoretical model identified the
role of trust as an intervening or mediating variable, where m mediates a relationship
between x and y (i.e. x → m → y). The independent variables (i.e. social-cognitive factors
x) were predicted to affect the dependent variables (i.e. knowledge sharing behaviors y)
indirectly, through the intervening variable (i.e. overall trust m). This would mean that
social-cognitive factor (SCF) variables (x) have a direct effect on knowledge sharing
behavior (KSB) (y) and a direct effect on overall trust (m); while overall trust (m) also
has a direct effect on knowledge sharing behavior (y). An intervening variable “delineates
the causal mechanisms producing the observed relationship between the focal
independent [i.e. social-cognitive factors] and dependent [i.e. knowledge sharing
behaviors] variables” (Aneshensel, 2002, p. 155). The intervening variable plays two
analytic roles: it is the dependent variable affected by the independent variables
(displayed in connection A in Figure 4.1) and the independent variable affected by the
dependent variables (displayed in connection B in Figure 4.1). Since it must serve a
connective function, both these connections are necessary for Otrust to be confirmed as a
mediating variable. Specifying Otrust as a meditating variable also suggests that the
connection between SCF and KSB is thought to be indirect (i.e. SCFs influences Otrust,
which in turn influences KSBs).
Figure 4.1 The Focal Relationship and Trust as a Mediating Variable
In addition to an indirect relationship, a mediating variable also has an effect on the direct
relationship that remains between the independent and dependent variables, after the
169
indirect relationship is taken into account. Tabachnick and Fidell (2007) called this
relationship (between the IV and DV after “controlling for” the mediator) the direct
effect.
According to Baron and Kenny (1986), four conditions are necessary for a variable to be
confirmed as a mediator:
1. There must be a significant relationship between the IV and the DV;
2. There must be a significant relationship between the IV and the mediator;
3. The mediator must still predict the DV after controlling for the IV and;
4. The relationship between the IV and the DV is reduced when the mediator is in
the equation.
If Hypothesis 17 was to be found valid, then the inclusion of the Otrust variable should
account for some or all of the relationship between SCF and KSB (connection C in
Figure 4.1). Aneshensel (2002) argued that an increase in the indirect component
decreased the direct component. A reduction in, or elimination of, the relationship
between the independent variables (SCF) and the dependent variables (KSB) means that a
part or all of the relationship is explained through the mediating variable (i.e. Otrust). If
the inclusion of Otrust entirely eliminates the relationship between IV and DV, the
relationship is said to be ‘fully elaborated’ and the mediation is said to be perfect (‘full’
or ‘complete’). “This pattern of results tends to occur because the dependent variable is
more proximal to the intervening variable than the focal independent variable and
because proximal effects tend to be stronger than distal effects” (Aneshensel, 2002, p.
164). However, if the inclusion of Otrust into the model only partially accounts for the
relationship, it is said to be ‘partially elaborated’, and the mediation is said to be
‘partial’. In these cases, the results would explain only part of the theory, leaving other
portions unexplained and directing attention to further development of theory. Finally, if
the inclusion of Otrust leaves the focal IV-DV relationship intact, it is said to be
‘unelaborated’. In these cases, it could be ruled out that Otrust is a mediating variable.
170
To test the mediating effect of Otrust, hierarchical multiple regression analysis (HMRA)
was used. In the first stage of the HMRA the knowledge sharing behavior variables were
regressed on the independent variables. In this first step, Otrust was excluded from the
model (i.e. KSB = β1*Age+ β2*Edu+ β3*Gen+ β4*R.Len+ β5*T.S.p+ β6*T.S.w+
β7*S.Lan+ β8*S.Vis+ β9*AgeDiff+ β10*EduGap+ β11*GenGap+ β12*O.Trust). However,
in the second step of the HMRA, the mediating variable, Otrust, was included as an
independent variable, along with the other independent variables (i.e. KSB = β1*Age+
β2*Edu+ β3*Gen+ β4*R.Len+ β5*T.S.p+ β6*T.S.w+ β7*S.Lan+ β8*S.Vis+ β9*AgeDiff+
β10*EduGap+ β11*GenGap+ β12*O.Trust). The contribution of Otrust as a mediating
variable to the explanation of KSB (y) was assessed using F tests for the increment in R2.
If significant, then βetas and t-tests for each of the individual variables were used.
The focus of the analysis is on what happens to the regression coefficient for the focal
relationships (SCF-KSB), when the mediating variable (Otrust) is added to the model.
Mediation is said to take place when the β of the IV-DV relationship decreases. The
mediation is said to be complete if the focal relationship is no longer significant, after
Otrust is introduced into the model. A partial mediation would still show a statistically
significant relationship, but would have a reduced β, as Otrust explains only part of the
relationship. Finally, the focal relationship is not found to be mediated by Otrust if the
regression coefficient for the relationship is essentially unchanged with the addition of
the mediating variable.
4.5.8.1 Mediating Effect of Overall Trust Results for Positive Referents (Group 1)
Figure 4.2 summarizes the statistically significant relationships found by testing
hypotheses 1-15 for positive referents (i.e. Group 1 or the co-worker the respondent
worked best with). As noted by Baron and Kenny (1986), there must be a significant
relationship between the IV and the DV and between the IV and the mediating variable,
to confirm a variable as a mediator. Of all of the significant relationships in Figure 4.2,
shared language and shared vision were the only two variables that satisfied the first two
conditions set out by Baron and Kenny (1986). Homophily variables, relationship length,
171
and tie strength variables did not meet the criterion of being mediating variables, as they
had no significant relationships with either Otrust or KSB.
Figure 4.2 Significant Relationships and Corresponding Hypotheses for Positive
Referents (Group 1) The final two conditions suggested by Baron and Kenny (1986) to confirm Otrust as a
mediator were tested using a 2-step hierarchical multiple regression analysis. In the
second model (step 2), the effect of Otrust was considered by adding it to the HMRA
Step 1 model, which included the rest of the IVs. The complete results of the HMRA for
positive referents appear in Table 4.71 and Table 4.72.
The results of the analysis showed that Otrust still predicted KSB. Further, the results
showed that the relationships between shared language and KSB, and between shared
vision and KSB, weakened when Otrust was introduced into the model. These results
suggested that Otrust, at least partially, mediated the relationships between shared
language and KSB, as well as between shared vision and KSB. The specifics of these
mediated relationships are discussed in the sections below.
172
Table 4.71 Results of the HMRA for Positive Referents (Group 1) (*p < .05, **p < .01, ***p < .001)
HMRA Model Change Summary
Change R2 F Change Sig. F Change KSB - WSO 0.014 4.576 0.034 KSB - WSE 0.01 2.957 0.087 KSB - WST 0.013 4.117 0.044 KSB - WUO 0.042 11.57 0.001 KSB - WUE 0.094 25.166 0 KSB - WUT 0.032 8.532 0.004 KSB - PRUK 0.043 14.365 0
Table 4.72 HMRA Model Change Summary for Positive Referents (Group 1) Group 1: The Overall Mediating Effect of Trust Between Shared Language and KSB
S. Language Model Summary Step 1 β T Adj. R2 F Value Sig N KSB - WSO 0.366 5.118*** 0.413 13.002 0 189 KSB - WSE 0.364 4.747*** 0.321 8.789 0 192 KSB - WST 0.339 4.638*** 0.388 11.837 0 189 KSB - WUO 0.127 1.590 0.274 7.431 0 188 KSB - WUE 0.213 2.541* 0.193 5.142 0 192 KSB - WUT 0.106 1.330 0.268 7.287 0 190 KSB - PRUK 0.204 2.777** 0.392 12.003 0 189
173
Step 2 β T Adj. R2 F Value Sig N Change R2 F Change Sig. F
Change KSB - WSO 0.32 4.330*** 0.424 12.540 0 189 0.014 4.576 0.034 KSB - WSE 0.325 4.074*** 0.329 8.789 0 192 0.01 2.957 0.087 KSB - WST 0.294 3.892*** 0.399 11.384 0 189 0.013 4.117 0.044 KSB - WUO 0.047 0.584 0.316 8.185 0 188 0.042 11.57 0.001 KSB - WUE 0.095 1.156 0.288 7.444 0 192 0.094 25.166 0 KSB - WUT 0.036 0.438 0.298 7.674 0 190 0.032 8.532 0.004 KSB - PRUK 0.123 1.667 0.434 13.030 0 189 0.043 14.365 0
Table 4.73 Group 1 HMRA Results for Shared Language on KSB (*p < .05, **p < .01, ***p < .001)
The Group 1 HMRA results for shared language (on KSB) are shown in Table 4.73 and
summarized below. Otrust was introduced into the model in Step 2 of the HMRA.
The Overall Mediating Effect of Trust Between Shared Language and Willingness to
Share Knowledge
The addition of Otrust made a small contribution to the explanation of WSO, as indicated
by the increment of R2 from the Step 1 Model (WSO ΔR2 = 3.3%; p < .001; WSE ΔR2 =
3%; p < .001; WST ΔR2 = 3.3%; p < .001). The addition of Otrust reduced the coefficient
for shared language (on WSO) by 12.6% (Δβ = .366 - .320). Mediation was present, but
only partially, as shared language remained statistically associated with WSO. The
results also showed that the addition of Otrust reduced the coefficient for shared
language with WSE by 10.7% (Δβ = .364 - .325) and with WST by 13.3% (Δβ = .339-
.294). In both these cases the mediation was present, but only partially, as relationships
between shared language and WSE and WST remained statistically significant.
The Overall Mediating Effect of Trust Between Shared Language and Willingness to Use
Knowledge
The addition of Otrust made a substantial contribution to the explanation of WUO, as
indicated by the increment of R2 from the Step 1 Model (WSO ΔR2 = 13.3%; p < .001;
WUE ΔR2 = 32.6%; p < .001; WUT ΔR2 = 10.7%; p < .001).
The coefficient for shared language was found to be not significantly associated with
WUO in the first model, or in the second, when Otrust was introduced. However, it was
found that the addition of Otrust reduced the coefficient for shared language with WUE
174
by 55.4% (Δβ = .213-.095) and, more interestingly, was no longer statistically significant.
This suggested that Otrust had a complete mediating effect for the relationship between
shared language and WUE. The coefficient for shared language was found to be not
significantly associated with WUT in either model.
The Overall Mediating Effect of Trust Between Shared Language and Perceived Receipt
of Useful Knowledge
Otrust made a moderate contribution to the explanation of PRUK, as indicated by the
increment of R2 from the Step 1 Model (PRUK ΔR2 = 9.9%; p < .001).
The addition of Otrust reduced the coefficient for shared language (on PRUK) by 39.7%
(Δβ = .204-.123). Further, because shared language was no longer statistically significant
after Otrust was introduced into the model, it can be concluded that Otrust had a
complete mediating effect for the relationship between shared language and PRUK.
Group 1: The Overall Mediating Effect of Trust Between Shared Vision and KSB
S. Vision Model Summary
Step 1 β T Adj. R2 F Value Sig N KSB - WSO 0.307 4.314*** 0.413 13.002 0 189 KSB - WSE 0.210 2.758** 0.321 8.789 0 192 KSB - WST 0.325 4.474*** 0.388 11.837 0 189 KSB - WUO 0.382 4.855*** 0.274 7.431 0 188 KSB - WUE 0.248 2.986** 0.193 5.142 0 192 KSB - WUT 0.382 4.858*** 0.268 7.287 0 190 KSB - PRUK 0.500 6.877*** 0.392 12.003 0 189
Step 2 β T Adj. R2 F Value Sig N Change R2 F Change Sig. F
Change KSB - WSO 0.24 3.106** 0.424 12.540 0 189 0.014 4.576 0.034 KSB - WSE 0.153 1.851 0.329 8.789 0 192 0.01 2.957 0.087 KSB - WST 0.260 3.293** 0.399 11.384 0 189 0.013 4.117 0.044 KSB - WUO 0.265 3.169** 0.316 8.185 0 188 0.042 11.57 0.001 KSB - WUE 0.072 0.846 0.288 7.444 0 192 0.094 25.166 0 KSB - WUT 0.282 3.358** 0.298 7.674 0 190 0.032 8.532 0.004 KSB - PRUK 0.385 5.049*** 0.434 13.030 0 189 0.043 14.365 0
Table 4.74 Group 1 HMRA Results for Shared Vision (*p < .05, **p < .01, ***p < .001)
The Group 1 HMRA results for shared vision (on KSB) are shown in Table 4.74 and
summarized below. Otrust was introduced into the model in Step 2 of the HMRA.
175
The Overall Mediating Effect of Trust Between Shared Vision and Willingness to Share
Knowledge
The addition of Otrust reduced the coefficient for shared vision (on WSO) by 21.8% (Δβ
= .307-.240). Mediation was present, but only partially, as shared vision remained
statistically associated with WSO. The results also showed that the addition of Otrust
reduced the coefficient for shared vision with WSE by 27.1% (Δβ = .210-.153) and with
WST by 19.2% (Δβ = .325-.260). In the case of WSE, shared vision was no longer
statistically significant after Otrust was introduced into the model. This suggested that
Otrust had a complete mediating effect for the relationship between shared vision and
WSE. For WST, the mediation was present, but only partially, as the relationship between
shared vision and WST remained statistically significant after Otrust was introduced into
the model.
The Overall Mediating Effect of Trust Between Shared Vision and Willingness to Use
Knowledge
The addition of Otrust reduced the coefficient for shared vision (on WUO) by 30.6% (Δβ
= .382-.265). Mediation was present, but only partially, as shared vision remained
statistically associated with WUO. The results also showed that the addition of Otrust
reduced the coefficient for shared vision with WUE by 71% (Δβ = .248-.072) and with
WUT by 26.2% (Δβ = .382-.282). In the case of WUE, shared vision was no longer
statistically significant after Otrust was introduced into the model. This suggested that
Otrust had a complete mediating effect for the relationship between shared vision and
WUE. For WUT, the mediation was present, but only partially, as the relationship
between shared vision and WUT remained statistically significant after Otrust was
introduced into the model.
The Overall Mediating Effect of Trust Between Shared Vision and Perceived Receipt of
Useful Knowledge
The addition of Otrust reduced the coefficient for shared vision (on PRUK) by 23% (Δβ
= .500-.385). Mediation was present, but only partially, as the relationship between
176
shared vision and PRUK remained statistically significant after Otrust was introduced
into the model.
4.5.8.2 The Mediating Effect of Overall Trust Results for Negative Referents (Group 2)
Figure 4.3 summarizes the statistically significant relationships found by testing
hypotheses 1-15 for negative referents (i.e. Group 2 or the co-workers the respondent did
not work well with). As noted by Baron and Kenny (1986), there must be a significant
relationship between the IV and the DV, and between the IV and the mediating variable,
to confirm a variable as a mediator. Of all of the significant relationships in Figure 4.3,
shared vision was the only variable that satisfied the first two conditions set out by Baron
and Kenny (1986) (i.e. homophily measures, shared language, relationship length, and
tie strength variables did not have significant relationships with either Otrust or KSB).
Figure 4.3 Significant Relationships and Corresponding Hypotheses for Negative
Referents (Group 2)
177
The complete results of the HMRA for Group 2 appear in Table 4.75 and Table 4.76. The
results of the analysis showed that Otrust still predicted KSB. Further, the results showed
that the relationship between shared vision and KSB weakened when Otrust was
introduced into the model. These results suggested that Otrust, at least partially, mediated
the relationship between shared vision and KSB with negative referents. The specifics of
these mediated relationships are discussed in the sections below.
Table 4.75 Results of the HMRA for Negative Referents (Group 2) (*p < .05, **p < .01, ***p < .001)
HMRA Model Change Summary Change R2 F Change Sig. F Change
KSB - WSO 0.034 8.112 0.005 KSB - WSE 0.039 9.103 0.003 KSB - WST 0.029 6.685 0.011 KSB - WUO 0.071 18.284 0 KSB - WUE 0.099 24.023 0 KSB - WUT 0.058 14.784 0 KSB - PRUK 0.165 41.128 0
Table 4.76 HMRA Model Change Summary for Negative Referents (Group 2)
178
Group 2: The Overall Mediating Effect of Trust Between Shared Vision and KSB
S. Vision Model Summary Step 1 β T Adj. R2 F Value Sig N KSB - WSO 0.302 3.993*** 0.171 4.531 0 189 KSB - WSE 0.250 3.264** 0.140 3.818 0 191 KSB - WST 0.300 3.941*** 0.160 4.246 0 189 KSB - WUO 0.431 5.690*** 0.205 5.331 0 186 KSB - WUE 0.261 3.315** 0.114 3.215 0 191 KSB - WUT 0.457 6.085*** 0.208 5.441 0 187 KSB - PRUK 0.092 1.156 0.077 2.422 0 189
Step 2 β T Adj. R2 F Value Sig N Change R2 F Change Sig. F
Change KSB - WSO 0.222 2.796** 0.203 4.996 0 189 0.034 8.112 0.005 KSB - WSE 0.165 2.064* 0.177 4.416 0 191 0.039 9.103 0.003 KSB - WST 0.227 2.825** 0.186 4.574 0 189 0.029 6.685 0.011 KSB - WUO 0.312 4.031*** 0.277 6.896 0 186 0.071 18.284 0 KSB - WUE 0.128 1.618 0.215 5.328 0 191 0.099 24.023 0 KSB - WUT 0.348 4.483*** 0.266 6.613 0 187 0.058 14.784 0 KSB - PRUK -0.082 -1.065 0.247 6.151 0 189 0.165 41.128 0 Table 4.77 Group 2 HMRA Results for Shared Vision (*p < .05, **p < .01, ***p < .001)
The Group 2 HMRA results for shared vision (on KSB) are shown in Table 4.77 and
summarized below. Otrust was introduced into the model in Step 2 of the HMRA.
The Overall Mediating Effect of Trust Between Shared Vision and Willingness to Share
Knowledge
The addition of Otrust made a substantial contribution to the explanation of WSO
indicated by the increment of R2 from the Step 1 Model (WSO ΔR2 = 16.8%; p < .001;
WSE ΔR2 = 22%; p < .001; WST ΔR2 = 15.6%; p < .001).
The addition of Otrust also reduced the coefficient for shared vision (on WSO) by 26.5%
(Δβ = .302-.222). Mediation was present, but only partially, as shared vision remained
statistically associated with WSO. The results also showed that the addition of Otrust
reduced the coefficient for shared vision with WSE by 34% (Δβ = .250-.165), and with
WST by 24.3% (Δβ = .300-.227). In both cases, the mediation was present, but only
partially, as relationships between shared vision and WSE and WST remained statistically
significant after Otrust was introduced into the model.
179
The Overall Mediating Effect of Trust Between Shared Vision and Willingness to Use
Knowledge
The addition of Otrust made a substantial contribution to the explanation of WUO as
indicated by the increment of R2 from the Step 1 Model (WUO ΔR2 = 25.6%; p < .001;
WUE ΔR2 = 46%; p < .001; WUT ΔR2 = 21.8%; p < .001).
The addition of Otrust reduced the coefficient for shared vision (on WUO) by 27.6% (Δβ
= .431-.312). Mediation was present, but only partially, as shared vision remained
statistically associated with WUO, after Otrust was introduced into the model. The results
also showed that the addition of Otrust reduced the coefficient for shared vision with
WUE by 51% (Δβ = .261-.128) and with WUT by 23.9% (Δβ = .457-.348). In the case of
WUE, shared vision was no longer statistically significant after Otrust was introduced
into the model. This suggested that Otrust had a complete mediating effect for the
relationship between shared vision and WUE. For WUT, the mediation was present, but
only partially, as the relationship between shared vision and WUT remained statistically
significant after Otrust was introduced into the model.
The Overall Mediating Effect of Trust Between Shared Vision and Perceived Receipt of
Useful Knowledge
The coefficient for shared vision was found to be not significantly associated with PRUK
in the first model, or in the second, when Otrust was introduced.
4.5.8.3 Summary of the Mediating Effects of Overall Trust
In summary, Otrust was found to mediate some of the effects of the independent
variables on the dependent variables. For positive referents, Otrust was found to have
varying degrees of a mediating effect between shared language and each of knowledge
sharing behaviors tested. In the same group, Otrust was found to also have a mediating
effect between shared vision and each of the knowledge sharing behaviors tested. For
negative referents, Otrust had a mediating effect between shared vision and two of the
three main KSBs (i.e. willingness to share knowledge and willingness to use knowledge).
A complete summary of the mediating effect of Otrust is presented in Table 4.78.
180
Group IV DV Type of
mediating effect 1 Shared Language Overall Willingness to Share Knowledge Partial 1 Shared Language Willingness to Share Explicit Knowledge Partial 1 Shared Language Willingness to Share Tacit Knowledge Partial 1 Shared Language Willingness to Use Explicit Knowledge Complete / Full 1 Shared Language Perceived Receipt of Useful Knowledge Complete / Full 1 Shared Vision Overall Willingness to Share Knowledge Partial 1 Shared Vision Willingness to Share Explicit Knowledge Complete / Full 1 Shared Vision Willingness to Share Tacit Knowledge Partial 1 Shared Vision Overall Willingness to Use Knowledge Partial 1 Shared Vision Willingness to Use Explicit Knowledge Complete / Full 1 Shared Vision Willingness to Use Tacit Knowledge Partial 1 Shared Vision Perceived Receipt of Useful Knowledge Partial 2 Shared Vision Overall Willingness to Share Knowledge Partial 2 Shared Vision Willingness to Share Explicit Knowledge Partial 2 Shared Vision Willingness to Share Tacit Knowledge Partial 2 Shared Vision Overall Willingness to Use Knowledge Partial 2 Shared Vision Willingness to Use Explicit Knowledge Complete / Full 2 Shared Vision Willingness to Use Tacit Knowledge Partial
Table 4.78 Summary of the Mediating Effects of Overall Trust Based on the analysis suggested by Baron and Kenny (1986) and the results of the
HMRA, hypothesis 17 was partially supported for positive referents through shared
language and shared vision, and partially supported for negative referents through shared
vision.
4.6 Summary of Hypothesis Tests
Table 4.79 presents a summary of the hypotheses tested, for an overview of the results of
the study.
Hypotheses Support Homophily 1 Age Homophily will be positively related to Trust
Not supported for either group
2 Age Homophily will be positively related to Knowledge Sharing Behavior
Partially supported for Group 1 (Willingness to Use Tacit Knowledge) and Not supported for Group 2
3 Educational Homophily will be positively related to Trust
Not supported for either group
4 Educational Homophily will be positively related to Knowledge Sharing Behavior
Not supported for either group
5 Gender Homophily will be positively related to Trust
Not supported for either group
6 Gender Homophily will be positively related to Knowledge Sharing Behavior
Not supported for either group. Partial support of the opposite - Willingness to Use (Explicit) in Group 1 and Perceived Receipt of Useful Knowledge in Group 2
181
Shared Language and Shared Vision 7 Shared Language will be positively related to Trust
Fully Supported for Group 1 (Overall Trust and each form) and Not Supported for Group 2
8 Shared Language will be positively related to Knowledge Sharing Behavior
Partially Supported for Group 1 (Overall Willingness to Share, Willingness to Share Explicit Knowledge and Willingness to Share Tacit Knowledge) and Partially Supported for Group 2 (Willingness to Share Explicit Knowledge)
9 Shared Vision will be positively related to Trust
Fully Supported for both groups (Overall Trust and each form)
10 Shared Vision will be positively related to Knowledge Sharing Behavior
Partially Supported for Group 1 (Overall Willingness to Share, Willingness to Share Tacit Knowledge, Overall Willingness to Use, Willingness to Use Tacit Knowledge and Perceived Receipt of Useful Knowledge) and Partially Supported for Group 2 (Overall Willingness to Share, Willingness to Share Explicit Knowledge, Willingness to Share Tacit Knowledge, Overall Willingness to Use and Willingness to Use Tacit Knowledge)
Relationship Length and Tie Strength 11 Relationship Length will be positively related to Trust
Not Supported in either group (Partial Support of the opposite for Group 2: Overall Trust, Integrity-Based Trust, and Benevolence-Based Trust)
12 Relationship Length will be positively related to Knowledge Sharing Behavior
Not supported for either group
13 Tie Strength will be positively related to Trust
Partially Supported for Group 1 both Prior to the project or matter (Benevolence-Based Trust) and While on the project or matter (Overall Trust, Integrity-Based trust and Benevolence-Based Trust). It is not supported for Group 2 Prior to or While on the project or matter.
14 Tie Strength will be positively related to Knowledge Sharing Behavior
Not Supported for Group 1 Prior to the project or matter or While on the project or matter. It is partially supported for Group 2 Prior to (Overall Willingness to Share, Willingness to Share Tacit Knowledge, Overall Willingness to Use Knowledge, Willingness to Use Explicit and Tacit Knowledge) and While on the project or matter (Perceived Receipt of Useful Knowledge)
Trust 15 Overall Trust will be positively related to Knowledge Sharing Behavior
Partially Supported for Group 1 (Overall Willingness to Share, Willingness to Share Tacit Knowledge, Overall Willingness to Use (both explicit and tacit), and Perceived Receipt of Useful Knowledge) and Fully Supported for Group 2
Collective Effect 16 Overall Trust and Social-Cognitive Factors explain Knowledge Sharing Behavior. Fully Supported for both groups Mediating Effect 17 Overall Trust will be a mediating variable between Social-Cognitive factors and Knowledge Sharing Behavior
Partially Supported for Group 1 (through Shared Language and Shared Vision) and Partially Supported for Group 2 (through Shared Vision)
Table 4.79 Summary of Data Analysis and whether the Hypotheses were Supported
182
4.7 Summary of Results by Research Question
This research study of trust and knowledge sharing behavior was guided by five research
questions. The following section presents the results of the data analysis, as it pertains to
the research questions.
Summary of Results for Research Question 1
The first research question asked “what are the significant relationships between social-
cognitive variables and trust?” As there were several social-cognitive variables tested in
the study, the answer to this research question varied based on social-cognitive variable.
For example, the data analysis showed that no statistically significant relationships
existed between age, education, or gender homophily and trust, in either group. Next,
shared language was significantly positively related to each trust variable in Group 1 and
not at all related to any trust variable in Group 2. Shared vision was significantly
positively related to each trust variable, in both groups. Next, relationship length was
negatively related to overall trust, integrity-based trust, and benevolence-based trust in
Group 2, and not at all significantly related to trust in Group 1. In Group 1, tie strength,
prior to the project, was positively related to benevolence-based trust. Also in this group,
tie strength, while on the matter or project, was positively related to overall trust,
integrity-based trust, and benevolence-based trust. No significant relationships between
tie strength (prior to or while on) and trust were found in Group 2.
