EMPLOYEE RESPONSE TO NEGATIVE WORKPLACE SITUATIONS:
DOES CULTURE MATTER?
By
Joo-Young Park

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Industrial Relations and Human Resources
2011

ABSTRACT
EMPLOYEE RESPONSE TO NEGATIVE WORKPLACE SITUATIONS:
DOES CULTURE MATTER?
By
Joo-Young Park

The purpose of this paper is to investigate how culture affects employee response to
negative workplace situations, by attempting to answer the following questions: (1) Does culture
affect employee response to negative workplace situations? (2) If so, how does culture affect
employee response to negative workplace situations? In other words, which culture-specific
values affect employee response to negative workplace situations? What key features of a nonwestern culture directly make a difference in the application of existing voice-related models?
To answer these questions, this dissertation suggested a comprehensive theoretical model
built on two theoretical grounds—the exit-voice theory proposed by Hirschman (1970) and the
culture model developed by Kozan (1997). This study started with Hirschman‘s initial
conceptualization of voice and Rusbult and colleagues (1982)‘s Exit-Voice-Loyalty-Neglect
(EVLN) construct for the theoretical framework to explain employee response to negative
workplace situations. To explore the cultural factors to be tested in this dissertation, Kozan‘s
culture model was used to suggest key cultural values in the Asian countries. This dissertation
suggested that in the Asian countries employees will be guided by the values of collectivism,
face-saving, and conflict-avoiding when they respond to negative workplace situations. In
addition, the predictors found in previous studies—including alternative quality, self-esteem,
Power Distance, and LMX quality—were added to the proposed model to examine their effect
on EVLN behaviors.

A field study was conducted in automotive facilities both in Korea and in the U.S. Based
on the data analysis, several interesting findings emerged. First, and most notably, for a ‗leaver,‘
who choose the exit option, culture does not matter such that 1) a country effect was not found in
the exit option; 2) none of the three harmony culture values have a significant association with
the exit option across countries. Second, for a ‗stayer,‘ who choose voice, loyalty, or neglect
option, culture does matter in that 1) country effect was observed in all non-exit options; 2)
culture-specific values, such as collectivism, face-saving, and conflict-avoiding were found to
affect employees will choose all non-exit options in the Korean sample, but not in the U.S.
sample. It supported that these three values guide and predict voice behaviors those belonging to
such a culture. Finally, aside from these culture-specific values, global factors working cross
cultures were found in both the Korean and the U.S. samples, such as LMX for the voice option
and alternative quality for the exit option. Additionally, the results of the data analysis show that
these EVLN responses can be categorized as constructive (the voice and loyalty option) or
destructive (the exit and neglect option), depending on the consequences of these behaviors.
Collectivism and face-saving have a significant and positive association with constructive
responses, while conflict-avoiding has a significant and positive association with the destructive
responses. Overall, this study presents strong empirical evidence of the effect of culture on
employee voice-related behaviors, increasing our understanding of employee voice behavior
across cultures.

Copyright by
JOO-YOUNG PARK
2011

To my parents

v

ACKNOWLEDGMENTS
On the completion of this dissertation, I am indebted to many people for supporting me.
To all those who have made this dissertation possible and because of whom my graduate
experience has been one I always will cherish, I express my sincere thanks.
First, my deepest gratitude is to my advisor, Dr. Richard N. Block, accompanied with me
throughout my doctoral journey. His inspirational guidance, support and encouragement gave me
energy to overcome many unexpected obstacles and finish this dissertation. In particular, it has
been a huge honor to be the last student to have him as my advisor before his retirement. As I
told him, he will be my advisor always.
I am heartily thankful to the rest of my committee, Dr. Michael Moore, Dr. Peter Berg,
and Dr. James Dulebohn. I thank them for their guidance and insight to better my work over the
years. Thanks also to Dr. Rubén Martinez for his extra work as a Dean's Representative for my
dissertation defense. I thank to Dr. Eun-Jong Shin; he has been my spiritual mentor for this long
journey. My appreciation is also extended to Dr. Dong-One Kim and Dr. Kiu-Sik Bae for their
generosity of offering their precious time whenever I had questions. In addition, I would like to
thank Dr. Eun-Sil Lee; her financial and emotional support in my very last challenging moment
helps me complete this study.
I would also like to thank the organizations that allowed me access to their employees
who took the time to participate in my research. I learned a lot from them, and without their help,
I could never conduct this study. I sincerely appreciate their contributions.
Thank you to Young-Hee Kang, Matt Piszczek, Kritkorn Nawakitphaitoon; they were my
fine colleagues in my doctoral program. Thanks especially to Kaumudi Misra; throughout my
doctoral study years, she shared with me so many moments. I would like to give a big hug to all

vi

my friends at MSU; they accompanied me at every stage of my path at MSU with love and
support. I would also like to thank all members of the Lansing Korean Tennis Club. What I have
learned from them is more than tennis skills; without them, I could never keep healthy physically
as well as mentally.
Of course, extra-most important, special thanks to my parents; Heung-Kook Park and
Guem-Soon Lee. Without their unconditional and endless love and support, none of this
dissertation would have been possible.
Last, thanks to holy spirits who have stayed with me and helped me see the light at the
end of the tunnel whenever I saw darkness in this long struggling journey.

vii

TABLE OF CONTENTS
LIST OF TABLES ......................................................................................................................... xi
LIST OF FIGURES ...................................................................................................................... xii
CHAPTER 1 INTRODUCTION .....................................................................................................1
1.1 Problem Statement .................................................................................................................1
1.2 Purpose of Study ....................................................................................................................4
1.3 Research Questions ................................................................................................................5
1.4 Overview of Dissertation .......................................................................................................5
CHAPTER 2 LITERATURE REVIEW ..........................................................................................7
2.1 Defining Employee Voice Behavior ......................................................................................7
2.1.1 Approaches of Employee Voice Behavior ......................................................................7
2.1.2 Hirschman‘s Definition of Voice ....................................................................................9
2.1.3 Features of Hirschman‘s Definition of Voice ...............................................................10
2.2 Exit-Voice-Loyalty-Neglect Construct ................................................................................12
2.2.1 Defining the EVLN Construct .......................................................................................12
2.2.2 Evidences of Construct Validity of the EVLN Construct .............................................15
2.2.3 Rationales to Use the EVLN Construct .........................................................................16
2.3 Predictors of Employee Voice Behavior ..............................................................................18
2.3.1 Individual Factors ..........................................................................................................19
2.3.2 Organizational Factors...................................................................................................23
2.4 Culture and Employee Voice Behavior ................................................................................26
2.4.1 Defining Culture ............................................................................................................26
2.4.2 Hofstede‘s Culture Study ..............................................................................................27
2.4.3 Culture in Employee Voice Behavior Study .................................................................28
2.5 Kozan‘s Culture Model ........................................................................................................30
2.5.1 Kozan‘s Three Culture Models .....................................................................................31
2.5.2 Empirical Evidences of Kozan‘s Culture Models .........................................................33
2.6 Summary of Literature Review ............................................................................................35
CHAPTER 3 MODEL ...................................................................................................................36
3.1 Research Focus .....................................................................................................................36
3.2 Proposed Model....................................................................................................................37
3.2.1 EVLN Construct as Dependent Variable ......................................................................38
3.2.2 Rationale for Applying Kozan‘s Culture Model to Employee Voice Behavior ............39
3.2.3 Cultural Values from Kozan‘s Culture Model ..............................................................41
3.2.4 Additional Predictors of Employee Voice Behavior .....................................................42
3.3 Hypotheses ...........................................................................................................................43
3.3.1 Collectivism...................................................................................................................44
3.3.2 Face-Saving ...................................................................................................................46
3.3.3 Conflict-Avoiding..........................................................................................................49
3.3.4 Alternative Quality ........................................................................................................53
3.3.5 Self-Esteem....................................................................................................................55

viii

3.3.6 Power Distance ..............................................................................................................57
3.3.7 LMX ..............................................................................................................................59
3.3.8 Control Variables ..........................................................................................................61
CHAPTER 4 METHODOLOGY ..................................................................................................63
4.1 Research Design ...................................................................................................................63
4.1.1 Rationale for Research Method .....................................................................................63
4.1.2 Sampling Strategies .......................................................................................................63
4.1.2.1. Selection of Country ..............................................................................................64
4.1.2.2. Selection of Subgroup ............................................................................................65
4.1.2.3. Employee Sample, Not Student Sample ................................................................65
4.2 Determination and Acquisition of Subject ...........................................................................66
4.3 Subjects ................................................................................................................................69
4.3.1 Rationale for Sample Size .............................................................................................69
4.3.2 Demographic Information of Subjects ..........................................................................70
4.4 Translation of Materials .......................................................................................................73
4.5 Measures ...............................................................................................................................73
4.5.1 Dependent Variables: the EVLN Construct ..................................................................73
4.5.2 Independent Variables ...................................................................................................75
4.5.3 Control Variables...........................................................................................................78
4.6 Analysis Strategies ...............................................................................................................79
4.6.1 Confirmatory Factor Analysis (CFA) ............................................................................79
4.6.2 Model Fit .......................................................................................................................80
4.6.3 Multiple Regression.......................................................................................................82
4.6.4 Chow Test......................................................................................................................83
CHAPTER 5 RESULTS ................................................................................................................85
5.1 Descriptive Statistics ............................................................................................................85
5.2 Correlations among Variables ..............................................................................................89
5.3 Results of Confirmatory Factor Analysis .............................................................................94
5.4 Regression Outcomes ...........................................................................................................96
5.4.1 Regression Analysis with the Total Sample ..................................................................97
5.4.1.1 Regression on Exit Option ......................................................................................97
5.4.1.2 Regression on Voice Option .................................................................................102
5.4.1.3 Regression on Loyalty Option ..............................................................................105
5.4.1.4 Regression on Neglect Option ..............................................................................109
5.4.2 Comparison between the Korean Sample and the U.S. Sample ..................................112
5.4.2.1 Regression on Exit Option ....................................................................................113
5.4.2.2 Regression on Voice Option .................................................................................115
5.4.2.3 Regression on Loyalty Option ..............................................................................117
5.4.2.4 Regression on Neglect Option ..............................................................................120
5.4.3 Additional Analysis .....................................................................................................123
5.4.3.1 Constructive and Destructive Responses with the Total Sample .........................125
5.4.3.2 Constructive and Destructive Responses in the Korean Sample and the U.S.
Sample ..................................................................................................................128

ix

CHAPTER 6 DISCUSSION ........................................................................................................132
6.1 Summary of Findings .........................................................................................................132
6.1.1 Findings from Regressions on EVLN with Total Sample ...........................................132
6.1.2 Findings from Comparison between the Korean Sample and the U.S. Sample ..........135
6.1.3 Findings from Regressions on Constructive and Destructive Response .....................137
6.2 Interpreting Findings ..........................................................................................................139
6.2.1 Research Question One: Does Culture Affect Employee Voice Behavior?................139
6.2.1.1 For a ‗Leaver,‘ Culture Does Not Matter .............................................................140
6.2.1.2 For a ‗Stayer,‘ Culture Does Matter .....................................................................143
6.2.2 Research Question Two: Which Culture Specific Values Affect Employee Voice
Behavior? ......................................................................................................................145
6.2.2.1 Collectivism, Face-saving, and Conflict-Avoiding in a Harmony Culture ..........145
6.2.2.2 The Effects of These Values Vary Depends on Each Response Option ..............146
6.2.2.3 Among Them, Conflict-Avoiding is the Most Widely Influencing Predictor in the
Harmony Culture ..................................................................................................147
6.2.3 However, There Are Still Universal Predictors Available across Cultures.................148
6.2.4 Dynamics within the EVLN construct with Constructive and Destructive dimension
..............................................................................................................................................148
6.3 Results of Hypotheses Testing ...........................................................................................150
CHAPTER 7 CONCLUSION......................................................................................................153
7.1 Summary of the Dissertation ..............................................................................................153
7.1.1 Summary of the Purpose of the Study and the Theoretical Background ....................153
7.1.2 Summary of Methodology...........................................................................................156
7.1.3 Summary of Results ....................................................................................................157
7.2 Contributions ......................................................................................................................159
7.2.1 Theoretical Contributions ............................................................................................159
7.2.2 Methodological Contributions .....................................................................................162
7.2.3 Empirical Contributions ..............................................................................................164
7.2.4 Practical Contributions ................................................................................................166
7.3 Limitations and Future Research ........................................................................................171
7.4 Concluding Remarks ..........................................................................................................182
APPENDICES .............................................................................................................................183
Appendix A: Survey (in English) .............................................................................................184
Appendix A: Survey (in Korean) .............................................................................................188
REFERENCES ............................................................................................................................193

x

LIST OF TABLES
Table 1. Three cultural models of conflict management .............................................................31
Table 2. Subject Demographic Statistics .....................................................................................71
Table 3. Items of EVLN construct ...............................................................................................74
Table 4. Items of Independent Variables .....................................................................................77
Table 5. Descriptive Statistics for All Variables for the Analyses ..............................................86
Table 6. Correlation Matrix .........................................................................................................90
Table 7. Regression Outcome on Exit with the total sample .......................................................98
Table 8. Regression Outcome on Voice with the total sample ..................................................104
Table 9. Regression Outcome on Loyalty with the total sample ...............................................106
Table 10. Regression Outcome on Neglect with the total sample ...............................................110
Table 11. Regression Outcome on Exit with the Korean sample and the U.S. sample ...............114
Table 12. Regression Outcome on Voice with the Korean sample and the U.S. sample ............116
Table 13. Regression Outcome on Loyalty with the Korean sample and the U.S. sample .........118
Table 14. Regression Outcome on Neglect with the Korean sample and the U.S. sample .........121
Table 15. Regression Outcome on Constructive vs. Destructive response with the total sample126
Table 16. Coefficient Outcome on Constructive vs. Destructive response with the Korean sample
and the U.S. sample .....................................................................................................................129
Table 17. Summary of Hypotheses Test ......................................................................................151

xi

LIST OF FIGURES
Figure 1. Model ..............................................................................................................................38
Figure 2. Outcome of Confirmatory Factor Analysis ....................................................................95

xii

CHAPTER 1
INTRODUCTION

This chapter includes the following sections: (1) Problem Statement; (2) Purpose of
Study; (3) Research Questions; and (4) Overview of Dissertation.

1.1 Problem Statement
The way in which employees respond to negative or problematic situations at the
workplace has been suggested to have an important impact on work-related behaviors, the
employees‘ relationship with the organization, and eventually the effectiveness of the
organization (Dundon et al., 2004; Farrell & Rusbult, 1992; Hagedoorn et al., 1999; Leck &
Saunders, 1992; Lewicki & Minton, 1992; Rusbult et al., 1988; Staw, 1984; Van den Bos & van
Prooijen, 2001). These behaviors also have implications for understanding the rights and
responsibilities of both managers and employees, teaching organizations how to encourage
desired employee behaviors and how to discourage undesired ones (Leck & Saunders, 1992).
The importance of employee voice has been supported in an impressive number of
previous studies over the years. Some of the most robust and accepted evidence of the
importance of employee voice has been observed in the organizational justice literature (e.g.
Brockner et al., 1998; De Cremer & Alberts, 2004; Folger, 1977; Folger et al., 1979; Lind &
Tyler, 1988; Lind, Kanfer, & Earley, 1990; Thibaut & Walker, 1975; Tyler, 1987; Tyler,
Rasinski, & Spodick, 1985; Van den Bos, 1999, 2005; Van den Bos & Spruijt, 2002). Numerous
studies have revealed that whether or not employees are allowed voice is an important factor for
these positive reactions since employees believe that voice will help them control their outcomes

１

by persuading the decision-maker to provide a better outcome, either a favorable outcome
(Leventhal, 1980) or an equitable outcome (Thibaut & Walker, 1978), which is categorized as
instrumental consequences (Leventhal, 1980; Thibaut & Walker, 1978). In addition, some
researchers (e.g. Lane, 1988; Lind & Tyler, 1988) have explained that employee voice is
important since it sends employees a positive signal about the organization that they belong to,
which means the symbolic and informational consequences of procedures or noninstrumental
consequences (Lind, Kanfer, & Earley, 1990).
Therefore, these studies have found that people show more positive reactions when they
are allowed an opportunity to voice their views and opinions in the decision-making process than
when they are denied participation (Van Den Bos & Spruijt, 2002). The instances of these
positive reactions include employees‘ perceptions of procedural fairness, attitude toward
supervisors, relational judgments about authorities, task performance, and measures of affect,
such as anger toward an official (e.g. De Cremer & Alberts, 2004; Lind, Kanfer, & Earley, 1990;
Lind & Tyler, 1988; Tyler, 1987; Van den Bos & Spruijt, 2002). These consistent findings have
confirmed employee voice as one of the most critical phenomena in an organization.
A stream of research that examines how employees respond to negative or problematic
situations in the workplace has determined that they react in one of four ways: they can leave the
organization—exit, express their opinion or idea to make the situation better—voice, continue to
support the organization—loyalty, or reduce their effort—neglect. This categorization is often
referred to as the EVLN construct: exit, voice, loyalty, and neglect. Since the publication of
Hirschman's Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations and States
(1970), the topic of employee voice has attracted academic interest across research domains such
as human resource management, organizational behavior, I/O psychology, and industrial

２

relations, as a means by which employees deal with negative or problematic events in the
workplace (Addison & Belfield, 2004; Batt et al., 2002; Beardwell, 1998; Benson, 2000; Dundon
et al., 2004; Farrell, 1983; Farrell & Rusbult, 1992; Dowding et al., 2000; Lewin & Mitchell,
1992; Roche, 2000; Rusbult et al., 1986, 1988; Withey & Cooper; 1989; Van Dyne et al., 2003).
For instance, in the field of procedural justice, researchers have convincingly suggested
that employee voice is one of the most important and robust findings as a key determinant of
procedural justice (Lind & Tyler, 1988; Tyler, 1987; Tyler & Lind, 1992; Van den Bos, Lind,
Vermunt, & Wilke, 1997), leading employees to positive reactions, such as enhanced perceived
procedural justice (Folger, 1977; Folger et al., 1979; Lind & Tyler, 1988; Lind et al., 1990;
Thibaut & Walker, 1975; Van den Bos & van Prooijen, 2001) and less negative affect such as
anger against officials (Tyler, 1987). In labor economics, researchers conclude that voice lowers
employee quit rates, raises employee job tenure, and enhances employer investment in human
capital and increases employee productivity (Bemmels, 1997; Boroff & Lewin, 1997; Freeman,
1980; Freeman & Medoff, 1984; Lewin & Mitchell, 1992; Miller & Mulvey, 1991).
In the organizational citizenship behavior (OCB) research, employee voice is regarded as
a critical issue since the challenges of global competition and a rapidly changing organizational
environment highlight the importance of organizational innovation, flexibility, productivity, and
responsiveness to changing external conditions (Van Dyne, 1994). New ideas from employees
facilitate continuous improvement (Nemeth & Staw, 1989; Van Dyne & LePine, 1998). For
instance, Sheppard and colleagues (1992) found that reporting a problem to a supervisor may
have more beneficial effects than avoiding the problem by staying away from work, because
reporting the problem may reduce distress and may alert the organization that some procedure or
practice is not working. Leck and Saunders (1992) also suggested that constructive responses

３

such as trying to improve working conditions, providing management with suggestions, and
alerting management to problems can increase organizational productivity and effectiveness. On
the other hand, less constructive responses such as quitting, absenteeism, or psychological
withdrawal may adversely affect the organization's profitability.
Despite impressive advances in the research on employee voice behavior over the course
of the four decades, there is still more to be uncovered for a better understanding of the EVLN
construct. One aspect of research that has received little attention is the effect of culture,
particularly national culture, on employee voice behavior at work (Botero & Van Dyne, 2009;
Edward & Greenberg, 2009; Landau, 2009; Lee & Jablin, 1992; Price et al., 2001; Xu et al.,
2005). Although a large number of previous studies suggest that the various cultural dimensions
influence work-related psychological and behavioral phenomena (Bontempo & Rivero, 1992;
Markus & Kitayama, 1991; Suh et al., 1998; Triandis et al., 1995; Wagner, 1995; Wasti, 2003),
very few studies have been published examining how culture affects employee voice behavior.
Therefore, further cultural research has been proposed to be one of the most deserving areas for
future research t on employee voice (Botero & Van Dyne, 2009; Edward & Greenberg, 2009;
Landau, 2009; Lee & Jablin, 1992; Price et al., 2001; Xu, Van de Vliert & Van der Vegt, 2005).

1.2 Purpose of Study
This study is aimed at examining if and how culture affects employee voice behavior as a
response to negative workplace situations. Considering that most, if not all, employee voice
behavior studies have been developed and tested with samples in so-called ―western countries,‖
such as the United States, these studies do not address the effects of possible differences in
national culture. For instance, it is plausible to believe that the EVLN construct, which is one of

４

frequently used theoretical frameworks of employee voice behavior, may operate differently in
Asian countries–which are heavily influenced by the Confucian principles of harmony and
hierarchical obligation (Kim & Park, 2003)–than it does in the developed countries of the West.
When applying these contexts to the EVLN construct, the exit and neglect option may be seen as
incompatible with the loyalty option, while the voice option may be seen as inconsistent with the
obligation of subordinate. Furthermore, we can investigate under-developed predictors of
employee voice behavior, especially those derived from key cultural values that influence
peoples‘ behaviors.