Summary of Results for Research Question 2
The second research question asked “what are the significant relationships between
social-cognitive variables and knowledge sharing behavior?” In Group 1, the data
analysis showed that similarity in age (age homophily) was positively related to
willingness to use tacit knowledge. No age homophily relationships were found in Group
2. Similarity in education (educational homophily) was also not statistically related to
knowledge sharing behavior, in either group. Next, similarity in gender (gender
homophily) was negatively related to willingness to use explicit knowledge from positive
referents, which suggested a gender heterogeneous effect. Gender homophily was also
183
found to have a negative effect on a respondent’s perception that the knowledge received
from the negative referents was useful.
The data analysis also showed that in Group 1 shared language was positively related to
overall willingness to share knowledge, willingness to share explicit knowledge, and
willingness to share tacit knowledge. In Group 2, shared language was only found to
have a positive effect on willingness to share explicit knowledge. In Group 1, shared
vision was positively related to overall willingness to share knowledge and willingness to
share tacit knowledge. A significant relationship was also found in that group, between
shared vision and overall willingness to use knowledge and willingness to use tacit
knowledge. Next, shared vision was found to have a positive effect on the respondent’s
perception that the knowledge received from positive referents was useful. In Group 2,
shared vision was positively related to overall willingness to share knowledge,
willingness to share explicit knowledge, and willingness to share tacit knowledge. Shared
vision also had a positive effect on a respondent’s overall willingness to use knowledge
and willingness to use tacit knowledge. The data analysis also showed that no significant
relationships existed between relationship length and any knowledge sharing behavior, in
either group. In addition, no significant relationships were found to exist between tie
strength (prior to or while on) and knowledge sharing behavior in Group 1. In Group 2
however, tie strength, prior to the project, was positively related to a respondent’s overall
willingness to share knowledge and willingness to share tacit knowledge. In this same
group, tie strength, prior to the project, was positively related to a respondent’s overall
willingness to use knowledge, willingness to use explicit knowledge, and willingness to
use tacit knowledge. Finally, in Group 2, tie strength (while on the project) was positively
related to a respondent’s perception that the knowledge received from the negative
referent was useful.
Summary of Results for Research Question 3
The third research question asked “what are the significant relationships between trust
and knowledge sharing behavior?” In Group 1, the data analysis showed that overall
trust was positively related to a respondent’s overall willingness to share knowledge and
184
willingness to share tacit knowledge. In the same group, the data analysis also showed
that overall trust had a positive effect on overall willingness to use knowledge,
willingness to use explicit knowledge, and willingness to use tacit knowledge. Finally, in
Group 2, overall trust was positively related to all knowledge sharing behaviors tested in
the study.
Summary of Results for Research Question 4
The fourth research question asked “what is the collective effect of the identified social-
cognitive variables and trust on knowledge sharing behavior?” The data analysis showed
that overall trust and social-cognitive factors explained each of the knowledge sharing
behaviors identified in the study, with both positive and negative referents. Notably, more
knowledge sharing behaviors could be explained by the identified IVs in relationships
with positive referents, than could be explained with negative referents. For example, in
Group 1, the collective model explained 42% of overall willingness to share knowledge,
32% of overall willingness to use knowledge, and 43% of perceived receipt of useful
knowledge. In the Group 2, the collective model explained 20% of overall willingness to
share knowledge, 28% of overall willingness to use knowledge, and 25% of perceived
receipt of useful knowledge.
Summary of Results for Research Question 5
The fifth research question asked “does trust act as a mediating variable between social-
cognitive variables and knowledge sharing behavior?” The data analysis showed that, in
Group 1, overall trust acted (to varying degrees) as a mediating variable between shared
language and knowledge sharing behavior, and as a mediating variable between shared
vision and knowledge sharing behavior. In Group 2, overall trust was found to be a
mediating variable between shared vision and knowledge sharing behavior (also to
varying degrees). Table 4.78 contains the complete details of the mediating effects.
185
Chapter 5: Discussion and Summary 5.0 Chapter Overview
As noted in the first chapter the effective sharing of knowledge within organizations
promises many benefits, both to the individual and to the firm. However, in reality, most
companies struggle with sharing knowledge (Ruggles, 1998), because promoting
knowledge sharing within their organizations is a technological and behavioral challenge.
The KM research acknowledges these challenges, by focusing on technological issues
and more recently on human and social factors (Hislop, 2003). Understanding the impact
of these human and social factors on knowledge sharing behavior assists researchers and
practitioners in understanding the question: what makes someone share their knowledge
with others in the firm? These “factors, that motivate people to codify and share
knowledge for the benefit of others, have been identified as a priority area for knowledge
research” (Hall, 2001, p. 140).
Since the mid 1960s, the management literature (e.g. Argyris (1964)) has identified trust
as an important social factor positively linked to the organizational performance. In more
recent literature, trust has been highly regarded as one of the main social factors
facilitating knowledge sharing behavior (Section 2.2). In addition to trust, other research
studies examining knowledge-sharing behavior have identified numerous social or
cognitive factors acting as motivators or inhibitors for knowledge sharing in
organizations. A select few of these social-cognitive factors have been connected to trust
in other studies. However, no study to date has examined the interrelatedness of the three.
For the research presented here, the phenomenon of interest was the knowledge sharing
behavior (KSB) of professionals involved in project-based group work, in a national
professional legal services firm. The study was specifically interested in the significant
direct and indirect relationships that exist between trust, identified social-cognitive
factors, and knowledge sharing behavior, in this setting. The direct and mediating effect
of trust50 was of specific interest, since the literature review identified it as a main factor
50 Trust was a composite variable based on Mayer et al.’s (1995) measure for perceived trustworthiness; operationalized as the sum of ability-based, integrity-based, and benevolence-based trust.
186
to influencing effective knowledge sharing. Also of interest was the collective effect of
social-cognitive factors, together with trust, on knowledge sharing behavior. Social-
cognitive factors tested in the analysis included homophily (gender, age, and education),
shared perspective (shared vision and shared language), tie strength (prior to and while
on the project), and relationship length. Knowledge sharing behaviors included
willingness to share knowledge (overall, explicit and tacit), willingness to use knowledge
(overall, explicit and tacit), and perceived receipt of useful knowledge. The conceptual
framework for the research study is presented in Figure 5.1.
Figure 5.1 Conceptual Framework
187
The overarching research question was: what are the factors that influence knowledge
sharing behavior directly and indirectly through trust? To better answer this research
question, five additional research questions were posed, as follows:
1. What are the significant relationships between social-cognitive variables and
trust?
2. What are the significant relationships between social-cognitive variables and
knowledge sharing behavior?
3. What are the significant relationships between trust and knowledge sharing
behavior?
4. What is the collective effect of the identified social-cognitive variables and trust
on knowledge sharing behavior?
5. Does trust act as a mediating variable between social-cognitive variables and
knowledge sharing behavior?
To answer these research questions, one of Canada’s largest multijurisdictional law firms
was selected as an appropriate site for study. As previously mentioned (Section 3.6) this
firm was selected for the collaboration between knowledge workers on project, its
knowledge intensive environment, and for the nature of the projects they assign.
Before the survey was finalized, a pre-test was conducted with twenty knowledge
workers not affiliated with the sample firm. The survey was then published on the web,
using a private web-based academic survey tool called Qualtrics. Once the survey was
online, it was pre-tested with an additional ten knowledge workers not affiliated with the
sample firm and three senior executives at the firm. Once the pre-tests were complete,
some question wording was adjusted, to match the vernacular of the firm. These changes
were made based on interviews with senior executives at the firm. After the changes were
in place, a senior partner at the firm sent a firm-wide email, asking a sample of
approximately 900 legal professionals and paralegals/law clerks, in six nationally
distributed offices, to participate in the survey. All the employees asked to participate in
the survey were knowledge workers engaged in knowledge-intensive legal project work,
188
the nature of which required a significant reliance on others for both explicit and tacit
forms of knowledge. Of the approximately 900 invitations sent, 775 were distributed to
‘legal professionals’ and 120 to ‘paralegals’ and ‘law clerks’. No administrative staff
received the invitation to participate.
The survey took approximately 15-20 minutes to complete and respondents were assured
that all their responses would remain completely confidential and they would remain
anonymous. The survey was divided into three sections: an individual section that asked
respondents to answer questions about themselves and their background; a co-worker
section that asked respondents to answer questions about a positive referent (i.e. someone
they worked best with on a recent project); and finally a co-worker section that asked
respondents to answer questions about a negative referent (i.e. someone they did not
work well with on a recent project). This approach of getting respondents to comment on
both a positive and negative referent was based on similar distinctions made in previous
research studies (McAllister, 1995; Tsui, 1984, 1986; Holste, 2003). The approach was
also motivated by a conceptual distinction in the type of relationships that occur within
these settings. For example, one of the interesting features of knowledge intensive
organizations is that their employees rarely have free choice in deciding with whom they
work, and with whom they are required to share knowledge. To achieve the project
objectives, in most instances, employees would be required to share knowledge with both
individuals they work well with and those with whom they do not work well.
In total, 275 surveys were completed by employees at this national multijurisdictional
law firm, for a response rate of 30.6%. Factor analysis was used to extract underlying
dimensions. Cronbach’s alpha was then used to assess the reliability of the scales used.
Next, each of the hypotheses was tested using statistical techniques. Initially, correlation
analysis was used to measure the bivariate relationships between the independent and
dependent variables, and then t-tests were used to test differences between the means of
the variables. Multiple regression analysis was used to examine the relationships between
the dependent variables and the focal independent variable, while controlling for all the
other variables in the model. If the results of the regression analysis contradicted the
189
results of the correlation analysis, or t-tests, the results of the regression analysis were
given priority and accepted as MRA controlled for all of the variables in the theoretical
framework of the study and, therefore, represented the more stringent analytical
framework. To test for the mediating effect of trust, hierarchical multiple regression
analysis was used in accordance with a testing method suggested by Baron and Kenny
(1986).
A summary of the research questions, associated hypotheses, and testing results appears
below. The discussion of these research findings is organized by research question and
presented in greater detail in the sections following. The chapter closes with research
contributions, implications for future research, and implications for practice.
RQ1: What are the significant relationships between social-cognitive variables and
trust?
o H1 Age homophily will be positively related to trust
• Not supported for either group
o H3 Educational homophily will be positively related to trust
• Not supported for either group
o H5 Gender homophily will be positively related to trust
• Not supported for either group
o H7 Shared language will be positively related to trust
• Supported for Group 1 and not supported for Group 2
o H9 Shared vision will be positively related to trust
• Supported for both Groups
o H11 Relationship length will be positively related to trust
• Not supported for either group (i.e. partial support of opposite for Group 2)
o H13 Tie strength will be positively related to trust
• Partially supported for Group 1 both prior to and while on the project; not
supported for Group 2 prior to or while on the project
190
RQ2: What are the significant relationships between social-cognitive variables and
knowledge sharing behavior?
o H2 Age homophily will be positively related to knowledge sharing behavior
• Partially supported for Group 1; not supported for Group 2
o H4 Educational homophily will be positively related to knowledge sharing
behavior
• Not supported for either group
o H6 Gender homophily will be positively related to knowledge sharing behavior
• Not supported for either group (i.e. partial support of the opposite in both
groups)
o H8 Shared language will be positively related to knowledge sharing behavior
• Partially supported for both groups
o H10 Shared vision will be positively related to knowledge sharing behavior
• Partially supported for both groups
o H12 Relationship length will be positively related to knowledge sharing behavior
• Not supported for either group
o H14 Tie strength will be positively related to knowledge sharing behavior
• Not supported for Group 1 prior to or while on the project; partially supported
for Group 2 both prior to and while on the project
RQ3: What are the significant relationships between trust and knowledge sharing
behavior?
o H15 Overall trust will be positively related to knowledge sharing behavior
• Partially supported in Group 1; supported in Group 2
RQ4: What is the collective effect of the identified social-cognitive variables and trust
on knowledge sharing behavior?
o H16 Overall trust and social-cognitive factors explain knowledge sharing behavior
• Supported for both groups
191
RQ5: Does trust act as a mediating variable between social-cognitive variables and
knowledge sharing behavior?
o H17 Overall trust will be a mediating variable between social-cognitive factors and
knowledge sharing behavior
• Partially supported for Group 1 (through shared language and shared vision);
partially supported for Group 2 (through shared vision)
5.1 Discussion of the Research Findings
The discussion of the research findings is organized and presented below, by the five sub-
research questions. Each section briefly discusses previous research findings, important
distinctions, notes the expected (hypothesized) results, presents the actual research
findings, and then discusses any divergence with previous findings.
5.1.1 Relationships Between Social-Cognitive Variables and Trust
Research Question 1:
What are the significant relationships between social-cognitive variables and trust?
Hypotheses tested for RQ1:
H1 Age homophily will be positively related to trust (not supported for either group)
H3 Educational homophily will be positively related to trust (not supported for either group)
H5 Gender homophily will be positively related to trust (not supported for either group)
H7 Shared language will be positively related to trust (supported for Group 1 and not
supported for Group 2)
H9 Shared vision will be positively related to trust (supported for both Groups)
H11 Relationship length will be positively related to trust (not supported for either group; with
partial support of opposite for Group 2)
H13 Tie strength will be positively related to trust (partially supported for Group 1 prior to
and while on the project; not supported for Group 2 prior to or while on the project)
To better understand the factors influencing knowledge sharing behavior, directly and
indirectly through trust, the research study began by exploring the significant
relationships between identified social-cognitive factors and trust. As previously
discussed, the construct of trust was based on Mayer et al.’s (1995; Mayer & Davis,
192
1999) proposed model for organizational trust, that includes a measure for the trustor’s
‘propensity to trust’ and that trustor’s ‘perceived trustworthiness’ of the trustee (i.e.
measured as the sum of ability-based, benevolence-based, and integrity-based trust in the
trustee). Initially, the study proposed to measure trust by considering the complete Mayer
et al.’s (1995) model, yet during data analysis it was discovered that the measure of
propensity to trust had low Cronbach’s alpha, which determined its exclusion from any
further analysis of trust. Without the inclusion of a measure for the trustor’s propensity to
trust, the aggregate measure of overall trust became very similar to Mayer et al.’s (1995)
measure for perceived trustworthiness, which captures a set of characteristics (i.e. ability,
benevolence, and integrity) the trustor (i.e. respondent) believed the trustee (i.e. co-
worker) to possess.
After an analysis of the literature on trust and knowledge sharing and an assessment of
the research setting, a total of eight social-cognitive factors were identified as appropriate
for study. These factors were further divided into three categories: homophily, shared
language/shared vision, and relationship length/tie strength. Seven hypotheses51 were
tested to answer the first research question. Each hypothesis predicted a positive
relationship between the social-cognitive factor and trust. A complete discussion of the
implications, relating to the first research question, is presented in the sections below.
5.1.1.1 Homophily and Trust
Based on a literature review of the different types of homophily effects on trust (Section
2.3.1), the research study initially sought to test status-based homophily for five ascribed
characteristics (i.e. age, gender, race/ethnicity, citizenship, and immigration) and two
acquired characteristics (i.e. education and martial status). However, after reviewing the
descriptive statistics from the survey data, it was discovered that birth country,
citizenship, ethnicity, and marital status did not have adequate variability or distribution
across the sample to accurately test for or identify homophily in the sample. Specifically,
85% of those who completed the survey were born in Canada, more than 98% had
Canadian citizenship, almost 87% identified themselves as, at least partially, Caucasian,
51 H1, H3, H5, H7, H9, H11 and H13
193
and over 60% were legally married. The exclusion of these characteristics left the study
exploring the role of homophily with a focus on two ascribed characteristics (i.e. age and
gender) and one acquired characteristic (i.e. education).
Age Homophily and Trust
Based on the existing research examining the impact of age homophily on trust (Section
2.3.1) it was anticipated that higher levels of trust would be found among participants
closer in age. Unexpectedly, this was found to be not true, as data analysis showed that
no significant relationships existed between age homophily and trust, in either group. One
possible explanation for this result may be that many of the previous studies, which
discovered age homophily affecting trusting behavior, were among school acquaintances
and friends, and not among co-workers. Age homophily would be expected in those types
of relationships, as schools traditionally organize ages together, into classes. On the other
hand, findings from Marsden’s (1988) study52 suggested that this type of homophily can
be found in non-educational settings. Notably, respondents in Marsden’s (1988) study
were more likely to have met in schools and social groups, rather than corporate
environments. Social groups and community events, such as schools, may have a
tendency of separating individuals into compatible age groups; at least more so than one
would expect this to occur in corporate environments. Based on this separation, it may be
possible that one’s closest friends, and significant others, with whom they discuss
important matters, would be similar to them in age. On the other hand, in knowledge-
intensive organizational settings, employees are assigned into networks and work
relationships based on their skill, experience, and knowledge domain. In this corporate
environment, age plays little to no role in who will work with whom on a project.
Additionally, in social environments people have free will to befriend and “discuss
important matters” with whomever they please. In corporate environments, like the one in
this study, working relationships are far more rigid, rarely having this same fluidity.
Often, project teams are assembled based on either the position of knowledge workers in
a practice group, on availability, by client request, or by the choice of a senior executive.
52 Marsden’s (1988) study asked respondents to identify those persons with whom they "discussed important matters" within six months prior to the interview. Respondents were then asked more specific questions about the first five people named, including age, education, race/ethnicity, religion, and gender.
194
Considering how the forming and functioning of corporate relationships are different
from those in other social environments, it is reasonable to assume age homophily would
not exist in such environments.
A more recent study (Toh & Srinivas, 2011), examining trust between expatriates and
host-country nationals at MNCs, offers a possible substantiation for this line of reasoning.
Despite expecting to, the authors (2011) found no significant interaction effects between
perceived interpersonal similarity (including observable demographics, such as gender
and ethnicity53) and scores for interpersonal trust. Similar to this study, Toh and Srinivas
(2011) expected to discover (based on the existing literature) that trust in an individual
(the expatriate) by their co-worker (the host-country national) would be moderated by the
extent to which the two parties were interpersonally similar. Since the relationships
studied by Toh and Srinivas (2011) were in a similar professional setting, it is reasonable
to assume that the influence of homophily on trust would not be present in such
environments. Future research should focus on making this environmental distinction.
Gender Homophily and Trust
Based on the existing research (Section 2.3.1), it was anticipated that, in this study,
higher levels of trust would be found among those individuals similar in gender.
Unexpectedly, this was found to be not true, as data analysis showed that no significant
relationships existed between gender homophily and trust, in either group. While
unexpected, this result happens to be consistent with Toh and Srinivas’ (2011) study
mentioned in the previous section. In their research, examining gender as a perceived
interpersonal similarity, Toh and Srinivas (2011) also expected to find trust to be affected
by the extent to which two parties were interpersonally similar. However, like in this
study, they found no such relationship.
One possible explanation for the findings in this study was discussed in the previous
section, and involved inherently preventing or reducing the effect of gender homophily
53 Notably, age was not specifically mentioned in the Toh and Srinivas (2011) study, yet it is reasonable to assume that ten-year age intervals (similar to the ones used in this study) would qualify as a perceived interpersonal similarity or observable demographic.
195
through the practice of composing teams based on domain knowledge; a common
practice in professional settings. Another possible explanation for why this study did not
find gender homophily to have an effect on trust may be found in the nature of the
relationship between the ‘legal professional’ and the clerk (or paralegal). According to
interviews with senior partners and executives at the firm, the ‘legal professionals’ and
law clerks (or paralegals) working in the same practice area (e.g. intellectual property)
frequently interact on numerous matters. As these practice groups work over time on
numerous projects (i.e. matters), they build a rapport and a reliance on each other for
support and information. Therefore, the nature of this working relationship requires close
emotional ties, which must be rooted in trust.
Considering the nature of the relationship between the ‘legal professional’ and
clerk/paralegal, as well as the gender composition of the two groups, it is understandable
why gender homophily was not found in this study. For example, of the 180 ‘legal
professionals’ that completed the survey, there was a fairly even distribution among
gender; 52% were men (93/180) and 48% women (87/180). On the contrary, less than 9%
of the 89 clerks (or paralegals) who completed the survey were men (8/89)and 91%
women (81/89). Women in the sample outnumbered men in the clerk/paralegal role by
over 10 to 1. According to interviews with senior executives, this was also an accurate
representation of the firm’s actual gender distribution by role.
Notably, each legal matter (or project) utilizes a mixture of both ‘legal professionals’ and
clerks/paralegals who assist them. Since clerks and paralegals (who are mostly women)
are distributed across each practice group, and are assigned to each matter, each team was
statistically heterogeneous in composition. This group composition, based on industry
practice, may have helped in preventing the effect of gender homophily on trust.
However, it is worth noting that this group composition may be unique to the legal
setting, which makes it difficult to replicate in other professional firm settings, without a
predominantly female-occupied role assigned to each project.
196
Educational Homophily and Trust
Based on the findings in the literature review (Section 2.3.1) and on the legal setting of
the organization (i.e. the setting required the individuals who work there to have a similar
education in law), it was anticipated that, in this study, higher levels of trust would be
found among those individuals with similar educational levels. Unexpectedly, data
analysis showed no significant relationships between educational homophily and trust, in
either group.
One possible suggestion for why educational homophily was not discovered in this study
could be that respondents relied on, and had faith in, the organization to hire reliable and
skilled co-workers, regardless of the education level they achieved. Respondents may
have had a predisposed expectation of educational competence in their co-worker, based
on the idea that the firm would not hire someone who did not posses the necessary or
adequate skills and abilities to be trusted to do their job. This faith in the organization
was likely built over time, and may be a factor of the history of hiring, the size of the
firm, its vision, culture, or reputation.
Another possible suggestion for why educational homophily was not discovered could,
once again, stem from the relationship between ‘legal professional’ and clerk/paralegal.
For example, almost 96% (175 of 183) of the ‘legal professionals’ who completed the
survey achieved at least one professional degree. In contrast, 93% (83 of 89) of law
clerks/paralegals who completed the survey achieved a 4-year degree or less, with 61%
(54 of 89) of the law clerks/paralegals not completing a 4-year degree (i.e. having either
“some college” or only a high school degree). Despite this large disparity in education,
the frequent interaction between the ‘legal professional’ and law clerk/paralegal on legal
matters (i.e. projects) required a very close tie and working relationship, which, as
previously mentioned, must be rooted in trusting behavior. The make up of the project
teams requires individuals in the firm to trust both those co-workers with similar and with
different levels of education, to successfully complete their work. Based on this
composition of the project teams, it is understandable why educational homophily was
not found.
197
Summary of Homophily and Trust
Contrary to what has been theorized and occasionally found in other research settings,
homophily or similarity in age, gender, and education had no significant effect on a
respondent’s perceived trust in their co-workers, in this setting. Detailed discussions
suggesting why these research findings were not consistent have been presented in the
preceding sections, but generally relate to one or more of three reasons. First, how the
organization assembled project teams or workgroups were different from how they may
have been formed in other settings. This may have prevented or minimized the natural
selection of work partners through homophily. Second, homophily may have not been
found because respondents believed in the organization’s ability to hire domain
knowledgeable, competent individuals. Finally, the role and job description of legal
clerk/paralegal, as well as their relationship with ‘legal professionals’ on each matter (i.e.
project), may inherently have structured each project group to be diverse, with respect to
age, gender, and education. For these reasons, it is believed that no relationship between
trust and homophily were found.
5.1.1.2 Shared Perspective and Trust (Shared Language and Shared Vision)
The constructs of shared language and shared vision were intended to measure the extent
to which a shared perspective exists between respondent and co-worker. More
specifically, the study sought to identify the role that shared perspective played in the
trusting behavior between co-workers in the firm. Understanding this relationship is
important since, given the legal setting of the firm, it is reasonable to assume that when
contextual mismatches occur, trust in the person with whom these occurred (or at the
very least domain or skill specific (ability-based) trust) would decline. Contextual
mismatches and distrust among co-workers could be serious inhibitors to project
outcomes.
Considering the existing research (Section 2.3.2) and the essential need to have a shared
context, language, and vision on legal matters, it was expected that in this study, higher
levels of trust would be found among those individuals similar in shared language and
shared vision. The results with positive referents (i.e. those individuals the respondents
198
worked best with) validated these hypotheses, showing that both shared language and
shared vision were significantly positively related to overall trust, ability-based trust,
integrity-based trust, and benevolence-based trust.
The findings for positive referents were consistent with previous research and may be
justified by arguing that respondents would get more meaningful outcomes from the
interactions with co-workers they share language or vision with, especially if they find
the working relationship to be pleasant. The positive working relationship may also
encourage respondents to interact more frequently. More frequent, meaningful, and
pleasant interactions provided ample opportunities to build rapport and trust between co-
workers. Tsai and Ghoshal (1998) explained this by arguing that “when organization
members have the same perceptions about how to interact with one another, they can
avoid possible misunderstandings in their communications and have more opportunities
to exchange their ideas or resources freely” (p. 467). More frequent ideas and resource
exchanges provided ideal opportunities to develop trusting relationships.
For negative referents (i.e. the group of individuals respondents did not work well with),
shared vision was also significantly positively related to overall trust, ability-based trust,
integrity-based trust, and benevolence-based trust; however, no significant relationships
were found between shared language and any type of trust. This suggested that shared
language no longer had an effect on trust, when the respondent deemed the working
relationship between them and co-worker as poor. This was an interesting finding, since
it suggested that having a shared language might not be as important predictor of trust,
especially in lieu of the working relationship. One possible way to explain this result may
be to argue that, unlike shared vision, having a common language is a necessary
prerequisite for working in a legal practice and on a legal team. By virtue of inclusion in
the work group, each respondent may have assumed that their co-workers shared some
degree of common language with them; which did not guarantee they shared a common
goal, and certainly did not guarantee they trusted each other, especially when considering
the effect of the negative working relationship. Based on these findings, having a shared
language was not enough to affect trust, when respondents had to overcome a negative
199
working relationship. This finding also suggested the need for further research into the
effect of working relationship on trusting behavior between co-workers.
On the other hand, higher shared vision between respondents and co-workers led to
higher trust in all sampled co-workers, regardless of the working relationship. One
possible explanation for this may have to do with the nature of the legal environment. For
example, the firm’s legal practice-groups begin with organizationally driven, clearly
explicated, shared goals (in terms of legal outcomes) and have a common understanding
of the main concerns and purpose of each matter they work on. Sharing the firm’s overall
vision, as well as the groups’ vision, for the outcomes of the matter, seemed to provide an
excellent platform on which trust could be built. Shared vision also turned out to be a
much stronger predictor of trust than shared language, as it can be witnessed by the
minimal effect from the change in working relationship.
Summary of Shared Perspective and Trust
As expected, the results with positive referents showed a positive relationship between
shared perspective and trust; where higher amounts of both shared language and shared
vision led to a higher trust, by the respondent, in the co-worker they worked best with.
The results from data collected with negative referents showed slightly different results
than those in the first group, offering some interesting perspectives on the role the
working relationship has on shared perspective and trust, in this study. For example, the
results from both groups suggested that the effect shared vision had on trust transcended
the nature of the working relationship between co-workers, reflecting its importance to
trust. Also, the results indicated that sharing a common language with a co-worker was
not necessarily enough to facilitate trust in that person, and that the nature of the working
relationship between the two co-workers may play an important role. This can partially
be explained by arguing that a common language can often be expected on a legal project
and does not guarantee the establishment of a common goal or trust.
200
5.1.1.3 Structural Characteristics and Trust (Tie Strength and Relationship Length)
Based on the literature review (Sections 2.3.3 and 2.3.4), this research study sought to
test the effect of two structural factors on trust: tie strength and relationship length.
Relationship length was measured by asking respondents how long they knew each of the
two co-workers they mentally selected. Tie strength was indicated by a 4-item measure,
where two items indicated the closeness of the relationship and the other two indicated
interaction frequency. In addition, based on a perception that tie strength prior to the
project was conceptually different from tie strength while on the project and thus might
have a different effect on trust, each of the four tie strength items appeared twice on the
survey; once asking the respondent about their relationship with their two co-workers
prior to the project and once again to measure the same relationship while on the project.
Tie Strength and Trust
Based on the literature review (Section 2.3.3) it was expected that, in this study, higher
levels of trust would be found among those co-workers who had higher tie strength
between them. This was found to be not entirely true, as the results from positive
referents suggested that there was not a significant relationship between overall trust and
tie strength prior to the project. However, a positive relationship was discovered between
benevolence-based trust and tie strength, prior to the project. This suggested that higher
tie strength with positive referents, prior to the project, was driven by a belief that the co-
worker cared about the respondent’s interests and wanted to do well by the respondent.
That is, if one believed that their co-worker cared about their interests and wanted to do
well by them, they would tend to interact more with this individual and feel more
emotional closeness towards them.
Interestingly, stronger tie strength (prior to) had no effect on trust in the co-
worker’s domain specific skills and competencies (ability-based trust), or adherence to a
set of principles or values acceptable to the respondent (integrity-based trust). This
finding suggested reconsidering how respondents in the study assessed the competence of
co-workers. Perhaps, they assessed their co-workers’ competence with respect to the
outcomes of the interaction, rather than the interaction itself or an emotional intensity
201
felt. Stronger tie strength did not necessarily lead to a belief in the competence of co-
workers, because one interaction may have been all that was required. If a respondent
needed some particular knowledge, they may have gone to the co-worker, gotten the
knowledge, and found that it was good knowledge, without the need for additional
emotional intensity or more frequent interaction. Knowing it was good, useful knowledge
may have been all that was needed for trust to occur.
Stronger tie strength, while on the matter or project, was positively related to overall
trust, integrity-based trust, and benevolence-based trust in positive referent co-workers.
These results were similar to those found by Levin and Cross (2004), however, ability or
competence based trust was found to not be affected. As briefly mentioned above, there
may have been several reasons why co-workers interacted on a project. For example, the
projects may have involved multiple types of knowledge and only limited interactions
may have been needed to share this knowledge. However, respondents may have also
needed to interact to get additional support for their analysis and their opinions; to decide
how to manage and proceed with the project; or to express concerns and explore risks.
Reasonably, these types of interactions, and the associated emotional intensity and
frequent interaction, seemed more likely to take place with individuals in whom
respondents had benevolence-based or integrity-based trust (vs. ability-based).
In this study, tie strength had the greatest influence on benevolence-based trust, both
prior to and while on the project. Tie strength had the least influence on competence or
ability-based trust; showing no relationships prior to or while on the project. This seemed
to be in line with what has been discussed above. Just because domain knowledge needed
to be shared, it did not mean that it took much emotional intensity or interaction to share
it. Also, strong ties may have related to interactions that had a purpose other than for
knowledge sharing.
The effect on integrity-based trust changed between the two types of tie strength, where
no significant effect was found with tie strength prior to the project and one was found
with tie strength while on the project. In this case, a common set of principles and values
202
acceptable to the respondent (integrity-based trust) led to more frequent interactions and
higher emotional intensity between the co-workers. These results pointed to the fact that
these interactions may have had a purpose other than the transference of skills or
competences (e.g. support). These types of interactions would be far more important and
common in problem solving while on projects, and would rarely take place prior to the
specifics of a project are known.
No significant relationships between tie strength (prior to or while on the project) and
trust were found, with negative referents. This result was interesting, since it pointed out
that the nature of the working relationship clearly played a role in affecting the
relationship between trust and tie strength. One possible reason for this can be found in
associating benevolence / integrity-based trust and working relationship. Perhaps, there is
inherently a connection between thinking a co-worker would “do good” to the trustor
(benevolence-based trust) and positive working relationship. The same may be asked for
whether the co-worker’s set of principles are acceptable to the trustor (integrity-based
trust), since both were found significant with positive referents. Alternatively,
respondents may have interacted with negative referents because they were required to do
so, and not because they chose to. Respondents may have minimized interactions with
this group and, as a result, did not have the opportunity to assess, in the interaction,
whether the co-worker was trustworthy. Any interaction may have been purely
instrumental, just to get the task done, but not to ask any extraneous questions.
Relationship Length and Trust
Based on the literature review (Section 2.3.4) it was expected that, in this study, longer
relationship length would relate to higher levels of trust, by the respondent, in the co-
worker. This was found not to be the case for the first group, where the results of the data
analysis revealed that relationship length was not significantly related to any type of
trust. More interestingly, with negative referents, relationship length was found to be
negatively related to overall trust, integrity-based trust, and benevolence-based trust,
which was the opposite of what was hypothesized; highlighting an interesting distinction
203
between the two groups. As with tie strength, ability-based trust was not significantly
related, perhaps for similar reasons.
One possible explanation for the opposite effect found between relationship length and
trust with negative referents can stem from the nature of the corporate relationship. In a
corporate setting, employees participate in mandated, involuntary work groups, where
members do not always have the freedom to choose co-workers, or terminate a poor
working relationship. With positive referents, the involuntary nature of the working
relationship would not be nearly as important factor, since the respondents had a good
working relationship. With negative referents, however, the relationship can be both
involuntary and unpleasant, yet still needed to exist, for the project work to get done. If
the negative referent relationships were completely voluntary, it would be unlikely that a
respondent would continue to work and develop a relationship under conditions where
they felt they did not work well with the co-worker.
Ongoing involuntary relationships with negative referents may explain why the opposite
effect of relationship length on benevolence and integrity-based trust was found.
Benevolence-based trust is driven by a care for the interest of the co-worker and a want to
do good towards them. In poor relationships, it would be easy to understand why a
respondent would not have any interest in doing good towards a co-worker with whom
they do not work well. Alternatively, the respondents may have convinced themselves,
over time, that their principles, values, and beliefs (i.e. drivers for integrity-based trust)
are different from those of negative referent co-workers and that, perhaps, it was even
those perceived differences that caused the poor working relationship. Interestingly, with
negative referents, ability-based trust, which is driven by domain specific skills and
competencies, was found not to be negatively significant, suggesting that it was not a lack
of trust in the skills of the individual that decayed trust in this group.
Summary of Structural Variables and Trust
Tie strength and relationship length were analyzed for relationships between structural
characteristics and trust between co-workers. A holistic view of the data from both
204
groups revealed three interesting insights into these relationships. First, in certain
situations, structural characteristics had no significant effect on trust (i.e. tie strength in
Group 2 or relationship length in Group 1). Second, all structural characteristics
measured in the study had an influence only on integrity-based and benevolence-based
trust. Finally, and perhaps most interestingly, no structural characteristic variable had any
effect on ability-based trust.
5.1.2 Research Question 1 Summary
The first research question asked “what are the significant relationships between social-
cognitive variables and trust?” As there were several social-cognitive variables tested in
the study, the answer to this research question varied based on social-cognitive variables.
First, the findings showed there to be no relationships between any types of homophily
and trust. Second, the results showed partial support for a relationship between shared
perspective and trust; with shared vision being fully supported in both groups and shared
language depending on the nature of the working relationship (i.e. was found for positive
referents only). Finally, the results showed the structural variables to have a specific
effect on certain types of trust; where no effect was found on ability-based trust and a
significant effect was found for both integrity and benevolence-based trust. The nature of
the working relationship also played a role in how the structural variables had an impact
on trust.
5.1.3 Relationships Between Social-Cognitive Variables and Knowledge Sharing
Behavior
Research Question 2:
What are the significant relationships between social-cognitive variables and knowledge
sharing behavior?
Hypotheses Tested for RQ2:
H2 Age homophily will be positively related to knowledge sharing behavior
H4 Educational homophily will be positively related to knowledge sharing behavior
H6 Gender homophily will be positively related to knowledge sharing behavior
H8 Shared language will be positively related to knowledge sharing behavior
H10 Shared vision will be positively related to knowledge sharing behavior
205
H12 Relationship length will be positively related to knowledge sharing behavior
H14 Tie strength will be positively related to knowledge sharing behavior
The second research question explored the significant relationships between identified
social-cognitive factors and knowledge sharing behavior. Knowledge sharing behavior
(KSB) was operationalized as three combined variables measuring willingness to share
knowledge, willingness to use knowledge, and perceived receipt of useful knowledge. The
eight social-cognitive factors tested in the first research question were used, once again,
for the second: homophily, shared perspective (shared language and shared vision), and
structural factors (relationship length and tie strength). Seven hypotheses54 were tested to
answer the second research question, exploring the existence of a positive relationship
between the social-cognitive factors and knowledge sharing behavior.
A complete discussion of the implications relating to this question is presented in the
sections below.
5.1.3.1 Homophily and Knowledge Sharing Behavior
As with trust, certain homophily measures (i.e. birth country, ethnicity, and martial
status) were excluded from the analysis, due to bias in the data (see Section 5.1.1.1 for
details). The exclusion of these factors left the study exploring the role of homophily on
KSBs, based on two ascribed characteristics (age and gender) and one acquired
characteristic (education).
Age Homophily and Knowledge Sharing Behavior
Given the findings in the existing research (Section 2.4.1) on age homophily and KSB, it
was anticipated that, in this study, higher levels of knowledge sharing behavior55 would
be found among those individuals closer in age. Unexpectedly, this was found to be not
true, as data analysis showed that no significant relationships existed, in either group,
between age homophily and willingness to share knowledge, or between age homophily
and perceived receipt of useful knowledge. In addition, there was no significant
54 H2, H4, H6, H8, H10, H12 and H14 55 KSB is defined using three separate composite variables: sharing, use, and perceived usefulness
206
relationship between age homophily and overall willingness to use knowledge, in either
group. However, there was a significant positive relationship between age homophily and
willingness to use tacit knowledge, with positive referents (none with negative referents).
In Section 5.1.1.1, it was suggested that age homophily was not found in this study
because of the difference between how project groups are structured in corporate
environments and how they may be structured in educational and social environments.
Based on firm practice (or industry-based methodologies), each legal project team’s
composition (on each legal matter) almost always consists of a mix of senior executives,
junior executives, and support staff of varying ages. Therefore, one explanation for why
no relationships were found between age homophily and most knowledge sharing
behaviors could be due to the diverse representation of age groups on each legal matter.
Further, it could be argued that legal knowledge rapidly changes and, hence, age may not
be a good proxy for the quality of knowledge possessed by any employee. Thus,
employees on the project team did not tend to deem age as a reliable basis for either
knowledge sharing behavior, or for considering that the knowledge shared was useful.
There was, however, one noteworthy significant relationship, suggesting that respondents
were more likely to use the tacit knowledge supplied to them by someone they worked
well with, who was closer to them in age. One possible suggestion for this anomaly, from
the rest of the age homophily findings, may be related to the career stage of the
employee. Connelly and Kelloway (2003) argued that an employee’s “age and career
stage may affect their knowledge sharing behaviors through the size and utility of their
social networks; [where] experienced employees may simply be more able to share their
knowledge because they know more of the right people in the organization” (p. 297). In
this setting, it would seem that the longer an employee was at the firm (i.e. the older they
are), the more they used knowledge from others who have been there longer (who,
themselves, are older in age). Using Connelly and Kelloway’s (2003) reasoning, a
younger legal professional may simply not know where to turn to acquire useable tacit
knowledge, because they have an inadequate or ineffectual network. On the other side,
the senior legal professionals may not fully comprehend what tacit knowledge the junior
207
executives possess, which causes them to rely on their existing long-standing knowledge
contacts. Alternatively, these senior executive respondents may simply not want to risk
using non-explicit forms of knowledge from less experienced professionals.
Another possible reason age homophily may have been associated with the sharing of
tacit knowledge could be that the quality and usefulness of tacit legal knowledge needed
considerable time to mature and develop, and, hence, was more likely to be shared
between those of similar age and, presumably, experience. It could be postulated that this
tacit knowledge was generally easier to share between those individuals who had some
level of shared experience, either limited or extensive. Alternatively, no relationship was
found with regards to explicit knowledge, because it was much easier to develop and
share among co-workers with varying degrees of experience.
Gender Homophily and Knowledge Sharing Behavior
Based on the previous research (Section 2.4.1), it was anticipated that, in this study,
higher levels of knowledge sharing behavior would be found among those individuals
similar in gender. Unexpectedly, this was found to be not entirely true, as data analysis
only found small negative relationships between gender homophily and knowledge
sharing behavior. With positive referents, there were no relationships found between
gender homophily and willingness to share knowledge, overall willingness to use
knowledge, or perceived receipt of useful knowledge. However, a negative relationship
was found between willingness to use explicit knowledge and gender similarity (i.e. no
relationship was found with willingness to use tacit knowledge). This finding suggested
that respondents were more willing to use explicit knowledge from those positive referent
co-workers different from them in gender.
The results with positive referents seemed to further suggest that gender homophily did
not exist in the firm, possibly due to an industry practice of how groups are assembled for
projects. As previously mentioned in the gender homophily and trust (Section 5.1.1.1), a
significant reason for why gender homophily was not found may have been due to the
composition of legal project teams. Each team was comprised of ‘legal professionals’
208
supported by paralegals and legal clerks. However, the gender distribution of legal
professionals was far more heterogeneous (52% men / 48% women) than that of the
paralegals (9% men / 91% women), yet the two still heavily relied on each other and
actively worked together to bring legal matters to a close. The one significant negative
relationship with positive referents, between gender homophily and willingness to use
explicit knowledge, was likely the result of legal professionals relying on
clerks/paralegals, of the opposite gender, for their supporting legal documents (i.e.
explicit knowledge). This should come as no surprise, since, according to one senior
partner, a significant portion of the paralegal/clerk job description required them to gather
and organize more explicit forms of knowledge. Interestingly, this same relationship was
not present with negative referents, suggesting that respondents preferred to rely on
paralegals or clerks they worked well with, for their useable explicit knowledge.
With negative referents, there were no significant relationships between gender
homophily and willingness to share or use knowledge. There was, however, a negative
relationship between gender homophily and the perceived receipt of useful knowledge.
This finding suggested that respondents perceived knowledge from negative referents,
different from them in gender, as being more useful than knowledge from the co-workers
of the same gender. A reason for this relationship was not entirely clear, since the same
relationship was not found between respondents and positive referents. It might be
possible that there was an interference effect, if the respondents wanted to indicate that
gender was not the basis of their failure to work well with another individual. However,
the results did point to the need to conduct more research in this area and further explore
these potential relationships. As suggested by Connelly and Kelloway (2003), “the
impact of gender on knowledge sharing in organizations has thus far not received much
attention from academics who study knowledge sharing” (p. 300).
Educational Homophily and Knowledge Sharing Behavior
Based on previous research studies (Section 2.4.1) and on the legal environment (i.e.
participants that have education in the area of law), it was anticipated that, in this study,
higher levels of knowledge sharing behavior would be found among those individuals
209
who had similar educational achievements. Unexpectedly, this was found to be not true,
as data analysis showed no significant relationships between educational homophily and
knowledge sharing behavior, in either group.
Both possible suggestions, for why relationships between knowledge sharing behaviors
and educational homophily were not discovered in this study, have been previously
discussed (Section 5.1.1.1). To summarize, first, respondents may have had a belief that
the organization hired skilled co-workers, regardless of the education they achieved. This
confidence in the reliability of the organization’s hiring policies, combined with a display
of skill or competence, by the co-worker, may have been all that was needed for both
parties to be willing to participate in the knowledge exchange. The second, more likely,
reason for why educational homophily was not found, was the significant difference in
educational levels between ‘legal professionals’ and clerks/ paralegals. As previously
mentioned, the education level was, on average, much higher for a ‘legal professional’
than a law clerk/paralegal; yet these two groups interacted extensively on each legal
project, and heavily relied on each other for information, which was vital to the success
of the project.
Summary of Homophily and Knowledge Sharing Behavior
Contrary to what has been theorized and found in other research settings, homophily, or
similarity in age, gender, and education, had very little significant effect on a
respondent’s knowledge sharing behaviors, in this study. Interestingly, two of the three
significant effects discovered showed reverse gender homophily to take place (i.e.
willingness to use knowledge from positive referents and perceived receipt of useful
knowledge from negative referents), which was the opposite of what was expected and
found in previous research. As previously mentioned, the results suggested a need to
conduct more research in this area, to further explore these potential relationships. The
single positive homophilous relationship found between age homophily and willingness
to use tacit knowledge was explained by postulating that tacit knowledge was generally
easier to share between those individuals who had a level of shared experience and age.
210
5.1.3.2 Shared Perspective and Knowledge Sharing Behavior
Shared perspective was defined for the second research question in the same manner as it
was for the first research question. As such, shared perspective was measured using two
separate constructs: shared language and shared vision. A complete discussion of their
effect on knowledge sharing behavior is discussed in the sections below.
Shared Language and Knowledge Sharing Behavior
Considering both the existing literature (Section 2.4.2) and the assumption that shared
language was likely to be an important precondition for consideration of any important
legal matter, it was expected that, in this study, higher levels of knowledge sharing
behavior would be found among those individuals who shared a language. The results
with positive referents partially supported the hypothesis, showing that shared language
was significantly positively related to a respondent’s willingness to share knowledge
(overall, explicit and tacit) but not significantly related to a respondent’s willingness to
use knowledge, or a respondent’s perception that the knowledge they received from
positive referents was useful.
Several studies (Section 2.4.2) have found similar positive connections between
willingness to share knowledge (i.e. knowledge sharing behavior) and shared language.
Using these studies, one might extract various reasons explaining the effect of shared
language on willingness to share knowledge. For example, shared language may have
provided reassurance for the respondent that their time would not be wasted, since they
shared a domain with the co-worker (Tsoukas & Vladimirou, 2001; Nonaka, 1994) or
that the co-worker would be able to better interpret the knowledge (Henderson, 2005;
Zenger & Lawrence, 1989). In a legal setting, it may also be possible for shared language
to be instrumental in providing access to knowledge sharing opportunities. For example,
shared language may have been a necessary prerequisite for knowledge sharing to take
place, especially within a specific legal practice area (e.g. intellectual property, maritime
law, etc.). Respondents may have been less willing to share knowledge with those outside
of their practice group (i.e. those with whom they did not share a language), because they
211
felt it was a waste of time, or simply because they lacked access (which would be
consistent with the findings of Nahapiet and Ghoshal (1998)).
However, a reason for the lack of relationship between shared language and willingness
to use knowledge was unclear. Perhaps one reason may be elucidated from the work of
Hoe and McShane (2002), who found shared vision to be a strong predictor for informal
knowledge sharing, but not at all associated with informal knowledge acquisition (a
construct similar in nature to willingness to use knowledge). The authors suggested that
this may have been because “informal knowledge acquisition is a more passive or natural
activity whereas informal knowledge dissemination requires more active motivation
guided by shared vision” (p. 289). Extending this idea to shared language, one may argue
that informal knowledge sharing behavior, guided by shared language, may also require
more active motivation (as compared to knowledge use behavior).
The absence of a relationship between shared language and perceived receipt of useful
knowledge is difficult to explain, since it may be expected that the respondents felt that
shared language was a necessary prerequisite for inclusion in a legal practice, on a legal
team, or on a legal matter. Assuming this was the case, each respondent may have gone
into new projects feeling that their co-workers shared a common language with them, by
virtue of inclusion on the project. However, as evidenced by the results, sharing a
language was not a precondition to knowledge sharing, in this setting. Further research
would be required to explore this relationship in more detail.
For the most part, results from negative referents mimicked those from positive referents,
with a notable exception: shared language was no longer related to overall willingness to
share knowledge and willingness to share tacit knowledge. There was, however, a
significant positive relationship found between shared language and willingness to share
explicit knowledge. This result suggested that even when respondents felt there was not a
good working relationship between them and their co-worker, they were still willing to
share explicit forms of knowledge, if the two shared a language. This would confirm
earlier speculations as to the effect of participation in a domain specific legal practice or
212
matter. More interestingly, for poor working relationships, respondents were no longer
willing to share tacit knowledge, suggesting that they were no longer interested in
investing the extra time and effort needed to transfer tacit knowledge (Nonaka, 1994).
The results in this group suggested that shared language may not be a sufficient
precondition for the sharing of knowledge with negative referents, especially tacit
knowledge.
Shared Vision and Knowledge Sharing Behavior
Based on the literature review (Section 2.4.2), it was expected that, in this study, shared
vision would have a positive effect on knowledge sharing behavior. As predicted, with
positive referents, significant positive relationships were found between shared vision
and overall willingness to share knowledge, overall willingness to use knowledge, and
perceived receipt of useful knowledge. These findings were consistent with those made
by previous authors, and again highlighted the important role shared vision played in the
knowledge transfer process. Building on previous studies, this research aimed to capture
a more comprehensive understanding of the influence of shared vision on knowledge
sharing behavior, by distinguishing between the behaviors of a knowledge transmitter and
receiver (i.e. a willingness to share vs. a willingness to use). This research also
considered knowledge type (explicit vs. tacit) and perceived usefulness of knowledge
shared.
With respect to knowledge type, the results showed that, with positive referents, there
were significant relationships only between shared vision and willingness to share and
use tacit knowledge. In fact, when shared vision was analyzed with willingness to share
explicit knowledge and willingness to use explicit knowledge, no significant relationships
were found. This suggested that, with positive referents, shared vision had an important
influence on willingness to transfer (i.e. share and use) tacit knowledge. This was an
important finding, since tacit knowledge is routinely known as being difficult to transfer.
The results showed that the sharing of values or goals may be necessary pre-conditions to
the transfer of tacit knowledge. This finding also built on the work of Chiu, Hsu, and
Wang (2006), in demonstrating that having a shared vision among co-workers led to
213
more meaningful, quality exchanges between them (i.e. the transference of tacit
knowledge).
With negative referents, shared vision was also positively related to willingness to share
knowledge (overall, explicit and tacit) and overall willingness to use knowledge. In
addition, shared vision had a positive influence on willingness to use tacit knowledge;
willingness to use explicit knowledge was still found to be not significant. These findings
with negative referents, for the most part, confirmed those with positive referents and
perhaps highlighted an even more complete effect of shared vision on knowledge sharing
behavior with the influence on willingness to share explicit knowledge.
There was another interesting result with respect to the relationship between shared
vision and perceived receipt of useful knowledge. Of all the variables tested in the first
group, shared vision had the strongest effect on perceived receipt of useful knowledge.
However, in the second group, shared vision was found not to have any significant effect
on perceived receipt of useful knowledge. One possible suggestion for this difference
between the groups may have been because respondents could not get past the nature of
the poor working relationship, and generally saw a majority of knowledge from negative
referents as not being useful, even if the two co-workers happened to share a common
vision. Alternatively, this may have been influenced by the significant relationships
found between shared vision and willingness to share explicit knowledge. For example,
with positive referents, only tacit knowledge was found to be significant, which
respondents may have perceived as being more useful than explicit knowledge. However,
in the second group, shared vision was found to also influence willingness to share
explicit knowledge, which respondents may have viewed as being routine or necessary to
the project, more than useful to the outcomes of the matter. Further research would be
warranted to study the usefulness of tacit and explicit knowledge.
214
Summary of Shared Perspective and Knowledge Sharing Behavior (Shared Language
and Shared Vision)
Overall, having a shared perspective had a positive effect on knowledge sharing behavior
in the firm. As expected, the results showed a positive relationship between shared
perspective and willingness to share knowledge, in both groups; where higher degrees of
shared language or shared vision led to higher willingness by the respondent to share
knowledge with their co-worker, regardless of the working relationship. The results also
showed an expected positive influence of shared vision on willingness to use knowledge,
in both groups. Surprisingly, no significant relationships were found between shared
language and willingness to use knowledge, in either group.
Perhaps the most interesting finding was the effect of shared vision on perceived receipt
of useful knowledge. The results, with positive referents, suggested that having a shared
vision led to a significantly higher perception that the knowledge received from that co-
worker was useful. However, shared vision did not remain a significant influence with
negative referents, where the working relationship was deemed as poor. This suggested a
need to explore the effect of working relationships further, in subsequent research.
Finally, shared language was not found to influence perceived receipt of useful
knowledge in either group.
5.1.3.3 Structural Characteristics and Knowledge Sharing Behavior (Tie Strength and
Relationship Length)
The structurally oriented social-cognitive factors tested for this research question were
the same as the ones used in the first research question (i.e. tie strength and relationship
length). As with the first research question, tie strength was further analyzed by
separating it into two constructs: tie strength between co-workers prior to the project and
tie strength between co-workers while on the project. A complete discussion of the effect
of structural factors on knowledge sharing behavior is in the sections below.