1.3 Research Questions
The main question in this study is as follows:
(1) Does culture affect employee response to negative workplace situations?
(2) If so, how does culture affect employee response to negative workplace
situations? In other words, which culture-specific values affect employee
response to negative workplace situations? What key features of a non-western
culture directly make a difference in the application of existing voice-related
models?
To answer these questions, this study will examine the impacts of culture on the
individual employee response to negative workplace situations, using a comparative test with
two samples – in the United States and in Korea.

1.4 Overview of Dissertation

５

This chapter describes the research questions as well as the rationale for this study.
Chapter 2 begins with a review that discusses the EVLN construct, from Hirschman‘s definition
of voice to the EVLN construct, which is an extension of Hirschman‘s model. Then, findings on
predictors of employee voice behavior from existing related studies are listed. The chapter
addresses why culture matters in employee voice behavior and the limitations of the existing
related studies, and explores an alternative framework based on employment relations systems in
non-western cultures, in particular, Korea. To do so, the three culture models of conflict
management suggested by Kozan will be introduced as theoretical ground for the model
suggested in this dissertation.
Chapter 3 presents the suggested model and the hypotheses to be tested in this
dissertation. The focus of the research is described, which includes definition of the EVLN
construct, a discussion of the cultural models posited by Kozan, and the factors that affect
employee s‘ EVLN behavior. The chapter goes on to illustrate the rationale to use Kozan‘s
model, although his model is about conflict management. Finally, the research hypotheses based
on the proposed model are presented.
Chapter 4 describes the research design and methods. It includes an overview of the
research methods, especially detailed explanation about data collection and data analyses.
Chapter 5 presents the results of the analyses conducted in this dissertation. Chapter 6 provides a
summary of findings, interpreting findings, and result of hypotheses testing, mainly focusing on
discussing what these findings tell us throughout a various data analysis outcomes. Chapter 7
provides a brief overview of the study, as well as a summary and discussion of the findings. It
also includes the implications of the study, theoretically and empirically, as well as practically,
the limitations of the study, and future research questions.

６

CHAPTER 2
LITERATURE REVIEW

This chapter is divided into six sections: (1) Defining Employee Voice Behavior, (2)
Exit-Voice-Loyalty-Neglect Construct, (3) Predictors of Employee Voice Behavior, (4) Culture
and Employee Voice Behavior, (5) Kozan‘s Culture Model, and (6) Summary of Literature
Review.

2.1 Defining Employee Voice Behavior
Since there is no one standard definition of employee voice behavior in the literature
(Van Dyne & LePine, 1998), it is important to start with a clear conceptualization of employee
voice behavior by asking, ‗What does employee voice behavior mean in this study?‘

2.1.1 Approaches of Employee Voice Behavior
Employee voice studies have been conducted within various academic domains, such as
organizational behavior, I/O-psychology, labor economics, and management studies, to analyze
specific phenomena in those fields. Inevitably, the studies on definitions and measures of
employee voice have flourished (Landau, 2009).
OCB researchers, for instance, regard employee voice behavior as a specific type of
extra-role behavior (Frese et al., 1999; LePine & Van Dyne, 1998; Van Dyne et al., 1995;
Withey & Cooper, 1989; Zhou & George, 2001). It is ―promotive behavior that emphasizes
expression of constructive challenge intended to improve rather than criticize‖ (Van Dyne &
LePine, 1998, p. 109). Given this definition of employee voice behavior, researchers have further

７

investigated the predictors of employee voice behavior such as LMX, self-esteem, and style of
management (LePine & Van Dyne, 1998, 2001; Van Dyne et al., 2008; Zhou & George, 2001).
For instance, LePine and Van Dyne (1998) empirically supported that self-esteem and selfmanaged styles have positive relationships on employee voice behavior. Van Dyne and
colleagues (2008) found that LMX, defined as the quality of supervisor-subordinate relationships,
was an important contextual climate affecting employee voice behavior. Zhou and George
(2001) empirically found that job dissatisfaction encouraged employees‘ expression of voice.
Researchers in the procedural justice studies view employee voice as the opportunity to
present information relevant to a decision, thereby enhancing employees‘ judgments of
procedural justice (e.g. Brockner et al., 1998; Folger, 1977; Folger et al., 1979; Lind & Tyler,
1988; Lind, Kanfer, & Earley, 1990; Thibaut & Walker, 1975; Tyler, 1987; Tyler, Rasinski, &
Spodick, 1985; Van den Bos, 1999, 2005; Van den Bos & Spruijt, 2002), which was termed the
―process control effect‖ by Thibaut and Walker (1978) and the ―voice effect‖ by Folger (1977).
This research has consistently found that employee voice is one of the key determinants
enhancing one‘s perception of organizational justice, particularly procedural justice (Folger,
1977; Folger et al., 1979; Lind & Tyler, 1988; Lind et al., 1990; Thibaut & Walker, 1975; Van
den Bos & van Prooijen, 2001). Therefore, voice effect is recognized as one of the most
important and robust findings in the domain of procedural justice (Lind & Tyler, 1988; Tyler,
1987; Tyler & Lind, 1992; Van den Bos, Lind, Vermunt, & Wilke, 1997).
Labor economists have studied employee voice mainly as one example of a union effect,
using the exit-voice model to explain the impact of unions on employee turnover, interpreting
unions as the institutions of collective voice in the job market (Addison & Belfield, 2004; Batt et
al., 2002; Cotton & Tuttle, 1986; Freeman, 1976, 1980; Freeman & Medoff, 1984; Lewin &

８

Mitchell, 1992; Lincoln & Kalleberg, 1996; Miller & Mulvey, 1991; Wilson & Peel, 1991). They
found that employee voice, mainly indicated by whether or not the organization is unionized,
lowers employee quit rates, raises employee job tenure, and enhances employer investment in
human capital and increases employee productivity (Bemmels, 1997; Boroff & Lewin, 1997;
Freeman, 1980; Freeman & Medoff, 1984; Lewin & Mitchell, 1992; Miller & Mulvey, 1991).
Furthermore, there are several terms for employee voice that are used alternately to
describe other aspects of the concept (Dutton & Ashford, 1993; Howell & Higgins, 1990;
Morrison & Phelps, 1999; Parker, 1993; Van Dyne et. al., 2003), such as ‗Civic Virtue‘ (Graham,
1991; Organ, 1988; Robinson, 1996; Robinson & Morrison, 1995), ‗Advocacy Participation‘
(Van Dyne et al., 1994), and ‗Constructive Suggestions‘ (George & Brief, 1992; Zhou & George,
2001; Frese et al., 1999). Commonly, these terms refer to verbal expression of ideas, information,
and opinions with the positive motive of making cooperative contributions to the organization
(Van Dyne et al., 2003).

2.1.2 Hirschman‘s Definition of Voice
Hirschman‘s classic study of voice sand the assumption that
“Under any economic, social, or political system, individual, business firms, and
organizations in general are subject to lapses from efficient, rational, law-abiding, virtuous,
or otherwise functional behavior. No matter how well a society‟s basic institutions are
devised, failures of some actors to live up to the behavior which is expected of them are
bound to occur, if only for all kinds of accidental reason. Each society learns to live with a
certain amount of such dysfunctional or misbehavior” (p.1).
He then questions how the actors respond to ―the deterioration in the performance of a
firm or an organization, for unspecified, random causes which are neither so compelling nor so
durable as to prevent a return to previous performance levels‖ (p.4). From the options available,

９

to the actors, he decided one behavioral option among two options—exit and voice. With the
example of the product market, he describes each option, saying
“… in absolute or comparative deterioration of the quality of the product or service
provides... (1) some customers stop buying the firm‟s products or some members leave the
organization: that is the exit option….. (2) the firms „customers or the organization‟s
members express their dissatisfaction directly to management or to some other authority to
which management is subordinate or through general protest addressed to anyone who
cares to listen: that is the voice option (p.4)”

In more detail, Hirschman defined voice as ―any attempt at all to change, rather than to
escape from, an objectionable state of affairs, whether through individual or collective petition to
the management directly in charge, through appeal to a higher authority with the intention of
forcing a change in management, or through various types of actions and protest, including these
are meant to mobilize public opinion‖ (1970, p. 30). In particular, he defined voice as a response
to ―the deterioration in the performance of a firm or an organization, for unspecified, random
causes which are neither so compelling nor so durable as to prevent a return to previous
performance levels‖ (p.4). Since this dissertation does not intend to argue and explore all of these
divergent usages from various research disciplines, attention will be restricted to Hirschman‘s
initial conceptualization of voice above.

2.1.3 Features of Hirschman‘s Definition of Voice
Here are key observations from Hirschman‘s initial conceptualization of voice. First, he
assumes that there is a situation leading to the actor‘s behavioral choice. In particular, focusing
on employees‘ voice behavior, he assumes that employees start to respond in a certain way once
one has experienced deterioration, perceived or actual, at work. When there is a shock, an
incident, or an event the mechanism of response begins to operate. Therefore, this study will not

１０

consider the employees‘ tendency of voice in an equilibrium state at work; rather it will focus on
their tendency of voice when an experience breaks this equilibrium state.
Second, as Hirschman describes, there is no uniform theoretical terminology for the
situation‘s cause. Rather, he uses the term ―unspecified, random causes‖ (p. 4) and focuses on an
explanation of the situation in general. Hirschman‘s use of non-specific language, for example,
―the deterioration in the performance of a firm,‖ (p. 4) or ―objectionable state,‖ (p.30) has
allowed subsequent researchers to interpret his writings as they chose. For instance, when
describing a trigger of voice, researchers in the field of justice use the term ‗unfairness,‘
researchers of job satisfaction use the term ‗dissatisfaction‘ (Farrell, 1983; Farrell & Rusbult,
1992; Rusbult et al., 1988), and others use their own terms such as ‗problematic events‘
(Hagedoorn, 1999). Following Hirschman‘s conceptualization, it should be noted that it is not the
focus of this study to argue exactly what causes this situation. Rather, this dissertation uses the
term ‗a negative workplace situation‘ which is a situation occurring at work that an employee
does not agree with, want, or accept.
Third, Hirschman‘s definition of voice clearly implies that voice is intentional behavior
that is used to change one‘s status quo at work. He describes voice as a political action to make
management engage in a search for the causes and possible cures. For that reason, he points out
that voice is ―nothing but a basic portion and function of any political system, known as ‗interest
articulation‖ (p. 30). Hirschman‘s description of voice provides us with two main conceptual
findings: voice is any verbal expression of the employee and, furthermore, it is an intentional
verbal expression to influence the status quo.
In sum, across various research disciplines, it has been claimed that Hirschman‘s
conceptualization of voice is the initial and most influential general theory investigating

１１

employee voice at work (Addison & Belfied, 2004; Batt et al., 2002; Farrell & Rusbult, 1992;
Dowding et al., 2000; Lewin & Mitchell, 1992; Rusbult et al., 1988; Van Dyne, Ang, & Botero,
2003; Withey & Cooper; 1989). Therefore, this study uses Hirschman‘s conceptualization of
voice to answer what does employee voice mean in this study. The following section describes
the EVLN construct, which is derived from Hirschman‘s conceptualization of voice, to explore
employee response to negative workplace situations.

2.2 Exit-Voice-Loyalty-Neglect Construct
The EVLN construct is based on Hirschman's (1970) theoretical work and on Rusbult
and his colleagues' model of responses to dissatisfaction in close relationships (Rusbult,
Zembrodt, & Gunn, 1982). Although the initial EVLN construct has been developed to explain
personal relationships, it has been recognized as a useful conceptual framework for analyzing
employee responses to negative workplace situations as well (Farrell & Rusbult, 1985; Farrell et
al., 1990; Hagedoorn et al., 1999; Leck & Saunders, 1992; Liljegren et al., 2008; Rusbult &
Lowery, 1985; Rusbult et al., 1988).

2.2.1 Defining the EVLN construct
According to the EVLN construct, when employees confront negative workplace
situations, they select one of four available response options: exit, voice, loyalty, and neglect.

Exit
The first option is exit. The exit option in this typology is equivalent to voluntary
separation from the job to either leave the job and the firm or seek a transfer within the same
organization as a means of leaving the dissatisfying job. This "painful decision to withdraw or

１２

switch" (Hirschman, 1970, p. 81) requires considerable effort by the employee and usually
means that the employee believes the situation is unlikely to improve. Since the exit option
requires considerable effort on the part of the employee, it is regarded as ―uniquely powerful and
expected to produce a wonderful concentration of the mind for the abandoned employer‖
(Hirschman, 1970, p. 21). The exit option includes job movement both within and across
organizational boundaries, as well as a variety of cognitive activities that precede leaving, such
as searching for a different job, or even thinking about quitting (Farrell & Rusbult, 1985).

Voice
The second option is voice. In the same situation as above, others may not want to leave,
―because of the costs that may be involved‖ (Hirschman, 1970, p. 38). As a substitute for exit, as
well as a complement to it, they can decide to exercise their voices, when ―they are sufficiently
convinced that voice will be effective‖ (Hirschman, 1970, p. 37). The voice option is described
as ―any attempt at all to change, rather than to escape from, an objectionable state of affairs,
whether through individual or collective petition to the management directly in charge, through
appeal to a higher authority with the intention of forcing a change in management, or through
various types of actions and protest, including these that are meant to mobilize public opinion‖
(Hirschman, 1970, p. 30). The notion of voice is differentiated from exit by its aim to achieve
change ―from within‖ (Hirschman, 1970, p. 38) meaning that it always involves the decision to
―stick‖ with the deteriorating organization (Hirschman, 1970, p. 38).

Loyalty

１３

Some choose neither exit nor voice. Rather, they may stick with the organization they
belong to for a period of time before reacting to the problem, if at all. ―In a less rational, though
far from wholly irrational fashion…some may simply refuse to exit and suffer in silence,
confident that things will soon get better" (Hirschman, 1970, p. 38). ―An individual member can
remain loyal without being influential himself, but hardly without the expectation that someone
will act or something will happen to improve matters" (1970, p. 78-79). In other words, loyalty is
described as passively but optimistically waiting for conditions to improve: giving public and
private support to the organization, waiting and hoping for improvement, or practicing good
citizenship (Rusbult et al., 1988). Although there has been some confusion among researchers
about whether the label ―loyalty‖ in this model means solely an attitude (e.g. Mayes & Ganster,
1988; Minton, 1988) or a behavioral outcome (e.g. Rusbult & Farrell, 1983; Rusbult et al., 1988),
in this EVLN construct, loyalty has been regarded mainly as a behavioral outcome (see Farrell,
1983; Farrell & Rusbult, 1981; Farrell & Rusbult, 1985; Farrell, Rusbult, Lin & Bernthall, 1990;
Rusbult & Farrell, 1983; Rusbult, Farrell, Rogers, & Mainous, 1988).

Neglect
Rusbult and colleagues (1982) have developed and expanded Hirschman's original Exit,
Voice and Loyalty, or EVL model by adding a fourth option: neglect. The neglect option refers
to allowing conditions to deteriorate, resulting in lax and inattentive behavior (Rusbult,
Zembrodt, & Gunn, 1982). Neglect is differentiated from loyalty; the latter has an expectation of
recovery in the future; the individual who exhibits neglect accepts that recovery is not going to
happen (Rusbult et al., 1982; Withey & Cooper, 1989). Examples of such behavior include
reduced interest or effort, chronic lateness or absences, using company time for personal

１４

business, or increased error rate (Adler & Golan, 1981; Angle & Perry, 1981; Farrell & Robb,
1980; Farrell & Rusbult, 1985; Hammer, Landau, & Stern, 1981; Petty & Bruning, 1980). In
addition, temporary abandonment and psychological inattention can be monitored and identified
as responses to dissatisfaction (Farrell & Rusbult, 1985).

2.2.2 Evidences of Construct Validity of the EVLN Construct
Since Rusbult and colleagues (1982) suggested the EVLN construct, a large number of
studies have supported the EVLN construct in organizational settings, with a wide range of
methodologies—multidimensional scaling (Farrell, 1983), cross-sectional survey research
(Farrell & Rusbult, 1985; Withey & Cooper, 1989), secondary analysis of extant data sets
(Rusbult & Lowery, 1985), simulation and laboratory experimentation (Rusbult et al., 1988), and
panel research (Farrell et al., 1990). Overall, through the evolution of the framework and
empirical works, the EVLN model has long been one of the most influential works of a set of
categories for exploring how people exercise their voices as a response to negative or
problematic situations at work (Farrell & Rusbult, 1985; Farrell et al., 1990; Hagedoorn, et al.,
1999; Leck & Saunders, 1992; Liljegren et al., 2008; Rusbult & Lowery, 1985; Rusbult et al.,
1988).
However, one possible limitation to be addressed in the EVLN construct is that
researchers do not measure actual behavior; rather they measure these behaviors through
examples and theoretical behavior--in other words, the intention or propensity of these behaviors.
Despite of it, there is a reason that this construct has been widely accepted. Partly, it is because
measuring the actual behaviors would be unreasonable due to the inherent boundary conditions
of the EVLN construct. As described earlier, the precondition of this construct is that there is a

１５

situation leading to the actor‘s behavioral choice, particularly a negative workplace situation.
Therefore, we assume that the employee starts to think of a certain response to be enacted once
he or she has experienced this situation. There is a certain point at which the mechanism of
EVLN becomes operative. Therefore, in EVLN studies it is hard to measure actual behaviors
when there is a real situation leading to the respondent‘s EVLN behavior. For that reason, most
EVLN studies provided respondents with a scenario such as ―assume that there is a situation….
How would you….?‖
However, measuring the respondent‘s EVLN intention, not actual behavior should not be
automatically discounted. Researchers support that intention is not far from the actual behavior;
rather, it is the strongest precursor of the actual behavior. For instance, in turnover research, a
fundamental tenet is that a decision to withdraw from the workplace best portends subsequent
withdrawal (Hulin, Roznowski, & Hachiya, 1985; Mobely et al., 1979; Price & Mueller, 1986;
Susbult & Farrell, 1983). The meta-analysis outcomes also revealed that intention to quit best
predicted actual departure (e.g. Hom & Griffeth, 1995). Overall, to measure actual EVLN
behaviors, timing does matter due to the nature of the EVLN construct itself. For this reason,
many studies testing EVLN construct used the EVLN scale developed by Rusbult and his
colleagues (1988), measuring the tendencies toward each response (e.g. Farrell & Rusbult, 1992;
Farrell, Rusbult, Lin, & Bernthal, 1990; Hagedoorn et al., 1999; Thomas & Au, 2002; Van de
Vliert, 1997).

2.2.2 Rationales to Use the EVLN Construct
Overall, through the subsequent evolution of the framework and empirical works, the
EVLN model has become one of the most influential works of a set of categories for exploring

１６

how people respond to their deteriorating or problematic situations at work (Farrell, 1983; Farrell
& Rusbult, 1985; Farrell, Rusbult, Lin, & Bernthal, 1990; Fischer & Locke, 1992; Hargadorn et
al., 1999; Hulin, 1991; Leck & Saunders, 1992; Rusbult, Farrell, Rogers & Mainous, 1988;
Rusbult & Lowery, 1985; Withey & Cooper, 1989). Accordingly, this dissertation uses the
EVLN construct as a dependent variable based on its theoretical and empirical strength.
There are two main reasons for using EVLN versus another construct describing
employee voice behavior. First of all, compared with other constructs, the EVLN construct
provides a comprehensive explanation with detailed classifications of the four types of response.
Although a variety of notable concepts have emerged to explain employees‘ voice behaviors,
especially since Hirschman‘s study, there is a lack of consistent findings from previous empirical
literature on measuring and predicting voice. Most of these studies have focused specifically on
one part of the EVLN model, specifically voice, they regard ‗non-voice‘ as merely an absence of
voice; therefore they are limited in explaining what ‗non-voice‘ means (Lind et al., 1990).
Accordingly, it has been suggested that other perspectives should be applied to explain voice
behaviors. For example, the concept of employee silence as meaningful beyond simply an
absence of voice should be considered an important area of inquiry (see Milliken & Morrison,
2003; Morrison & Milliken, 2000; Pinder & Harlos, 2001). Interestingly, the EVLN construct
itself shows that there is not only the option of voice but also options other than voice; such as
exit, loyalty, and neglect as a response to the same situation, and these options are distinctively
different from one another. Therefore, this construct provides a further understanding of what
voice means and what ‗non-voice‘ mean as well. In doing so, the EVLN construct contributes to
a comprehensive understanding of employee voice behavior.