215
Tie Strength and Knowledge Sharing Behavior
Based on the literature review (Section 2.4.3), it was difficult to predict the relationships
between tie strength and knowledge sharing behavior, as previous studies have found
contradicting results. However, after considering the setting of this study and fairly
similar settings in previous research, it was expected that tie strength would be positively
related to knowledge sharing behavior. As previously mentioned, tie strength (prior to
the project) was analyzed with each of the knowledge sharing behaviors independently
from tie strength (while on the project).
Tie Strength (Prior to the Project)
The results of the analysis, with positive referents, showed that no significant
relationships were found between tie strength (prior to the project) and any knowledge
sharing behaviors. Thus, for this group, tie strength (prior to the project) was not
positively related to knowledge sharing behavior. The findings, in this group, were
contrary to prior research, since they showed neither weak nor strong ties to have a
significant effect on knowledge sharing behavior. The results, for this group, were also
counter to the work of Levin and Cross (2004), who found that tie strength, mediated by
trust, positively affected knowledge sharing behavior. One possible reason for the results
in this group may have been that, in this professional setting, knowledge sharing did not
depend on the development of close relationships, or on the frequency of interaction, but
on the characteristics of the project itself. Knowledge sharing may have been expected to
take place, as it was an appropriate part of the routine work process, and not because of
prior or present ties between co-workers. For example, the extent of interaction may have
depended on the nature of knowledge exchanged. Just because legal projects were
knowledge-intensive, it did not mean that there necessarily was intensive interaction
between the members of the project. The results, for this group, also suggested that other
factors influenced knowledge sharing behavior, with positive referents, and that the
number of previous interactions or feelings of closeness did not play a role.
The previous suggestion was also supported by the findings, with negative referents,
where stronger tie strength, prior to the project, had a positive influence on knowledge
216
sharing behavior. Specifically, in this group, stronger ties, prior to the project, led to
higher overall willingness to share knowledge on the project. This result was peculiar,
even if it was consistent with previous research, since it suggested that strong ties
developed in prior work encounters superseded a negative current working relationship,
to influence knowledge sharing behavior. On the other hand, the results made sense, since
respondents may be more willing to share their knowledge with those negative referents
co-workers with whom they had more prior interaction, or had more prior feelings of
closeness towards. In these cases, the respondents would be more familiar with these co-
workers and could share their knowledge with less effort, or time, invested in working
with someone they felt they did not work well with.
With negative referents, tie strength (prior to the project) was also found to significantly
influence willingness to share tacit knowledge, yet had no effect on willingness to share
explicit knowledge. These findings were consistent with prior research, which also found
strong ties to positively influence tacit knowledge sharing. One possible reason for this
may have been because, as previously discussed, sharing explicit knowledge was
considered a part of normal work routines. This is further reinforced by the fact that the
group members on a particular project likely had very similar backgrounds, with respect
to the interpretation and codification of explicit knowledge. The same cannot be said of
tacit knowledge.
In the same group, stronger ties, prior to the project, also led to more effective knowledge
sharing behavior, in the form of higher willingness to use knowledge (overall, explicit
and tacit). Like in the previous section, strong ties developed on prior work encounters
had more effect on knowledge sharing behavior than the nature of the working
relationship between the co-workers. As previously mentioned, this can be explained by
suggesting that the respondents were more familiar with these co-workers, than those
they had less contact with and, hence, would more easily (or quickly) be able to use
knowledge from them. Finally, stronger ties, prior to the project, had no effect on the
perception that the knowledge received was useful, in either group.
217
Tie Strength (While on the Project)
Unlike prior to the project, stronger ties (while on the project) were found to not have any
effect on a respondent’s willingness to share or willingness to use knowledge. However,
with negative referents, stronger ties, while on the project, led respondents to perceiving
knowledge from these co-workers as being more useful. Even though the same
relationship was not found with positive referents, this result with negative referents was
consistent with the literature and made logical sense, since the existence of a strong tie
may have given respondents more confidence in the knowledge that their co-worker
provided. Strong ties may also have led to more emotional attachment, which may have
influenced the confidence a respondent had in the negative referent co-worker’s
knowledge.
Summary of Tie Strength and Knowledge Sharing Behavior
The research produced a number of noteworthy findings from examining the
relationships between tie strength and each of the knowledge sharing behaviors with
negative referents, but found no significant relationships with positive referents. Making
the distinction between positive and negative referents turned out to be valuable in this
study, since significant relationships were only found with negative referents. With
negative referents, relationships were found between tie strength prior to the project and
willingness to share and use knowledge; and between tie strength while on the project
and perceived receipt of useful knowledge. Some suggestions have been made above to
explain why this relationship varied, but further research would be needed to explore this
relationship in more detail.
Finally, the results confirmed that tie strength (prior to the project) and tie strength
(while on the project) were separate constructs; validating their inclusion as control
variables. The two constructs had different effects on knowledge sharing behavior, even
in the same group (i.e. as witnessed by the results with negative referents). More research
is needed to fully explore these differences, but future studies should consider the
importance of making this distinction by controlling for the two variables, along with
218
considering the nature of the working relationship (i.e. by incorporating a positive and
negative referent group).
Relationship Length and Knowledge Sharing Behavior
Even though very few research studies were found that tested the effect of relationship
length on knowledge sharing behavior, it was believed that, based on the existing studies
with similar constructs (Section 2.4.4), relationship length would be positively related to
knowledge sharing behavior. However, the results from both groups suggested that
relationship length had no effect on any knowledge sharing behavior. This result was
interesting, since it suggested that respondents were willing to share and use explicit and
tacit knowledge both with co-workers they knew a long time and co-workers they only
knew a short time. In addition, respondents did not perceive the knowledge received from
co-workers they knew a long time as being any more useful than the knowledge received
from co-workers they knew only a short time. One possible reason for this may have to
do with project group composition. Group composition was primarily determined by the
firm, who assigned workers to a legal practice-group or legal matter, based on their
domain knowledge, skill, and availability, and not how long they may have known their
co-workers. The resulting legal teams still had to share and use each other’s knowledge,
to successfully accomplish their work. Such a scenario could have resulted in effective
knowledge sharing behaviors with varying relationship lengths between co-workers,
explaining the lack of relationships found.
Summary of Structural Factors and Knowledge Sharing Behavior (Tie Strength and
Relationship Length)
Overall, the structural factors tested in this study had minimal effect on knowledge
sharing behavior, especially after considering the nature of the working relationship. For
example, relationship length had no significant effect on knowledge sharing behavior, in
either group, and tie strength only had a significant positive effect with negative
referents. Tie strength had no effect on any knowledge sharing behavior with positive
referents. With negative referents, tie strength prior to the project had a positive effect on
both willingness to share knowledge and willingness to use knowledge with that co-
219
worker. In the same group, tie strength, while on the project, led to a higher perception
that the knowledge received was useful. These findings suggested that tie strength may
have had only a limited effect on knowledge sharing behavior, and was useful only in
relationships with negative referents. Notably, the results did suggest the importance of
considering both tie strength prior to the project and tie strength while on the project, as
their relationships with knowledge sharing behaviors varied. The same can be said for
highlighting the importance of studying a respondent’s relationship with both positive
and negative referents.
5.1.4 Research Question 2 Summary
The second research question asked “what are the significant relationships between
social-cognitive variables and knowledge sharing behavior?” The results of the study
showed there to be a weak partial support for a relationship between homophily (i.e. age
and gender) and willingness to use knowledge, but only with positive referents. The
results also showed a partial support for a positive relationship between shared language
and willingness to share knowledge, in both groups. Next, support was found for a
positive relationship between shared vision and all of the knowledge sharing behaviors,
with positive referents. With negative referents, shared vision had a positive influence on
willingness to share knowledge and willingness to use knowledge. There was no
evidence of any relationships between relationship length and any knowledge sharing
behavior, in either group. There was also no evidence of any relationship between tie
strength (prior to or while on the project) and any knowledge sharing behavior, with
positive referents. However, with negative referents, the results showed there to be
positive relationships between tie strength and all the knowledge sharing behaviors.
5.1.5 Relationships Between Trust and Knowledge Sharing Behavior
Research Question 3:
What are the significant relationships between trust and knowledge sharing behavior?
Hypothesis Tested for RQ3:
H15 Overall trust will be positively related to knowledge sharing behavior
220
The third research question explored the significant relationships between trust and
knowledge sharing behavior. Knowledge sharing behavior was operationalized in the
same manner as it was in the second research question (i.e. measuring willingness to
share knowledge, willingness to use knowledge, and perceived receipt of useful
knowledge). Similarly, the measure of trust was operationalized in the same manner as it
was for the first research question56. One hypothesis exploring the existence of a positive
effect of overall trust on knowledge sharing behavior was tested to answer the third
research question.
Based on the evidence previously discussed in the literature review on trust and
knowledge sharing behavior (Section 2.2.2), it was expected that if respondents had
higher overall trust in co-workers, then they would also have higher willingness to share
knowledge with them, higher willingness to use knowledge from them, and a higher
perception that knowledge received from them had useful outcomes.
Trust and Willingness to Share Knowledge
The results of the study found that overall trust had a positive influence on overall
willingness to share knowledge, with both positive and negative referents. These findings
were similar to other research that looked at relationships between various types of trust
and willingness to share knowledge (Holste, 2003; Tsai & Ghoshal, 1998; Hinds &
Pfeffer, 2003; Van den Hooff & Van Weenen, 2004; Lin, 2007; Hislop, 2003). These
results were also similar to other research that found trust to have a positive effect on
knowledge sharing behaviors (Renzl, 2008; Ho, Kuo, Lin, & Lin, 2010; Ho, Kuo, & Lin,
2011) and activities (Andrews & Delahaye, 2000).
For positive referents, trust had a positive influence on willingness to share tacit
knowledge, but provided no benefit for sharing explicit or codified knowledge (i.e. no
significant relationships were found). These results were similar to those found in
previous research (e.g. Levin & Cross, 2004; Yang & Farn, 2009; Lin, 2007).
56 The measure of overall trust was based on Mayer et al.’s (1995) measure for perceived trustworthiness, which is a set of three trust related characteristics (i.e. ability, benevolence, and integrity) the trustor (i.e. respondent) believed the trustee (i.e. co-worker) to possess.
221
Interestingly, these results contrasted those of Holste (2003), who found trust to also have
a significant positive effect on a willingness to share both explicit and tacit knowledge,
with positive referents.
One possible explanation for the findings with positive referents may be that respondents
had a higher base level of confidence that the knowledge shared would be used
appropriately by these co-workers, which was independent of trust. With a higher level of
confidence in the positive referent’s ability to internalize the knowledge, there may have
been less need for trust to exist between the individuals. Willingness to share explicit
knowledge may also be considered routine work practice between these individuals, as
respondents would not choose to avoid interacting with positive referents, like they may
with negative ones. This explanation was further justified by the results with negative
referents, where trust had a positive influence on both types of knowledge, with a slightly
stronger influence on willingness to share explicit knowledge, than on willingness to
share tacit knowledge57. In this group, the respondent’s negative feeling towards this co-
worker may have caused them to have less confidence that shared knowledge would be
used appropriately, making trust more important for both forms of knowledge.
Trust and Willingness to Use Knowledge
The results of the study also found that overall trust had a positive influence on overall
willingness to use knowledge, from both positive and negative referents. These findings
were similar to other research, which looked at relationships between various types of
trust and constructs similar to willingness to use knowledge (Holste, 2003; Andrews &
Delahaye, 2000; Tsai & Ghoshal, 1998; Zand, 1972; McAllister, 1995; Bromiley &
Cummings, 1995). Trust also had a noticeably stronger positive effect on overall
willingness to use knowledge, than it did on overall willingness to share knowledge, in
both groups. In addition, the results showed trust to have a positive influence on
willingness to use both explicit and tacit knowledge, with a slightly stronger influence on
tacit knowledge, in both groups. Based on Holste’s (2003) work, this was the expected
result for negative referents, but not for positive ones, where he found trust to have a 57 Holste (2003) found trust to have a greater effect on willingness to share tacit knowledge than on willingness to share explicit knowledge in this group
222
moderately greater impact on willingness to use tacit knowledge than on willingness to
use explicit knowledge.
In this study, one possible explanation for the relationships between trust and willingness
to use knowledge may have to do with the level of risk associated with each form of
knowledge. For example, in a legal environment, it may be less risky to use someone’s
tacit knowledge, than his or her explicit knowledge. This can be a challenge, since the
transfer and use of explicit knowledge can be considered quite routine in this
environment. For example, explicit knowledge is more likely to be directly used, and
reused, in project decisions and deliverables. Explicit knowledge is also more likely to be
used by clients, managers, and partners, in evaluating the projects and employees. When
knowledge is made explicit, it is written and made formal, forcing the employee to put
the knowledge “on the record”, which carries accountability and social risk. The
respondent inherently takes a risk by using a co-worker’s explicit knowledge, because
they must ultimately be accountable for the knowledge used. There is less risk that tacit
knowledge would be shown as being used inappropriately, because tacit knowledge is
more difficult to represent in formal legal project deliverables.
Trust and Perceived Reciept of Useful Knowledge
The results from this study found trust to have a positive influence on perceived receipt
of useful knowledge, in both groups. In fact, the influence of trust on perceived receipt of
useful knowledge was the strongest of all the knowledge sharing behaviors tested in the
study. Between the two groups, trust had a significantly stronger impact on perceived
receipt of useful knowledge from negative referents; however, this relationship was
significant in both groups. The finding in this study was similar to the one discovered by
Levin and Cross (2004), although their hypotheses did not set out to test the direct
relationship between trust and perceived receipt of useful knowledge.
As with willingness to use knowledge, one possible explanation for this result may also
involve a respondent’s assessment of risk, which is known as an integral part of
organizational trust (Mayer et al., 1995; 1999). For example, higher levels of trust, by a
223
respondent, in their co-worker, may have reduced feelings of risk associated with the
quality of knowledge shared (Mayer et al., 1995). A reduction of this risk, associated with
the quality of knowledge, increased the respondent confidence in the co-workers
knowledge. Together, an increased confidence and decreased risk in the co-worker’s
knowledge, would expectedly raise perceptions that the knowledge received was useful to
the respondent, their project, and the firm.
5.1.6 Research Question 3 Summary
Overall, as expected, trust had a significant positive influence on the combined
knowledge sharing behaviors in the study (i.e. overall willingness to share, overall
willingness to use, and perceived receipt of useful knowledge). Of the knowledge sharing
behaviors, trust had the weakest influence on overall willingness to share knowledge,
with a similar weak influence on willingness to share tacit knowledge, in both groups.
The influence of trust on willingness to share explicit knowledge was only significant for
negative referents. Trust was also found to have a strong positive influence on overall
willingness to use knowledge, willingness to use explicit knowledge, and willingness to
use tacit knowledge. Finally, of all the knowledge sharing behaviors, trust had the largest
positive influence on perceived receipt of useful knowledge, which was significant for
both groups, but about 50% stronger in relationships with negative referents, versus those
with positive referents.
5.1.7 Collective Effect of Social-Cognitive Factors and Trust on Knowledge Sharing
Behavior
Research Question 4:
What is the collective effect of the identified social-cognitive variables and trust on
knowledge sharing behavior?
Hypotheses Tested for RQ4:
H16 Overall trust and social-cognitive factors explain knowledge sharing behavior
The fourth research question explored the collective effect of overall trust and social-
cognitive factors on each of the knowledge sharing behaviors. Knowledge sharing
behavior, overall trust, and each of the social-cognitive factors were operationalized in
224
the same manner as they were in the previous research questions. One hypothesis
explored how overall trust and social-cognitive factors explain knowledge sharing
behavior.
It was difficult to predict the resulting effect of the analysis for this research question
based on previous research, since no known organizational study examined the collective
effects of trust, and the specific social-cognitive factors identified in this study, on
knowledge sharing behavior. Also, no other known study has looked at the effect of this
number of social-cognitive factors and trust on knowledge sharing behavior. The research
discussed in the literature review has often identified trust and numerous interrelated
social-cognitive factors as motivators and inhibitors of knowledge sharing behavior, but
have not examined more than few of these constructs collectively.
For the purposes of presenting and discussing the findings relating to this research
question and hypothesis, a social capital framework was adopted. The factors in the study
were organized in an analytical and conceptual way, similar to the one used by Nahapiet
and Ghoshal (1998) in their classification of social capital components or dimensions.
Since the factors used in this study were conceptually similar to the ones used by
Nahapiet and Ghoshal (1998), they could be classified using the same three theoretical
categories: relational (i.e. trust and homophily), cognitive (i.e. shared language and
shared vision), and structural (i.e. tie strength and relationship length). Importantly, this
representation was not conceptually different from the original theoretical model of this
study (Figure 5.1). This representation was only a slightly different way of organizing
and presenting the results, for the purpose of discussion. It was recognized in this
research, as it was by Nahapiet and Ghoshal (1998) that although these three categories
were separated analytically, many of their features may be argued to be highly
interrelated.
Figure 5.2 (Positive Referent Group) and 5.3 (Negative Referent Group) are visual
representations of the findings of this study, with respect to the collective effect of trust
and social-cognitive factors on knowledge sharing behavior. The sections following the
225
two figures offer discussion of the fourth research question, elaborating on each specific
knowledge sharing behavior (i.e. willingness to share knowledge, willingness to use
knowledge, and perceived receipt of useful knowledge). The discussion of this research
question closes with a more holistic view of the factors and their effect on knowledge
sharing behavior, as a whole.
Figure 5.2 Collective Effect of Factors on Knowledge Sharing Behavior for Positive
Referents (Group 1) (*p < .05, **p < .01, ***p < .001)
226
Figure 5.3 Collective Effect of Factors on Knowledge Sharing Behavior for Negative
Referents (Group 2) (*p < .05, **p < .01, ***p < .001)
Collective Effect of Social-Cognitive Factors and Trust on Willingness to Share
Knowledge
In this study, trust, shared language, and shared vision were found to be the most
important factors to influence willingness to share knowledge, with tie strength (prior to
the project) displaying a minimal effect with negative referents.
Beginning with the relational dimension (Figures 5.2 and 5.3), the findings showed that
the effect of trust on overall willingness to share knowledge was significant and of
relatively similar strength, for both groups. These findings confirmed previous research
studies discussed in the literature review (Section 2.2.2). Specifically, with positive
referents, higher trust had a positive effect on overall willingness to share knowledge and
willingness to share tacit knowledge, with little to no effect on explicit knowledge
sharing. In the second group, trust had a positive effect on overall willingness to share
227
knowledge, willingness to share explicit knowledge, and willingness to share tacit
knowledge. Homophily, was not found to have any effect on willingness to share
knowledge of any type, in either group.
The cognitive dimension (Figures 5.2 and 5.3) was found to have the strongest effect on
willingness to share knowledge, in both groups. Specifically, with positive referents,
shared language had a positive effect on overall willingness to share knowledge,
willingness to share explicit knowledge, and willingness to share tacit knowledge. With
negative referents, shared language was not as important, with no influence on overall
willingness to share knowledge or willingness to share tacit knowledge, and only a small
influence on willingness to share explicit knowledge.
Shared vision had a stronger influence on willingness to share knowledge, than overall
trust. However, as with overall trust, with positive referents, shared vision had a positive
influence on overall willingness to share knowledge and willingness to share tacit
knowledge, with no effect on explicit knowledge sharing. With negative referents, shared
vision had a positive influence on overall willingness to share knowledge, willingness to
share explicit knowledge, and willingness to share tacit knowledge. Notably, the effect on
willingness to share tacit knowledge was almost twice the effect on willingness to share
explicit knowledge (Figure 5.3). This result suggests that having a shared vision is
generally important in the sharing of tacit knowledge.
With respect to the structural dimension (Figures 5.2 and 5.3), the effect on willingness to
share knowledge was minimal. For example, neither tie strength nor relationship length
were found to have any influence on willingness to share knowledge, with positive
referents. However, with negative referents, tie strength prior to the project was found to
have a small, yet significant, influence on overall willingness to share knowledge and
willingness to share tacit knowledge, with no effect on explicit knowledge. This result
suggested that one possible method of encouraging respondents to share their tacit
knowledge with negative referents may be to create opportunities for prior interactions,
as well as possibilities to develop feelings of closeness. Finally, relationship length was
228
not found to have any significant influence on any form of willingness to share
knowledge, with negative referents.
Collective Effect of Social-Cognitive Factors and Trust on Willingness to Use Knowledge
In this study, trust and shared vision were found to have the greatest influence on
willingness to use knowledge, in both groups; with tie strength (prior to the project)
showing a small positive effect, in relationships with negative referents (Figures 5.2 and
5.3). A couple of additional small significant relationships were found between
homophily and willingness to use knowledge, but only for relationships with positive
referents.
Trust was the single most important factor influencing willingness to use knowledge in
this study. In both groups, overall trust was found to have a positive influence on overall
willingness to use knowledge, willingness to use explicit knowledge, and willingness to
use tacit knowledge. These findings showed the importance of trust to the receiving end
of the knowledge sharing process. The findings also showed that respondents would be
equally willing to use knowledge from co-workers they had both good and bad working
relationships with, as long as the respondents trusted them. In other words, the nature of
the working relationships in this study had no effect on the influence of trust on
willingness to share knowledge.
Shared vision was the second most influential factor on willingness to use knowledge.
Specifically, in both groups, shared vision had a positive influence on overall willingness
to use knowledge and willingness to use tacit knowledge, with no effect on willingness to
use explicit knowledge. This was an important finding, since it, once again, showed
shared vision to be a key factor in influencing effective knowledge sharing behavior,
especially for the difficult and time-consuming transfer of tacit knowledge.
A few other factors were found in the study to have a small effect on willingness to use
knowledge. First, age homophily was found to have a small positive effect on willingness
to use tacit knowledge, with positive referents. Next, gender homophily was found to
229
have a small negative effect (heterogeneous effect) on willingness to use explicit
knowledge from positive referents. Finally, a small positive effect was found with
negative referents between tie strength (prior to the project) and overall willingness to
use knowledge, willingness to use explicit knowledge, and willingness to use tacit
knowledge. These findings showed small effects, but were interesting because they
seemed to depend on the nature of the working relationship. For example, each of the
homophily effects were only found with positive referents, having no effect with negative
referents. On the other hand, tie strength (prior to the project) had no effect with positive
referents, but impacted willingness to use both types of knowledge with negative
referents. Prior interactions or feelings of closeness may have provided confidence in the
negative referent’s knowledge, or minimized feelings of risk associated with using the
knowledge. Notably, these same interactions were not required with positive referents.
These findings demonstrated the role working relationship played in impacting
knowledge sharing behavior, and made a case for the inclusion of this distinction in
future studies. Future studies are also needed to understand the impact of working
relationship on willingness to use knowledge, in alternative organizational settings.
Collective Effect of Social-Cognitive Factors and Trust on Perceived Receipt of Useful
Knowledge
As with the other two knowledge sharing behaviors, overall trust proved to play a strong
role in influencing perceived receipt of useful knowledge. In fact, of all the factors tested,
overall trust had the strongest effect on perceived receipt of useful knowledge, and was
the only factor to show a significant positive effect in both groups (Figures 5.2 and 5.3).
Interestingly, the effect of overall trust on perceived receipt of useful knowledge was
considerably stronger with negative referents, than it was with positive. This strong effect
from overall trust warranted its consideration, as a central factor, in influencing the
effective transfer of knowledge in the form of useful outcomes. This was especially true
for conditions where the working relationship between the co-workers was negative.
The other factors found to influence perceived receipt of useful knowledge seemed to
highly depend on the nature of the working relationship. For example, with positive
230
referents, shared vision had a strong significant effect on perceived receipt of useful
knowledge. However, with negative referents, shared vision had no effect. Next, tie
strength (while on the project) had a small positive effect on perceived receipt of useful
knowledge from negative referents, which was not found to exist with positive referents.
Finally, there was a small reverse effect of gender homophily (or gender heterogeneous
effect) on perceived receipt of useful knowledge from negative referents, but none from
positive referents. Some explanations have already been suggested as to why these results
have occurred, although additional research would be needed to explore the unique effect
of working relationships on perceived receipt of useful knowledge in alternative
organizational settings.
5.1.8 Research Question 4 Summary
Adopting a more holistic view of knowledge sharing behavior, the results revealed
overall trust to be the single most important factor to influence knowledge sharing
behavior. Overall trust had a positive effect on all three knowledge sharing behaviors, in
both groups, and was the single most important factor for influencing willingness to use
knowledge and perceived receipt of useful knowledge. The effect of overall trust on
willingness to share knowledge was also significant, but less than one third that of overall
trust on the other knowledge sharing behaviors.
Shared vision was found to be the second most important single factor influencing
effective knowledge sharing behavior; showing a strong positive influence on all the
knowledge sharing behaviors tested in both groups, with only one exception (i.e. having a
shared vision had no influence on the respondent’s perception that the knowledge
received from negative referents was useful). Shared vision was also the single most
important factor found to influence a respondent’s willingness to share knowledge.
Having a shared language was more important, between the respondent and the positive
referent, in influencing effective knowledge sharing behavior. This was especially the
case with the effect of shared language on the respondent’s willingness to share
231
knowledge. However, sharing a language with negative referents did not seem to have
much of an influence58 on effective knowledge sharing behavior.
Tie strength was quite statistically significant between the respondent and the negative
referent; and not at all statistically significant with the positive referent. In fact, between
the respondent and the negative referent, tie strength was one of only two factors (trust
being the other) to have a positive effect on all the tested knowledge sharing behaviors.
On the contrary, relationship length was not found to have any influence on any
knowledge sharing behavior, in either group.
Age and gender homophily were found to have a small effect (positive and negative) on
knowledge sharing behavior, although these results were relatively weak and not
consistent across the three behaviors and two groups, to claim that homophily played a
significant role in effective knowledge sharing behavior. Numerous reasons have been
suggested in this chapter for why some of these results have occurred and, in some cases,
additional research was suggested, because the legal setting may have not been an
appropriate industry for examining status homophily.
5.1.9 Mediating Effects of Trust Between Social-Cognitive Factors and Knowledge
Sharing Behavior
Research Question 5:
Does trust act as a mediating variable between social-cognitive variables and knowledge
sharing behavior?
Hypothesis Tested for RQ5:
H17 Overall trust will be will be a mediating variable between social-cognitive factors and
knowledge sharing behavior
The fifth research question in the study explored the mediating effect of trust between
identified social-cognitive factors and knowledge sharing behavior. Knowledge sharing
behavior and overall trust were operationalized in the same way as they were in the
58 There was a small positive effect (β = .146) of shared language on willingness to share explicit forms of knowledge with the individuals the respondent did not work well with.
232
previous research questions. A total of eight social-cognitive factors were tested for this
research question. These were also the same ones used in the previous research questions
(i.e. 3 x homophily, shared language, shared vision, relationship length and 2 x tie
strength). To answer this research question, one hypothesis was tested, which explored
the existence and strength of the mediating effect of overall trust between social-
cognitive factors and knowledge sharing behavior.
Based on the limited research studies available (Sections 2.2.2 and 2.4.4), it was expected
that trust would, at least, have a partial mediating effect on some relationships between
social-cognitive factors and knowledge sharing behaviors. However, it was difficult to
predict the presence or the strength of the mediation effects between specific social-
cognitive factors and knowledge sharing behaviors, because previous studies did not use
similar measures, or include as many independent variables in their analysis. Predicting
the effect of working relationships was also difficult, as the only previous study that
examined trust and knowledge sharing behavior for positive and negative referents
(Holste, 2003) did not measure any other variables similar to the social-cognitive factors
tested in this study. For these reasons, strategies to identify the mediating effects of
overall trust, in the study, were mostly exploratory in nature.
To identify and test the mediating effects that overall trust had on the relationships
between social-cognitive factors and knowledge sharing behavior, logic developed by
Baron and Kenny (1986) was applied. As first step, the direct effect of each social-
cognitive factor was tested against the mediating variable (overall trust) and each of the
dependent variables (knowledge sharing behaviors). To qualify for inclusion, the social-
cognitive factor needed to have a significant relationship with both overall trust and at
least one knowledge sharing behavior. As it can be seen in Figure 5.4, for positive
referents, two social-cognitive factors met this criterion (i.e. shared language and shared
vision), and for negative referents the conditions were only met by one social-cognitive
factor (i.e. shared vision).
233
Figure 5.4 Summary of the Overall Mediating Effects of Trust for Positive Referents
(Group 1, diagram on left) and Negative Referents (Group 2, diagram on right) (*p < .05, **p < .01, ***p < .001)
Mediating Effect of Trust Between Shared Language and Knowledge Sharing Behavior
Numerous researchers have suggested that co-workers who share a language have greater
trust in each other (Section 2.3.2) and are more effective in their knowledge sharing
behavior (Section 2.4.2). However, no research has specifically explored the mediating
effect of trust, or the potential interactions among the three concepts. The results of this
study found that, with positive referents, overall trust had a partial mediating effect
between shared language and overall willingness to share knowledge. These results
suggested that, overall trust helped explain the effect of shared language on knowledge
sharing behavior between individuals and the co-workers they worked well with. In other
words, trust partially took the place of shared language in increasing willingness to share
knowledge. Overall trust was also found to have a partial mediating effect between
shared language and willingness to share explicit knowledge, and between shared
language and willingness to share tacit knowledge.
Interestingly, overall trust was found to not have a mediating effect between shared
language and overall willingness to use knowledge. However, overall trust did have a
complete or full mediating effect between shared language and willingness to use explicit
knowledge. This result suggested that overall trust helped explain the effect of shared
language on willingness to use explicit knowledge from positive referents. These results
reaffirm the previous suggestion made in Section 5.1.5 that explicit knowledge may have
234
more visible associated risk in the legal setting. A similar relationship was not found for
willingness to use tacit knowledge.
Finally, overall trust was found to have a complete mediating effect between shared
language and perceived receipt of useful knowledge, from positive referents. This result
suggested that when trust was higher among co-workers who worked well together, there
was less need for them to share a language, in order to perceive their co-worker’s
knowledge as being useful. This result also suggested that trust helped to explain the
effect of shared language on perceived usefulness of knowledge received. Practically,
this finding highlights the important role trust plays in perceived knowledge outcomes in
the firm.
One possible explanation for these results is that trust affected the extent to which
respondents were forthcoming about their lack of shared language with co-workers. By
doing this, they brought attention to the need to bridge these language barriers,
effectively creating a condition for successful knowledge sharing behavior. This would,
to some extent, explain the partial mediating effect of trust between shared language and
willingness to share knowledge, and the complete mediating effect of trust between
shared language and perceived receipt of useful knowledge. However, it does not explain
the results with willingness to use knowledge, where it was unclear why overall trust was
found to not have a mediating effect for relationships between shared language and
overall willingness to use knowledge, and between shared language and willingness to
share tacit knowledge. One possible explanation for the mediating effect of trust between
shared language and willingness to use explicit knowledge was previously alluded to, and
may have to do with the inherent risk associated with explicit knowledge, in this setting.
Further research would be needed to explore these types of relationships, in alternative
settings.
One interesting point worth noting was how the nature of the working relationship played
a significant role in affecting overall trust, as a mediating variable. Specifically, overall
trust was only found to have a mediating effect between shared language and knowledge
235
sharing behaviors, with positive referents, with no significant effects found with negative
referents. The primary reason for this was because no direct relationships were found
between shared language and trust, in this group. Without the existence of a direct
relationship, mediation could not exist. Earlier, it was suggested that having a shared
language may have been a necessary prerequisite for working on a legal project and, by
virtue of inclusion in the work group, each respondent may have assumed that their co-
worker shared some degree of language with them. This, however, did not guarantee that
the two co-workers trusted each other, or that trust would play any role. On the other
hand, there were significant relationships found between shared language and knowledge
sharing behaviors, in this group. This could have been because respondents only involved
negative referents in knowledge sharing behaviors, when the two shared a language, to
simplify and quicken the exchange (e.g. as it was a negative one and respondents did not
want to have extended interaction with someone they did not work well with). Additional
research is needed to further explore the effect of working relationship.
Mediating Effect of Trust Between Shared Vision and Knowledge Sharing Behavior
As with shared language, a number of previous researchers have suggested that co-
workers who share a vision have greater trust in each other (Section 2.3.2) and are more
effective in their knowledge sharing behavior (Section 2.4.2). However, no researchers
have specifically explored the mediating effects of trust between shared vision and
knowledge sharing behavior. Interestingly, unlike with shared language, the results from
this study showed the nature of the working relationship to not matter with shared vision
(i.e. a mediating effect was found in both test groups).
The results of the study found that overall trust had a partial mediating effect between
shared vision and overall willingness to share knowledge, with both positive and negative
referents. This result suggested that overall trust helped explain the effect of shared
vision on willingness to share knowledge with both positive and negative referents.
Overall trust was also found to have a complete mediating effect for the relationship
between shared vision and willingness to share explicit knowledge, with positive
referents, and a partial mediating effect for the relationship between shared vision and
236
willingness to share explicit knowledge, with negative referents. Next, overall trust was
found to have a partial mediating effect for the relationship between shared vision and
willingness to share tacit knowledge, in both groups.
Overall trust was also found to have a partial mediating effect between shared vision and
overall willingness to use knowledge, in both groups. Next, overall trust was found to
have a complete mediating effect between shared vision and willingness to use explicit
knowledge, in both groups. In addition, overall trust was found to have a partial
mediating effect for the relationship between shared vision and willingness to use tacit
knowledge, in both groups. Overall trust had the greater mediating effect on willingness
to use explicit knowledge and the effect on both types of knowledge was noteworthy.
This result suggested that overall trust helped explain the effect of shared vision on
willingness to use knowledge from both positive and negative referents.
As with shared language, one possible explanation for the above findings may have been
that trust shaped the extent to which respondents were forthcoming about their lack of
shared vision, creating a condition (i.e. willingness or need) for knowledge sharing
behavior. In other words, with higher trust, there was more transparency for the need to
create opportunities to establish a unified vision. These opportunities to build a shared
vision manifested themselves in the form of increased willingness to share and use
knowledge. In either case, the effect of trust on willingness to share and use knowledge,
in lieu of shared vision, really exhibited its role in organizational knowledge sharing
behavior. This was especially true, because similar mediating effects were found in both
groups (i.e. with both positive and negative referents).
Finally, overall trust was found to have a partial mediating effect between shared vision
and perceived receipt of useful knowledge, with positive referents. Interestingly, overall
trust was found to not have any mediating effects between shared vision and perceived
receipt of useful knowledge, with negative referents. Earlier in the chapter it was
suggested that one possible reason for this may have been that respondents could not get
past the nature of the poor working relationships and generally saw a majority of
237
knowledge from negative referents as not being useful, even if the two co-workers
happened to share a common vision.
5.1.10 Research Question 5 Summary
Discovering the mediating effects of overall trust between social-cognitive variables and
knowledge sharing behaviors was largely an exploratory exercise, as many of these
relationships have not been investigated in prior research studies. Interestingly, in one
study by Levin and Cross (2004), where similar variables were used, different results
were discovered59.
Of all the relationships tested, overall trust had the strongest mediating effect between
shared vision and willingness to use knowledge. Only slightly weaker was the mediating
effect of overall trust between shared vision and willingness to share knowledge.
Interestingly, both these effects did not depend on the nature of the working relationships,
and were found to occur in both test groups. In both groups, when there was less trust,
respondents used shared vision to decide whether they were willing to participate in
knowledge sharing behaviors with their co-worker.
Overall trust was also found to have a partial mediating effect on the relationship
between shared vision and perceived receipt of useful knowledge, and full mediating
effect between shared language and perceived receipt of useful knowledge. In both these
cases, the mediating effect was only found with positive referents. When respondents did
not trust these co-workers, they used shared language or shared vision as basis for
deciding how useful the positive referent’s knowledge was to the respondent, the project,
or the firm. Interestingly, this same mediating effect was not found with shared language
or shared vision, with negative referents. Even though possible explanations for the lack
of relationships found with negative referents were given above, these results were
counter-intuitive and worthy of further study. Finally, overall trust was found to have a
partial mediating effect between shared language and willingness to share knowledge,
59 Levin and Cross (2004), who controlled for homophily and other knowledge related factors, found trust to mediate a relationship between tie strength and perceived receipt of useful knowledge; a finding not discovered in either group of this study.
238
and between shared language and willingness to use explicit knowledge; yet again only
with positive referents. The lack of significant results with negative referents was, again,
somewhat counter intuitive and worthy of further study.
With respect to knowledge type, overall trust was found to have a complete mediating
effect on three of the four explicit knowledge sharing behaviors relationships tested, with
a partial mediating effect on the fourth. On the contrary, overall trust only had a partial
mediating effect on three of the four tacit knowledge sharing behaviors tested, with no
effect on the fourth. These results suggested that trust had a slightly stronger mediating
effect on behaviors involving the sharing and using of explicit knowledge, versus those
behaviors involving the sharing and using of tacit knowledge. This may be due to the
proposed associated risk involved with explicit knowledge at the firm. Trust also seemed
more likely to have a mediating effect with positive referents, as opposed to negative
ones, but further research would be required to specifically explore why this was the case.
5.2 Discussion of the Main Findings and Research Contributions
The study set out to better understand knowledge sharing behavior in the context of
knowledge intensive organizational project group work. Of specific interest was the
investigation of the extent to which social-cognitive factors influence knowledge sharing
behavior directly and indirectly, through trust. Also of interest are the direct effects trust
has on knowledge sharing behavior and the collective effects of social-cognitive factors
and trust on knowledge sharing behavior. The following are the main findings and
contributions of this study.
1. Shared vision, shared language, and tie strength were found to have the greatest
influence on a respondent’s trust in their co-worker.
Among all the variables examined in the study, the three factors that were found to
have the strongest effect on respondent’s trust in their co-workers were shared vision,
shared language, and tie strength. Of these three variables, shared vision had the
greatest effect on trust, and was found to influence the perceived trustworthiness of
239
both positive (i.e. those co-workers the respondent worked best with) and negative
(i.e. those co-workers the respondent did not work well with) referents. This finding
reflect the importance of creating shared goals, shared concerns, and shared purpose
among co-workers working on a project together, to establish conditions where trust
can thrive (Levin, Whitener, & Cross, 2006).
Shared language and tie strength were also found to increase trust in a co-worker, but
only in cases where the working relationship between respondent and co-worker was
positive. Creating a shared language entails building an environment where co-
workers are able to easily understand each other, convey information between each
other, and come to agreement (Levin, Whitener, & Cross, 2006).
2. Trust and shared vision were found to have the greatest influence on a respondent’s
knowledge sharing behaviors.
Among all the variables examined in the study, the two factors found to have the
strongest effect on organizational knowledge sharing behavior were trust and shared
vision. Trust had the greatest influence on knowledge sharing behavior, and was
found to positively influence all three knowledge sharing behaviors tested in the
study, with both positive and negative referents. Shared vision also had a strong
positive influence on all the knowledge sharing behaviors with positive referents and
with two of the three knowledge sharing behaviors with negative referents.
It is also worth mentioning that shared language was found to have a positive
influence on two of the knowledge sharing behaviors, with positive referents; and tie
strength developed prior to the project influenced two of the knowledge sharing
behaviors, with negative referents. Developing tie strength required the two co-
workers to have prior social interaction and experience feelings of “closeness”
towards each other (Levin & Cross, 2004; Marsden & Campbell, 1984).
240
3. Trust was found to have a mediating effect between shared vision and knowledge
sharing behavior, as well as between shared language and knowledge sharing
behavior.
This study was one of only few to measure the mediating effect between the
identified social-cognitive factors and knowledge sharing behaviors. The previous
studies that have examined this effect have not done so with this number of social-
cognitive factors, or with such a comprehensive conceptualization of knowledge
sharing behavior. This study found that, when perceived trustworthiness was high in
the referent co-worker, there was less need for the respondent and the co-worker to
have a shared vision. Alternatively, when there was less trust between co-workers,
shared vision was required for them to participate in effective knowledge sharing
behavior with each other. Trust also had a mediating effect between shared language
and knowledge sharing behavior, but this was only found in relationships between
respondents and positive referents.
4. Knowledge sharing behavior was influenced by a combination of trust and one or
more social-cognitive factors.
The collective models revealed that each identified knowledge sharing behavior was
influenced by a combination of trust and one or more of the identified social-
cognitive factors for both positive and negative referents. A fairly large proportion of
knowledge sharing behaviors could be explained by these factors in the first group.
However, in the second group, only a small number were explained. For example,
with positive referents the social-cognitive factors explained more of the knowledge
sharing behaviors, than trust on its own. On the other hand, with negative referents,
the collective model showed that social-cognitive factors made a slightly lower or
equal contribution to that of trust alone, in explaining knowledge sharing behavior.
5. In addition to the empirical research contributions, this study made a number of
methodological contributions in the form of conceptualizations for knowledge sharing
241
behavior, trust, and tie strength. Also, it has provided a more nuanced and focused
analysis, by factoring for knowledge type and co-worker working relationship.
First, this study adopted a more comprehensive conceptualization of knowledge
sharing behavior than in previous knowledge sharing research, one that includes
characteristics of both the knowledge sender and knowledge receiver. In
conceptualizing knowledge sharing behavior, this study also considered the perceived
impact of the shared knowledge with respect to usefulness to the individual, the
project, and the firm.
The study also considered explicit and tacit forms of knowledge, a distinction only a
few prior studies have made when examining knowledge sharing behavior. This was
an important distinction, since the ease with which one may be able to transfer the
two types of knowledge may greatly differ (Nonaka & Takeuchi, 1995). Further, the
factors that motivate a worker to share one type of knowledge may not be the factors
that motivate the transfer of the other (as the results of the study have shown). The
same could be said for the distinction made between tie strength prior to the project
and tie strength while on the project, the inclusion of which was motivated in this
study by the perception that they could be conceptually differentiated. For example,
the frequency with which two co-workers interacted could significantly vary, based
on the project requirements. Thus, it would also be reasonable to assume that the
closeness two co-workers felt towards each other could change, as they continue to
work together and interact.
Since earlier studies have utilized 207 different psychometric trust measures of
organizational trust (McEvily & Tortoriello, 2011), it was important for this study to
adopt a conceptually sound measure for organizational trust, in order to set precedent
for future studies, and to compare the results with previous studies. Following the
advice of McEvily and Tortoriello (2011), who conducted the most complete analysis
of trust measures, this research adopted Mayer, Davis, and Schoorman’s (1995)
model for organizational trust.
242
Finally, very few studies have made the distinction between positive and negative
working relationships. This was an important distinction, since it enhanced the
richness and authenticity of this study, by highlighting the different influence
independent variables had on dependent variables based on the nature of the working
relationship. To highlight these, a summary of the unique relationships is provided in
Table 5.1. In addition, taking this approach provided an opportunity for testing the
identified factors and relationships in two theoretical test groups, in one study.
Independent
Variable Dependent Variable Positive
Referents Negative Referents
Age Homophily Willingness to Use Tacit Knowledge X Gender Homophily Willingness to Use Explicit Knowledge X Gender Homophily Perceived Receipt of Useful Knowledge X Shared Language All types of Trust X Shared Language Overall Willingness to Share and
Willingness to Share Tacit knowledge X
Shared Vision Perceived Receipt of Useful Knowledge X Shared Vision Willingness to Share Explicit Knowledge X Relationship Length
Overall, Integrity, and Benevolence-Based Trust
X
Tie Strength-Prior Benevolence-Based Trust X Tie Strength-While on
Overall, Integrity, and Benevolence-Based Trust
X
Tie Strength-Prior Overall Willingness to Share, Willingness to Share Tacit, Overall Willingness to Use, Willingness to Use Explicit, Willingness to Use Tacit
X
Tie Strength-While Perceived Receipt of Useful Knowledge X Overall Trust Willingness to Share Explicit Knowledge X
Table 5.1 Significant Relationships Unique to the Nature of the Working Relationship Employees are rarely given a choice in the selection of co-workers. Hence, they
sometimes work with others with whom they have not previously had a positive working
relationship. Although organizations may wish to build project teams of individuals who
have previously worked well together, this may be neither possible nor optimal.
243
5.3 Limitations of the Study
There are a few limitations of the study that should be acknowledged. First, the study
may have limited external validity, as it only surveyed knowledge workers in one
organization, within one industry. All survey respondents worked on legal matters in the
role of either a legal professional or paralegal/law clerk. As it was previously suggested,
in some cases, the legal industry practices and routines may have influenced the make up
and outcomes of the working relationships, differently from what one might see in other
organizational settings. Examples of this may be the risk associated with sharing and
using explicit knowledge, or the unique relationship between legal professional and law
clerk/paralegal. It is also noteworthy that 87% of respondents identified themselves, at
least partially, as Caucasian and 98% Canadian in citizenship. Even so, this is not
believed to affect the relationship between trust, social-cognitive factors, and knowledge
sharing behavior. Future research should explore these relationships in other
organizational contexts.
The study was also limited in that it only measured self-reported data (i.e. expressed
willingness and perceived usefulness) for knowledge sharing behavior, and not actual
exchange of knowledge, actual employee sharing behaviors, or actual project outcomes.
Future research could attempt to correlate findings from both expressed and actual
knowledge sharing. However, as previously suggested, measuring actual knowledge
sharing or knowledge sharing behavior may prove difficult. Alternatively, future research
could find ways to determine if expressed willingness to share or use knowledge actually
translates into the sharing of knowledge or if the perceived useful knowledge actually has
tangible useful outcomes for the individual, project, or the firm. These techniques were
not employed in this study, as they would have violated the anonymity of the
respondents.
5.4 Implications for Practice
The most significant implication for practitioners, from this study, is that effective
knowledge sharing behavior among knowledge workers is predominately dependent on
the development of co-worker trust and shared vision. Further, that trust between co-
244
workers was primarily influenced by shared language and shared vision. Although other
factors were found to be significant, these three (i.e. trust, shared language, and shared
vision) were the most influential, as witnessed by the direct and indirect relationships
between them. Therefore, one possible suggestion to encourage effective knowledge
sharing behavior is to nurture and promote behaviors and practices that create trust,
shared language, and shared vision among co-workers. Management support is important;
since management must establish and clarify objectives, encourage workers to
participate, and, most importantly, reward workers for their time (i.e. for activities to
create and share language, vision, trust, or knowledge).
Looking specifically at knowledge sharing networks, Abrams, Cross, Lesser, and Levin
(2003) performed a review of existing literature and conducted interviews in 20
organizations, to better understand ways to nurture trust in professional organizations.
Based on their research, the authors identified behaviors and practices that managers may
use in their organizations for promoting trust. Table 5.1 summarizes the authors’ trust
building suggestions and associated managerial actions (p. 67). The authors suggest that
certain behaviors and practices promote different kinds of trust. For example, items 1, 5,
8, and 9 promote benevolence-based trust, item 10 promotes competence-based trust (i.e.
ability-based trust), and the rest of the items promote both (items 2, 3, 4, 6, and 7).
However, the authors warn practitioners that “the right set of trust builders to focus on is
likely unique for each organization [and that] the specific environment that one is in will
dictate which trust builders offer the greatest potential to improve trust” (p. 74).
Ultimately, the authors suggest experimenting with the trust builders, to see which work
best for the organization. Interestingly, the findings from the present research validate the
authors’ suggestion for promoting shared language and shared vision as ways of
nurturing trust (i.e. item 6). For this reason, practical and managerial suggestions for
nurturing shared language and shared vision, in an organizational setting, are briefly
discussed below.
245
Trust Builder Managerial Actions Trustworthy Behaviors 1. Act with discretion • Be clear about what information you are expected to keep confidential.
• Don’t reveal information you have said you would not and hold others accountable for this.
2. Be consistent between word and deed
• Be clear about what you have committed to do, so there is no misunderstanding. • Set realistic expectations when committing to do something, and then deliver.
3. Ensure frequent and rich communication
• Make interactions meaningful and memorable. • Consider having some face-to-face (or at least telephone) contact. • Develop close relationships.
4. Engage in collaborative communication
• Avoid being overly critical or judgmental of ideas still in their infancy. • Don’t always demand complete solutions from people trying to solve a problem. • Be willing to work with people to improve jointly on their partially formed ideas.
5. Ensure that decisions are fair and transparent
• Make sure that people know how and why personnel rules are applied and that the rules are applied equally. • Make promotion and rewards criteria clear-cut, so people don’t waste time developing a hidden agenda (or trying to decode everyone else’s).
Organizational Factors 6. Establish and ensure shared vision and language
• Set common goals early on. • Look for opportunities to create common terminology and ways of thinking. • Be on the lookout for misunderstandings due to differences in jargon or thought processes.
7. Hold people accountable for trust
• Explicitly include measures of trustworthiness in performance evaluations. • Resist the urge to reward high performers who are not trustworthy. • Keep publicizing key values such as trust—highlighting both rewarded good examples and punished violations—in multiple forums.
Relational Factors 8. Create personal connections
• Create a “human connection” with someone based on non-work things you have in common. • Maintain a quality connection when you do occasionally run into acquaintances, including discussing non-work topics. • Don’t divulge personal information shared in confidence.
9. Give away something of value
• When appropriate, take risks in sharing your expertise with people. • Be willing to offer others your personal network of contacts when appropriate.
Individual Factor 10. Disclose your expertise and limitations
• Make clear both what you do and don’t know. • Admit it when you don’t know something rather than posture to avoid embarrassment. • Defer to people who know more than you do about a topic.
Table 5.2 Summary of Abrams, Cross, Lesser and Levin’s Trust Builders and Associated Managerial Actions (2003, p. 67)
The results from numerous research studies (described in Section 2.3.2) plus the one
conducted here, suggest that having a shared language and shared vision increases trust
among co-workers. Further, the findings in the present study, as well as others, suggest
that having a greater degree of shared vision among co-workers also increases effective
246
knowledge sharing behavior. For this reason, it is important for practitioners to
understand how to create and nurture project and work environments where there are
high degrees of shared vision and shared language (along with trust).
Research by Abrams, Cross, Lesser, and Levin (2003) suggests that one possible method
for establishing and ensuring shared perspective is to initiate all new projects with clearly
explicated goals, with respect to group objectives. These explicated goals and group
objectives require co-workers to share their understandings, opinions, and perceptions
around a common reference point (Daft & Lengel, 1986). This also provides the team
with a unified mandate, preventing them from misinterpreting and misdirection. Abrams
et al. (2003) also suggest using the start of the project as a time to clarify unique
terminology, to make sure individuals from different functional backgrounds place
similar meanings on important words, phrases, and concepts. Carlile (2002), as well as
Mäkelä and Brewster (2009), suggest that using flowcharts and other boundary objects
may help in developing a shared resource, through which project members can have a
common language and common reference point. A boundary object is an “object that
serves to coordinate the perspectives of various constituencies” (Wenger, 1998, p. 106).
Other examples include blueprints, prototypes, process models, and legal precedents or
opinions.
Another way the literature suggests to nurture and promote shared language and shared
vision is to expose co-workers to each other’s domain (Carlile, 2002, 2004; Nonaka &
Takeuchi, 1995). According to Mäkelä and Brewster (2009), this allows project members
to better understand each other’s tacit knowledge, codes, and language systems. The
authors (2009; Mäkelä, 2007) suggest that this type of interaction leads to an inside view
of the co-workers practice, which “facilitates an understanding of the discourse and
expected ways of behaving” (2009, p. 596).
The difficult reality in knowledge intensive project groups is that managers cannot fully
anticipate or predict what the nature of the working relationships between co-workers
may be. On every project, it must be assumed that there may be both positive and
247
negative working relationships. Therefore, any strategy that is adopted must have a
positive influence in both types of relationships. Although this study has identified
several other factors that influence knowledge sharing behavior; trust, shared vision, and
shared language (in that order) are the most important to promote and nurture in
organizational settings where individuals must work with both positive and negative
referents.
5.5 Conclusion and Future Work
The phenomenon of interest in the present study was the knowledge sharing of
employees in organizations. Specifically, the research examined knowledge workers, on
projects, in a professional service firm. Of particular interest were trust and other social
and cognitive factors, which have been found to influence or inhibit knowledge sharing
behavior, in this setting.