１７

Secondly, as described earlier, a variety of notable concepts have emerged, and continue
to emerge, which focus on employee voice behavior from varying perspectives. Despite the
flourishing of studies on employee voice behavior, there is lack of consistent findings from
previous empirical literature on measuring and predicting voice, because voice is a multidimensional construct and empirical research has focused on more general forms of voice as a
unitary concept (Lind et al., 1990). In contrast, the EVLN construct has earned its validity
through various methods. Subsequent studies have supported the exit, voice, loyalty, and neglect
typology in organizational settings, with a wide range of methodologies—multidimensional
scaling (Farrell, 1983), cross-sectional survey research (Hargadorn, et al., 1999; Farrell &
Rusbult, 1985; Lee & Jablin, 1992; Withey and Cooper, 1989), secondary analysis of extant data
sets (Rusbult & Lowery, 1985), simulation and laboratory experimentation (Rusbult, Farrell,
Rogers, & Mainous, 1988), and panel research (Farrell, Rusbult, Lin, & Bernthal, 1990). Overall,
since the EVLN construct should be a useful typology for building predictive theories (e.g.
Farrell, 1983), this dissertation use the EVLN construct to study predictors of employee voice
behavior.

2.3 Predictors of Employee Voice Behavior
A large volume of studies have previously been conducted to investigate which factor
predicts an employee‘s exercise of voice as a response to negative workplace situations. Largely,
the studies are divided by two types of factors: individual factors and organizational factors (for
review of voice predictors, see Landau, 2009).
Individual factors include job satisfaction (Farrell & Rusbult, 1992; LePine & Van Dyne,
1998; Rusbult et al., 1988; Withey & Cooper, 1989), investment size (Farrell & Rusbult, 1992;

１８

Rusbult et al., 1988), alternative quality (Farrell & Rusbult, 1992; Fleishman & Edwin, 1962;
Lammers, 1969; Larson & Fukami, 1985; McLaughlin & Butler, 1974; Muchinsky &
Maassarani, 1980; Pfeffer & Lawler, 1980 ; Rusbult et al., 1988; Watson, 1981; Zald & Berger,
1978; Miceli & Near, 1989), self-efficacy (Avery, 2003; Landau, 2009; Parker, 1993), selfesteem (Brockner et al., 1998; LePine & Van Dyne, 1998), and locus of control (Pinder & Harlos,
2001; Withey & Cooper, 1989). Organizational factors mainly deal with organizational
mechanisms conducive to employee voice, provided by the organization both formally–for
instance, providing grievance procedures, suggestion boxes, open door policies, and lunch with
the CEO--as well as informally–for instance, talking to supervisors on a causal basis (Gorden et
al., 1988; Kassing, 2000; Landau, 2009). In particular, researchers have emphasized the
importance of supervisors in employees‘ voice (Botero & Van Dyne, 2009; Brown et al., 2005;
Chia et al., 2000; Detert & Burris, 2007; Janssen et al., 1998; Kassing, 2000; Landau, 2009;
Saunders et al., 1992; Van Dyne et al., 2008; Walumbwa & Schaubroeck, 2009).

2.3.1 Individual Factors

Alternative Quality
Offering a new theory predicting employees‘ response to declining job satisfaction (exit,
voice, loyalty, and neglect), Rusbult and colleagues (1988) suggested alternative quality as a key
predictor of these responses. They refer to alternative quality as ―the degree that employees have
high-quality alternatives to their current job‖ (p. 604), then illustrate that good alternatives such
as attractive alternative job opportunities, the possibility of early retirement, or the acceptable
option of not working, promote motivation to do something about work problems and serve as a

１９

source of power for effecting change. Therefore, greater alternative quality should be associated
with greater tendencies toward active reactions to dissatisfaction (exit or voice) and lesser
tendencies toward passive response (loyalty or neglect) (Farrell & Rusbult, 1992; Rusbult et al.,
1988). Also, the possession of higher-quality alternatives appears to be associated with greater
tendencies toward various types of employee voice behaviors such as whistle-blowing (Miceli &
Near, 1989), grievance filing (Fleishman & Edwin, 1962; Muchinsky & Maassarani, 1980), and
constructive social movements (Lammers, 1969; Zald & Berger, 1978), with lesser tendencies
toward loyalty behaviors such as job commitment (McLaughlin & Butler, 1974; Pfeffer &
Lawler, 1980), and neglectful acts such as absenteeism and slow-down behaviors (Larson &
Fukami, 1985; Watson, 1981).

Self-Esteem
Some studies have shown that self-esteem has a positive relationship with employee
voice behavior (LePine & Van Dyne, 1998). Global self-esteem is the degree of positive selfworth that an individual ascribes to him- or herself (Brockner, 1988) and it indicates the degree
to which individuals believe they are capable, significant, successful, and worthy (Coopersmith,
1967, p. 5). Accordingly, researchers have found that people who are relatively high in selfesteem tend to believe that their viewpoints are correct and that their actions make a difference,
which lead to employee voice behavior (Brockner & Elkind, 1985; Brockner et al., 1998; LePine
& Van Dyne, 1998).

Job Satisfaction

２０

Rusbult and colleagues (1988) suggested that job satisfaction is also as a key predictor of
EVLN responses. They suggested that employees who are generally satisfied with their jobs
should feel optimistic about the possibilities for improving working conditions, and should feel
more strongly motivated to restore favorable conditions. Their empirical study supported that
high job satisfaction encouraged voice and loyalty and discouraged exit and neglect. Subsequent
studies also supported that job satisfaction has a positive relationship with voice behavior
(Farrell & Rusbult, 1992; Rusbult et al., 1988; Withey & Cooper, 1989). Overall, job satisfaction
has been the primary antecedent included in models of employee voice behavior to date (LePine
& Van Dyne, 1998).

Investment Size
Rusbult and colleagues (1988) also suggested investment size as a predictor of EVLN
behaviors. They referred to investment size as the ―resources an employee has put into a job that
have become intrinsic to that position, such as years on the job, nonportable training, familiarity
with the organization and supervisor, nonvested retirement funds, and resources that were
originally extraneous but have become connected to the job, such as convenient housing, travel
arrangement, and friends at work‖ (p.604). They asserted that employees who are highly invested
in their jobs should not terminate the relationship with the organization, and they should
therefore be more highly motivated to react constructively when conditions at work decline.
Therefore, they suggested that high investment size encouraged voice and loyalty and
discouraged exit and neglect. In addition to the Rusbult and colleagues study, subsequent studies
have also supported investment as a predictor of employees‘ response to dissatisfaction (e.g.
Farrell & Rusbult, 1992).

２１

Locus of Control
Withey and Cooper (1989) suggested locus of control as one aspect of efficacy of action
that affects employees‘ response to dissatisfaction. They suggested that people who have an
internal locus of control, a belief that their actions make a difference, will be more likely to
choose actions that most directly seek to alter the source of their dissatisfaction, such as exit and
voice; and people who have external locus of control will be more likely to choose more passive
responses such as loyalty and neglect. Similarly, Pinder and Harlos (2001) suggested that people
with an internal locus of control will be more likely to exercise voice than those with an external
locus of control.

Self-Efficacy
Other studies have examined the relationship between self-efficacy and employee voice
behavior (Avery, 2003; Landau, 2009; Parker, 1993). These studies suggested that self-efficacy,
which refers to the confidence that people have that they are competent and that their task efforts
will be effective (Landau, 2009), may influence employees‘ choice to exercise voice, because it
influences people‘s choices about whether and how to act by influencing whether they believe
their behavior can really make a difference. Common findings showed that self-efficacy has a
positive relationship with employee voice behavior propensity (Landau, 2009), willingness to
dissent (Parker, 1993), and the value individuals place on voice (Avery, 2003).

Other Predictors

２２

Withey and Cooper (1989) tested the possibility of improvement of the current situation
as a predictor of the EVLN behavior. They found that people thought more about exit when they
lacked the possibility of improvement of the current situation, which means the expectation that
their current situation would improve based on the belief that the culture of the organization they
belong to is supportive of change. LePine and Van Dyne (2001) found that among big-five
personality traits, which include extroversion, conscientiousness, neuroticism, agreeableness, and
openness, except for openness, the four personality traits, extroversion (r = .30),
conscientiousness (r = .26), neuroticism (r = -.12), and agreeableness (r = -.16), have significant
relationships with voice behavior. Molseed (1989) found that females were more likely to engage
in supportive silences, indicated by smiling and nodding of the head to encourage collaboration,
whereas males tended to use inexpressive silences (i.e., not responding to information requests)
in an attempt to compete with and dominate others. Similarly, Tannen (1990) found that females
are more likely to be silent in a conflicting situation at work. LePine and Van Dyne (1998) found
that whites engaged in more voice than non-whites, and that participants with at least some
college education engaged in more voice than those with no college education.

2.3.2 Organizational Factors
Although the literature review so far has focused on person-centered antecedents of
employees‘ voice behavior, organizational factors also have been found to influence employees‘
voice behavior. Generally speaking, organizational factors include characteristics of
organizations, individuals and their role in an organization, and other environmental factors
including social information at work. Organizational factors are important determinants of
people‘s thoughts and behaviors at work since people often spend a large amount of time

２３

understanding and interpreting information from their social context to determine their attitude
and behavior. In this line, several researchers have emphasized the importance of organizational
factors in interpreting employee voice. In employee voice literature, mechanisms provided by the
organization can be provided both formally—such as providing grievance procedures, suggestion
boxes, and open door policies—as well as informally—such as talking to supervisors on a casual
basis (Gorden et al., 1988; Kassing, 2000; Landau, 2009). The following section reviews the
examples of organizational factors that have been supported as predictors of employee voice
behavior.

Management Style: Self-managed vs. Traditional
LePine and Van Dyne (1998) suggested that management style, especially whether it is
self-managed or traditional, affects employees‘ voice behaviors. Since self-managed work
groups enhances each member's sense of involvement and responsibility, self-managed work
group leads to the member‘s broader involvement in group decision making compared to those
who work in traditional groups based on hierarchy.

LMX: Leader Member Exchange
Other studies suggested LMX, which refers to the quality of supervisor–subordinate
relationships (Dansereau et al., 1975), as an important contextual climate affecting employee
voice behavior (Botero & Van Dyne, 2009; Kassing, 2000; Van Dyne et al., 2008). They
explained that high-LMX, characterized as mutual trust, respect, reciprocal influence, loyalty,
liking, and a sense of obligation with their leaders (Graen & Uhl-Bien, 1995), provides
employees with more opportunities to speak up, exchange information or ideas with their

２４

supervisors, and use more communication channels compared to those in low-LMX. Other
research suggests that employees who have high LMX are likely to be more direct when
communicating with supervisors than employees with low-LMX (Krone, 1991, 1992). Similarly,
Kassing (2000) found that subordinates who perceived their relationships with supervisors as
high-quality reported using significantly more articulated dissent.

Leadership Characteristics, Especially Supervisors
Many studies support the importance of the leadership role, especially the characteristics
of the supervisor, to the subordinates‘ voice behaviors (Brown et al., 2005; Chia et al., 2000;
Detert & Burris, 2007; Janssen et al., 1998; Saunders et al., 1992; Walumbwa & Schaubroeck,
2009). Even though these studies examined somewhat different definitions of the characteristics,
the common finding was that the degree to which leaders support the employees‘ voice
influences the subordinates‘ willingness to voice. For instance, some researchers have suggested
that ethical leadership, defined as ―the demonstration of normatively appropriate conduct through
personal actions and interpersonal relationships, and the promotion of such conduct to
subordinates through two-way communication, reinforcement, and decision-making‖ (Brown et
al., 2005, p. 120), positively influences employee voice behavior, providing subordinates with
voice (Walumbwa & Schaubroeck, 2009). Since ethical leaders encourage their members to
voice opinions and suggestions about work-related matters as well as ethical issues, ethical
leadership is significantly related to members‘ willingness to report problems to management
(Brown et al., 2005). Detert and Burris (2007) found that manager openness was positively
related to voice, particularly for high-performing subordinates. Similarly, Saunders and

２５

colleagues (1992), Janssen and colleagues (1998), and Chia and colleagues (2000) found that
employees‘ willingness to voice was encouraged by good voice managers.

Organizational Climate of Voice Mechanism
Peirce and colleagues‘ (1989) work on the deaf ear syndrome suggested that
organizational norms can influence employees‘ direct expression of their dissatisfaction.
Morrison and Milliken (2000) suggested that an organizational climate includes (a) patterns of
organizational policies and structures, such as open-door policies, grievance systems, regular
team meetings, suggestion boxes, and hotlines, (b) belief structures of top management teams,
and (c) the process of collective sense-making and communication. They found that
organizational climate leads an employee‘s decision to speak. Similarly, Pinder and Harlos
(2001) suggest that organizational climate, and in particular a climate of injustice characterized
by unclear reporting structures, high centralization, low formalization, authoritarian management
styles, poor communication, poorly-conducted performance reviews, and erratic decision making,
may play a significant role in employees‘ willingness to exercise voice or to remain silent. In this
same vein, Spencer (1986) found that the number of voice mechanisms was positively related to
employees‘ expectations for problem resolution, because voice mechanisms can signal to
employees that management wants them to speak up and will take their concerns seriously.

2.4 Culture and Employee Voice Behavior

2.4.1 Defining Culture

２６

Culture has been studied across disciplines, attracting anthropologists, sociologists,
psychologists, and even business professionals for several decades. However, due to the fuzzy
nature of culture itself, the term ‗culture‘ has been defined in various ways (see Linda Smircich,
1983). Therefore, the literature review on cultural factors across disciplines in this dissertation is
far from exhaustive. Instead, it summarizes how culture is defined and approached in this
dissertation across different fields. It goes on to specifically entail a review of voice studies
dealing with culture.

2.4.2 Hofstede‘s Culture Study
In most culture-related social science research, researchers begin by referring to the
initial work of Hofstede (1980), because it has been recognized as one of the most important and
popular theories of culture study. In this work, Hofstede defined culture as ―the collective
programming of the mind that distinguishes the members of one category of people from those
of another‖ (p. 25) and described the core elements in culture as values, which are defined as
―broad tendencies to prefer certain states of affairs over others‖ (1980, p. 19).
According to Hofstede, values in certain societies represent patterns of thinking, feeling,
and acting; distinguishable by those who belong to the society. In addition, he assumed that each
person carries a certain amount of mental programming, which is stable over time and means that
a person will show more or less the same behaviors in similar situations. Hofstede divided
culture mainly based on nationality; he assumed that people from the same country would share
the same culture and would exhibit similar behaviors. In an attitude survey of 117,000 employees
of a large U.S. multinational corporations (later identified as IBM), he extracted four dimensions
with which he could classify the 40 different countries represented. These dimensions were

２７

labeled

‗individualism-collectivism,‘

‗power

distance,‘

‗uncertainty

avoidance,‘

and

‗masculinity-femininity.‘ Later, he added another dimension, named ‗long- versus short-term
orientation.‘ This Hofstede conceptualization of culture with a finite number of dimensions has
found favor with various social science researchers and has led to numerous studies using one or
more of the dimensions to explain observed differences between nationalities (Botero & Van
Dyne, 2009; Costigan et al., 2006; Landau, 2009; Xu et al., 2005). It should be noted that this
study is concerned primarily with national culture as it relates to organizational behavior.

2.4.3 Culture in Employee Voice Behavior Study
A large volume of previous studies have shown that various cultural dimensions have
been suggested as influencing work-related psychological and behavioral phenomena (Bontempo
& Rivero, 1992; Markus & Kitayama, 1991; Suh et al., 1998; Triandis et al., 1995; Wagner,
1995; Wasti, 2003). Unfortunately, there are very few studies in the literature of voice examining
how culture affects employees‘ responses to a problematic situation at work.

Power Distance
Research examining how culture affects an individual‘s voice behavior commonly uses
the concept of power distance, which is one of Hofstede‘s (1980) four dimensional frameworks
for cultural differences (collectivism-individualism, uncertainty avoidance, power distance, and
masculinity-femininity). Power distance (PD) refers to the extent to which the less powerful
members in an organization expect and accept that power is distributed unequally (Hofstede,
1980). In work contexts, employees with higher PD orientations believe that supervisors should
have a large amount of power over them and employees with lower PD orientations believe that

２８

all people should be relatively equal and that employees have the right to speak up with their
ideas and opinions (Hofstede, 1980).
Since employees in low power distance cultures are more likely than employees in high
power distance cultures to believe that they should have some input on the decision-making
process, and in high power distance cultures employees are less likely to engage in extra-role
behaviors (Costigan et al., 2006), and may be afraid to disagree with their superiors (Landau,
2009), PD has implications on employee willingness to voice (Botero & Van Dyne, 2009).
Individuals with a high power distance orientation are unlikely to believe that they have anything
important to say, and even if they do have an idea, they will not believe that it is their
responsibility to speak up (Landau, 2009). These findings supported that high power distance has
negative relationships with voice (Botero & Van Dyne, 2009; Landau, 2009; Xu et al., 2005).

Cross-Cultural Comparison
Price and colleagues (2001) examined the relationship between the value of voice and the
magnitude of voice, that is, voice which spans different phases of the decision-making process,
with samples from the U.K., Mexico, the Netherlands, and the U.S. They found similar shapes
across the four countries: direct, monotonic, and nonlinear. However, they used undergraduate
students who were age 21 or younger, not employees. Therefore, it is questionable that they
would respond with the same behaviors as those who are employed full-time, with more
experience in the workplace.
Botero and Van Dyne (2009) examined the effects of culture, particularly individual‘s
value orientation of ‗power distance‘ (PD), which is a measure of the extent to which differences
in power and status (Hofstede, 1980), on the relationship between LMX and employee voice

２９

behavior in the U.S. and Colombia. In this study, they found that not all patterns of voice
outcomes are similar in the two countries: for instance, it showed that both in the U.S. and
Colombia, LMX and PD predicted voice but, contrary to the findings in the U.S., in Colombia,
the interaction effect of LMX and PD was not found to predict voice, implying possible new
explanations for voice, especially in the different cultural settings. In addition, they suggested a
comparative study using the Asian culture for future research.
The only research on exit-voice using Asian countries is Lee and Jablin‘s study (1992)
examining the generalizability of the EVLN construct using samples from Korea, the US, and
Japan. This study supported the applicability of the theory to the Korean sample, demonstrating
that, with respect to exit-voice, Korea and the U.S. are more similar than Korea and Japan.
However, there are several limitations: like Price and his colleagues (2001), they used student
sample staying in the U.S. who were not employed. Thus, it is possible that these respondents
were not representative of the Korean workers employed in Korean firms in Korea.
In sum,very few studies have examined the cross-cultural generality of the features of
employee voice behavior and the role of cultural values in predicting employee voice behavior in
different cultural settings (e.g. Botero & Van Dyne, 2009; Lee & Jablin, 1992; Price et al., 2001).
Given these gaps, there is a need to test how employee voice mechanism works across culture
and to develop a model to help understand the effect of culture on employee voice behavior.

2.5 Kozan’s Culture Model
Kozan (1997) developed three cultural models to provide a holistic theoretical framework
for analyzing how culture affects the whole process of conflict management. The basic notion of
his model is that conflict management cannot simply be interpreted as a behavior in and of itself,

３０

but takes place within the overall process of managing conflict, which consists of antecedent
conditions; cognition, including thoughts and emotions, behavior, and outcomes; and the role of
third parties. He suggests that culture can cause variation in any of these areas.

2.5.1 Kozan‘s Three Culture Models
Kozan identified three cultural models—a confrontational model, a regulative model, and
a harmony model—and described the means, such as abstraction and shared meaning of conflict
management process, by which each model is held together. Based on the key cultural features in
each model, he suggested that cultures show differences in terms of five distinct conflict
management styles: competing, accommodation, collaboration, avoiding, and compromise; Table
1 summarizes the key features of each model.