The overarching research question was what are the factors that influence knowledge
sharing behavior directly and indirectly through trust? To elaborate this research
question, three additional questions were posed. What are the significant relationships
between trust and knowledge sharing behavior? What are the significant relationships
between social-cognitive variables and trust? And, what are the significant relationships
between social-cognitive variables and knowledge sharing behavior? To better
understand the direct and indirect effects of social-cognitive variables and trust on
knowledge sharing behavior, two additional research questions were posed. The first
examined the collective effect social-cognitive factors and trust had on knowledge
sharing behavior (i.e. what is the collective effect of the identified social-cognitive
variables and trust on knowledge sharing behavior?). The second additional research
question was concerned with the indirect or intervening effect trust had between social-
cognitive factors and knowledge sharing behavior (i.e. does trust act as a mediating
variable between social-cognitive variables and knowledge sharing behavior?).
The conceptual framework of the study built on previous research studies, and drew from
theoretical foundations from the organizational behavior, psychology, information
248
studies, sociology, and management literature on organizational trust and knowledge
sharing. The framework identified the most significant factors found to influence
organizational knowledge sharing behavior directly and indirectly through trust. Specific
factors examined, along with trust, included homophily, shared language, shared vision,
tie strength, and relationship length. The study made methodological contributions in the
form of conceptualizations for knowledge sharing behavior, trust, and tie strength. Also,
it provided a more nuanced and focused analysis, by factoring for knowledge type and
co-worker working relationship.
An exploratory approach was used to conduct empirical research on knowledge workers,
at a large professional service firm. The study collected quantitative data from 275
participants, using a web-based survey as a primary research instrument. The survey was
active from January 26th, 2011 to February 25th, 2011 (31 days) and respondents could
participate at anytime between those dates. The goal of the survey was to understand
which of the social and/or cognitive factors influenced knowledge sharing behavior,
directly and indirectly through trust, which was done by testing the identified hypotheses.
Statistical analyses were used to test the hypotheses, which assisted in answering the
overarching and sub-research questions. Statistical analyses of the data included factor
and reliability analysis, correlation analysis, t-tests, and multiple regression analysis. The
mediating (i.e. indirect) effect of trust was tested using hierarchical multiple regression
analysis. Additional data were gathered from site visits and 90-120 minutes un-
structured interviews with senior partners and the Director of Knowledge Management at
the firm. The additional data collected from the site visits and interviews were used to
further understand and interpret the findings of the study.
The results of the study showed that, of all the factors examined, the three found to have
the strongest effect on respondent’s trust in their co-workers were shared vision, shared
language, and tie strength. Furthermore, the two factors found to have the strongest effect
on organizational knowledge sharing behavior were trust and shared vision. Trust was
also found to have a mediating effect between shared vision and knowledge sharing
behavior, and between shared language and knowledge sharing behavior. In addition,
249
collectively, trust and one or more social-cognitive factors influenced knowledge sharing
behavior.
Throughout the study there have been several implications made for future research. The
most notable included exploring the influence of homophily in the corporate versus social
environments; further distinguishing between tie strength prior to the project from tie
strength while on the project; considering the nature of working relationships (i.e.
positive versus negative); testing the framework in other professional firm contexts; and
finally distinguishing between the reported knowledge sharing behavior and the actual
knowledge sharing behavior.
Significant implications for practitioners were that effective knowledge sharing behavior
among co-workers requires a nurturing manager to work on developing co-worker trust
and shared vision and that a manager wanting to promote trust between co-workers must
nurture shared language and shared vision.
250
References
Abrams, L. C., Cross, R., Lesser, E., & Levin, D. Z. (2003). Nurturing interpersonal trust
in knowledge-sharing networks. Academy of Management Executive, 17(4), 64-
77.
Adler, P. S. (2002). Market, hierarchy, and trust: The knowledge economy and the future
of capitalism. In C. W. Choo & N. Bontis (Eds.), The strategic management of
intellectual capital and organizational knowledge (pp. 23-46). New York: Oxford
University Press.
Alavi, M., & Leidner, D. E. (2001). Review: Knowledge management and knowledge
management systems: Conceptual foundations and research issues. MIS
Quarterly, 25(1), 107-136.
Andrews, K. M., & Delahaye, B. L. (2000). Influences on knowledge processes in
organizational learning: The psychosocial filter. Journal of Management Studies,
37(6), 797-810.
Aneshensel, C. S. (2002). Theory-based data analysis for the social sciences. Thousand
Oaks, CA: Pine Forge Press.
Argyres, N. S. (1999). The impact of information technology on coordination: Evidence
from the B-2 "stealth" bomber. Organization Science, 10(2), 162-180.
Argyris, C. (1964). Integrating the individual and the organization. New York, NY:
Wiley.
Babič, F., & Wagner, J. (2006). Modeling of knowledge creation processes based on
activity theory. Paper presented at the WIKT2006: Workshop on Intelligent and
Knowledge oriented Technologies from http://webocrat.fei.tuke.sk/fei-
cit/publikacie/babic_WIKT2006.pdf
Banai, M., & Reisel, W. D. (1999). Would you trust your foreign manager? An empirical
investigation. The International Journal of Human Resource Management, 10(3),
477-487.
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.
251
Bartol, K. M., & Srivastava, A. (2002). Encouraging knowledge sharing: The role of
organizational reward systems. Journal of Leadership and Organizational
Studies, 9(1), 64-77.
Bell, D. (1973). The coming of post-industrial society: A venture in social forecasting.
New York: Basic Books.
Bielby, W. T., & Baron, J. N. (1986). Men and women at work: Sex segregation and
statistical discrimination. American Journal of Sociology, 91(4), 759-799.
Blackler, F. (2002). Knowledge, knowledge work, and organization. In C. W. Choo & N.
Bontis (Eds.), The Strategic Management of Intellectual Capital and
Organizational Knowledge (pp. 47-64). New York: Oxford University Press.
Blackler, F., Crump, N., & Mcdonald, S. (2000). Organizing processes in complex
activity networks. Organization, 7(2), 277-300.
Blau, P. M., Beeker, C., & Fitzpatrick, K. M. (1984). Intersecting social affiliations and
intermarriage. Social Forces, 62(3), 585-605.
Blau, P. M., Blum, T. C., & Schwartz, J. E. (1982). Heterogeneity and intermarriage.
American Sociological Review, 47(February), 45-62.
Blau, P. M., Ruan, D., & Monika, A. (1991). Interpersonal choice and networks in China.
Social Forces, 69(4), 1037-1062.
Blum, T. C. (1984). Racial inequality and salience: An examination of Blau's theory of
social structure. Social Forces, 62(3), 607-617.
Boer, N-I., van Baalen, P. J., & Kumar, K. (2002). An activity theory approach for
studying the situatedness of knowledge sharing. Paper presented at the 35th
Hawaii International Conference on System Sciences Hilton Waikoloa Village,
Big Island, Hawaii.
Bohrnstedt, G. W., & Knoke, D. (1994). Statistics for social data analysis. Itasca,
Illinois: F.E. Peacock Publishers.
Boisot, M. H. (1998). Knowledge assets: Securing competitive advantage in the
information economy. New York: Oxford University Press.
Boisot, M. H. (2002). The creation and sharing of knowledge. In C. W. Choo & N. Bontis
(Eds.), The strategic management of intellectual capital and organizational
knowledge (pp. 65-77). New York: Oxford University Press.
252
Boon, S. D., & Holmes, J. G. (1991). The dynamics of interpersonal trust: Resolving
uncertainty in the face of risk. In R. A. Hinde & J. Groebel (Eds.), Cooperation
and prosocial behavior (pp. 190-211). Cambridge, MA: Cambridge University
Press.
Bornhorst, F., Ichino, A., Kirchkamp, O., Schlag, K. H., & Winter, E. (2010). Similarities
and differences when building trust: The role of cultures. Experimental
Economics, 13, 260-283.
Bracken, L. J., & Oughton, E. A. (2006). ‘What do you mean?’ The importance of
language in developing interdisciplinary research. Transactions of the Institute of
British Geographers, 31(3), 371–382.
Brass, D. J. (1985). Men's and women's networks: A study of interaction patterns and
influence in an organization. The Academy of Management Journal, 28(2), 327-
343.
Brass, D. J., Galaskiewicz, J., Greve, H. R., & Tsai, W. (2004). Taking stock of networks
and organizations: A multilevel perspective. Academy of Management Journal,
47(6), 795-817.
Brewer, M. B. (1979). In-group bias in the minimal intergroup situation: A cognitive-
motivational analysis. Psychological Bulletin, 86, 307-324.
Brewer, M. B. (1981). Ethnocentrism and its role in interpersonal trust. In M. B. Brewer
& B. E. Collins (Eds.), Scientific Inquiry and the Social Sciences (pp. 345-359).
New York: Jossey-Bass.
Bromiley, P., & Cummings, L. L. (1995). Transaction costs in organizations with trust. In
R. Beis, B. Sheppard & R. Lewicki (Eds.), Research on negotiations in
organizations (Vol. 5, pp. 219-247). Greenwich, CT.
Brown, J. S., & Duguid, P. (1991). Organizational learning and communities of practice:
A unified view of working, learning and innovation. Organization Science, 2(1),
40-56.
Brown, J. S., & Duguid, P. (1998). Organizing knowledge. California Management
Review 40(1), 90-111.
Brown, J. S., & Duguid, P. (2000). Balancing act: Capturing knowledge without killing it.
Harvard Business Review, 78(3), 73–80.
253
Burt, R. (1992). Structural holes. Cambridge, MA: Harvard University Press.
Byrne, D. E. (1971). The attraction paradigm. New York: Academic Press.
Campbell, K. E. (1988). Gender differences in job-related networks. Work and
Occupations, 15(2), 179-200.
Campbell, K. E., Marsden, P. V., & Hurlbert, J. S. (1986). Social resources and
socioeconomic status. Social Networks, 8(1), 97-117.
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 trans- forming: An integrative
framework for managing knowledge across boundaries. Organization Science,
15(5), 555-568.
Chan, D. (2002). Maintaining professional competence: Impact of organizational and
individual factors on the updating activities of public reference librarians.
University of Toronto, Toronto.
Chang, C-W., Huang, H.-C., Chiang, C.-Y., Hsu, C.-P., & Chang, C.-C. (2011). Social
capital and knowledge sharing: Effects on patient safety. Journal of Advanced
Nursing, 20. doi: 10.1111/j.1365-2648.2011.05871.x
Chatman, J. A., & Flynn, F. J. (2001). The influence of demographic heterogeneity on the
emergence and consequences of cooperative norms in work teams. Academy of
Management Journal, 44, 956-974.
Chattopadhyay, P. (1999). Beyond direct and symmetrical effects: The influence of
demographic dissimilarity on organizational citizenship behavior. Academy of
Management Journal, 42(3), 273-287.
Chiu, C-M., Hsu, M-H., & Wang, E. T. G. (2006). Understanding knowledge sharing in
virtual communities: An integration of social capital and social cognitive theories.
Decision Support Systems, 42, 1872-1888.
Choo, C. W. (1998). The knowing organization: How organizations use information to
construct meaning, create knowledge, and make decisions. New York: Oxford
University Press.
Choo, C. W. (2000). Working with knowledge: How information professional help
organizations manage what they know. Library Management, 21(8), 395-403.
254
Choo, C. W., & Bontis, N. (2002). The strategic management of intellectual capital and
organizational knowledge. New York: Oxford University Press.
Clegg, S., & Bailey, J. R. (Eds.). (2008). International encyclopedia of organization
studies. Australia: Sage Publications.
Cohen, J., & Cohen, P. (1983). Applied multiple regression / correlation analysis for the
behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Cohen, W. M., & Levinthal., D. A. (1990). Absorptive capacity: A new perspective on
learning and innovation. Administrative Science Quarterly, 35(1), 128-152.
Coleman, J. S. (1988). Social capital in the creation of human capital. American Journal
of Sociology, 94(Supplement), S95-S120.
Comrey, A. L., & Lee, H. B. (1992). A first course in factor analysis. Hillsdale, New
Jersey: Lawrence Erlbaum Associates.
Connelly, C. E., & Kelloway, E. K. (2003). Predictors of employees' perceptions of
knowledge sharing cultures. Leadership & Organization Development Journal,
24(5/6), 294-301.
Constant, D., Sproull, L., & Kiesler, S. (1996). The kindness of strangers: The usefulness
of electronic weak ties for technical advice. Organization Science, 7(2), 119-135.
Criscuolo, P., Salter, A., & Ter Wal, A. (2010, June 16 - 18). The role of proximity in
shaping knowledge sharing in professional service firms. Paper presented at the
Summer conference on opening up innovation: Strategy, organization and
technology, London Business School.
Cross, R., & Cummings, J. N. (2004). Tie and network correlates of individual
performance in knowledge-intensive work. Academy of Management Journal,
47(6), 928-937.
Cross, R., & Sproull, L. (2004). More than an answer: Information relationships for
actionable knowledge. Organization Science, 15(4), 446-462.
Curseu, P. L., & Schruijer, S. G. L. (2010). Does conflict shatter trust or does trust
obliterate conflict? Revisiting the relationships between team diversity, conflict,
and trust. Group Dynamics: Theory, Research, and Practice, 14(1), 66-79.
Cyr, S., & Choo, C. W. (2010). The individual and social dynamics of knowledge
sharing: An exploratory study. Journal of Documentation, 66(6), 824 - 846.
255
Daft, R. L., & Lengel, R. H. (1986). Organizational information requirements, media
richness and structural design. Management Science, 32(5), 554-571.
Dalkir, K., & Wiseman, E. (2004). Orgnizational storytelling and knowledge
management: A survey. Storytelling, Self, Society, 1(1), 57-73.
Davenport, T. H., & Prusak, L. (1998). Working knowledge: How organizations manage
what they know. Boston: Harvard Business School Press.
De Carolis, D. M. (2002). The role of social capital and organizational knowledge in
enhancing entrepreneurial opportunities in high-technology enviornments. In C.
W. Choo & N. Bontis (Eds.), The strategic management of intellectual capital
and organizational knowledge (pp. 699-709). New York: Oxford University
Press.
De Long, D. W., & Fahey, L. (2000). Diagnosing cultural barriers to knowledge
management. Academy of Management Executive, 14(4), 113-127.
Deutsch, M. (1960). The effect of motivational orientation upon trust and suspicion.
Human Relations, 13(1), 122-139.
DeVellis, R. F. (1991). Scale development. Newbury Park, CA: Sage Publications.
Dirks, K. T., & Ferrin, D. L. (2002). Trust in leadership: Meta-analytic findings and
implications for research and practice. Journal of Applied Psychology, 87(4), 611-
628.
Duguid, P. (2005). "The art of knowing": Social and tacit dimensions of knowledge and
the limits of the community of practice. The Information Society, 21(2), 109-118.
duPlessis, T., & duToit, A. S. A. (2006). Knowledge management and legal practice.
International Journal of Information Management, 26(5), 360-371.
Earl, M. (2001). Knowledge management strategies: Toward a taxonomy. Journal of
Management Information Systems, 18(1), 221-233.
Earley, P. C., & Mosakowski, E. (2000). Creating hybrid team cultures: An empirical test
of transnational team functioning. Academy of Management Journal, 43(1), 26-
49.
Elliott, S., & O'Dell, C. (1999). Sharing knowledge & best practices: The hows and whys
of tapping your organization's hidden reservoirs of knowledge. Health Forum
Journal, 42(3), 34-38.
256
Empson, L. (2001). Fear of exploitation and fear of contamination: Impediments to
knowledge transfer in mergers between professional service firms. Human
Relations, 54(7), 839-862.
Engeström, Y. (1999). Innovative learning in work teams: Analyzing cycles of
knowledge creation in practice In Y. Engeström, R. Miettinen & R.-L. Punamäki
(Eds.), Perspectives on activity theory (pp. 377-404). Cambridge, UK: Cambridge
University Press.
Evans, M. M. (2008). Determination of research readiness: Knowledge sharing: An
empirical study of the role of trust in an organizational setting. University of
Toronto.
Feld, S. L. (1982). Structural determinants of similarity among associates. American
Sociological Review, 47, 797-801.
Fiol, C. M. (1994). Consensus, diversity, and learning in organizations. Organization
Science, 5(3), 403-420.
Fischer, C. S. (1977). Networks and places: Social relations in the urban setting. New
York: Free Press.
Fischer, C. S. (1982). To dwell among friends. Chicago: University of Chicago Press.
Galaskiewicz, J., & Shatin, D. (1981). Leadership and networking among neighborhood
human service organizations. Administrative Science Quarterly, 26(3), 434-448.
Ganesan, S. (1994). Determinants of long-term orientation in buyer-seller relationships.
Journal of Marketing, 22(1), 1-19.
Geringer, M. J. (1988). Partner selection criteria for developed country joint ventures.
Business Quarterly, 53(1), 67-84.
Gherardi, S. (2001). From organizational learning to practice-based knowing. Human
Relations, 54(1), 131-139.
Ghoshal, S., Korine, H., & Szulanski, G. (1994). Interunit communication in
multinational corporations. Management Science, 40(1), 96-110.
Goman, C. K. (2002, June 10). Why people don’t tell you what they know. Paper
presented at the IABC International Conference, Hyatt Regency, Chicago, IL
USA.
257
Granovetter, M. S. (1973). The strength of weak ties. American Journal of Sociology,
78(6), 1360-1380.
Granovetter, M. S. (1982). The strength of weak ties: A network theory revisited. In P. V.
Marsden & N. Lin (Eds.), Social Structure and Network Analysis (pp. 105-130).
Beverly Hills, CA: Sage.
Grant, R. M. (2002). The knowledge-based view of the firm. In N. B. Chun Wei Choo
(Ed.), The strategic management of intellectual capital and organizational
knowledge (pp. 133-148). New York: Oxford University Press.
Haas, M. R., & Hansen, M. T. (2007). Different knowledge, different benefits: Toward a
productivity perspective on knowledge sharing in organizations. Strategic
Management Journal, 28(11), 1133-1153.
Habermas, J. (1971). Knowledge and human interests. Boston: Beacon.
Hall, H. (2001). Input-friendliness: Motivating knowledge sharing across intranets.
Journal of Information Science, 27(3), 139-146.
Hansen, M. T. (1999). The search-transfer problem: The role of weak ties in sharing
knowledge across organization subunits. Administrative Science Quarterly, 44(1),
82-111.
Henderson, J. K. (2005). Language diversity in international management teams.
International Studies of Management and Organization, 35(1), 66-82.
Hendricks, P. (1999). Why share knowledge? The influence of ICT on the motivation for
knowledge sharing. Knowledge and Process Management, 16(2), 91-100.
Hinds, P. J., & Pfeffer, J. (2003). Why organizations don't "know what they know":
Cognitive and motivational factors affecting the transfer of expertise. In M.
Ackerman, V. Pipek & V. Wulf (Eds.), Beyond knowledge management: Sharing
expertise (pp. 3-26). Cambridge, MA: MIT Press.
Hislop, D. (2003). Linking human resource management and knowledge management via
commitment: A review and research agenda. Employee Relations, 25(1/2), 182-
202.
258
Hitt, M. A., Bierman, L., Shimizu, K., & Kochhar, R. (2001). Direct and moderating
effects of human capital on strategy and performance in professional service
firms: A resource-based perspective. The Academy of Management Journal,
44(1), 13-28.
Ho, L-A., Kuo, T-H., & Lin, B. (2011). How social identification and trust influence
organizational online knowledge sharing. Internet Research, 22(1), 4-28.
Ho, L-A., Kuo, T-H., Lin, C., & Lin, B. (2010). The mediate effect of trust on
organizational online knowledge sharing: An empirical study. International
Journal of Information Technology & Decision Making, 9(4), 625-644.
Hoe, S. L., & McShane, S. L. (2002). Leadership antecedents of informal knowledge
acquisition and dissemination. International Journal of Organisational
Behaviour, 5(10), 282-291.
Hogg, M. A., & Terry, D. J. (2000). Social identity and self-categorization processes in
organizational contexts. Academy of Management Review, 25(1), 121-140.
Holste, J. S. (2003). A study of the effects of affect-based trust and cognition-based trust
on intra-organizational knowledge sharing and use. Unpublished Ph.D.
Dissertation, Regent University, Virginia Beach, Virginia.
Huber, G. P. (1991). Organizational learning: The contributing processes and the
literatures. Organization Science, 2(1), 88-115.
Huckfeldt, R. R., & Sprague, J. (1995). Citizens, politics and social communication:
Information and influence in an election campaign (Cambridge studies in public
opinion and political psychology). New York: Cambridge University Press.
Husted, K., & Michailova, S. (2002). Diagnosing and fighting knowledge-sharing
hostility Organizational Dynamics, 31(1), 60-73.
Ibarra, H. (1992). Homophily and differential returns: Sex differences in network
structure and access in an advertising firm. Administrative Science Quarterly,
37(3), 422-447.
Ibarra, H. (1993). Personal networks of women and minorities in management: A
conceptual framework. Academy of Management Review, 18(1), 56-87.
Ibarra, H. (1995). Race, opportunity, and diversity of social circles in managerial
networks. Academy of Management Journal, 38(3), 673-703.
259
Ibarra, H. (1997). Paving an alternative route: Gender differences in managerial
networks. Social Psychology Quarterly, 60(1), 91-102.
Johnson, J. L., Cullen, J. B., Sakano, T., & Takenouchi, H. (1996). Setting the stage for
trust and strategic integration in Japanese - U.S. cooperative alliances. Journal of
International Business Studies, 27(5, Global Perspectives on Cooperative
Strategies), 981-1004.
Johnson-George, C., & Swap, W. C. (1982). Measurement of specific interpersonal trust:
Construction and validation of a scale to assess trust in a specific other. Journal of
Personality and Social Psychology, 43(6), 1306-1317.
Kalleberg, A. L., Knoke, D., Marsden, P. V., & Spaeth, J. L. (1996). Organizations in
America: Analysing their structures and human resource practices. Thousand
Oaks, CA: Sage.
Kalmijn, M. (1998). Intermarriage and homogamy: Causes, patterns and trends. Annual
Review of Sociology, 24, 395–421.
Kalmijn, M., & Vermunt, J. K. (2007). Homogeneity of social networks by age and
marital status: A multilevel analysis of ego-centered networks. Social Networks,
29(1), 25-43.
Keller, R. T. (1994). Technology-information processing fit and the performance of R&D
project groups: A test of contingency theory. The Academy of Management
Journal, 37(1), 167-179.
Kogut, B., & Zander, U. (1992). Knowledge of the firm, combinative capabilities, and the
replication of technology. Organization Science, 3, 383-397.
Krackhardt, D. (1992). The strength of strong ties: The importance of philos in
organizations. In N. Nohria & R. G. Eccles (Eds.), Networks and organizations:
Structure, form and action (pp. 216-239). Boston, MA: Harvard Business School
Press.
Kramer, R. M. (1999). Trust and distrust in organizations: Emerging perspectives,
enduring questions. Annual Review of Psychology, 50, 569-598.
Latour, B. (1987). Science in action: How to follow scientists and engineers through
society. Cambridge, Mass.: Harvard University Press.
260
Laumann, E., Gagnon, J. H., Michael, R. T., & Michaels, S. (1994). The social
organization of sexuality: Sexual practices in the United States. Chicago:
University of Chicago Press.
Lave, J. (1993). Situating learning in communities of practice. In J. L. L. Resnick, & T.
Teasley (Ed.), Perspectives on socially shared cognition (pp. 63-84). Washington
D. C.: American Psychological Association.
Lave, J., & Wenger, E. (1991). Situated learning: Legitimate peripheral participation.
Cambridge, Mass.: Cambridge University Press.
Lee, R. (2009). Social capital and business and management: Setting a research agenda.
International Journal of Management Reviews, 11(3), 247-273.
Leonard, D., & Sensiper, S. (2002). The role of tacit knowledge in group innovation. In
C. W. Choo & N. Bontis (Eds.), The strategic management of intellectual capital
and organizational knowledge (pp. 485-499). New York: Oxford University
Press.
Leont'ev, A. N. (1978). Activity, consciousness, and personality. Englewood Cliffs, New
Jersey: Prentice-Hall.
Levin, D., Cross, R., & Abrams, L. (2003). Why should I trust you? Predictors of
interpersonal trust in a knowledge transfer context. Rutgers Business School.
Levin, D. Z., & Cross, R. (2004). The strength of weak ties you can trust: The mediating
role of trust in effective knowledge transfer. Management Science, 50(11), 1477-
1490.
Levin, D. Z., Walter, J., & Murnighan, J. K. (2011). Dormant ties: The value of
reconnecting. Organization Science, 22(4), 923-939.
Levin, D. Z., Whitener, E. M., & Cross, R. (2004). Perceived trustworthiness of
knowledge sources: the moderating impact of relationship length. Paper presented
at the Best Papers Proceedings of the Academy of Management, New Orleans.
Levin, D. Z., Whitener, E. M., & Cross, R. (2006). Perceived trustworthiness of
knowledge sources: The moderating impact of relationship length. Journal of
Applied Psychology, 91(5), 1163-1171.
261
Lewicki, R. J., & Bunker, B. B. (1995). Trust in relationships: A model of development
and decline. In B. B. Bunker & J. Z. Rubin (Eds.), Conflict cooperation and
justice: Essays inspired by the work of Morton Deutsch (pp. 133-174). San
Francisco, CA: Jossey-Bass.
Lewicki, R. J., & Bunker, B. B. (1996). Developing and maintaining trust in work
relationships. In T. R. T. R.M. Kramer (Ed.), Trust in organization: Frontiers of
theory and research (pp. 114-139). Thousand Oaks, CA: Sage.
Lewis, J. D., & Weigert, A. (1985). Trust as a social reality. Social Forces, 63(4), 967-
985.
Limerick, D., & Cunnington, B. (1993). Managing the new organisation: A blueprint for
networks and strategic alliances. Chatswood, NSW: Business and Professional
Publishing.
Lin, C-P. (2007). To share or not to share: Modeling tacit knowledge sharing, its
mediators and antecedents. Journal of Business Ethics, 70, 411-428.
Lincoln, J. R., & Miller, J. (1979). Work and friendship ties in organizations: A
comparative analysis of relational networks. Administrative Science Quarterly,
24, 181-199.
Louch, H. (2000). Personal network integration: Transitivity and homophily in strong-tie
relations. Social Networks, 22(1), 45–64.
Maccoby, E. E. (1998). The two sexes: Growing up apart, coming together. Cambridge,
MA: Harvard University Press.