Table 1. Three cultural models of conflict management
Components of
Conflict Management

Harmony
Model

Confrontational
Model

Regulative
Model

Highly competitive
Low competitiveness
Low competitiveness
work environment
1) Antecedent conditions due to observance of
due to extensive rules
due to individualistic
mutual obligations
of procedures
goals
Analytical definition
Thoughts Holistic definition of Analytical definition of conflict in terms of
conflict in
of conflict, in terms
universalistic
particularistic terms of sub-issues
principles
2) Cognition
Emotions
3) Behaviors
4) Outcome Criteria

Suppression of
negative emotions

Expression of
negative emotions

Expression of
negative emotions

Avoidance and
accommodation

Confrontation and
compromise

Avoidance or forcing

Face-saving concern

Due process concerns Due process concerns

Source: Kozan (1997)
３１

Table 1. (cont‘d)
Components of
Conflict Management

Harmony
Model

Confrontational
Model

Regulative
Model

NonFrequent, intrusive,
managerial
informal
third parties
5) ThirdParty

Infrequent, planned, Formal appeal
non-intrusive
systems

Managerial
Mediatorial
Intervention

Facilitational or
autocratic

Restructuring or
laissez-Faire

Reason, fairness
(equity)

Reason, general
principles (equality)

Emphasis

Harmony, shame

Source: Kozan (1997)
According to Kozan‘s theory, in the harmony model, potential conflict is handled by
various non-confrontational means in order to maintain group harmony. This model has mostly
been found in associative or collectivistic culture such as Asian countries. This cultural context
stresses harmony within the group and puts a high value on consensus or absence of conflict.
Therefore, this context emphasizes interdependence. When handling conflict, people in the
harmony culture emphasizes normative reasoning rather than an individual means/ends analysis,
reflecting social expectations and the appropriateness of an act, for instance whether or not their
behaviors are desirable and appropriate based on the social norm that they belong to. In addition,
the harmony model involves rules restricting the display of negative emotions during conflicts.
Considering these characteristics, people in the harmony model stress cooperative behavior:
accommodating and avoiding, rather than competing and assertiveness, especially in terms of
face-saving concerns (both self-face and other-face concern).
In the confrontational model, conflict management is governed by the norms of fair play,
mutual concessions, and compromise. This model is most likely to be found in an individualistic
and low uncertainty avoidance culture, such as U.S. These cultures emphasize the individual and

３２

aggressive pursuit of one‘s goals. In this culture, people stress pragmatic reasoning in conflict
management with empirically observable facts. This model does not discourage negative
emotions. Reflecting these characteristics, people in the confrontational model typically have
tendency of confrontation and compromise with a problem-solving approach. Third parties in the
confrontational model are likely to play a less crucial role than in the harmony model and
managers are likely to frequently play a third party role, more autocratic (inquisitorial or
motivational) role when resolution becomes crucial to performance. As regards conflict
resolution, important criteria include the fairness of a process as well as the fairness of an
outcome, in terms of short-term consequences.
In the regulatory model, conflict management is guided by universalistic principles and
rules with bureaucratic arrangements that show characteristics of an abstractive, co-subjective, or
high uncertainty avoidance, and individualistic culture. This model is likely to be found in
continental European countries, and emphasizes the role of organizational structure such as
bureaucratic rules to control competition and conflict. In addition, this model is designed to leave
the emotions out of conflict management. In the regulative model, people stress either avoidance
or authoritative command, and the negotiation style is more rigid since concessions may be
viewed as deviations from principle. As for third parties in the regulatory model, with the
tendency of high uncertainty avoidance, people rely on bureaucratic means such as hierarchical
referral or formal appeal mechanisms. As a result, this model puts a high value on procedural
justice in the resolution, with detailed written procedures outlining due process in cases of
dispute, especially in terms of short-term consequences.

2.5.2 Empirical Evidences of Kozan‘s Culture Models

３３

Subsequent studies support Kozan‘s culture model (Gelfand et al., 2007). For instance,
persons in individualistic nations, which endorse a direct confrontational model, prefer to resolve
conflicts using their own expertise and training (Smith et al., 1998), forcing conflict resolution
styles (Holt & DeVore, 2005), and tend to focus on integrating interests (Tinsley, 1998, 2001).
Germany is consistent with a regulative model, in part due to its values of explicit contracting
(Tinsley, 1998, 2001). By contrast, individuals in collectivistic cultures which use a harmony
model prefer avoidance and withdrawal (Holt & DeVore, 2005; Ohbuchi et al., 1999). For
instance, Friedman and colleagues (2006) showed that the Chinese tend to avoid conflict because
they are concerned that direct conflict will hurt their relationship with the other party. The
Chinese are more sensitive to hierarchy than Americans, and avoidance behavior is heightened
when the other party is of higher status.
In sum, Kozan‘s conflict management model provides a holistic approach to studying
conflict management across cultures, illustrates the process of conflict management, and shows
how the component of each culture model make sense in certain styles of conflict management.
Interestingly, the components described for each style of conflict management provide plausible
clues as to how each component can affect employees‘ response to dissatisfaction at work.
Of course, conflicting situations and negative situations at work are not exactly same.
However, it can be assumed that employees will show similar behavioral traits when they decide
how to respond to conflicting situations and negative situation at work them, and that their
thoughts or behaviors shown in the process of conflict management can be applied to explain the
distinct style of an employee‘s response to a negative situation at work. In particular, this study
will apply the harmony model to explore what values affect employees‘ responses to

３４

dissatisfaction at work across cultures, since it provides implications of cognitive and behavioral
features, especially in the Asian cultures.

2.6. Summary of Literature Review
Reviewing the literature reveals key findings in employee voice behavior research, and
there is still much more to explore on employee voice behavior. First of all, despite continuous
development of voice literature across disciplines, particularly to investigate what affects
employees‘ voice behavior, this literature does not fully examine how culture affects employees‘
voice behaviors through a theoretical lens. Second, there are few efforts to provide overreaching
explanations of how these individual, organizational, and even cultural predictors work
simultaneously in an individual‘s exercise of voice behaviors. Considering the complex and
multidimensional nature of voice behavior, it is assumed that when employees choose a certain
type of response to a negative situation at work, they do not consider one specific factor, such as
an individual factor or organizational factor. Rather, they put all available factors together and
calculate all possible options based on these factors. For that reason, it can be argued that a more
comprehensive approach to examining predictors of voice behavior is needed. This dissertation
is therefore aimed at 1) developing and testing cultural predictors of employees‘ voice behavior
within a cohesive theoretical ground and based on that, 2) expanding research on predictors of
employees‘ voice behavior by examining three different categories of predictors: individual,
organizational, and cultural factors. The next chapter proposes this model.

３５

CHAPTER 3
MODEL

This chapter discusses the research model to be examined in this dissertation. This
chapter includes the following sections: (1) Research Focus, (2) Proposed Model, and (3)
Hypotheses.

3.1 Research Focus
Before proceeding any further, it is important to clarify the focus of this dissertation. First,
as far as a definition of employee voice behavior, as described in the literature review, there is no
single, clear referent. This dissertation does not intend to argue and explore all of the definitions
in divergent usages from various research disciplines. Therefore, to make the research focus
clear, attention will be restricted to Hirschman‘s initial definition of voice: ―any attempt at all to
change, rather than to escape from, an objectionable state of affairs, whether through individual
or collective petition to the management directly in charge, through appeal to a higher authority
with the intention of forcing a change in management, or through various types of actions and
protest, including these that are meant to mobilize public opinion‖ (1970, p. 30). In other words,
voice

is a way of responding to ―the deterioration in the performance of a firm or an

organization, for unspecified, random causes which are neither so compelling nor so durable as
to prevent a return to previous performance levels‖ (1970, p.4). In this line, the EVLN construct
which is the extended version of Hirschman‘s work is used as dependent variable in this study.
Second, Kozan‘s harmony model will be used as the theoretical framework to explore
how culture affects employee voice behavior in this dissertation. The main purpose of this study,

３６

however, is not to test the validity of Kozan‘s three culture models as a whole. Rather, Kozan‘s
harmony culture model will be tested to provide theoretical grounds for cultural differences in
employees‘ response to dissatisfaction at work. In addition, to compare two distinct cultures, a
sample from Korean will be used as an example of a harmony culture since Kozan categorizes
the Asian countries as example of a harmony culture; and the U.S. will be used as an example of
non-harmony culture because Kozan categorizes the U.S. as other culture, particularly
confrontational culture.
Third, as shown in the literature review, a wide range of predictors of employee voice
have been studied in various disciplines. The purpose of this study is not to test a complete
model of all of these factors, but to examine an underdeveloped factor, culture. However, this
study does not attempt to suggest that culture is more important factors to predict employee
voice behavior than any other factors. Rather, this study attempts to provide comprehensive
understanding of employee response to negative workplace situation saying that cultural factor
are also important as individual and organizational factors are. Therefore, this dissertation
includes four predictors of employee voice: 1) alternative quality, 2) self-esteem, 3) power
distance orientation, and 4) LMX. The details of each factor will be discussed in the next chapter.

3.2 Proposed Model
Figure 1 shows the graphic representation of the proposed model. As shown in Figure 1,
the model consists of four components: 1) EVLN construct as dependent variable, 2) cultural
variables, 3) additional variables as voice predictors, and 4) control variables. These four
components are discussed in detail.

３７

Figure 1 Model
Harmony Culture Factors
Collectivism
A

Face-Saving

EXIT

Conflict-Avoiding
VOICE
B

Other Factors
LOYALTY

Alternative Quality
Self-Esteem

NEGELCT

Power Distance
LMX

3.2.1 EVLN Construct as Dependent Variable
When employees feel or experience ―the deterioration in the performance of a firm or an
organization, for unspecified, random causes which are neither so compelling nor so durable as
to prevent a return to previous performance levels‖ (Hirschman, 1970, p.4) at work, they will
respond in various ways. This typology includes four response categories: exit, voice, loyalty,
and neglect.
As described in the literature review, the exit option means voluntary separation or
turnover from the job, including job movement both within and across organizational boundaries,
as well as a variety of cognitive activities that precede leaving such as searching for a different
job or thinking about quitting (Farrell & Rusbult, 1985). The voice option refers to ―any attempt
at all to change, rather than to escape from, an objectionable state of affairs, whether through

３８

individual or collective petition to the management directly in charge, through appeal to a higher
authority with the intention of forcing a change in management, or through various types of
actions and protest, including these that are meant to mobilize public opinion‖ (p. 30). Loyalty
describes passively but optimistically waiting for conditions to improve--giving public and
private support to the organization, waiting and hoping for improvement, or practicing good
citizenship (Rusbult et al., 1988). The neglect option means allowing conditions to deteriorate
with lax and disregardful behavior such as reduced interest or effort, chronic lateness or absences,
using company time for personal business, or increased error rate (Adler & Golan, 1981; Angle
& Perry, 1981; Farrell & Robb, 1980; Farrell & Rusbult, 1985; Hammer, Landau, & Stern, 1981;
Petty & Bruning, 1980).
As Hirschman (1970) illustrated, when deciding which option among these four
behaviors will be selected, employees will consider a variety of individual and situational factors,
as well as other values regulating the person‘s behavior. In particular, as supported by previous
cross-cultural research, they will be influenced by the dominant culture in their society; thus,
they will be more likely to choose the option encouraged by their culture and less likely to
choose the option discouraged by their culture. Particularly considering the increasing diversity
of the workforce and the globalization of organizations, this could have a significant impact on
whether and how employees exercise their voice. Therefore, investigating the cultural factor as
predictor of employees‘ voice should be promising area for the future research.

3.2.2 Rationale for Applying Kozan‘s Culture Model to Employee Voice Behavior
The primary purpose of this model is to examine how cultural values affect employees‘
response to a problematic situation at work. Culture theory is rich in new taxonomies of cultural

３９

values (House et al., 2004; Schwartz, 1994; Smith et al., 1996), beliefs (Bond et al., 2004), and
norms (Gelfand et al., 2006b), that affect employees‘ behaviors at work. Thus, this dissertation
attempts to use a holistic approach to illustrate the process of employees‘ choice of certain
behavioral option among four distinct options. This dissertation draws on Kozan‘s harmony
model as a framework for the analysis of culture, using three cultural values that have been
proposed in the literature as dominant.
There are two main reasons to apply Kozan‘s model in this study. First, these two
behaviors–conflict management behavior and response to negative workplace situations–share
similar traits; both lead to behaviors that are 1) discretionary, meaning not required by
organizations or management, and not outlined in one's job description (Van Dyne et al., 1995),
2) challenge-oriented, which means they aim to change the status quo (LePine & Van Dyne,
1998; Morrison & Phelps, 1999; Van Dyne et al., 1995), and 3) potentially risky, since the
behavior may (a) be associated with discomfort (Milliken et al., 2003); (b) gain a negative public
image or label (Ashforth & Humphrey, 1995; Milliken et al., 2003); or (c) damage relationships
with others and thus destroy social capital (Adler & Kwon, 2002). Consequently, it can be
assumed that the behavioral traits that are present in the way that employees manage their
conflicts will also be the way that employees respond to a negative situation at work. Kozan s
model can be a useful framework providing theoretical clues to understanding the overall process
through which employees respond to negative situations at work.
Second, Kozan argues that existing dimensions of culture in cross-cultural literature have
not been fully explored. He suggests that what is needed is a comprehensive framework that
relates a cultural typology to the conflict management process as a whole. As described in the
literature review, he argues that conflict management cannot be simply be interpreted as a

４０

behavior, but takes place within the overall process of managing conflict, which includes 1)
antecedent conditions, 2) cognition (thoughts and emotions), 3) behavior, 4) outcomes, and 5)
the role of third parties. He suggests that culture can cause variation in any of these five
components. Therefore, Kozan‘s model provides a holistic theoretical framework for analyzing
how culture affects each stage of the entire conflict management process. Similarly, it can be
assumed that culture can make variation in any component of the overall process of employees‘
decision to choose certain option among exit, voice, loyalty, and neglect options to respond to
negative workplace situations. In particular, this study attempts to use Kozan’s harmony model
since it provides implications for cognitive and behavioral features, especially in the Asian
cultures.

3.2.3 Cultural Values from Kozan‘s Culture Model
According to Kozan‘s harmony model, when employees encounter a work-related
problematic situation, they tend to emphasize normative reasoning regarding social expectations
concerning the appropriateness of the act. They will consider how their behaviors will be judged
in the organization, rather than prioritizing their individual interests, particularly if the behavior
is perceived to impair harmony in the organization. This view will strongly influence the
employees‘ choice of behavioral response to the situation at work. Because the harmony model
puts a high value on collectivism and face-saving, employees who feel dissatisfaction at work
will choose a response that is consistent with these values.
In summary, to investigate what predicts employee voice behavior, this dissertation
draws on Kozan‘s culture model of conflict management: this dissertation takes his holistic
approach to predicting a certain type of behavior when responding to conflicting situations at

４１

work. Rather than interpreting the way employees handle the situation simply as a behavior itself,
Kozan views conflict management as the overall process considering the antecedent, cognition,
and behavior as a whole. Following this line of reasoning, the model proposed in this dissertation
considers the way employees respond to their deteriorating situation at work as an overall
process.
In addition, this dissertation takes Kozan‘s approach of applying culture to predict certain
behavioral traits. Rather than investigating certain types of cultural values, such as power
distance and collectivism, his model discusses how culture affects each stage of the overall
process of conflict management. For instance, in the harmony model, he extracts several key
values encouraged in that type of culture and explains how these values result in certain
behaviors. It can be hypothesized that key values encouraged in cultures represented by the
harmony model, such as collectivism, face-saving, and conflict-avoiding, will predict employees‘
voice behavior. The following chapter will suggest the testable hypotheses, describing how these
two cultural values from the harmony model affect employees‘ choice of one of the EVLN
typologies.

3.2.4 Additional Predictors of Employee Voice Behavior
As shown in the literature review, a wide range of predictors of employee voice behavior
have been studied in various disciplines. As discussed earlier, the purpose of this study is not to
test a complete model of all of these factors. Rather, this study attempts to examine relatively
underdeveloped cultural factors, and to expand the model by combining cultural factors with
other predictors of employee voice that have been supported in the literature. As researchers of

４２

culture have discussed (e.g. Schwartz, 1999), there is individual variation in value priorities since
value priorities are a product both of shared culture and of unique personal experience.
Therefore, this dissertation includes four predictors of employee voice: 1) alternative
quality, 2) self-esteem, 3) power distance orientation, and 4) LMX. Based on the literature
review of factors that affect employees‘ voice behavior, alternative quality is tested since it has
been supported as a significant factor of EVLN construct (e.g. Farrell & Rusbult, 1992; Lee &
Jablin, 1992; Rusbult et al., 1988). Self-esteem will also be included since it is an example of an
individual factor that has been supported in many employees‘ voice behavior literature (e.g.
Brockner & Elkind, 1985; Brockner et al., 1998; LePine & Van Dyne, 1998). Power distance is
an example of a cultural dimension that has been tested as a factor of employees‘ voice behavior
(e.g. Botero & Van Dyne, 2009; Landau, 2009; Xu et al., 2005). Finally, LMX will be tested in
this study as an additional factor of employees‘ voice behavior, since it has been found as an
organizational or contextual factor of employees‘ voice behavior (e.g. Botero & Van Dyne,
2009; Kassing, 2000; Van Dyne et al., 2008).

3.3 Hypotheses
In this dissertation, the independent variables affecting employees‘ response to a negative
situation at work are categorized as two factors: (1) Cultural factors including collectivism, facesaving, and Conflict-Avoiding; and (2) other factors such as a) alternative quality, b) self-esteem,
c) power distance orientation, and d) LMX. Figure 3.1 shows the graphic representation of the
proposed model. As the figure shows, the model consists of three components: cultural factors,
other factors, and the EVLN construct.

４３

First, the relationship between the key values in the harmony culture ––collectivism,
face-saving, and conflict-avoiding, and the EVLN construct will be discussed (Path A). Then, the
relationship between other factors such as individual and organizational factors and the EVLN
construct will be discussed (Path B). The details of the hypotheses with each factor will be
described below.

3.3.1 Collectivism
In the harmony model, individuals tend to mainly be affected by the value of collectivism,
which puts a strong emphasis on interdependence and harmony (Kozan, 1997). The concepts of
individualism and collectivism have served as a powerful theoretical construct in cross-cultural
research across various disciplines (Hofstede, 1980; Hui & Triandis, 1986; Leung, 1987;
Miyahara et al., 1998; Ting-Toomey, 1988; Triandis, 1988). Collectivism refers to the tendency
of in-group members to be concerned about the consequences of their behavior and to be willing
to sacrifice personal interests for collective interest (Leung, 1987). In the collectivism culture,
maintaining harmony within one‘s social environment is a key virtue (Hofstede, 1991).
Individuals in groups are dependent upon others as much as others are dependent upon
them through mutual obligations; all have an interest in fulfilling their roles and maintaining
harmony. High emphasis on the value of harmony means that social and moral values precede
individual and competence values (Redding et al., 1994). Thus, the stress is not so much upon an
individual and his/her interest, but on the maintenance of the collectivity and the continuation of
harmonious relationships of the members within the group. Therefore, when deciding how to
behave, members in groups tend to emphasize not an individual means/ends analysis but the
social expectations which hold groups together and determine the appropriateness of an act

４４

(Fishbein & Ajzen, 1975). To this end, role-appropriate behaviors–placing value on maintaining
a stable social structure through relationships, organizations, and institutions-–is encouraged.
Overall, the value of collectivism leads to compromises harmony and collectivity (Kirkbride et
al., 1991).
Accordingly, in a collectivistic culture, employees will respond to the negative situation
that they want to have changed or improved, based on the criteria of whether or not their
behavior complies with the key values embedded in a collectivism culture: for instance, they
may consider whether their behavior impairs harmony and relationship within the organization.
Therefore, if the behavior is interpreted as destructive towards harmony and relationships within
the organization, this behavior will be discouraged. On the other hand, if the behavior is
interpreted as constructive behavior that maintains or improves harmony and relationships within
the organization, this behavior will be encouraged. Consequently, when employees confront a
negative situation at work that needs changing or improvement, they will select a certain
behavior based on these criteria. In particular, among exit, voice, loyalty, and neglect, they will
choose based on whether the behavior is consistent with the key values emphasized in the
collectivism culture: maintaining or improving harmony and relationships within the
organization.
Overall, in the harmony model, where collectivism is highly valued, employees will be
more likely than otherwise to choose options that are considered not to hurt or dissolve the
relationship with the organization. Consequently, employees in the collectivism culture will
likely be encouraged to choose the option of loyalty since it is interpreted as a constructive
behavior within the organization, aimed at maintaining or reviving the employee‘s relationship
with the organization (Hagedoorn et al., 1999; Liljegren et al., 2008; Rusbult et al., 1982).

４５

Loyalty will allow the employee to maintain the relationship with the organization while
continuing to fulfill his or her obligations, and may not have any room to hurt the relationship.
Therefore, loyalty will be more encouraged in high collectivistic culture. Voice will also be
encouraged since voice is also interpreted as a constructive response, basically intended to make
the situation better, and it alerts management to problems (Farrell & Rusbult, 1983).
Exit and neglect will be discouraged in a collectivistic culture since they are interpreted
as behaviors destructive to the relationship between employee and the organization that the
employee belongs to. Rather, ext and neglect aim to diminish or terminate the relationship
between the organization and the employee (Liljegren et al., 2008; Rusbult et al., 1982). The exit
option will be interpreted to aim to terminate the relationship with the organization, thus it
cannot be interpreted as a constructive behavior in any case. Therefore, in a highly collectivistic
culture, the exit option will be not be encouraged at all.
Similarly, in the highly collectivist culture, the neglect option will not be encouraged
since it does not aim to make and support satisfactory organizational relations. In particular, it
will be discouraged when it is interpreted as a destructive behavior with lax and disregardful
behavior among the employees, such as lateness, absenteeism and high error rates (Farrell, 1983;
Liljegren et al., 2008; Rusbult et al., 1982). Consequently,

HP 1. In the harmony culture, while collectivism will have a positive association with voice and
loyalty, it will have a negative association with exit and neglect.

3.3.2 Face-Saving

４６

In the harmony model, one of the criteria for judging the effectiveness of behavior is
face-saving concern (Kozan, 1997), which is regarded as a means for fostering harmony (Moore,
1967). ‗Face‘ refers to an individual‘s claimed sense of favorable social self-worth in a relational
and network context (Ting-Toomey, 1988) that a person wants others to have of her or him
(Ting-Toomey & Kurogi, 1998). Face is considered as interpersonal self-worth issues and otheridentity consideration issues for all communicative behaviors (Ting-Toomey & Kurogi, 1998), as
a consequence of position in society. By nature, face is a concept that has simultaneous affective,
cognitive, and behavioral layers, associated with respect, honor, status, reputation, credibility,
competence, family/network connection, loyalty, trust, relational indebtedness, and obligation
issues (Ting-Toomey & Kurogi, 1998).
According to Ting-Toomey‘s (1988) face-negotiation theory, it is assumed that (1) people
try to maintain and negotiate face in all communication situations; (2) the concept of ―face‖ is
especially problematic in vulnerable interpersonal situations (such as request, embarrassment, or
conflict situations) when the situated identities of the communicators are called into question; (3)
the cultural variability dimension, such as individual-collectivism, influences members‘ selection
of various facework behaviors. Facework behaviors refer to clusters of communicative behaviors
that are used to enact self-face and to uphold, challenge/threaten, or support the other person‘s
face (Ting-Toomey, 1988).
The face-saving norm regulates responsibilities and interpersonal relationships in the
society (Tse et al., 1988), including protecting one‘s pride, status and honor (Kozan, 1997), as
well as the preservation of the face of the other (Miyahara et al., 1998; Redding et al., 1994).
Concern with another‘s face is readily translated into a high degree of cooperativeness and a

４７

lower degree of assertiveness. Members find it quite difficult to aggressively pursue their
interests at the expense of the interest of the in-group (Kozan, 1997).
During the decision-making process, especially in conflict situations, cultures that value
face-saving consider aggressive behavior as damaging the face of the other and feel it is
shameful to disturb interpersonal harmony (Moore, 1967). To this point, in conflict situations,
confrontations are regarded as face-threatening (Friedman et al., 2006) and people are normally
hesitant about engaging in such behavior (Kirkbride et al., 1991). Overall, in cultures with a high
degree of face -saving, individuals value the maintenance of a public face, and prefer the use of a
collaborative style to resolve conflict (Cushman & King, 1985). They therefore use more passive
strategies in handling conflict (Wolfson & Norden, 1984).
Accordingly, in a harmony model, employees respond to their negative situation at work
based on the criteria of whether or not their behaviors comply with their face-saving concern, or
in other words, whether or not their behaviors will hurt the employees‘ self-face or any other‘s
face in the organization. If the behavior is interpreted as a behavior enhancing or saving their
face, this behavior will be encouraged, thus employees prefer to choose this behavior.
Consequently, when employees confront a negative situation at work that they want to change or
improve, among exit, voice, loyalty, and neglect they choose the behavior consistent with the key
value: face-saving.
In the harmony model, which puts a high value on face-saving, loyalty is encouraged
since it does not hurt the employee‘s face. Because attachment to the organization is encouraged,
the employee who chooses the loyalty option will continue to put effort for the organization,
refusing to perform a confrontational action toward solving it. Therefore, the loyalty option is
interpreted as ‗cooperative behavior‘ by other members in the organization, without leading to

４８

negative or unfavorable judgment from other members in the organization. The voice option may
also be encouraged since it is interpreted as a cooperatively-orientated behavior, containing the
positive motive of making a cooperative contribution to the organization (Van Dyne & LePine,
1998). It will lead to enhancing other‘s favorable judgment about the employee, for example
―that employee is very active in making the organization or the work better.‖
On the other hand, employees will be likely to be discouraged from choosing the option
of exit and neglect since they will be concerned that it be interpreted as ‗threatening‘ behavior,
damaging the employee‘s face in the organization as well as disturbing the group or interpersonal
harmony. As described earlier, exit aims to terminate the tie with the organization, thus it can be
interpreted as aggressive and destructive behavior damaging the harmony within the organization.
For instance, exit can damage the employee‘s face by leading to other‘s unfavorable judgment
about the employee, such as ―that employee is not ‗us‘ anymore since he/she wants to leave our
organization.‖
Likewise, the neglect option will discouraged since it would hurt the person‘s face in the
organization. The employees‘ neglectful behaviors such as lateness, absenteeism, and high error
rates can be interpreted as unfavorable or disregardful behaviors and might lead to other
members‘ unfavorable judgment of the employee, such as ―he or she does not work hard,‖ or
―that employee is not a good fit for our organization.‖ Consequently,

HP 2. In a harmony culture, while face-saving will have a positive association with loyalty and
voice, it will have a negative association with exit, and neglect.

3.3.3 Conflict-Avoiding

４９

In the harmony model, integrating and avoiding behaviors to handle conflict is likely to
be encouraged (Kozan, 1997). Conflict-avoiding is defined as refusing recognition of a conflict
and engagement in any active action toward its resolution (Ohbuchi & Takahashi, 1994). As a
style of conflict management, conflict avoidance is characterized by unassertive and
uncooperative behavior in which contact with the other party is avoided to prevent visible shows
of conflict, by withdrawing from the situation or suppressing the urge to engage in conflict while
not actually cooperating with the other party (Rollinson & Broadfield, 1998). Many crosscultural studies in conflict management reveal a common empirical finding supporting a
culture‘s preference for a conflict-avoiding style, categorized as a harmony culture and found
particularly in Asian cultures (Bond & Wang, 1983; Chiu & Kosinski, 1994; Dyer & Song, 1997;
Friedman et al., 2006; Kirkbride et al., 1991; Miyahara et al., 1998; Morris et al., 1998; Tang &
Kirkbride, 1986; Trubisky et al., 1991).
More specifically, Chiu and Kosinski (1994) found that the Chinese—who are
categorized as a harmony culture—are more likely to support conflict avoidance than U.S.
respondents, categorized as a confrontational model. Other studies have also suggested that
Chinese culture promotes an indirect and avoidance style of handling conflicts (Bond & Wang,
1983). Tang and Kirkbride (1986) found that the Chinese are more likely to be on the avoiding
style than British respondents. Similarly, Trubisky and his colleagues (1991) found that
Taiwanese participants rely on an indirect avoidance style more than U.S participants. Morris
and colleagues (1998) also found that Chinese participants rely more on an avoidance style for
conflict management. In addition, Dyer and Song (1997) found that in when handling conflict,
Japanese respondents were less likely to engage in forcing behaviors than U.S. respondents,
supporting the strong influence of Confucian values. Kirkbride and colleagues (1991) also found

５０

that the Chinese are likely to prefer compromising and therefore avoiding addressing conflicts at
work. Furthermore, in their empirical study, Friedman and his colleagues (2006) suggested that
the Chinese highly prefer to avoid conflict, because they expect that direct conflict will hurt their
relationship with the other party, with greater concern for the other party in a conflict situation,
and a more direct approach will be interpreted as aggression and threatening to the relationship.
Similarly, the study of Miyahara and colleagues (1998) showed that in communication style,
Koreans focus on relationship with others in their organization.
Although the concept of conflict-avoiding is a unique concept to explain a behavioral
style of dealing with a conflict in an organization, it can be suggested that this type of behavioral
style will be reflected when handling a negative situation at work. Conflict is defined as
difference among perceptions, beliefs, and goals of organization members (Stroh et al., 2002)
and conflict-avoiding can be regarded as a style of avoiding the expression of differences among
organization members in perceptions, beliefs, and goals within the organization. Following this
logic, in a harmony model, with a conflict-avoiding style, employees will respond to their
negative situation at work in the way that avoids explicit expression that is not consistent with
their mental status quo.
In high conflict-avoiding culture, one is encouraged to pursue integrating behaviors in
handling a problematic situation at work and to refuse a confrontational action toward solving it.
Accordingly, employees will be more likely to choose loyalty since there is no explicit
expression: rather, the loyalty option is regarded as a passive or subtle response (Farrell, 1983;
Farrell & Rusbult, 1985; Hagedoorn et al., 1999; Rusbult et al., 1988; Withey & Cooper, 1989).
Since a loyal employee believes that the situation will get better without his or her involvement,
the employee may choose one example of the conflict-avoiding attitudes that Thomas (1992)

５１

listed: withdrawal, isolation, indifference, or ignorance. Furthermore, as mentioned earlier, the
loyalty option can be enacted by being silent (Hirschman, 1970), and patiently waiting for the
situation to improve (Saunders, 1992). Similarly, in high conflict-avoiding orientation, the
neglect option is frequently exercised since neglect is characterized by using a passive response
to sidestep the conflict (Ting-Toomey & Oetzel, 2001).
On the other hand, employees are likely to be discouraged from choosing the exit and
voice options since they will be concerned that these options will be interpreted as expressions of
disagreement or discontent with others in the organization or with the organization itself.
Previous studies support empirically that voice and exit are active responses, since they deal
directly with the problematic situation (Farrell, 1983; Farrell & Rusbult, 1985; Hagedoorn et al.,
1999; Rusbult et al., 1988; Withey & Cooper, 1989). In particular, leaving the organization can
be interpreted as the employee not having any more willingness to compromise the differences to
others in the organization or to the organization. Furthermore, in the exit option, even though the
employee does not leave actually leave, thinking of leaving or letting others know that the
employee is thinking of leaving can be interpreted as aggression threatening the organization.
For that reason, acting under a high conflict-avoiding tendency, employees will not choose exit
option.
Similarly, the voice option is discouraged in a high conflict-avoiding culture, since, in
most cases, the voice option accompanies any type of, in particular, verbal expression of
disagreement with the status quo that the organization maintains. The voice option means
expressions of constructive challenge and recommending modifications to standard procedures
even when others disagree (Saunders, 1992; Van Dyne & LePine, 1998). Considering that
Thomas (1992) describes conflict-avoiding as a style of avoiding explicit discussion of the

５２

conflict, the voice options will be discouraged by the high conflict-avoiding trait, which
emphasized refusing engagement in any active action toward its resolution (Ohbuchi &
Takahashi, 1994). In particular, the voice option will be discouraged when the voice behavior is
interpreted as an expression of negative feelings, which is discouraged in the harmony model.
Consequently,

HP 3. In a harmony culture, while conflict-avoiding will have a positive association with loyalty
and neglect, it will have a negative association with exit and voice.

3.3.4 Alternative Quality
Offering a new theory predicting employees‘ response to declining job satisfaction
among exit, voice, loyalty, and neglect, Rusbult and colleagues (1988) proposed that alternative
quality is one of the most important predictors. Alternative quality is defined as ―the degree that
employees have high-quality alternatives to their current job‖ (p. 604). They suggested that to the
degree that employees have high-quality alternatives to their current job, they should have the
propensity to engage in active responses. They illustrated that good alternatives–such as
attractive alternative job opportunities, the possibility of early retirement, or the acceptable
option of not working–promote motivation to do something about work problems and serve as a
source of power for effecting change, because if the situation gets worse or ends, the employee
has acceptable options such as leaving the organization. Low alternative quality means that the
employee does not have a good alternative to the current job, thus the employee is not available
to proactively involve him- or herself in the situation to improve it. Instead, the employee
remains passive (Rusbult et al., 1988).

５３

Previous empirical studies have yielded findings that say good alternatives promote exit
behaviors such as quitting and intent to quit (Dreher & Dougherty, 1980; Farrell & Rusbult,
1981; Mowday et al., 1984; Price & Mueller, 1981; Schneider, 1976; Stumpf & Hartman, 1984)
and voice behaviors such as whistle-blowing (Miceli & Near, 1984), grievance filing (Fleishman
& Harris, 1962; Muchinsky & Maassarani, 1980), and constructive social movements (Lammers,
1969; Zald & Berger, 1978). Other studies supported that good alternatives discourage
tendencies toward loyalty behaviors such as job commitment (McLaughlin & Butler, 1974;
Pfeffer & Lawler, 1980), and lessen tendencies toward neglectful acts such as absenteeism and
slow-down behaviors (Larson & Fukami, 1985; Watson, 1981). In contrast, other researchers
found that an employee with poor alternatives has a propensity for high levels of loyal behaviors
like job commitment (McLaughlin & Butler, 1974; Pfeffer & Lawler, 1980) and for neglectful
responses like absenteeism and slow-down behaviors (Behrend, 1953; Crowther, 1957; Larson &
Fukami, 1985; Owens, 1966; Watson, 1981; Youngblood, 1984).
Despite the differences in the details of the outcome behavior, the results have been
highly consistent: greater alternative quality should be associated with greater tendencies toward
active reactions to dissatisfaction (exit or voice) and lesser tendencies toward passive responding
(loyalty or neglect). In contrast, the absence of good alternatives promotes motivation to wait
quietly for conditions to improve (remain loyal) or to passively allow conditions to worsen
(engage in neglect). Consequently,

HP 4. While alternative quality will have a positive association with exit and voice, it will have a
negative association with loyalty and neglect across cultures.

５４

3.3.5 Self-Esteem
Some studies found that self-esteem had a positive relationship with voice (Brockner &
Elkind, 1985; Brockner et al., 1998; LePine & Van Dyne, 1998; Van Dyne et al., 1995), because
self-esteem is an important predictor of attitudes and behavior in both work and non-work
contexts (Brockner, 1988; Wells & Marwell, 1976), especially behavior that is somewhat risky to
the self (LePine & Van Dyne, 1998). Self-esteem refers to the degree of positive self-worth that
an individual ascribes to him- or herself (Brockner, 1988); it indicates the degree to which
individuals believe they are capable, significant, successful, and worthy (Coopersmith, 1967, p.
5). Researchers suggested that a higher level of self-esteem may encourage employees to take a
risk and to behave with confidence in the belief that they can make meaningful change at work
(Brockner & Elkind, 1985; Brockner et al., 1998; LePine & Van Dyne, 1998).
The key notion of self-esteem as an important predictor of voice is that, as discussed in
the literature review, by nature, voice behavior may be accompanied by proactive and assertive
non-conformance; voice behavior can be interpreted as change-oriented and challenging
behavior, therefore risky to employees because organizations often view dissent as nonconforming behavior (Van Dyne et al., 1995). In addition, speaking up and engaging in voice can
be interpreted as threatening cohesiveness and can result in sanctions against those who express
divergent opinions (Asch, 1956; Festinger, 1957; Hackman, 1992; Nemeth, 1986).
Accordingly, when all other factors are equal, people‘s positive beliefs that they can
provide meaningful input can make them more motivated than otherwise to use voice. Consider,
for example, individual differences in self-esteem. People who are relatively high in self-esteem
(high SEs) tend to believe that their viewpoints are correct and that their actions make a
difference. In contrast, their low self-esteem counterparts (low SEs) are less likely to maintain

５５

such beliefs (Brockner, 1988). If self-esteem is related to individuals' beliefs about their
capability to provide meaningful input (leading high SEs to be more motivated to have voice in
the decision process), it is reasonable that high SEs will be more influenced by the level of voice.
In this vein, researchers suggested that self-esteem is an important personality
determinant of voice behaviors. For instance, Van Dyne and colleagues (1995) empirically
revealed that self-esteem is as an important antecedent to voice behavior. Near and Miceli (1985)
and Miceli and Near (1992) also suggested that low self-esteem individuals may be less likely to
get involved in controversial situations at work and high self-esteem individuals might be more
likely to get involved in attempts to reform the situation. Similarly, Brockner and colleagues
(1998) also predicted that one determinant of employees‘ motivation to exercise their voice is the
expectation that their input will be influential; the more confident employees are of their
capability to provide input that may influence important outcomes, the more motivated they will
likely be to voice.
Accordingly, this dissertation proposes that employees with high self-esteem will be
more proactive in engaging their environments (voice) or continuing to support the organization
(loyalty), rather than setting aside (neglect). In contrast, low self-esteem can be interpreted as
thinking that one is not significant, successful, or worthy within the organization. In other words,
employees with low self-esteem think that they are powerless and they cannot make any change.
This negative belief will discourage them from behaving more actively within the organization.
For that reason they may not want to take the risk possibly caused by expressing their voice.
Rather they will be more likely to withdraw a proactive response to a deteriorating situation at
work (such as neglect). In the worse case, they will consider leaving the organization to find an
alternative that will make them capable, significant, successful, and worthy. Consequently,

５６

HP 5. While self-esteem will have a positive association with voice and loyalty, it will have a
negative association with neglect and exit across cultures.

3.3.6 Power Distance
Some research has examined how culture affects individuals‘ voice behavior and the
common concept that has been used is power distance. Power distance is one of Hofstede‘s
(1980) culture model for cultural differences (the four being collectivism-individualism,
uncertainty avoidance, power distance, and masculinity-femininity), which is one of the most
cited culture theory in organizational behavior literature, although not without its‘ critics
(Albers-Miller & Gelb, 1996; Egan & Bendick, 2007). Power distance (PD) refers to the extent
to which the less powerful members in an organization expect and accept that power is
distributed unequally (Hofstede, 1980). The basic notion of PD is that the organizational
hierarchyin which people are placed at the organizations that they belong to is very important to
all of the members, and power holders are entitled to privileges not available to those lower in
the hierarchy (Hofstede, 1983). Therefore, less powerful members in an organization view
inequality among persons in different positions of formal power as a natural (and even desirable)
aspect of the social order (Brockner et al., 2001).
A high power distance culture legitimizes differences of power between those who are in
high power positions versus those who are in low power positions in the decision-making
process (Brockner et al., 2001). In work contexts, employees with higher PD orientations believe
that supervisors should have a large amount of power over them. In contrast, employees who
have lower power than their supervisors are more constrained by prescribed role expectations; in

５７

high power distance cultures, they may be afraid to disagree with their superiors (Landau, 2009).
This means that even though they disagree with their supervisor or have an opinion that is
different from that of their supervisor, they will obey their supervisor‘s opinion (remain loyal).
In contrast, in a low power distance culture, norms reduce power differences among
people in positions with varying levels of formal decision-making power (Brockner et al., 2001).
People who have legitimate decision-making power are more likely to share their power with
those who have less power. Therefore, people who have less power are more likely to believe
that they should have a voice in decision-making processes, or at least more likely to believe that
they should have a voice in decision-making processes than would be the case in high power
distance cultures (Brockner et al., 2001). In work contexts, employees with lower PD
orientations believe that all people should be relatively equal and that they should have some
input into the decision-making process, and therefore have the right to speak up with their ideas
and opinions (Hofstede, 1980).
Accordingly, employees who have high power distance will respond to negative
situations at work by considering whether or not their behavior complies with organizational
hierarchy, in other words, whether or not their behaviors will challenge the authority in the
organization that they belong to. If the behavior is interpreted as a behavior threatening the
authority of the organization, they will not choose the behavior. Individuals with a high power
distance orientation are unlikely to believe that they have anything important to say, and even if
they do have an idea, they will not believe that it is their responsibility to speak up (Landau,
2009). Therefore, they will be less likely to choose the voice option. Previous researchers have
frequently supported the idea that high power distance has negative relationships with voice
(Botero & Van Dyne, 2009; Landau, 2009; McNab & Worthley, 2008; Xu et al., 2005).

５８

As described above, since people who have high power distance orientation believe that
employees who have lower power should obey the authority of the organization, they will be
more likely to refuse a confrontational action aimed at solving a deteriorating situation in the
organization they belong to. Rather, they will remain loyal, continuing to fulfill their obligations
or at least withdraw from a confrontational situation (neglect), rather than expressing
disagreement. In the worse case, they will consider escaping the authority by leaving the
organization (exit). Therefore, when employees confront a negative situation at work that they
want to change or improve, among exit, voice, loyalty, and neglect, higher power distance will
lead to exit, loyalty, and neglect options while discouraging the voice option. Consequently,

HP 6. While power distance orientation will have a positive association with exit, loyalty, and
neglect, it will have a negative association with voice across cultures.

3.3.7 LMX: Leader Member Exchange
Researchers have emphasized the importance of the supervisor‘s role to employee voice,
especially LMX (Botero & Van Dyne, 2009; Brown et al., 2005; Chia et al., 2000; Detert &
Burris, 2007; Janssen et al., 1998; Kassing, 2000; Landau, 2009; Saunders et al., 1992; Van
Dyne et al., 2008; Walumbwa & Schaubroeck, 2009). LMX refers to the quality of supervisor–
subordinate relationships (Dansereau et al., 1975), and researchers have suggested that LMX is
an important contextual climate affecting employees‘ voice behaviors (Botero & Van Dyne,
2009; Kassing, 2000; Van Dyne et al., 2008). High-quality LMX relationships are characterized
by mutual trust, professional respect, reciprocal influence, liking, and a sense of obligation
toward leaders (Graen & Uhl-Bien, 1995; Liden & Maslyn, 1998). On the contrary, low-LMX

５９

means that employees have more formal and restricted relationships with their supervisors that
are based on economic exchange, and are characterized by low trust, low support, and few
rewards (Botero & Van Dyne, 2009). These employees have limited access to their supervisors,
fewer resources, and restricted information.
Previous research has supported that LMX quality affects employees at the workplace in
various areas, such as employee satisfaction, promotions, performance ratings, OCBs, and
communication behaviors (Gerstner & Day, 1997; Kamdar & Van Dyne, 2007; Liden, Sparrowe,
& Wayne, 1997). LMX research has commonly found that LMX quality has a positive
relationship with employees‘ voice behavior (Botero & Van Dyne, 2009; Van Dyne, Kamdar, &
Joireman, 2008; Walumbwa & Schaubroeck, 2009). According to the research, high-quality
LMX relationships provide employees with more opportunities to speak up, exchange
information or ideas with their supervisors, and use more communication channels, compared to
those in low-LMX relationships (Landau, 2009), encouraging employees to exercise their voice.
In particular, considering that voicing to a supervisor can be risky (because the supervisor
usually has the power to influence an employee‘s work status, e.g. work assignments, pay raises,
and promotions), employees are only likely to voice to management if they perceive that the
benefits of speaking up outweigh the costs, and that their suggestions will be treated seriously
(Landau, 2009). In addition, Saunders, Sheppard, Knight, and Roth (1992), Janssen, de Vries,
and Cozijnse (1998), and Chia, Landau, and Ong (2001) found that employees‘ willingness to
voice was facilitated by supervisors who were good voice managers 1) to the extent employees
feel they can bring their concerns to their supervisors without being penalized (Saunders et al.
1992) and 2) to the extent to which supervisors are prompt and willing to take action to deal with
the issues voiced by employees (Saunders, Sheppard, Knight & Roth, 1992).