Mäkelä, K. (2007). Knowledge sharing through expatriate relationships: A social capital
perspective. International Studies in Management and Organization, 37(3), 108-
126.
Mäkelä, K., Andersson, U., & Seppala, T. (2012). Interpersonal similarity and knowledge
sharing within multinational organizations. International Business Review, 21(3),
439-451.
Mäkelä, K., & Brewster, C. (2009). Interunit interaction contexts, interpersonal social
capital and the differing levels of knowledge sharing. Human Resource
Management, 48(4), 591–613.
262
Makela, K., Kalla, H. K., & Piekkari, R. (2007). Interpersonal similarity as a driver of
knowledge sharing within multinational corporations. International Business
Review, 16(1), 1-22.
Manev, I. M., & Stevenson, W. B. (2001). Nationality, cultural distance, and expatriate
status: Effects on the managerial network in a multinational enterprise. Journal of
International Business Studies, 32(2), 285-303.
Marsden, P. V. (1987). Core discussion networks of Americans. American Sociological
Review, 52, 122-131.
Marsden, P. V. (1988). Homogeneity in confiding relations. Social Networks, 10(1), 57-
76.
Marsden, P. V., & Campbell, K. E. (1984). Measuring tie strength. Social Forces, 63,
482-501.
Marx, K. (1932). The Economic and Philosophical Manuscripts: The Third Manuscript
Written: between April and August 1844 in Paris (G. B. I. 1974, Trans.). First
Published: 1932 in German, by the Institute of Marxism-Leninism in the Union of
Soviet Socialist Republics.
Mayer, R. C., & Davis, J. H. (1999). The effect of the performance appraisal system on
trust for management: A field quasi-experiment. Journal of Applied Psychology,
84(1), 123-136.
Mayer, R. C., Davis, J. H., & Schoorman, F. D. (1995). An integrative model of
organizational trust. The Academy of Management Review, 20(3), 709-734.
McAllister, D. J. (1995). Affect and cognition-based trust as foundations for interpersonal
cooperation in organizations. The Academy of Management Journal, 38(1), 24-59.
McCauley, D. P., & Kuhnert, K. W. (1992). A theoretical review and empirical
investigation of employee trust in management. Public Administration Quarterly,
16(2), 265-284.
McDermott, R., & O'Dell, C. (2001). Overcoming cultural barriers to sharing knowledge.
Journal of Knowledge Management, 5(1), 76-85.
McEvily, B., & Tortoriello, M. (2011). Measuring trust in an organizational context:
Review and recommendations. Journal of Trust Research, 1(Special Issue), 23-
63.
263
McFadyen, M. A., & Cannella, A. A. (2004). Social capital and knowledge creation:
Diminishing returns of the number and strength of exchange relationships.
Academy of Management Journal, 47(5), 735-746.
McPherson, J. M., & Smith-Lovin, L. (1987). Homophily in voluntary organizations:
Status distance and the composition of face-to-face groups. American
Sociological Review, 52(3), 370-379.
McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds of a feather: Homophily in
social networks. Annual Review of Sociology, 27, 415-444.
Mehra, A., Kilduff, M., & Brass, D. J. (1998). At the margins: A distinctiveness approach
to the social identity and social networks of underrepresented groups. The
Academy of Management Journal, 41(4), 441-452.
Mital, M., Isreal, D., & Agarwal, S. (2010). Information exchange and information
disclosure in social networking web sites. The Learning Organization, 17(6), 479-
489.
Nahapiet, J., & Ghoshal, S. (1998). Social capital, intellectual capital, and the
organizational advantage. Academy of Management Review, 23(2), 242-266.
Nonaka, I. (1991). The knowledge-creating company. Harvard Business Review,
69(November-December), 96-104.
Nonaka, I. (1994). A dynamic theory of organizational knowledge creation. Organization
Science, 5(1), 14-37.
Nonaka, I. (2002). A dynamic theory of organizational knowledge creation. In C. W.
Choo & N. Bontis (Eds.), The strategic management of intellectual capital and
organizational knowledge (pp. 437-462). New York: Oxford University Press.
Nonaka, I., & Takeuchi, H. (1995). The knowledge-creating company: How Japanese
companies create the dynamics of innovation. Oxford: Oxford University Press.
Oetzel, J. G. (1995). Intercultural small groups: An effective decision-making theory. In
R. L. Wiseman (Ed.), Intercultural communication theories (pp. 247-270).
Newbury Park, CA: Sage.
Ojha, A. K. (2005). Impact of team demography on knowledge sharing in software
project teams. South Asian Journal of Management, 12(3), 67-78.
264
Orlikowski, W. J. (1993). CASE tools as organizational change: Investigating increment.
MIS Quarterly, 17(3), 309-340.
Ouchi, W. G. (1980). Markets, bureaucracies, and clans. Administrative Science
Quarterly, 25(1), 129-141.
Penley, L. E., & Hawkins, B. (1985). Studying interpersonal communication in
organizations: A leadership application. Academy of Management Journal, 28(2),
309-326.
Polanyi, M. (1962). Personal knowledge. London, UK: Routledge and Kegan Paul.
Polanyi, M. (1966). The tacit dimension (1st ed.). Garden City, New York: Doubleday
and Company.
Polanyi, M., & Prosch, H. (1975). Meaning. Chicago, IL: University of Chicago Press.
Popielarz, P. A. (1999). (In)voluntary association: A multilevel analysis of gender
segregation in voluntary organizations. Gender and Society, 13(2), 234-250.
Reagans, R. (2005). Preferences, identity, and competition: predicting tie strength from
demographic data. Management Science, 51(9), 1374–1383.
Renzl, B. (2008). Trust in management and knowledge sharing: The mediating effects of
fear and knowledge documentation. Omega, 36, 206-220.
Riege, A. (2005). Three-dozen knowledge-sharing barriers managers must consider.
Journal of Knowledge Management, 9(3), 18-35.
Ring, P. S., & van de Ven, A. H. (1994). Developmental processes of cooperative
interorganizational relationships. The Academy of Management Review, 19(1), 90-
118.
Roberts, K. H., & O'Reilly III, C. A. (1974). Failures in upward communication in
organizations: Three possible culprits. Academy of Management Journal, 17(2),
205-215.
Rogers, E., & Bhowmik, D. K. (1970). Homophily-heterophily relational concepts for
communication research. Public Opinion Quarterly, 34(4), 523-538.
Rogers, E. M. (1995). Diffusion of innovations (5th ed.). New York: Anchor Day Books.
Ruggles, R. (1998). The state of the notion: Knowledge management in practice.
California Management Review, 40(3), 80-89.
Ryle, G. (1949). The concept of mind. London, UK: Hutchinson & Company.
265
Schlenker, B., Helm, B., & Tedeschi, J. T. (1973). The effects of personality and
situational variables on behavioral trust. Journal of Personality and Social
Psychology, 25(3), 419-427.
Schoorman, F. D., Mayer, R. C., & Davis, J. H. (1996). Organizational trust:
Philosophical perspectives and conceptual definitions. Academy of Management
Review, 21(2), 337-340.
Schoorman, F. D., Mayer, R. C., & Davis, J. H. (1996). Empowerment in veterinary
clinics: The role of trust in delegation. Paper presented at the 11th annual meeting
of the Society for Industrial and Organizational Psychology.
Shore, L. M., Cleveland, J. N., & Goldberg, C. B. (2003). Work attitudes and decisions as
a function of manager age and employee age. Journal of Applied Psychology, 88,
529-537.
Shrum, W., Cheek Jr., N. H., & Hunter, S. M. (1988). Friendship in school: Gender and
racial homophily. Sociology of Education, 61(4), 227-239.
Sitkin, S. B., & Roth, N. L. (1993). Explaining the limited effectiveness of legalistic
"remedies" for trust/distrust. Organization Science, 4(3), 367-392.
Smith-Lovin, L., & McPherson, M. (1993). You are who you know: A network
perspective on gender. In P. England (Ed.), Theory on gender. Feminism on
theory: Social institutions and social change (pp. 223-241). New York: Aldine.
Spender, J. C. (1994). Knowing, managing and learning: A dynamic managerial
epistemology. Management Learning, 25(3), 387-412.
Spender, J. C. (1996). Making knowledge the basis of a dynamic theory of the firm.
Strategic Management Journal, 17(Winter Special Issue), 45-62.
Spender, J. C. (1996). Organizational knowledge, learning and memory: Three concepts
in search of a theory. Journal of Organizational Change Management, 9(1), 63-
78.
Statistics Canada (2006). 2006 census questions: And reasons why the questions are
asked Retrieved December 3, 2009, from
http://www12.statcan.ca/english/census06/info/questions/index.cfm?S=11
Suchman, L. (1987). Plans and situated actions: The problem of human-machine
communications. Cambridge, UK: Cambridge University Press.
266
Swap, W., Leonard, D., & Abrams, L. (2001). Using mentoring and storytelling to
transfer knowledge in the workplace. Journal of Management Information
Systems, 18(1), 95-114.
Szulanski, G. (1995). Unpacking stickiness: An empirical investigation of the barriers to
transfer best practices in the firm. Academy of Management Journal, 38(Special
Issue), 437-441.
Szulanski, G. (1996). Exploring internal stickiness: Impediments to the transfer of best
practice within the firm. Strategic Management Journal, 17(Winter Issue), 27-43.
Szulanski, G., Cappetta, R., & Jensen, R. J. (2004). When and how trustworthiness
matters: Knowledge transfer and the moderating effect of causal ambiguity.
Organization Science, 15(5), 600-613.
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics. Boston: Pearson.
Thompson, M. P. A., & Walsham, G. (2004). Placing knowledge management in context.
Journal of Management Studies, 41(5), 725-747.
Thomson, J. A. K. (1955). Aristotle: Nichomachean ethics. London, UK: Penguin.
Toh, S. M., & Srinivas, E. S. (2011). Perceptions of task cohesiveness and organizational
support increase trust and information sharing between host country nationals and
expatriate coworkers in Oman. Journal of World Business, In Press. Retrieved
from http://www.rotman.utoronto.ca/facbios/file/JWB_Toh&Srinivas2011.pdf
Tolman, C. W. (1988). The basic vocabulary of activity theory. Activity Theory, 1, 14-20.
Toulmin, S. (1999). Knowledge as shared procedures. In Y. Engeström, R. Miettinen &
R.L. Punamäki (Eds.), Perspectives on activity theory (pp. 53-64). Cambridge,
UK: Cambridge University Press.
Triandis, H. C. (1960). Cognitive similarity and communication in a dyad. Human
Relations, 13(1), 175-183.
Tsai, W., & Ghoshal, S. (1998). Social capital and value creation: The role of intrafirm
networks. The Academy of Management Journal, 41(4), 464-476.
Tsoukas, H. (2005). What is organizational knowledge? In H. Tsoukas (Ed.), Complex
knowledge: Studies in organizational epistemology (pp. 117-140). New York:
Oxford University Press.
267
Tsoukas, H. (2005). Do we really understand tacit knowledge? In H. Tsoukas (Ed.),
Complex knowledge: Studies in organizational epistemology (pp. 141-161). New
York: Oxford University Press.
Tsoukas, H., & Vladimirou, E. (2001). What is organizational knowledge? Journal of
Management Studies, 38(7), 973-993.
Tsui, A. S., & O'Reilly, C. A., III (1989). Beyond simple demographic effects: The
importance of relational demography in superior-subordinate dyads. Academy of
Management Journal, 32, 402-423.
Tsui, A. S-Y. (1984). A role-set analysis of managerial reputation. Organizational
Behavior and Human Performance, 34(1), 64-96.
Tsui, A. S-Y. (1986). Interpersonal affect and rating errors. Academy of Management
Journal, 29(3), 586-599.
Tyler, T. R., & Degoey, P. (1996). Trust in organisational authorities: The influence of
motive attributions on willingness to accept decisions. In R. M. Kramer & T. R.
Tyler (Eds.), Trust in organisations: Frontiers of theory and research (pp. 331-
356). Thousand Oaks, CA.: Sage.
Uslaner, E. M. (2011). Trust, diversity, and segregation in the United States and the
United Kingdom. Comparative Sociology, 10(2), 221-247.
Uzzi, B. (1996). The sources and consequences of embeddedness for the economic
performance of organizations: The network effect. American Sociological Review,
61(4), 674-698.
Uzzi, B. (1997). Social structure and competition in interfirm networks: The paradox of
embeddedness. Administrative Science Quarterly, 42, 35-67.
Uzzi, B., & Lancaster, R. (2003). Relational embeddedness and learning: The case of
bank loan managers and their clients. Management Science, 49(4), 383-399.
Van De Ven, A., & Johnson, P. E. (2006). Knowledge for theory and practice. Academy
of Management Review, 31(4), 802-821.
van Duijn, M. A. J., van Busschbach, J. T., & Snijders, T. A. B. (1999). Multilevel
analysis of personal networks as dependent variables. Social Networks, 21(2),
187-210.
268
vanDenHooff, B., & vanWeenen, F. d. L. (2004). Committed to share: Commitment and
CMC use as antecedents of knowledge sharing. Knowledge and Process
Management, 11(1), 13-24.
Verbrugge, L. M. (1977). The structure of adult friendship choices. Social Forces, 56,
576-597.
Vygotsky, L. (1978). Mind in society (M. Cole, V.John-Steiner, S. Scribner, & E.
Souberman ed.). Cambridge,MA: Harvard University Press.
Wenger, E. (1998). Communities of practice: Learning, meaning, and identity.
Cambridge, MA: Cambridge University Press.
Wigg, K. M. (1997). Knowledge management: An introduction and perspective. Journal
of Knowledge Management, 1(1), 6-14.
Wittgenstein, L. (1953). Philosophical investigations (G.E.M. Anscombe & R. Rhees,
Trans.). Oxford: Blackwell Publishers.
Worchel, P. (1979). Trust and distrust. In W. G. Austin & S. Worchel (Eds.), Social
psychology of intergroup relations (pp. 174-187). Monterey, CA Brooks/Cole
Publishing.
Wright, E. O. (1997). Class counts: Comparative studies in class analysis. Cambridge,
MA: Cambridge University Press.
Yang, S., & Farn, C. (2009). Social capital, behavioural control, and tacit knowledge
sharing: A multi-informant design. International Journal of Information
Management, 29(3), 210-218.
Yuasa, Y. (1987). The body: Toward an Eastern mind-body theory (N. Shigenori & T. P.
Kasulis, Trans.). Albany, NY: State University of New York Press.
Zaheer, A., McEvily, B., & Perrone, V. (1998). Does trust matter? Exploring the effects
of interorganizational and interpersonal trust on performance. Organization
Science, 9(2), 141-159.
Zand, D. E. (1972). Trust and managerial problem solving. Administrative Science
Quarterly, 17(2), 229-239.
Zenger, T. R., & Lawrence, B. S. (1989). Organizational demography: The differential
effects of age and tenure distributions on technical communication. The Academy
of Management Journal, 32(2), 353-376.
269
Appendix List of Appendices:
A.1 Knowledge Definition Components A.2 Survey Instrument A.3 Firm-‐wide Invitation Email A.4 University of Toronto Ethics Approval Letter
A.1 Knowledge Definition Components
Activity (Action, Practice, and Process)
Polanyi (1966) advanced the notion that knowledge was action-orientated by arguing that
all knowing requires skillful action60. According to Spender (1996b), Polanyi was one of
the first to discuss knowledge as “a form of abstraction that can only be known,
evidenced and communicated through action” (p. 54). This perspective is now commonly
accepted in some form by numerous other social scientists and knowledge management
theorists, including Nonaka and Takeuchi (1995); Choo (1998); Davenport and Prusak
(1998); Leonard and Sensiper (2002); Suchman (1987); Wigg (1997); Tsoukas (2005a,
2005b); Blackler (2002); Spender (1996b); Bartol and Srivastava (2002); Habermas
(1971); Van De Ven and Johnson (2006); and Elliott and O’Dell (1999).
Tsoukas (2005a) refers to a similar component of knowledge, when he explains
knowledge acquisition as “a process for incorporating new experiences and information”
(p. 118). For Nonaka (2002) knowledge is “a dynamic human process of justifying
personal beliefs” (p. 438). A similar view of ‘knowledge as process’61 is also shared by
Latour (1987); Thompson and Walsham (2004); Spender (1996b); and Blackler (2002).
Gherardi (2001) propagates a similar pragmatic theory of knowing, where practice
connects ‘knowing’ with ‘doing’. For Gherardi (2001), “knowledge consequently does
not arise from scientific ‘discoveries’; rather, it is fabricated by situated practices of
knowledge production and reproduction, using the technologies of representation and
mobilization employed by scientists” (p. 136).
60 For the purpose of this research, the terms activity and action will be considered synonymous. 61 ‘Practice based theory of knowing’ (Blackler, 2002; Thompson & Walsham, 2004)
270
The terms practice and process are commonly used interchangeably in the organizational
science and knowledge management literature when defining knowledge. As opposed to
colloquial usage of action, process and practice imply some degree of standardization and
specialization. In this respect, both processes and practices are comprised of a set of
actions (action oriented and action dependent), and have a methodical/standardized
objective. This understanding of process and practice may be used synonymously with
activity, something commonly done in activity theory (e. g. Leont’ev, 1978; Blackler,
Crump, & McDonald, 2000).
Perhaps the best illustration of the activity component of knowledge is represented in a
quote from Blackler (2002) who said that “rather than thinking of knowledge as a thing
that people possess, it is more helpful to analyze knowing as something that people do”
(p. 63).
Domain Situation (Context)
The second component of the proposed definition of knowledge focuses on the idea of
being situated within a socially constructed domain. This section will focus on the idea
of domain situation and the principle of social construction will be discussed in the next
component.
To better understand domain situation, one must first consider the ‘channel’ through
which information is being communicated. ‘Channels’ can consist of physical objects,
such as textbooks, maps, prototypes, or architectural models, or more conceptual objects
such as language, metaphor, analogies, or mental models (Vygotsky 1978; Tsoukas,
2005a; Blackler, 2002; Toulmin, 1999). Traditionally, these more conceptual
representations are referred to as ‘instruments’ (Vygotsky, 1978) or ‘cultural tools’
(Tsoukas, 2005a), yet physical objects may be thought of in a similar way.
The specifics within these ‘channels’ derive their meaning for a particular field or
expertise. Using an quote from Toulmin’s (1999), “language has a definite meaning only
when it is related to a given constellation of practical activities…we understand the
271
meaning of the word strike only if we are familiar with the game of baseball” (p. 59).
Using another example from a mathematics textbook, it may be proposed that algebraic
concepts are learned or understood by the students through the use of algebraic equations,
metaphors, and analogies.
The domain or context (i.e. expertise or field) influences and supports the meaning or
understanding of the information. Using the baseball example, if the expressions
“stealing” or “striking” are taken out of the ‘playing baseball’ context, their meaning
dramatically changes. “Playing baseball […] provides the background against which the
word strike has this meaning. The shared intelligibility of any utterance requires it to
have a standard place in a specific practical context” (Toulmin, 1999, p. 60). A specific
context helps to determine interpretation. Thompson and Walsham (2004) explain
context as “the relationally situated ingredients through which knowing occurs” (p. 735).
The appropriate interpretation of the information depends on the nature of the context, as
well as the nature of the community within that context (Duguid, 2005; Leonard &
Sensiper, 2002). According to Nonaka (1994) “what makes sense in one context, can
change or even lose its meaning when communicated to people in a different context” (p.
30). To effectively interact with a domain, one must learn to decode from the perspective
of that domain and community (Duguid, 2005). This idea is consistent with Wittgenstein
(1953), who argues that meaning and practical implication depend on their use and on the
framework in which they exist. Wittgenstein (1953), Spender (1996a), Thompson and
Walsham (2004) would argue that the meaning of all knowledge is tied up within the
context of its development. This reasoning is also supported by Lave’s (1993) and
Blackler’s (2002) argument that knowledge may not be divorced from context and
transmitted as simply abstract data.
According to Bell (1973), individuals use theory to generalize from one context to
another. It is this generalizing and use of theory that assists in exercising judgment
(especially in a foreign domain). Since all people have different ‘relationally situated
ingredients’ or different influencing (meaning-supporting) contexts, it is clear why no
272
two agents can ever share exactly the same meaning. According to Boisot (2002), if
people do not share similar meanings, then they will operate with ‘different conceptual
schemes’ (p. 72) causing distortion and rejection of new knowledge.
To interpret meaning within a domain, a person must understand the interrelated contexts
of that domain. For example, to understand the meaning of “the fastball struck out the
third baseman”, the agent must understand both the context of playing baseball as well as
the context use of the English language. This same phrase uttered at a baseball game in
Japan may not have the same interpreted meaning, since the examples only share one of
the two minimum required contexts to interpret meaning from that statement.
Being able to act prudently and correctly within a particular domain implies learning to
make distinctions and connections to the contexts influencing that domain (Tsoukas,
2005a; Wenger, 1998; Van De Ven & Johnson, 2006). In most cases, this learning
requires the person to be a member of the community operating in the domain. This
sentiment is echoed in Lave and Wenger’s (1991) definition of community as
“participation in an activity system about which participants share understandings
concerning what they are doing…the social structure of [their] practice, its power
relations, and its conditions for legitimacy” (p. 97).
Social Construction
Social construction theory62, as it applies to knowledge, finds its roots in early Russian
psychology (Leont'ev, 1978; Vygotsky, 1978) and economics (Marx, 1932). Vygotsky
(1978) and his colleagues argued that human consciousness is both shaped by social
experiences and mediated by culturally established tools. Marx (1932) argued that “the
eye has become a human eye, just as its object has become a social, human object – an
object made by man for man”. Using Leont’ev (1978), the argument is elaborated:
“Isolated activity cannot be understood apart from social ties... Entering into contact with each other, people formulate a language
62 Social Construction Theory is defined using Spender (1996a) as “the idea that the individual’s consciousness and thinking are fashioned socially” (p. 69).
273
that serves to represent the objects, the means, and the very process of work itself. [W]ords, the language signs, are not simply replacements for things, their conditional substitutes. Behind philological meanings is hidden social practice, activity transformed and crystallized in them; only in the process of this activity is objective reality revealed to man” (p. 18).
This premise of socially constructed consciousness is foundational to the work in activity
theory, and is so influential that it is found in their definition of consciousness itself:
“Man is born into the world of objects created by previous generations, and is formed as such only in the process of learning how to use them to a definite end. The mode of his relation to reality is not determined directly by his bodily organisation (as in the case with animals), but by the habits of practical activity acquired solely through communication with other people63 (Tolman, 1988, p. 16).
Influenced by this perspective, many theorists have turned towards social construction
and activity theory to understand and define knowledge, and theorize on how it is created
and disseminated. In general, the social construction viewpoint defines knowledge as ‘the
social practice of knowing’ (Boer, van Baalen, & Kumar, 2002). Gherardi (2001) argues
that “learning and knowing are mediated by social relations [...] knowledge resides in
social relations […and] knowing is part of a surrendering to a social habit” (p. 133).
Nonaka (1994) claims that knowing is something that emerges through continuous
dialogue among practitioners. In fact, Nonaka’s (2002) ‘socialization’64 refers to a
process of creating knowledge through shared experience.
Individuals learn to exercise judgment through a process of socialization, which is based
on a socially constructed shared context (Tsoukas, 2005a). New knowledge is socially
constructed by and becomes meaningful to the community within which it was
constructed (Boer, van Baalen, & Kumar, 2002). In other words, abstract formulations
ultimately depend on collective, socially accepted definitions (Tsoukas, 2005a; Polanyi,
1966; Toulmin, 1999; Blackler, Crump, & McDonald, 2000; Nonaka & Takeuchi, 1995;
63 Extract from Tolman’s (1988) The basic vocabulary of Activity Theory 64 Socialization is borrowed from Nonaka’s (2002) proposed model for knowledge creation; the SECI model.
274
Spender, 1996). This is what Wittgenstein (1953) called ‘forms of life’ (Lebensformen),
or what Grant (2002) referred to as 'common knowledge'.
Social constructionists understand knowledge to be a product of a collective, something
that is developed communally, over time (Brown & Duguid, 1991; 1998; 2000; Blackler,
2002; Leonard & Sensiper, 2002). In other words, it is the “outcome of people working
together, sharing experiences, and constructing meaning out of what they do” (Choo,
2000, p. 395). Some of these theorists (Leonard & Sensiper, 2002; Polanyi, 1966;
Tsoukas, 2005a) argue that ‘personal knowledge’ exists in collaboration with ‘collective
knowledge’, or that each person socialized within the collective encompasses the
knowledge of the collective. Other theorists (De Carolis, 2002; De Long & Fahey, 2000;
Boer, van Baalen, & Kumar, 2002; Boisot, 2002; Spender, 1994; Brown & Duguid,
1998) believe that socially constructed ‘collective’ knowledge resides (or is embedded) in
and is the possession of the collective itself (independently from the individuals
comprising the collective). This latter perspective suggests that knowledge supersedes
any one individual, and is greater than the sum of the individual knowledge within the
collective.
Developmental Capacity (Personal Ability and Capacity)
If developmental capacity is to be acknowledged as a component of knowledge, one must
first accept Polanyi and Prosch’s (1975) argument that all knowledge is personal and
subjective, despite the fact that its meaning is constructed socially. The authors suggest
that “all knowing is personal knowing” (Polanyi & Prosch, 1975, p. 44). Tsoukas (2005a)
makes a similar argument stating that “all knowledge is personal knowledge” (p. 126).
According to Nonaka (2002), knowledge has a subjective nature, which, at a fundamental
level, is created by the individual. Other authors argue that knowing is mediated through
personal human judgment (Tsoukas, 2005a; 2005b; Polanyi & Prosch, 1975) or a
personal exercising of reason (Spender, 1996b).
Personal judgment involves ‘applying abstract representations of the world’ and making
assessments of the existing gaps within those representations (Tsoukas, 2005b). Personal
275
judgments are not guided by strict rules, but through a skillful exercise of the body and
mind, guided by the senses (Polanyi, 1966; Polanyi & Prosh, 1975; Tsoukas, 2005b).