６０

Given this previous research, we can expect that when employees respond to negative
workplace situations to be changed or improved, high-quality LMX relationships will encourage
employees to choose voice options, providing employees with contexts in which they are willing
to speak up and express their ideas for change, as well as with better sense that communication is
open. Similarly, employees who have high-quality LMX relationships will be more likely to
choose loyalty option, as a way of reciprocating a good relationship with their supervisor,
providing them with mutual trust, professional respect, reciprocal influence, liking, and a sense
of obligation. For this reason, destructive behaviors that can possibly hurt the good relationship
with the supervisor, such as exit and neglect, are discouraged. Consequently,

HP 7. While LMX will have a positive association with voice and loyalty, it will have a negative
association with exit and neglect across cultures.

3.3.8 Control Variables
In addition to the eleven variables above, a list of demographic characteristics–such as
nationality, ethnicity, age, gender, and education–as well as work-related characteristics–such as
tenure in the organization, and occupational type–have been found to or will possibly be related
to employee voice behaviors.
First, the nationality of the respondent was included based on Hofstede‘s culture study.
According to his culture theory, people from the same country would exhibit similar behaviors;
therefore the nationality of the individual is a strong factor in predicting an individual‘s
behaviors. In particular, considering that Kozan (1997) indicated Asian countries, Korea in this
dissertation, as examples of harmony culture, whether or not participants are Korean will be

６１

added as a control variable. Second, ethnicity was included as a control variable, since previous
voice behavior studies such as LePine and Van Dyne‘s (1998) study found that whites engaged
in more voice than non-whites. Third, age was included as a control variable since previous
research has shown that older employees are more likely to speak up than younger employees
(Luchak, 2003). In addition, research has shown that males are more likely to voice than females
(LePine & Van Dyne, 1998), and employees with higher levels of education are more likely to
voice than those with less education.
Among work-related characteristics, tenure in the organization was included because
previous research has also shown that those with greater tenure are more likely to voice than
those with shorter tenure (Stamper & Van Dyne, 2001). In addition, occupational type has been
found to be related to employee voice behaviors. For instance, Near and Miceli (1996) found that
whistleblowers tended to hold professional jobs.
Overall, although these control variables are not the primary focus of the analysis,
previous studies have supported their relationships with employee voice behavior, and this study
includes these variables.

６２

CHAPTER 4
METHODOLOGY

To fulfill the objectives of this study, this chapter discusses the research methodology
that was used to answer the proposed research hypotheses. This chapter includes the following
sections: (a) research design including rationale for research method and sampling strategies, (b)
determination and acquisition of subjects, (c) translation of materials, (d) measures, and (e)
analysis strategies.

4.1 Research Design

4.1.1 Rationale for Research Method
To test the hypotheses, field research with a cross-sectional survey was conducted. This
survey method is considered most appropriate because of its ability to provide a broad,
generalizable set of findings by obtaining responses from many people. A survey is often used
for its effectiveness at examining relationships between factors (Trochim, 2001).

4.1.2 Sampling Strategies
In this dissertation, a cross-sectional survey will be conducted with two samples--the U.S.
sample and the Korean sample–from employees of Korean organizations in the automotive
manufacturing industry, operating in the U.S. and in Korea respectively. Sampling strategies for
this study were developed 1) based on guide of research design, especially for cross-cultural
research, and 2) on findings from cross-cultural research of employee voice.

６３

Sampling is at the core of data collection and a number of sampling types have been
described in the literature (see Boehnke, Lietz, Schreier, & Wilhelm, 2011), the reality, however,
is that many researchers frequently take a convenience sampling. The main reason for this may
be accessibility, especially considering cost and time. They argue that, practically, it is
impossible to create a complete sampling frame and draw a random sample without auxiliary
information (Boehnke, Lietz, Schreier, & Wilhelm, 2011). Researchers often select particular
cultures for their cross-cultural research because of their familiarity with a certain culture or
because they already know somebody in that particular country. Despite these practical
considerations, it is suggested that all choices should be made based on theoretical grounds, as
well as practical grounds (Boehnke, Lietz, Schreier, & Wilhelm, 2011; Lyons, & Chryssochoou,
2000). In this dissertation, the key strategy for sampling is based on guidance given by Lyons
and Chryssochoou (2000).

4.1.2.1. Selection of Country
According to Lyons and Chryssochoou‘s (2000) cross-cultural research, there are two
stages of sampling; first, determining which cultural groups will be used, and second,
determining which subgroup within each culture will be used. First, as to which cultures/nations
are to be used, they suggest that this decision should be based on theoretical grounds.
Researchers should decide which variables/aspects of culture are likely to have an effect on the
target behavior and then choose cultures which are likely to vary among those particular
dimensions. This dissertation has attempted to examine three key variables extracted from a
harmony culture, based on Kozan‘s (1997) culture model. Since Kozan (1997) stated that Asian
countries are examples of a harmony model, this dissertation will take Korea as a main culture

６４

group. This is highly compatible with Kozan‘s (1997) harmony model. The U.S. will be used as
an example of a non-harmony model since Kozan (1997) mentions the U.S. as an example of
other cultures.

4.1.2.2. Selection of Subgroup
As for which subgroup within each culture will be involved in the research, the
researcher should try to match the subgroup to the research purpose across the selected cultures
in the first stage of the sample selection (Lyons & Chryssochoou, 2000): For instance, if the
research is on cultural differences in consumer behavior, it will not make sense to compare
middle class adults in one culture with working class adults in another. The research focus of this
dissertation is to explore certain employees‘ behaviors at work, particularly the effects of culture
on employees‘ voice behavior. Therefore, this dissertation will use employee samples from the
U.S. and Korea. Furthermore, by selecting a sample in the same industry and within the country,
this dissertation attempts to make these two countries‘ samples as compatible with each other as
possible. Therefore, in this dissertation, the subjects will be restricted to employees at the Korean
companies in the automotive manufacturing industry in the U.S. and Korea, in an effort to
exclude other possible contextual factors such as industry characteristics and the nationality of
the company.

4.1.2.3. Employee Sample, Not Student Sample
This dissertation will use an employee sample, not a student sample, based upon the
literature review on voice. For instance, to suggest the appropriate methodology for voice
research, Miceli and Near (2005) argue that using college student samples is not appropriate for

６５

investigating how and why people will speak up at work. They argue that, given the highly
contextual nature and the complex decision-making processes involved in employee behavior,
experimental methods with college students cannot capture many of the relevant dynamics that
are required when making the kind of career-relevant evaluations that feature in decision to
engage in certain behavior at work.
Another finding from previous researchers is that the findings from studies with varying
methods are quite consistent when they use employee samples. For instance, Van Dyne and
LePine‘s (1998) longitudinal multi-group results demonstrated that voice behavior is stable over
six months. Van Yperen and colleagues (2000) used two methods, 1) asking questions with a
general statement such as, ―When you encounter a problem or a problematic event at work, how
likely is it that you…?‖ and 2) an experimental vignette study with a fictitious situation. They
concluded that the results were consistent with other field studies, and determined that the
reasons for the consistent results may be the use of employee sample in the vignette study, and
the fact that the vignette confronted employees with an actual and realistic problematic event. In
sum, considering the dynamics caused by the highly contextual nature and the complex decisionmaking process involved in employee voice behavior, it is appropriate to use an employee
sample, rather than student sample. Accordingly, this dissertation will use an employee sample.

4.2 Determination and Acquisition of Subjects
Data was collected from two samples--the U.S. sample and the Korean sample–from
employees of Korean organizations in the automotive manufacturing industry, operating in the
U.S. and in Korea. The data was collected in fall 2010. As for deciding which organization to use,
one consideration in selecting a target organization was that the organization needed to hire a

６６

number of local employees at both sites. In the process of selecting initial candidate
organizations, I found that many Korean organizations do not have enough local employees in
the U.S. to compare these two cultural groups. Rather, they mainly dispatch their Korean
employees from Korea to the U.S. site on a regular rotation, or some prefer to hire a person
familiar and embedded within Korean culture--such as Korean students in U.S. schools or
Korean-Americans--to work in the U.S. For that reason, one criteria of selecting target
organizations was whether or not the organization had local employees at both sites for cultural
comparison. Overall, I selected target organizations based on the following criteria: 1) Korean
organizations, 2) operating in U.S. facilities as well as in Korea, 3) with enough local employees
at both sites for cross-cultural comparison, 4) in the same industry. Therefore, I listed the Korean
organizations from the automotive industry with operating manufacturing facilities in both sites.
Initially, I had contacted these organizations to request consent by presenting them with
the project, including the nature and methods of research, a couple of months prior to the
anticipated launch date. The problem was that most of them hesitated to approve the survey, for
two main reasons. First, they did not approve the survey because they were concerned that the
organization would be easy to identify, even though confidentiality was promised. Second, they
did not want to participate in this survey in particular, because, in their opinion, it addresses
many sensitive issues. For instance, some organizations said employee voice behaviors and
relationships with a supervisor are not comfortable issues that they want discussed publicly.
Some also declined this survey since the organization‘s union did not approve it, due to the
sensitivity of the items in the survey.
Even after granting permission to survey subjects, the organizations placed very strict
limits on the interactions. For instance, in one case, access was only provided to limited spaces

６７

such as the cafeteria and only during the lunch period. In addition, some organizations, both in
the U.S. and in Korea, cancelled the survey immediately before the anticipated date without any
advance notice. To compensate for this unanticipated reduction in potential sample size, I also
needed to recruit subjects from additional organizations in the same industry.
Finally, four facilities representing three organizations granted permission: two facilities
in the U.S. and two in Korea. The details of the accessible time and place at each site, however,
were dependent upon the conditions that each organization required for the permission, due to
their high concern. One organization in the U.S. asked me to organize a one-time pizza lunch for
the employees. Another allowed me to solicit surveys in the cafeteria during the daytime for a
period of two days. In the permitted place during the permitted time, questionnaires were
distributed to the employees who came to the site during that time. The employees were asked to
complete the questionnaire and place their responses in the box left for them before leaving the
site. At the end of the permitted time, the box was sealed and retrieved.
Regardless of the variations in the schedules and the restriction of administering the data
due to the organizations‘ situational restrictions, the three things ensured across sites were that 1)
all participants in the research were totally voluntary, 2) data collection was conducted by myself
alone, without HR managers or other managers in the organization, and 3) the survey did not
allow individuals to be identified in any case. The survey does not label respondents by name or
by other information which might allow participants to identify themselves or others, and no data
that would allow respondents to be identified as an individual will be shared with anyone at the
organization.

６８

An additional consideration in the data collection at the U.S. sites is that since the
1

purpose of collecting data in the U.S. site is to collect data from the non-Korean employees,
questionnaires were distributed only to the non-Korean employees at the U.S. site. To do so,
before distributing a questionnaire, I asked the Asian employees whether or not they were
Korean. By doing so, I refrained from distributing it to the Korean employees at the U.S. site.
For that reason, even though some respondents from the U.S. site did not identify their ethnicity,
there is no possibility that they can be Korean. 543 questionnaires were distributed and 510 valid
responses were collected (93.9 percent response rate).

4.3 Subjects
4.3.1 Rationale for Sample Size
510 valid responses were collected as data to be used in this dissertation. As for sample
size, although there is a good deal of debate around the issue of sample size, there is little
theoretical guidance or ―rule of thumb‖ to determine an adequate sample, especially in crosscultural research. Determination of sample size based on a power analysis (Cohen, 1977) is
difficult considering the nascent stage of the literature in cross-cultural employee voice research.
Lee and Jablin (1992), in their cross-cultural study of the EVLN construct, utilized 225 students-100 Korean students, 80 U.S. students, and 45 Japanese students from a university in the U.S.-to examine the generalizability of the EVLN construct. Botero and Van Dyne (2009) used a
convenience sample of 247 employees, 109 in the U.S. and 138 in Colombia, to examine LMX
and PD as predictors of voice. To investigate features of the value function for voice across
1

To avoid confusion or further discussion about nationality versus ethnicity, Korean, in this
dissertation, is restricted to those who have grown up and work in Korea and consider it their
native country. For instance, the Korean-Americans or Korean immigrants in the U.S. are not
referred to as Korean in this dissertation.
６９

countries, Price and colleagues (2001) used a sample of 511 students: 118 students in Great
Britain, 185 students in Mexico, 99 students in the Netherlands, and 109 students in the United
States.
One factor to be considered in sample size is derived from Hogarty, Hines, Kromrey,
Ferron, and Mumford‘s suggestion (2005): they suggested that there should be less emphasis on
sample size ―rule of thumb‖ and greater emphasis on sound research design. Considering that
this dissertation attempted to develop a sound research design as described in the sampling
strategy, and that there is no consistent ―rule of thumb‖ in this research area, as shown in the
literature review above, I anticipated that 510 respondents would be sufficient for this phase of
the research.
Another guidance for sample size is brought from validity test studies of related
psychological constructs. In a criterion of validity test study of newly developed organizational
justice measures, Colquitt (2001) used a sample of 337 employees from a manufacturing
organization. In a study to evaluate the validity of a new measure of organizational citizenship
behavior, Becker and Randall (1994) used a sample of 112 restaurant employees within a field
setting. Hinkin (1998), and Guadagnoli and Velicer (1988), suggest that 200 subjects should be
sufficient for internal consistency and factor structure analysis. Accordingly, 510 respondents is
sufficient for this dissertation to conduct internal consistency, and factor structure analysis.

4.3.2 Demographic Information of Subjects
Table 2 shows the demographic statistics of the subjects in this dissertation. The sample
consisted of 415 Korean employees in the Korean sites and 95 non-Korean employees in the U.S.
sites; 433 employees are male and 21 employees are female; in education, 310 employees (68

７０

percent) have a high school diploma or GED. The average age was 39.4 years and the average
length of employment was 12.8 years. 386 employees (77 percent) fall into the
Executive/Management, Sales/Marketing, and Administrative/Clerical job categories.
Additionally, ethnicity was included as a response, however, this data failed to obtain
sufficient variation. For instance, among 510 respondents, 30 respondents are Caucasian (seven
percent), 19 respondents are African American (four percent), and two respondents are Hispanic.
Since there is insufficient variation in the responses to generate results, ethnicity was excluded
from the data analysis.

Table 2 Subject Demographic Statistics
Org
Valid N

1

2

3

4

Location

KOR

KOR

U.S.

U.S.

Population

575

320

78

65

Distribution
Response

510

Age

475

Tenure

474

Gender

505

Education

503

Total
1,038

329
110
61
43
543
312
103
55
40
510
94.8% 93.6% 90.2% 93.0% 93.9%
Mean
39.43 39.59 37.95 40.57 39.4
S.D.
5.16
6.04 10.59 9.19 10.04
Mean
14.8
14.0
2.03
3.9
12.8
S.D.
5.15
6.22
1.14
4.88
3.61
1
7
7
6
21
Female
0%
8%
21%
20%
5%
296
86
27
24
433
Male
100% 92%
79%
80%
95%
0
1
0
3
4
Some high school
0%
3%
0%
10%
1%
23
15
8
310
High school diploma or 264
GED
89%
25%
44%
27%
68%
28
7
13
7
55
Some college
9%
8%
38%
23%
12%
3
56
5
11
75
College degree
1%
60%
15%
37%
17%
0
6
0
3
9
Master‘s degree or
higher
0%
6%
0%
10%
2%

７１

Table 2 (cont‘d)
Org
Valid N

Job

504

1

2

3

4

Total

4
71
3
10
88
1)
Executive/Management,
Sales/Marketing, and
1.3% 70.3% 5.5% 26.3% 17.5%
Administrative/Clerical
0
10
6
14
30
2) R&D and
Technical/Professional .0%
9.9% 10.9% 36.8% 6.0%
306
20
46
14
386
3) Skilled Crafts and
Laborer
98.7% 19.8% 83.6% 36.8% 76.6%
0
1
0
3
4
Others
0%
3%
0%
10%
1%

As shown in Table 2, the data reveals some notable points. First of all, the table shows
that across sites, most of the respondents are males in the skilled crafts or laborer job category;
this is possibly due to the labor force demographics in manufacturing industry. Second, the
Korean sample shows relatively longer tenure than in the U.S.; that is possibly because the
Korean automotive facilities in the U.S. were established much later than those in Korea, as
subsidiaries of the Korean company and in Korea, life-long employment still somewhat remains.
Finally, as for job categories, Organization 2 in Korea shows a high composition of the following
job categories: Executive/Management, Sales/Marketing, and Administrative/Clerical (70.3%).
This is possibly due to the restriction of access. As described earlier in the description of
determination and acquisition of subjects, one organization only allowed me to one cafeteria
among a couple on the site. It is possible that this available cafeteria is located close to the
managerial are of the facility, therefore a relatively larger number of managerial or clerical
employees in the plant participated in the survey than at other sites. The limitations of sampling
will be addressed further in the section on the limitations of the study.

７２

4.4 Translation of Materials
Since all of the materials were created in English, they were translated into Korean for
the Korean sample. All survey items were translated into Korean by five bilingual speakers who
have work experience. Subsequently, a back translation was prepared by the researcher.
Adjustments were made when there were differences between the original items and the backtranslated items, to keep all items equivalent and consistent with the English version. Most of the
English version scales used in this study had previously been used by other researchers, therefore
consistency between these measures was necessary for comparison with previous studies.

4.5 Measures
4.5.1 Dependent Variables

EVLN
To measure employees' responses to unsatisfactory work situations, an existing scale of
the EVLN construct was used from Rusbult and colleagues‘ (1988) EVLN measurement. As
discussed in the literature review, although they measure intention rather than actual behavior,
this study used their measurement since it is most commonly used in later studies (such as Lee &
Jablin, 1992; Rusbult et. al.; Turnely & Feldman, 1998, 1999; Withey & Cooper, 1989) testing
the EVLN model, and this measure produced a satisfactory reliability value. The items are shown
in Table 3.
In more detail, consistent with previous studies, the exit, voice, loyalty, and neglect were
measured with generalized response tendencies rather than responses to a particular dissatisfying
incident (Rusbult et al., 1988). It is also consistent with the Hirschman (1970) initial

７３

conceptualization of voice, which did not specify a certain situation, but rather responded to ―the
deterioration in the performance of a firm or an organization, for unspecified, random causes
which are neither so compelling nor so durable as to prevent a return to previous performance
levels‖ (p.4). Therefore, in the questionnaires, the EVLN scales were introduced with the
following statement: ―Now consider a situation in which you think something important in your
job needs to be changed or improved. For instance, consider a situation in which you think that
you are being treated unfairly at work. In the above situation, what would you do?‖ Participants
were asked to indicate their level of agreement with each statement, using a 7-point Likert scale
(1= Not at all, and 7= Great Extent).
The items for each variable were summed and averaged. Crombach‘s alpha for exit is .75,
voice is .80, loyalty is .56, and neglect is .88 respectively.

Table 3 Items of EVLN Construct
Variables Items
Exit

Voice

Loyalty

Neglect

I would think about quitting my job.
I would let my supervisor know that I was thinking about quitting.
I would accept another job offer.
I would quit my current job even without a job offer.
I would go to my immediate supervisor to discuss the problem.
I would ask my co-workers for advice about what to do.
I would talk to another member of management about how I felt about the
situation.
I would try to solve the problem by suggesting changes in the way work was
supervised in the office.
I would hang in there and wait for the problem to go away.
I would stick with my job.
I would still think that my job was as good as most.
I would wait patiently for the problem to disappear.
I would lose motivation to do my job as well as I might otherwise.
I would call in sick occasionally because I didn't feel like working.
I would take a lot of breaks.
I would show up late because I wasn't in the mood for work.
I would put less effort into my job.
７４

4.5.2 Independent Variables

Collectivism
Collectivism was measured by three items generated based upon the work of Hofstede
(1980, 1991). Hofstede‘s original items had to be modified to suit to measure work-related
values. Existing studies based directly or indirectly on Hofstede‘s work (e.g. Earley, 1993; Hui,
1984; Hui & Triandis, 1986; Trandis et al., 1995) generated additional items. Participants were
asked to indicate their level of agreement with each statement, using a 7-point Likert scale (1=
Not at all, and 7= Great Extent). The items were summed and averaged to create a single
measure in which higher score indicates a higher level of collectivism. Crombach‘s alpha for
collectivism is .70.

Face-Saving
Four items from Oetzel and Ting-Toomey (2003) regarding face-concern were used to
assess the participants‘ level of agreement, measured with a 7-point Likert scale (1= Not at all,
and 7= Great Extent). The face-saving score was created by averaging across all four items, to
create a single measure. Crombach‘s alpha for face-saving is .81. A higher score indicates higher
levels of face-saving.

Conflict-Avoiding
Conflict-avoiding was measured by four items of Morris and colleagues‘ (1998)
measurement, with a 7-point Likert scale (1= Not at all, and 7= Great Extent). The items were

７５

summed and averaged to create a single measure. Cronbach‘s alpha for conflict-avoiding is .88.
A higher score indicates higher levels of conflict-avoiding.

Alternative Quality
Two items from Farrell and Rusbult‘s (1992) study were used to measure participants‘
perception of the quality of alternatives with a 7-point Likert scale (1= Not at all, and 7= Great
Extent). The items were summed and averaged to create a single measure. Crombach‘s alpha for
alternatives quality is .90. A higher score indicates higher levels of alternative quality.

Self-Esteem
Self-esteem was assessed with five items from the Organization-Based Self-Esteem Scale
(OBSE, Pierce et al., 1989). The OBSE is defined as "the self-perceived value that individuals
have of themselves as organizational members acting within an organizational context" (Pierce et
al., 1989, p. 625). Therefore, it seemed more worthwhile to assess participants' self-esteem,
specifically in the organizational setting, than other measures, which stem from a history of
organizational, interpersonal, and systemic experiences (Brockner et al., 1998). For each item,
participants were asked to indicate their level of agreement using a 7-point Likert scale (1= Not
at all, and 7= Great Extent). The items were summed and averaged to create a single measure.
Crombach‘s alpha for self-esteem is .88. A higher score indicates higher levels of self-esteem.

Power Distance
Six items for Power Distance were generated based on Hofstede‘s (1980, 1991) work.
Hofstede‘s original items had to be modified to suit to measure work-related values (e.g.

７６

Dorfman & Howell, 1988). The items were addressed using a 7-point Likert scale (1= Not at all,
and 7= Great Extent). The results were summed and averaged to create a single measure.
Crombach‘s alpha for power distance is .83. A higher score indicates higher levels of power
distance.

LMX
Five items were used from Graen and Uhl-Bien‘s (1995) study to assess how participants
perceive the quality of leader-member exchange. For each item, participants were asked to
indicate their level of agreement using a 7-point Likert scale (1= Not at all, and 7= Great Extent).
The items were summed and averaged to create a single measure. Crombach‘s alpha for LMX
is .88. A higher score indicates higher levels in the quality of the leader-member exchange.

Table 4. Items of Independent Variables
Variables
Items
Individuals should sacrifice self-interest for the group that they belong to.
Group success is more important than individual success.
Collectivism
Individuals should pursue their goals after considering the welfare of the
group.
Face-saving

It is important to be sensitive to another person‘s self-worth.
I am concerned with helping the other person to maintain his/her credibility.
I am concerned with not bringing shame to myself.
I am concerned with protecting my self-image.

Conflictavoiding

I believe it is better to keep negative opinions to ourselves rather than create
hard feelings.
I avoid being put "on the spot" by keeping conflict to ourselves.
I try to stay away from disagreements.
I usually avoid open discussions of my differences with the other person.
I generally avoid an argument.

７７

Table 4. (cont‘d)
Variables

Items

Alternative
Quality

If you left this job, would your next job probably be better than the job you
have now?
How confident were you that you would find a satisfactory job if you had quit
this job?

Self-Esteem

I am important around here.
I am trusted around here.
I am efficient around here.
I am valuable around here.
I am cooperative around here.

Power
Distance

It is better not to disagree with management decisions.
When my supervisor makes a decision with which I disagree I prefer to accept
the decision rather than question it.
I believe that it is not right to disagree with my boss.
People in higher positions should make most decisions without consulting
people in lower positions.
People in higher positions should not ask opinions of people in lower positions
too frequently.
People in lower positions should not disagree with decisions made by people
in higher positions

LMX

My supervisor is investing a great deal in my career.
I often share my good ideas with my supervisor.
I am open to my supervisor‘s suggestions regarding my work.
My working relationship with my supervisor is better than average.
My working relationship with my supervisor is very effective.

4.5.3 Control Variables
Participants were asked questions about their demographic characteristics such as
nationality, ethnicity, age, gender, and education, as well as work-related characteristics such as
tenure with the organization, and occupational type.
Country variable was also coded as a dichotomous variable where 1 = Korea and 0 =
other. Ethnicity was listed with five responses: Caucasian, Asian, African/black, Hispanic/
Latino, and Other. In the analysis, ethnicity was recoded as a dichotomous variable where 1 =
White and 0 = other. Age was asked with a single question: ―indicate your age (in years),‖ and

７８

gender was requested with two categories: 1) female and 2) male. Education had five possible
answers: 1) some high school, 2) high school diploma or GED, 3) some college, 4) college
degree, and 5) masters degree or higher. Job tenure was evaluated with a single item, ―How long
have you worked for your current organization (in years)?‖ Occupational type consisted of three
categories: 1) Executive/Management, Sales/Marketing, and Administrative/Clerical, 2) R&D
and Technical/Professional, and 3) Skilled Crafts and Laborer. In the analysis, occupational type
was recoded as three dummy variables 1) whether or not the respondent falls under the category
of Executive/Management, Sales/Marketing, and Administrative/Clerical, 2) whether or not the
respondent falls under R&D and Technical/Professional, and 3) whether or not the respondent is
considered Skilled Crafts and Laborer.

4.6 Analysis Strategies
In this dissertation, two statistical methods were employed for the analyses of the
proposed model: confirmatory factor analysis and multiple regression. The purpose of using
confirmatory factor analysis is to conduct construct validity tests in order to examine if items
used in this dissertation actually measure the latent variable they are supposed to assess. The
purpose of using multiple regression method is to examine the relationships between each
dependent variable and multiple independent variables. Confirmatory factor analysis was
conducted using AMOS 19 and multiple regression is conducted using SPSS 19.

4.6.1 Confirmatory Factor Analysis (CFA)
Substantive use of SEM has been growing across disciplines, including psychology, the
social sciences, and biology. The main reason for the popularity of SEM is that 1) SEM provides

７９

researchers with a comprehensive method for assessing and modifying complex theoretical
multivariable models (e.g. Bentler, 1983), and 2) SEM takes into account measurement error that
is ubiquitous in most disciplines, and which typically contain latent variables (Raykov &
Marcoulides, 2006). Therefore, SEM is used mainly 1) when researchers are interested in
evaluating the extent to which particular instruments actually measure a latent variable they are
supposed to assess, called construct validation, and 2) when researchers are interested in
exploring potential relationships between variables of interest for theory development (Raykov
& Marcoulides, 2006).
In particular, this dissertation used confirmatory factor analysis (CFA), since it is best
suited for construct validation, to examine the items used in this dissertation to measure
independent variables. Confirmatory factor analysis (CFA) is the most common method to test
the construct validity of survey items. The difference between explanatory factor analysis (EFA)
and confirmatory factor analysis (CFA) is that EFA does not necessarily require prior theory,
while CFA is conducted based on pre-established theory.

4.6.2 Model Fit
In CFA analysis, the model was tested for validity by using goodness-of-fit indices.
Goodness-of-fit (GOF) indicates how well the specified model reproduces the covariance matrix
among the indicator items. In this dissertation, the model fit for CFA will be evaluated with a
root mean square error of approximation (RMSEA) and the Incremental Fit Index (IFI).
Although there are other indices for evaluating model fit, the analyses utilized the above two
indices since they are most commonly recommended and used by organizational researchers. In

８０

particular, considering the strengths and weaknesses of each index, it is better to use multiple
indices for examining fit in a comprehensive way.
The RMSEA is a measure of the discrepancy between the sample, S, and the modelimplied matrix per degree of freedom, derived from the χ2 test and expressing the degree of fit of
model to data, taking both sample and model size into account (Byrne, 2001). The key feature of
the RMSEA among the fit indices is that the RMSEA is least affected by sample size, unlike
many other fit indices that are sample-dependent or have a characteristic of their distribution,
such as the mean, depending on sample size (Raykov & Marcoulides, 2006). Bentler (1990) and
Marsh, Balla and Hau (1996) recommend the RMSEA since it is complemented by three relative
goodness-of-fit indices: the Non-Normed Fit Index (NNFI), the Incremental Fit Index (IFI), and
the Comparative Fit Index (CFI). For this reason, the RMSEA has recently become a popular
index of model fit (Raykov & Marcoulides, 2006). It has been suggested that a value with a
RMSEA of ≤0.05 is considered to indicate a good model fit for the analyzed data (Byrne, 2001).
A value of about 0.08 or less for the RMSEA would indicate an acceptable model fit with a
reasonable error of approximation. If the value is equal to or greater than 0.1, the model would
not be considered a good fit.
The Tucker-Lewis Index (TLI) is an example of relative fit indices comparing a chisquare for the model tested to one from a so-called ―baseline‖ model, or a tested model that
specifiest all measured variables as uncorrelated (there are no latent variables) and which should
always have a very large chi-square (poor fit) (Raykov & Marcoulides, 2006). Although other
baseline models can be used, this is not often seen in practice. TLI has generally been used with
a conventional cutoff in which values larger than .90 are considered good-fitting models.

８１

4.6.3 Multiple Regression
To examine the proposed relationships between dependent variables and independent
variables in this dissertation, multiple regression was conducted. Multiple regression analysis is
one of the most widely used of all statistical methods, since multiple regression is an appropriate
method for analyzing the relationship between a single dependent variable and several
independent variables. In this dissertation, the dependent variables in the analysis were the exit,
voice, loyalty, and neglect options. Each dependent variable was regressed upon the following
independent variables: collectivism, face-saving, conflict-avoiding, alternative quality, selfesteem, power distance, and LMX, as well as control variables including nationality, age,
education, tenure in the organization, and occupational type.
In particular, in order to develop the ―best‖ subset of predictor variables, a series of
regressions was performed. The three step regression method develops a sequence of regression
models, at each step adding or deleting potential predictor variables. The criterion for adding or
deleting a predictor variable can be stated equivalently in terms of error sum of square reductions,
coefficient of partial relations, t statistics, or F statistic. Therefore, stepwise search procedures
end with the identification of a single regression model as ―best.‖

²

The best model is selected based on R value, which is a summary statistic of the impacts

of all of the independent variables taken together. The model with the highest R

² among the

produced models, is the best combination of predictors of the models; therefore, the model with

²

the highest R is selected as the best model. Then the change in R

８２

² is assessed, to check whether

²

or not the model has significant changes in R . The F value for the "Change Statistics" shows the
significance level associated with adding the variable for that step.
In this analysis, dummy country variables and some demographic variables (such as
education, tenure, and job category) that showed statistically significant relationships with
dependent variables were entered first (Model 1), followed by other independent variables
(alternative quality, self-esteem, power distance, and LMX)(Model 2), then the three main
harmony culture variables (collectivism, face-saving, and conflict-avoiding)(Model 3). The
model summary and beta coefficient values with the significance level were provided separately
for each of the dependent variables, including exit, voice, loyalty, and neglect. In addition, these
stepwise regressions were conducted with 1) the total sample, 2) the Korea sample, and 3) the
U.S. sample, respectively, to see any differences of regression outcomes by sample composition.
In addition, since data was collected in two organizations for each country sample, a dummy
organization variable was added as a control variable to see if there was an organization effect
within the country sample.

4.6.4 Chow Test
Finally, to examine whether the coefficients in two linear regressions on different data
sets were equal, the Chow test was conducted. The Chow test is a statistical and econometric test
invented by economist Gregory Chow to examine whether the coefficients in two linear
regressions on different data sets are equal, defined as "the standard F test for the equality of two
sets of coefficients in linear regression models " (Chow, 1960).

８３

More specifically, Chow (1960) suggested that homoskedasticity of errors is assumed,
although this can be doubted since we are open to the possibility that the parameter vector (b)
has changed.
RSSR = the sum of squared residuals from a linear regression in which b 1 and b 2 are
assumed to be the same
SSR 1 = the sum of squared residuals from a linear regression of sample 1
SSR 2 = the sum of squared residuals from a linear regression of sample 2
b has dimension k, and there are n observations in total
Then the F statistic is:
((RSSR- SSR1- SSR2)/k ) / ((SSR1+ SSR2)/(n-2k).
In this dissertation, the Chow test was conducted to examine if the coefficients in the
linear regressions models are the same in the Korean sample and the U.S. sample.

８４

CHAPTER 5
RESULTS

This chapter presents the results of the analyses conducted in this dissertation. It includes
(1) Descriptive Statistics, (2) Correlations among Variables, (3) Confirmatory Factor Model of
Harmony Culture, and (4) Regression Outcomes.

5.1 Descriptive Statistics
The means, standard deviations, and other descriptive statistics of all variables are shown
in Table 5. First, for the three culture values, the mean of the composite scale measuring the
degree of an individual‘s collectivism is 5.00 on a scale of one to seven, indicating that
respondents in the sample hold a relatively higher degree of collectivism. For the Korean sample,
the means are 4.19 and for the U.S., 4.53. The result of t-test indicates that this difference has
statistical significance. What is interesting about the finding is that respondents in the U.S. had a
higher degree of collectivism than in Korea.
This may be due to a firm selection effect: according to the logic of person-organization
fit literature, culture is an important factor in determining how well an individual fits an
organizational context (e.g. Kilmann, Saxton, and Serpa, 1986; Chein, 1985). The aspects both of
the individual (such as values and expectations) and situation (such as incentive system and
norms) combine to influence a focal individual‘s attitudinal and behavioral response to a given
situation (e.g. Chatman, 1989; Schneider, 1987; Terborg, 1981). Based on the above notion of
person-organization fit, it may be that the high collectivism score in the U.S. respondents is
observed because all U.S. respondents are those who are working in the Korean company; they

８５

were selected based on their degree of fit with the values the organization stands for, such as
high collectivism orientation. Or they may have high score of collectivism because they are
influenced by a socialization program provided by the organization designed to help newcomers
to achieve high levels of person-organization fit (Kristof, 1996). Hence it is likely that they
would score higher on a collectivism score than the average U.S employees. However, since the
average tenure of the U.S. respondents is much shorter (three years) than that of the Korean
respondents (14 years), in can be interpreted that the U.S. respondents may not have negative
experiences in the current organization, therefore, it may be questionable whether or not this high
score collectivism represents their embedded values.
Table 5. Descriptive Statistics for All Variables for the Analyses
N
Mean
Independent Variables
Total
510
4.25
*
Korea
415
4.19
Collectivism
U.S.
95
4.53
Total
510
5.19
Face-Saving
Korea
415
5.22
U.S.
95
5.08
Total
510
4.12
Conflict-Avoiding
Korea
415
4.09
U.S.
95
4.23
Total
510
5.28
Self-Esteem
Korea
415
5.28
U.S.
95
5.28
Total
510
3.27
*
Korea
415
2.89
Alternative Quality
U.S.
95
4.9
Total
509
4.51
LMX
Korea
414
4.45
U.S.
95
4.76
Total
510
5.28
*
Korea
415
5.28
Power Distance
U.S.
95
3.13
*

S.D

Min.

Max.

1.29
1.25
1.42
1.14
1.12
1.24
1.34
1.31
1.46
1.07
0.99
1.36
1.74
1.59
1.40
1.16
1.11
1.35
1.07
0.99
1.29

1
1
1
1
1
2
1
1
1
1
1
1
1
1
1
1
1
2
1
1
1

7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7

t-test result with significant difference between Korea and U.S. at the 0.05 level (2-tailed).

８６

Table 5. (cont‘d)
N

Mean

S.D

Min.

Max.

510
415
95
510
415
95
510
415
95
510
415
95
509
414
95

2.88
2.78
3.29
4.55
4.59
4.36
4.38
4.51
3.8
3.26
3.44
2.5

1.20
1.15
1.32
1.28
1.31
1.12
1.12
1.00
1.41
1.34
1.25
1.44

1
1
1
1
1
1
1
1
1
1
1
1

7
7
7
7
7
7
7
7
7
7
7
7

Dependent Variables
Exit

*

Voice
*

Loyalty

Neglect*

Valid N (listwise)
*

Total
Korea
U.S.
Total
Korea
U.S.
Total
Korea
U.S.
Total
Korea
U.S.
Total
Korea
U.S.

t-test result with significant difference between Korea and U.S. at the 0.05 level (2-tailed).
The mean of the composite scale measuring the degree of an individual‘s face-saving is

5.19 on a scale of one to seven. It indicates that employees in the sample are relatively highly
concerned with their face at workplace. The means are 5.22 and 5.08 for the Korean and the U.S.
samples respectively. The mean of the composite scale measuring the degree of an individual‘s
conflict-avoiding is 4.12 on a scale of one to seven, indicating that individual employees in the
sample have a relatively higher tendency of avoiding conflict at work. For the Korean sample,
the mean is 4.09 and for the U.S. it is 4.23.
Second, regarding other independent variables, the mean of the composite scale
measuring the degree of an individual‘s self-esteem is 5.28 on a scale of one to seven, indicating
that respondents in the sample have a relatively high degree of self-esteem. The mean for both
the U.S. and the Korean sample is 5.28. The mean of the composite scale measuring the degree
of an individual‘s alternative quality is 3.27 on a scale of one to seven. This indicates that
８７

employees in the sample think that they have relatively low-quality of alternatives to their
current job. The means are, for the Korean sample, 2.89, and for the U.S., 4.9, and the result of ttest indicates that this difference has statistical significance. The mean of the composite scale
measuring the degree of an individual‘s LMX is 4.61 on a scale of one to seven, indicating that
respondents in the sample have a moderate-quality relationship with their supervisors. The
means are: 4.45 for the Korean sample, 4.76 for the U.S. The mean of the composite scale
measuring the degree of an individual‘s power distance is 3.22 on a scale of one to seven,
indicating that respondents in the sample hold a relatively lower degree of power distance. The
mean is 3.54 in the Korean sample and 3.13 in the U.S. sample.
For the dependent variables, the mean of the composite scale measuring the degree of an
individual‘s likelihood of exit is 2.88 on a scale of one to seven, indicating that respondents in
the sample hold a relatively lower tendency of exit. The means are 2.78 and 3.29 for the Korean
and the U.S. samples, respectively, and the result of t-test indicates that this difference has
statistical significance. The mean of the composite scale measuring the degree of an individual‘s
voice is 4.55 on a scale of one to seven. This indicates that employees in the sample have a
moderate degree of voice option. For the Korean sample, the mean is 4.59 and for the U.S.
sample, 4.36. The mean of the composite scale measuring the degree of an individual‘s
likelihood for loyal behavior is 4.38 on a scale of one to seven, indicating that individual
employees in the sample have a moderate likelihood of loyalty. For the Korean sample, the mean
is 4.51 and for the U.S. sample it is 3.8., and the result of t-test indicates that this difference has
statistical significance. The mean of the composite scale measuring the degree of an individual‘s
tendency of neglect is 3.26 on a scale of one to seven, indicating that respondents in the sample
have a relatively lower likelihood of the neglect option. For the Korean sample, the mean is 3.44

８８

and for the U.S., 2.5, and the result of t-test indicates that this difference has statistical
significance.
In terms of the spread of a set of observations, Table 5 shows that all of the independent
and dependent variables have a standard deviation value of around 1. These results indicate that
the data points tend to be very close to the mean, showing a tightly concentrated distribution,
which is possibly driven by the characteristics of respondents in Korea. As shown in subject
demographic statistics, 81% of responses were collected in Korea, 95% of respondents are male,
68% are high school graduates, and 77% are employed in skilled crafts or laborer positions. Due
to these concentrated distributions of respondents, a question could be raised as to whether the
findings from this sample can represent the population of employees in each country, or if these
findings can be generalized to other samples. In the section discussing the limitations of this
study, more details will be provided, including suggestions for future research.

6.2 Correlations among Variables
Table 6 provides the effect size and significance levels for all of the zero-order
correlations between all measured variables. The results indicate that the exit option is
significantly correlated with alternative quality, self-esteem, education, tenure, job category of
R&D and Technical/Professional, job category of Skilled Crafts and Laborer, and Korea. The
alternative quality, education, and job category of R&D and Technical/Professional variables are
positively associated with the exit option; while self-esteem, tenure, job category of R&D and
Technical/Professional, job category of Skilled Crafts and Laborer, and Korea are negatively
associated with exit. Some of the correlations are consistent with several of the model‘s
hypotheses: alternative quality was predicted to have a positive association with the exit option.