Tsoukas (2005a) describes personal judgment as being associated with the prefix re- (re-
order, re-arrange, and re-design). Judgment involves the personal ability to draw
distinctions or to divide the world into ‘this’ and ‘that’. According to Tsoukas (2005a),
exercising judgment involves “the ability of an individual to draw distinctions and the
location of the individual within a collectively generated and sustained domain of action”
(p.120). For example, a medical student must both be a part of the medical domain and be
able to draw distinctions/make judgments within that domain65. Bell (1973) and Boisot
(1998) take a similar theoretic approach, arguing that knowledge can be conceptualized
as a set of probability distributions that guide reasoned judgment and orient actions.
Being able to select relevant categories for abstraction (draw distinctions) requires the
individual to have prior knowledge of the context/domain in question (Boisot, 2002).
Knowing involves a configuration of context (Thompson & Walsham, 2004) or a unique
integrated set of particulars for which each agent is subsidiarily aware (Tsoukas, 2005a).
Knowing involves continuous interaction with the outside world (Nonaka, 2002), yet
each person’s representation and understanding of that world is personal and different
(Polanyi, 1966). For example, a differential equation cannot alone predict an unknown
function of a variable; it is the application and use of differential equations, in
engineering, physics, and economics, that allows us to make reasoned judgments and
create new knowledge.
If knowing is in fact personal and requires the individual to understand how to exercise
judgments based on historically situated and collectively constructed domains, then it can
be reasoned that biological and developmental capacity can influence the cognitive
ability needed to process new information, draw distinctions, and understand relevant
concepts. Brain functions influence conceptual skills and cognitive abilities. Biological or
physical changes to the brain may influence an individual’s knowledge state. Similarly,
65 Personal judgments within one domain will enviably encounter messiness and complexity once they are introduced into another domain.
276
changes to physical make up or motor abilities may also impact an individual’s ability to
understand and instrumentalize new knowledge.
Potentiality
The reference to potentiality is often alluded to in literature, but rarely explicitly stated,
since the concept of potentiality is difficult to frame. One way to view the potentiality of
knowledge is by considering its dependence on the rest of the definition. In other words,
domain, social construction, and developmental capacity are partial determinants of the
potential value of knowledge. Any variance in these partial determinants will cause some
utility change in the person’s knowledge state. Perhaps this is best explained through
example.
Two students who score the same on a test (getting the same answers right and wrong)
may be considered equal in ability, but are quite different in capability for further
learning. It is naïve to think of these students as having the same knowledge states. As
Boisot (2002) asserts, “no two agents possess identical mental schemas, they will
therefore assimilate and accommodate new knowledge in different ways […] external
data that different agents receive may be identical, what actually gets absorbed by each as
knowledge will differ” (p. 73).
Take, for example, two driver’s education students, sitting at the wheel of a car, for their
first lesson. Neither student has ever driven a car, although student A is an avid gamer
who enjoys playing first person driving simulators. Student B, who does not own a
gaming console, made sure they memorized the rules of the road handbook provided in
class. Student B also pays close attention in class and is attentive to drivers when they are
the passenger. During their first hands-on lesson, each student will process new data
differently, partially due to the differences in their prior experiences. Student A may
relate their driving experience to a stock of knowledge primarily formed out of video
game experiences, whereas student B may relate to a reference from the handbook.
277
Boisot (2002) argues that “two agents can never achieve identical dispositions to act and
hence identical knowledge states” (p. 68). Even if the stimuli are identical on the day of
the drive (instructions, supervision, car, driving conditions), student A and B will achieve
different knowledge states. Further, the different knowledge states may vary in effect (or
value), as expressed through action. This is primarily due to their existing different
knowledge states and the way the individuals interpret the meaning of new stimuli. Even
though the two students may have some knowledge state resonance, because of the
differences in how they process data and their prior experiences, they will never reach
identical knowledge states. Similar ideas have been echoed by Cohen and Levinthal’s
(1990) absorptive capacity66 and Vygotsky’s (1978) zone of proximal development67.
Potentiality is also expressed through the ever changing and evolving nature of
knowledge itself. Knowledge undergoes construction and transformation in its use
(Blackler, 2002). When new stimulus (information) is introduced, the existing knowledge
states undergo consolidation or modification. Knowledge is both constructed (created)
and destroyed (forgotten, or made obsolete) during this process (Boisot, 1998; 2002).
This constant transformation makes knowing a ‘continually emergent process’
(Thompson & Walsham, 2004, p. 735). Since knowledge is situated and personal, it will
inevitably change as the situation around the person evolves and develops (Blackler,
2002). The changing situation around the person alters the situated knowledge they
possess, and so on, in a cyclical fashion. Blackler (2002) refers to this type of knowing as
mediated where “changes associated with new information…transform[s] the contexts of
action” (p. 59). Spender (1994) argues that this circular process of ‘learning’ continues,
as long as there is activity. Theoretically, this could be never-ending. In discussing
organizational knowledge creation, Nonaka (2002) mirrors similar sentiments arguing
that knowledge creation is a “never-ending, circular process” (p. 451). Since knowledge
is constantly changing and evolving, the potential knowledge state of the person, at any
given time, may be of more or less value to the firm, and is always in flux.
66 A theory used to measure a firm's ability to value, assimilate, and apply new knowledge 67 A theory expressing the distance between a person’s actual developmental level as determined through independent problem solving and their potential development as determined through problem solving under guidance or in collaboration with more capable peers
278
For some activity theorists, the idea of potentiality is not new. Vygotsky, Luria, and
Leont’ev wrote extensively on the potential of human cognition. In his discussion of
organizational transformation (organizational knowledge creation/learning), Engeström
(1999) also referred to potentiality, with his concept of ‘expansive learning cycles’. He
argued that “miniature cycles of innovative learning should be regarded as potentially
expansive. A large-scale, expansive cycle of organizational transformation always
consists of small cycles of innovative learning” (Engeström, 1999, p. 385). Extending this
idea, it may be argued that any instance of a knowledge state (commonly referred to as a
piece of knowledge) may be only a fraction or ’small cycle‘ of knowing (where knowing
is a large scale expansive cycle). Potentiality is present in knowing, because these small
cycles will have varying, constantly evolving, degrees of effect (or value) on overall
knowing. This idea is consistent with activity theory, where knowing is constantly
developing in non-static activity systems. Knowledge is created or destroyed, as
contradictions and tensions emerge between the elements within the activity system
(Blackler, 2002; Babič & Wagner, 2006). As more activity systems interlink and more
information is introduced, the person’s potential to alter their existing knowledge state, in
a valuable way, increases.
279
A.2 Survey Instrument
Knowledge Sharing Survey: Important information for participants in the survey
(Survey web page designed to resemble institutional letterhead) We are conducting a research study that examines knowledge sharing behavior in group-based work. We are interested in the experiences you and other knowledge workers have with sharing knowledge during everyday activities involved in matter or firm-related project work. The study is led by Max Evans, a Doctoral Candidate at the Faculty of Information (iSchool), University of Toronto, and is part of a doctoral thesis. The research is supervised by Dr. Chun Wei Choo, professor at the Faculty of Information, University of Toronto and Dr. Anthony Wensley, professor at the Rotman School of Management, University of Toronto.
You are invited to complete questionnaire survey, which has three parts. The first is an individual section, which asks you to answer questions about yourself and your background; the second asks you to answer questions about two co-workers: someone you worked best with on a matter or law firm related project the project and someone you did not work well with on a matter or law firm related project. The final section asks a few questions about the general work environment at FIRM NAME. For most questions, you simply select a numbered response that best matches your opinion.
It should take approximately 20 minutes to answer the questions.
The risks associated with completing the survey are minimal and are no greater than those you may encounter in everyday work life. Please note that:
• The questionnaire is completely voluntary. • Your answers will be treated confidentially and with anonymity. • The questionnaire will not ask you to identify yourself, your co-workers, or the
name of the matter or project you are referencing. • No personnel records will be used and no matching or personal characteristics
will be made. Your identity will not be revealed in the reporting of the study’s results.
• The collected data will be housed on a secure server at the University of Toronto. • FIRM NAME will not have access to any raw data. Only members of the
academic research team will have access to this data. • You have the right to withdraw consent and discontinue your participation at any
time. There are no penalties or consequences if you choose not to participate or if you choose to withdraw.
The researchers intend to publish the study’s results in scholarly journals. In all publications, including the summary report, the identity of participants will remain confidential. Should you have any questions or concerns about these procedures or the project in general, please feel free to contact Max Evans ([email protected], 416-
280
854-6896), Dr. Chun Wei Choo ([email protected], 416-978-5266), or Dr. Anthony Wensley ([email protected], 905-569-4733). If you have any questions about your rights as a participant, please contact the University of Toronto Office of Research Ethics ([email protected], 416-946-3273). A summary of results would be made available to individual participants upon request. Please email the researcher if you are interested. As a small token of our appreciation, we will also give you a $5 gift card to COFFEE SHOP NAME at the end of the survey. To begin the questionnaire, simply click on the “I consent” button below. Clicking on the “I consent” button gives us your consent for participation. Doing so indicates that you agree to the following statements:
1. I have freely volunteered to participate. 2. I have been informed in advance about the nature of the questionnaire, what my
tasks will be, and what procedures will be followed. 3. I have been given the opportunity to ask questions and have had my questions
answered to my satisfaction. 4. I understand that the information I provide will be treated confidentially and with
anonymity. My identity will not be revealed in the reporting of the study’s results. 5. I am aware that I have the right to withdraw consent and discontinue participation
at any time.
<I Do Not Consent> and <I Consent> buttons on survey web page Section 1: Individual Section Please answer the questions in this section about yourself. 1. Age
21-30 31-40 41-50 51-60 Over 61 Prefer not to say
2. Sex Male Female Prefer not to say
3. What Country were you born in?
Select Country Prefer not to say
4. What Country are you a citizen of? Select Country Prefer not to say
281
5. Ethnicity (Mark more than one or specify, if applicable.) White Chinese South Asian (e.g., East Indian, Pakistani, Sri Lankan, etc.) Black Filipino Latin American Southeast Asian (e.g., Vietnamese,
Cambodian, Malaysian, Laotian, etc.) Arab West Asian (e.g., Iranian, Afghan, etc.) Korean Japanese Other Prefer not to say
6. What is the highest level of educational qualification that you have achieved?
High School / GED Some College / 2-year College Degree 3 or 4-year University Degree Professional Degree (JD/LL.B.,MD) Masters Degree Doctoral Degree
7. What is your Marital Status?
Never legally married (single) Legally married (and not separated) Separated, but still legally married Divorced Widowed Common Law Prefer not to say
8. Which department are you a part of?
Business Litigation
9. Which office do you primarily work in?
Toronto / Region of Waterloo Vancouver Calgary Montreal Ottawa
10. What is your role within the firm?
Law Clerk / Paralegal Articling Student Associate Partner
11. Approximately how many years have you worked for FIRM NAME?
Months/Yrs
12. Approximately how many years have you worked in your current role? Months/Yrs
282
13. Please indicate your level of agreement with each of the following 8 statements: One should be very cautious with strangers.
1 2 3 4 5 Strongly Strongly Disagree Agree
Most experts tell the truth about the limits of their knowledge. Most people can be counted on to do what they say they will do. These days, you must be alert or someone is likely to take advantage of you. Most salespeople are honest in describing their products. Most repair people will not overcharge people who are ignorant of their specialty. Most people answer public opinion polls honestly. Most adults are competent at their jobs.
For the questions in the next section you are asked to mentally select TWO (2) co-workers with whom you shared a completed matter or firm-related project.
• The first co-worker should be someone you have WORKED BEST with on the matter or firm-related project you shared.
• The second co-worker should be someone you did NOT WORK WELL with on the matter or firm-related project you shared.
• Both co-workers need not come from the same matter or firm-related project, but it is important that you interacted with both individually on whichever matter or firm project you shared.
With BOTH with these people in mind, respond to each of the following questions twice, once for the person you worked best with and once for the person you did not work well with. Co-worker Section – Please respond to each of the questions on this page twice, once for the person you worked best with and once for the person you did not work well with. 14. Age
Person I WORKED BEST with 21-30 31-40 41-50 51-60
Over 61 I am not able to assess
Person I DID NOT WORK WELL with 21-30 31-40 41-50 51-60
Over 61 I am not able to assess 15. Sex Person I WORKED BEST with Male Female Person I DID NOT WORK WELL with Male Female 16. Country of Birth Person I WORKED BEST with Select Country I don’t know Person I DID NOT WORK WELL with Select Country I don’t know
283
17. Country of Citizenship Person I WORKED BEST with Select Country I don’t know Person I DID NOT WORK WELL with Select Country I don’t know 18. Ethnicity (Mark more than one, if applicable.) Person I WORKED BEST with
White Chinese South Asian (e.g., East Indian, Pakistani, Sri Lankan, etc.) Black Filipino Latin American Southeast Asian (e.g., Vietnamese, Cambodian, Malaysian, Laotian, etc.) Arab West Asian (e.g., Iranian, Afghan, etc.) Korean Japanese Other I don’t know
Person I DID NOT WORK WELL with
White Chinese South Asian (e.g., East Indian, Pakistani, Sri Lankan, etc.) Black Filipino Latin American Southeast Asian (e.g., Vietnamese, Cambodian, Malaysian, Laotian, etc.) Arab West Asian (e.g., Iranian, Afghan, etc.) Korean Japanese Other I don’t know
19. Marital Status Person I WORKED BEST with Never legally married (single) Legally
married (and not separated) Separated, but still legally married Divorced Widowed
Common-Law I don’t know Person I DID NOT WORK WELL with Never legally married (single) Legally
married (and not separated) Separated, but still legally married Divorced Widowed
Common-Law I don’t know 20. Please select the highest level of educational qualification that each of your co-
workers has achieved (if known) Person I WORKED BEST with High school / GED Some college / 2-
year College Degree 3 or 4-year University Professional Degree (JD/LLB, MD)
Masters Degree Doctoral Degree I don’t know
Person I DID NOT WORK WELL with High school / GED Some college / 2-year College Degree 3 or 4-year University
Professional Degree (JD/LLB, MD) Masters Degree Doctoral Degree I don’t know
21. Approximately, how many years has each co-worker worked for FIRM NAME? Person I WORKED BEST with Months/Yrs I don’t know Person I DID NOT WORK WELL with Months/Yrs I don’t know 22. Approximately, how many years has each co-worker worked in their current role? Person I WORKED BEST with Months/Yrs I don’t know Person I DID NOT WORK WELL with Months/Yrs I don’t know
284
23. Approximately, how long have you known each of the co-worker you selected? Person I WORKED BEST with < 1 month 1 < 3 months 3 < 6
months 6 < 9 months 9 months < 1 year 1 < 1.5 years 1.5 < 2 years 2 < 3
years 3 < 4 years 4 < 5 years 5 < 6 years 6 < 7 years 7 < 8 years 8 < 9 years 9 < 10 years 10 < 15 years Over 15 years
Person I DID NOT WORK WELL with < 1 month 1 < 3 months 3 < 6 months 6 < 9 months 9 months < 1 year
1 < 1.5 years 1.5 < 2 years 2 < 3 years 3 < 4 years 4 < 5 years 5 < 6 years 6 < 7 years 7 < 8 years 8 < 9 years 9 < 10 years 10 < 15 years Over 15 years
24. Prior to working with each of the co-workers on the matter or firm-related project
you shared, how close was your working relationship? Person I WORKED BEST with No Prior Contact Distant Somewhat
Distant Somewhat Close Close Very Close
Person I DID NOT WORK WELL with No Prior Contact Distant Somewhat Distant Somewhat Close Close Very Close
25. Prior to working with each of the co-workers on the matter or firm-related project
how often did you communicate? Person I WORKED BEST with Never Once every 3 months or less
Once every 2nd month Once a month Twice a month Once a week Twice a week Daily
Person I DID NOT WORK WELL with Never Once every 3 months or less Once every 2nd month Once a month Twice a month Once a week Twice a week Daily
26. Prior to working with each of the co-workers on the matter or firm-related project to
what extent did you typically interact? Person I WORKED BEST with To No Extent To Little Extent To
Some Extent To a Great Extent To a Very Great Extent
Person I DID NOT WORK WELL with To No Extent To Little Extent To Some Extent To a Great Extent To a Very Great Extent
285
27. My relationship with each of the co-workers I mentally selected was a very intense, strong relationship prior to working on the matter or firm related project we shared.
Person I WORKED BEST with 1 2 3 4 5 Strongly Strongly Disagree Agree
Person I DID NOT WORK WELL with 1 2 3 4 5 Strongly Strongly Disagree Agree
28. While working on the matter or firm-related project you shared, how close was your
working relationship with each of the co-workers you mentally selected? Person I WORKED BEST with Distant Somewhat Distant
Somewhat Close Close Very Close Person I DID NOT WORK WELL with Distant Somewhat Distant
Somewhat Close Close Very Close 29. While working on the matter or firm-related project you shared, how often did you
communicate with each of the co-workers you mentally selected? Person I WORKED BEST with Once every 3 months or less Once every
2nd month Once a month Twice a month Once a week Twice a week Daily
Person I DID NOT WORK WELL with Once every 3 months or less Once every 2nd month Once a month Twice a month Once a week Twice a week Daily
30. While working on the matter or firm-related project you shared, to what extent did
you typically interact with each of the co-workers you mentally selected? Person I WORKED BEST with To No Extent To Little Extent To
Some Extent To a Great Extent To a Very Great Extent
Person I DID NOT WORK WELL with To No Extent To Little Extent To Some Extent To a Great Extent To a Very Great Extent
31. My relationship with each of the co-workers I mentally selected was a very intense,
strong relationship while working on the matter or firm related project we shared. Person I WORKED BEST with 1 2 3 4 5
Strongly Strongly Disagree Agree
Person I DID NOT WORK WELL with 1 2 3 4 5 Strongly Strongly Disagree Agree
286
32. Please indicate your level of agreement with each of the statements below. 1 2 3 4 5
Strongly Strongly Disagree Agree
Person I WORKED BEST with
Person I DID NOT WORK WELL with
- I believed that this person approached his or her job with professionalism and dedication. - Given his or her track record, I saw no reason to doubt this person’s competence and preparation. - This person is very capable of performing his/her job. - This person is known to be successful at the things he/she tries to do. - This person has much knowledge about the work that needs done. - I feel confident about this person’s skills. - This person has specialized capabilities that can increase performance. - This person is well qualified. - I can rely on this person not to make my job more difficult by careless work. - This person is very concerned about my welfare. - My needs and desires are very important to this person. - This person would not knowingly do anything to hurt me.
33. Please indicate your level of agreement with each of the statements below.
1 2 3 4 5 Strongly Strongly Disagree Agree
Person I WORKED BEST with
Person I DID NOT WORK WELL with
- This person really looks out for what is important to me. - This person would go out of his or her way to help me. - This person would go out of his or her way to make sure I am not damaged or harmed in this relationship. - I feel like this person cares what happens to me.
287
- This person would always look out for my interests. - I feel like this person is on my side. - This person has a strong sense of justice. - I never have to wonder whether this person will stick to his/her word. - This person tries hard to be fair in dealings with others. - This person’s actions and behaviors are not very consistent. - I like this person’s values. - Sound principles seem to guide this person’s behavior. 34. Please indicate your level of agreement with each of the statements below.
1 2 3 4 5 Strongly Strongly Disagree Agree
Person I WORKED BEST with
Person I DID NOT WORK WELL with
- I could understand completely what this person meant when he or she was talking. - I was familiar with the jargon/terminology that he or she used. - It felt like we could communicate on the same "wavelength”. - I felt like this person and I were working toward completely different goals. - I assumed that this person and I cared about the same issues. - I believed that this person and I shared a commitment to a common purpose. - I believed that this person and I shared the same ambitions and vision. - I believed that this person and I shared enthusiasm about pursuing the collective goals and mission of the whole organization.
288
35. Please indicate your level of agreement with each of the statements below. 1 2 3 4 5
Strongly Strongly Disagree Agree
Person I WORKED BEST with
Person I DID NOT WORK WELL with
- I would take the initiative to provide this individual with useful tools I have developed (e.g. precedents, memos, client information, industry information). - I would allow this individual to spend significant time observing and collaborating with me in order for him/her to better understand and learn from my work. - I would willingly share with this person rules of thumb, tricks of the trade, and other insights into the work of my office and that of the organization I have learned. - I would willingly share my new ideas with this individual. - I would willingly share with this individual the latest organizational rumors, if significant.
36. Please indicate your level of agreement with each of the statements below.
1 2 3 4 5 Strongly Strongly Disagree Agree
Person I WORKED BEST with
Person I DID NOT WORK WELL with
- I would eagerly receive and use tools developed by this person including precedents, memos, client information and industry information. - I would welcome the opportunity to spend significant time observing and collaborating with this individual in order for me to better understand and learn from his/her work. - I would welcome and use any rules of thumb, tricks of the trade, and other insights they have learned. - I would eagerly receive and consider any new ideas this individual might have. - I would tend to believe organizational rumors shared by this individual and would use such knowledge as appropriate.
289
37. The information I received from each of the co-workers made (or is likely to make) the following contribution to:
1 2 3 4 5 Very Very Negative Positive
Person I WORKED BEST with
Person I DID NOT WORK WELL with
- Client satisfaction with the matter / project. - The matter / projects quality. - The project team's overall performance. - The overall success of FIRM NAME. - The cost and/or the time it took to complete the portion of the matter/project I am responsible for. - My individual performance on the matter/project.
290
38. Please indicate your level of agreement with each of the following statements about the general work environment at FIRM NAME68.
1 2 3 4 5 Strongly Strongly Disagree Agree
- I feel safe to openly say what I am really thinking to anyone else in FIRM NAME, regardless of their position. - I feel safe to question others' decisions without fear of negative consequences. - My colleagues regularly ask me for my opinion on client matters that they are working on. - I regularly ask my colleagues for their opinions on client matters that I am working on. - The knowledge that I need to work on files is readily available to me. - The precedents that I need to work on files are readily available to me. - Information flows freely within my regional practice group. - Information flows freely within my national practice group. - Colleagues freely share legal information that would be helpful on client matters. - Colleagues freely share industry information that would be helpful on client matters. - Colleagues freely share client information that would be helpful on client matters. - Client contact partners freely share client information with others within the firm. - I know where to go to access the information that I need to perform my work on client matters. - Information flows freely between those in management and those who are not in management. - I feel that I have access to all of the information that I need to be an effective professional.
68 Item was included at the request of the firm. However, the responses for the item were not analyzed in the present study.
291
A.3 Firm-wide Invitation Email Subject: Knowledge Sharing within FIRM NAME - Survey of FIRM NAME professionals/Sondage sur le partage des connaissances chez FIRM NAME Importance: High
La version française de ce message suit la version anglaise.
You are invited to complete a survey that is going to all FIRM NAME professionals and law clerks. The survey is meant to measure how we use information collaboratively within the firm. The survey is being undertaken by Max Evans in the Faculty of Information Studies at the U. of T. with the consent and support of FIRM NAME. The survey will form part of Max's doctoral dissertation.
The survey has three parts. The first is an individual section, which asks you to answer questions about yourself and your background; the second asks you to answer questions about two coworkers: someone you worked best with on a matter or law firm related project and someone you did not work well with on a matter or law firm related project. The final section asks a few questions about the general work environment at FIRM NAME.
The individual survey responses are anonymous and confidential. Max will only share with us aggregate results.
We feel that this will be an excellent baseline study for us to see how well we collaborate. Perhaps a similar survey will be in order after we complete our SharePoint based portal, which will provide for greater collaboration tools for our professionals.
Note that there is a $5 reward in the form of a Starbucks coffee certificate for completing the survey. All you will have to do to redeem the certificate is print out the last page of the survey (see sample below) and bring it to the receptionist in your office. (Toronto participants should go to the reception desk on the 47th floor, behind the opaque glass doors.) They will be able to exchange the printout for a physical card. Please note one giftcard per participant.
We estimate that the survey will take between 15-20 minutes to complete. For most questions you simply select a numbered response that best matches your opinion.
To begin the survey, please click on the following link:
http://survey.qualtrics.com/SE/?SID=SV_6ukUUwsHo9Aqlne
Thank you in advance for your participation.
Veuillez remplir le sondage (voir le lien ci-dessous) qui est envoyé à tous les professionnels et stagiaires de FIRM NAME. L’objectif du sondage est de mesurer de
292
quelle manière , au cabinet, nous utilisons l’information en collaboration. Max Evans, de la Faculté des études en information de l’Université de Toronto, se charge du sondage, avec le consentement des cadres supérieurs de FIRM NAME et le soutien de l’équipe de la Gestion du savoir du cabinet. Le sondage fera partie de la thèse de doctorat de Max.
Le sondage compte trois parties. Dans la première, on vous demande de répondre à des questions personnelles et à des questions sur vos antécédents; dans la deuxième, on vous demande de répondre à des questions au sujet de deux collègues : 1) un collègue avec lequel vous avez travaillé de façon très satisfaisante, dans le cadre d’un dossier ou d’un projet relié au cabinet, et 2) un collègue avec lequel vous n’avez pas eu de relations de travail heureuses, également dans le cadre d’un dossier ou d’un projet relié au cabinet. Dans la troisième partie, on vous pose quelques questions au sujet du milieu de travail en général, chez FIRM NAME.
Les réponses que vous donnez au sondage sont anonymes et confidentielles. Max ne nous fera part que des résultats globaux.
À notre avis, il s’agit là d’une excellente étude de base qui reflétera bien la collaboration qui règne au cabinet.
Veuillez prendre note qu’une récompense de 5 $, à savoir une carte Starbucks, est remise à chaque personne qui répond au sondage. Pour obtenir la carte, imprimez simplement la dernière page du sondage et remettez-la à la réception principale de votre bureau (dans le cas du bureau de Toronto, veuillez vous rendre à la réception du 47e étage, derrière les portes de verre opaque, du côté ouest). On échangera alors votre page imprimée contre une carte Starbucks. Veuillez noter que chaque personne qui répond au sondage ne peut recevoir qu’une seule carte.
Nous croyons qu’il faudra de 15 à 20 minutes pour répondre au sondage. Dans la plupart des cas, vous n’avez qu’à sélectionner le numéro qui correspond le mieux à la réponse que vous souhaitez donner.
Pour commencer à répondre au sondage, veuillez cliquer sur le lien suivant :
http://survey.qualtrics.com/SE/?SID=SV_6ukUUwsHo9Aqlne
Je vous remercie à l’avance de votre participation.
293
A.4 University of Toronto Ethics Approval Letter