８９

Table 6. Correlation Matrix
1
1 EXIT
2 VOICE

1.00
0.01
**

3 LOY

-.24

4 NEG

.12
-0.02

5 COLR
6 FAC

**

-0.01

7 CON

0.05

8 ALT

.42

2

4

5

6

7

8

9

1.00
.13

**

-.12
.18

**

**

.32

*

**

**

-.22
-0.02

**

9 SEL

3

**

.09

*

**

1.00
*

.15
0.08

-.101
-0.08

**

**

.22

**

.22
-0.02

**

**

-.22

¹
12 AGE²
13 EDU³
11 LMX

1.00
-0.03

1.00

0.01

0.01

0.05

1.00

**

**

-0.08

0.01

1.00

**

-0.05

-0.03

**

0.08

.35

4
14 TEN
15 Manag
16 Tech
17 Labor
18 Korea

-0.02

.30

**

0.05

-.14

-0.05

-0.09

0.01

-0.08

.12

**

0.07

0.01

-0.07

0.01

*

0.03

-0.04

0.06

0.00

0.07

0.03

.25

**

0.04

**

0.00

.17

**

0.01

0.09

0.01

-.36

**

0.03

**

0.00

**

0.00

**

-0.01

**

0.00

.11
-.15

0.05
.15

**

-.11
-.16

*

**

.24

**

**

-0.01

0.00

-0.07

*

-.10

-.16

**

.28
0.03

-.22

**

.14

1.00

-.16
-0.01

10 PD

.23

1.00

.27

.17

.17

**
**

**

.12
-0.02

.11
.28

0.01

0.04

0.03

-0.08

-0.03

0.07

**

**

*

.25

.27

-.10

. Correlation is significant at the 0.01 level (2-tailed).

*

. Correlation is significant at the 0.05 level (2-tailed).
4
N=510; N=509, N=475; N=503; N=474

²

**

0.00

**

¹

*

³

９０

.44
-0.05
.30

**

0.03
0.02
-0.04
0.05

.48

-.17

0.02
0.04
-0.03
-0.04

.16
.25

-.26
-.45

**

Table 6. Correlation Matrix (cont‘d)
10
11
1 EXIT
2 VOICE
3 LOY
4 NEG
5 COLR
6 FAC
7 CON
8 ALT
9 SEL
10 PD
1

¹
12 AGE²
13 EDU³

.15

4
14 TEN
15 Manag

.23

11 LMX

16 Tech
17 Labor
18 Korea

-.04
**

.08
**
**

.15
.01
-.12
.14

**

**

12

13

14

15

16

17

18

1
.05
.12

**

-.01
.07
.12

**

-.15

**

-.10

*

1
-.10
.54

*

**

-.04
-.01
.050
.013

1
-.26
.63
.31

**

1

**
**

-.73
-.13

-.09
-.22

**
**

.21
.62

*

**

**
**

1
-.11

**
**

-.81
.05

1
-.44
-.31

**
**

1
.14

**

1

**

. Correlation is significant at the 0.01 level (2-tailed).

*

. Correlation is significant at the 0.05 level (2-tailed).
4
N=510; N=509, N=475; N=503; N=474

¹

²

³

However, none of the harmony culture variables, such as collectivism, face-saving and
conflict-avoiding, are found to be significantly correlated with exit. The Korean variable is found
to have a negative correlation with the exit option. In terms of demographic variables, the exit
option is found to be significantly correlated with education, tenure, and the R&D and
Technical/Professional, and Skilled Crafts and Laborer job categories. Education is found to
have a positive association with the exit option, while tenure is found to have a negative
association with the exit option.

９１

Among job categories, R&D and Technical/Professional is found to have a positive
association with the exit option, while Skilled Crafts and Laborer is found to have a negative
association with the exit option.
However, none of the harmony culture variables, such as collectivism, face-saving and
conflict-avoiding, are found to be significantly correlated with exit. The Korean variable is found
to have a negative correlation with the exit option. In terms of demographic variables, the exit
option is found to be significantly correlated with education, tenure, and the R&D and
Technical/Professional, and Skilled Crafts and Laborer job categories. Education is found to
have a positive association with the exit option, while tenure is found to have a negative
association with the exit option. Among job categories, R&D and Technical/Professional is
found to have a positive association with the exit option, while Skilled Crafts and Laborer is
found to have a negative association with the exit option.
The voice option is significantly correlated with collectivism, face-saving, conflictavoiding, self-esteem, power distance, and LMX. Conflict-avoiding and power distance are
negatively associated with the voice option, while collectivism, face-saving, self-esteem, and
LMX are positively associated with the voice option. The directions of correlation are consistent
with the hypotheses of the model: conflict-avoiding and power distance were predicted to be
negatively associated with voice option, while collectivism, face-saving, self-esteem, and LMX
were predicted to be positively associated with the voice option. However, contrary to the
hypotheses of the model, alternative quality does not show a significant association with the
voice option. No demographic variable (sex, age, education, tenure, or job) is found to be
significantly correlated with the voice option.

９２

The loyalty option is found to be significantly correlated with collectivism, conflictavoiding,

alternative

quality,

self-esteem,

power

distance,

tenure,

R&D

and

Technical/Professional job category, and Korea. Collectivism, alternative quality, and R&D and
Technical/Professional job category are negatively associated with the loyalty option, while
conflict-avoiding, self-esteem, power distance, tenure, and Korea are positively associated with
the loyalty option. Some of the directions of correlation are consistent with the hypotheses of the
model: alternative quality was predicted to be negatively associated with the loyalty option,
while conflict-avoiding was predicted to be positively associated with the loyalty option.
However, some of the directions of correlation are not consistent with the hypotheses of the
model; contrary to the findings from this correlations matrix, collectivism was predicted to be
positively associated with the loyalty option. Face-saving does not show a significant association
with the loyalty option. The Korea variable is found to have a positive correlation with the
loyalty option. In terms of demographic variables, the loyalty option is significantly correlated
with tenure, and with the R&D and Technical/Professional job category. Tenure is found to have
a positive association with the loyalty option, and the R&D and Technical/Professional job
category is found to be negatively related to the loyalty option.
The neglect option is found to have significant correlations with collectivism, conflictavoiding,

self-esteem,

power

distance,

LMX,

tenure,

the

Executive/Management,

Sales/Marketing, and Administrative/Clerical job category, and Korea. Conflict-avoiding, power
distance, sex, tenure, the Executive/Management, Sales/Marketing, and Administrative/Clerical
job category, and Korean are positively associated with the neglect option, while collectivism,
self-esteem, and LMX are negatively associated with the neglect option. Some directions of
correlation are consistent with the hypotheses of the model: collectivism was predicted to be

９３

negatively associated with the neglect option, while conflict-avoiding was predicted to be
positively associated with the neglect option. However, contrary to the hypotheses in the model,
face-saving, and alternative quality do not show a significant association with the neglect option.
In terms of demographic variables, neglect option is significantly correlated with tenure, the
Executive/Management, Sales/Marketing, and Administrative/Clerical job category, and Korea.
These three variables are found to have a positive association with the neglect option.

5.3 Results of Confirmatory Factor Analysis
In this dissertation, CFA was performed to confirm the theoretical distinctiveness of
seven independent variables: collectivism, face-saving, Conflict-Avoiding, alternative quality,
self-esteem, power distance, and LMX. The analysis started with the model in which thirty items
(from V1 to V30) are loaded into seven independent variables: collectivism, face-saving,
Conflict-Avoiding, alternative quality, self-esteem, power distance, and LMX.
Figure 2 shows the results from the analysis. The numbers to the right of the variables
(from COL1 to LMX5) are errors, while the numbers to the left of the variables are t-values of
each path coefficient. As the figure shows, all of the t-values of the coefficients are significant.
Two fit indices in this model, RMSEA (0.06) and TLI (0.91), confirm that this model is
acceptable. The values of other fit indices in this model, such as CFI (0.93) and IFI (0.93), also
confirm that this model is good fit. Overall, the result of CFA in this dissertation establishes the
discriminant validity of seven independent variables: collectivism, face-saving, ConflictAvoiding, alternative quality, self-esteem, power distance, and LMX.

９４

Figure 2. Outcome of Confirmatory Factor Analysis
Collectivism

1.00

COL1
COL2

ⓔ 0.73

COL3

1.42

1.08

ⓔ 1.29

FAC1
1.00
Face-Saving

ⓔ 1.28

.81
1.90

FAC2

ⓔ 1.29
ⓔ 1.69

Conflict-Avoiding
1.04

ⓔ .67

CON1
1.00

ⓔ .34

FAC4

1.29

FAC3

ⓔ 1.07

1.07 CON2
1.16
1.29
CON3
1.03
CON4
CON5

Alternative Quality

1.00

ALT1

3.86

ALT2
SEL1

1.00
Self-Esteem
.07

SEL2
1.02
.73
.54

SEL3
SEL4
SEL5
PD1

1.00
Power Distance

.74
1.11
1.12
1.19

1.48

PD2
PD3
PD4
PD5
PD6
LMX1

LMX

1.00
9
1.25

.99
1.02

LMX2
LMX3

1.22 LMX4
LMX5

ⓔ .81
1.29
ⓔ .54
1.29

ⓔ .96
1.29

ⓔ 1.29
1.29

ⓔ 1.29
1.29

ⓔ 2.66
1.29
ⓔ .36
1.29
1.29
ⓔ .16
1.29
ⓔ .84
1.29

ⓔ 1.13
1.29

ⓔ 1.32
1.29

ⓔ 1.53
1.29

ⓔ 1.69
1.29
1.29
ⓔ 1.16
1.29
1.29

1.29
ⓔ 1.36
1.29

ⓔ 1.29
1.29

ⓔ .77
1.29
1.29
ⓔ 1.45
1.29

1.29
ⓔ 1.27
1.29
1.29
1.29
1.29
ⓔ .98
1.29
1.29
ⓔ .38
1.29
1.29
ⓔ .34
1.29

1.29

９５

1.29
1.29
1.29

5.4 Regression Outcomes
To test the hypotheses, multiple regressions were conducted. The dependent variables in
the analysis are the exit, voice, loyalty, and neglect options. Each dependent variable was
regressed upon the following independent variables: collectivism, face-saving, conflict-avoiding,
alternative quality, self-esteem, power distance, and LMX. In addition, some demographic
variables that showed statistically significant relationships with dependent variables (such as
country, education, tenure, and job category) in the zero-order correlation matrix were added in
this regression analysis.
More specifically, multiple regressions were conducted on three different levels. First,
with the total sample, a series of stepwise regressions were conducted to assess the contribution
of control variables (Step 1), some independent variables other than the three cultural variables
(Step 2), and the three cultural variables (Step 3), respectively. Demographic variables showing
statistically significant relationships with dependent variables in the zero-order correlation matrix
(e.g. nationality, education, tenure, and job category) were entered first (Model 1), followed by
other independent variables (alternative quality, self-esteem, power distance, and LMX) (Model
2), then the three main harmony culture variables, collectivism, face-saving, and conflict-

²

avoiding(Model 3). Next, the change in R was assessed to test if these variables improve the
predictions of this model significantly.
Second, regressions were conducted with the Korean sample and the U.S. sample
separately, to see if there were any different findings of regression outcomes between the Korean
sample and the U.S. sample. In addition, since data was collected in two organizations for each
country sample, a dummy organization variable was added as a control variable to see if there
was an organization effect within the country sample. Additionally, to see if there was a
９６

statistically significant difference of regression outcome between the Korea sample and the U.S.
sample.

5.4.1 Regression Analysis with the Total Sample

5.4.1.1 Regression on the Exit Option
Table 7 presents the multiple regression analysis summary predicting the exit option with

²

the total sample. According to the outcome analysis of this table, Model 3 has the highest R ,
which means that Model 3 is the best combination of predictors among the three models for the

²

exit option. The R of Model 3 is 0.21, which means that the linear regression explains 21% of
the variance in the data, even after taking into account the number of predictor variables in this
model. However, the R

² of Model 3 brought the increase of the R² up to 0.01, which shows that

the three cultural variables did not explain additional statistically significant variances in the exit
option.
This table also provides the parameters of the model and their significance levels.
Although all parameters for each model are interesting up to a point, the most interesting is
Model 3, since it was found to be the best combination of predictors out of the three models.
Therefore, we will look only at the outcome of Model 3.
First, Table 7 shows that the country variable does not have a significant association with
the exit option, which means that the nationality of the respondent, especially whether or not the
employee is Korean, does not make any significant difference to their propensity for the exit

９７

option. Based on this finding, we can expect that there will not be a significant difference of
regression outcome between the Korean sample and the U.S. sample.

Table 7. Regression Outcome on Exit with the Total Sample

Model 1
β

Model 2
β

T
**

(Constant)

-3.53

2.90

Korea
Education
Tenure
Managerial
Professional
Laborer
Alternative Quality

-0.08
-0.01
-0.07
-0.03
0.06
-0.11

-1.34
-0.07
-1.16
-0.07
0.27
-0.26

²
**

**

0.02
-0.01
0.01
-0.23
-0.08
-0.26
0.43

0.05
3.82

3.10

0.41
-0.17
0.15
-0.65
-0.38
-0.66
9.04

-0.16
0.00
-0.01

R
∆F
p < 0.001,

β

t

-3.59

Self-Esteem
Power Distance
LMX
Collectivism
Face-Saving
Conflict-Avoiding

***

Model 3

-3.66
0.09
-0.17

***

0.21

***

***

23.52

***

t
**

-3.41

2.91

0.03
-0.02
0.00
-0.26
-0.10
-0.30
0.43

0.55
-0.29
-0.04
-0.75
-0.47
-0.78
8.97

-0.18
-0.01
-0.02
0.01
0.07
0.03

-3.93
-0.17
-0.32
0.19
1.48
0.67

***
***

0.21
1.00

*

p < 0.01, p < 0.05, N=471

Second, and more importantly, the regression outcome indicates that none of the three
culture variables, collectivism, face-saving, and conflict-avoiding, have a significant relationship
with the exit option. As a result, we can conclude that the exit option is not affected by these
three culture values. Accordingly, we can expect that parts of this study‘s hypotheses about the
possible significant relationship between these three culture variables and the exit option will be
rejected. In addition, considering that the country variable does not make any significant

９８

difference in the exit option, we can expect that these associations will not be significant
regardless of the country of the respondents.
Last, according to the regression outcome, only alternative quality (t = 8.97, p<.000) and
self-esteem (t = -3.93, p<.000) have a significant relationship with the exit option in Model 3.
More specifically, alternative quality was found to have a positive relationship with exit, which
means that the more people tend to think that they have good alternatives to their current job, the
more likely they are to choose the exit option. Therefore, this finding supports the part of the
hypothesis that says alternative quality will have a positive association with exit option, since
high-quality alternatives to the employee‘s current job will serve as a source of power for change.
In particular, according to the standardized beta values of these two variables, alternative quality
(β = .42) has a stronger impact to predict exit than self-esteem (β = -.18). In conclusion,
alternative quality is the strongest predictor of the exit option.
In contrast, self-esteem was found to have a negative relationship with exit. This means
that when more people perceive themselves to be valued in their organization, their propensity
toward the exit option will decrease. This supports the part of the hypothesis that says alternative
quality will have a negative association with the exit option. Therefore this finding supports that
low self-esteem, which means that their negative belief that they are not significant, successful,
and worthy in the organization, will discourage them from being more active in the organization,
since they think that they are powerless and they cannot make any change. This negative belief
will encourage them leave the organization to find an alternative making them capable,
significant, successful, and worthy.
In sum, the key finding of this regression outcome is that culture does not affect the exit
option; neither country variable, nor any of the three cultural values have any statistically

９９

significant association with the exit option. Therefore, we can conclude that cultural values
themselves do not directly affect employees‘ propensity to choose the exit option in a
deteriorating situation at work.
Why did these cultural values fail to gain statistical support as predictors of the exit
option? Why did these cultural values not affect the choice of exit option? This is possibly due to
1) the nature of the exit option itself, and 2) some mediators which have not been taken into
consideration in this dissertation, which could affect this association. First, literature on turnover
research gives us important implications for answering these questions: according to turnover
literature, turnover itself is a very complex phenomenon (Peterson, 2004), especially considering
the possible negative consequences for the quitter. Turnover researchers suggested that turnover
can lead to negative consequences for the leaver as well as the organization, as well as positive
consequences. For instance, Mobley (1982) summarized that the negative consequences for the
leaver include 1) loss of seniority and fringe benefits, 2) transition stress in a new job, 3)
relocation costs, 4) termination of personal and family social networks, 5) loss of valued
community services, and 6) disruption of spouse‘s career. In this line, Mobley (1982) suggested
that these warnings can help prospective leavers to make wise decisions about turnover as well
as deterring their exits. Therefore, although an employee may confront a deteriorating situation
at work, a decision to leave the organization will be considered carefully with consideration of
all possible advantages and disadvantages.
Of course, one may argue that other behavioral options— voice, loyalty, and neglect—
can bring some disadvantages to the person who chooses one of these options. For instance, the
neglect behavior may lead to a negative evaluation from the person‘s managers or coworkers.
However, the bottom line is that there is a huge difference in the magnitude of these possible

１００

negative consequences when compared with leaving the organization. For example, people do
not need to worry of the possible loss of seniority and fringe benefits as long as they stay with
the organization. Therefore, we can conclude that the exit option is totally different from the
other three options, all of which entail continuity of the employment relationship in the current
organization. This means that when people choose the exit option, they need to consider the
possible negative consequences for themselves. However, these consequences are not the
subjects of considerations for the other behavioral options—voice, loyalty, and neglect— since
the other three options assume that the individual will stay in the organization, rather than leave.
Therefore, the exit option is distinct from the other three options in terms of the consequences
the option.
Second, previous research has demonstrated that cultural variables are not significant
predictors of the exit option; the latest meta-analysis showed that there are 45 predictors that
have been examined (Griffeth, Hom, & Gaertner, 2000). They revealed that, in turnover or
withdrawal processes, proximal precursors were among the best predictors. The proximal
predictors include job satisfaction, organizational commitment, job search, comparison of
alternatives, withdrawal cognitions, and quit intentions. They also found that more distal
predictors, such as characteristics of the work environment, i.e., job content, stress, work group
cohesion, autonomy, leadership, and—to a lesser extent—distributive justice and promotional
chances, have small effect on the termination process. By the same logic, cultural values
themselves are possibly too distal to directly affect employees‘ exit-related behavior. This
analysis supports the finding that more realistic and economic considerations, such as job
alternative quality, strongly affect employees‘ propensity for the exit option.

１０１

5.4.1.2 Regression on the Voice Option
The next table summarizes the multiple linear regression outcomes and overall fit
statistics predicting the voice option with the total sample. In Table 8, Model 3 had the highest

²

²

R . In Model 3, the R is 0.24, which means that the regression explains 24% of the variance in
the data, even after taking into account the number of predictor variables in this model. Therefore,
we can conclude that Model 3 is the best combination of predictors out of the three models. In
other words, the three cultural variables explain the additional variance in the voice option with
statistical significance.
This table also provides the parameters for the model and their significance levels.
Similar to the regression outcome predicting exit option, we will look only at the outcome of
Model 3, since it was found to be the best combination of predictors among the three models.
Table 8 shows that the country variable (t = 2.62, p<0.05), has a significant and positive
association with the voice option. This means that the country of origin of the respondent makes
a statistically significant difference in their choice of the voice option; employees in Korea are
likely to choose voice option. In addition, based on this finding, we can expect the outcome of
regression on the voice option to be different between the Korea and the U.S. samples.
Next, in reference to the culture variables: collectivism, face-saving, and conflictavoiding, all three were found to have a significant relationship with the voice option.
Collectivism (t = 2.70, p<.01) has a positive relationship with the voice option; in other words,
the more people value collectivism, the more they are likely to choose voice option. It supports
the part of this study‘s hypothesis that says collectivism will have a positive association on the
voice option, since the voice option is interpreted as a constructive response, aiming at
improving the situation by alerting management to problems.
１０２

Face-saving (t = 4.34, p<.000) also has a positive relationship on the voice option; in
other words, the more people are concerned with face-saving, the more they are likely to choose
voice option to change or fix a situation, by, for example, going to an immediate supervisor to
discuss the problem, asking co-workers for advice about what to do, talking to another member
of management about their feeling about the situation, or trying to solve the problem by
suggesting changes in the way work is supervised in the office. A more interesting finding is that
face-saving orientation was found to be the strongest predictor of the voice option, among all
other predictors examined in this study. Based on this finding, it can be expected that the part of
this study‘s hypothesis that says face-saving will have a positive association with voice will be
supported.
In addition, conflict-avoiding (t = -2.45, p<.05) was found to have a negative relationship
with the voice option. This means that the higher the degree of conflict-avoiding, the less likely
people are to choose voice option. In other words, people with high conflict-avoiding believe that
it is better to keep negative opinion to themselves and try to ignore conflicts, staying away from
disagreement. Based on this finding, we can expect that the part of this study‘s hypothesis that
says conflict-avoiding will have a negative association with voice will be supported.
Last, as far as predictors other than the three culture variables, the regression outcome
indicates that power distance (t = -2.99, p<.01) and LMX (t = 4.22, p<.000) have significant
association with the voice option. Power distance was found to have a negative relationship with
the voice option: the higher the degree of power distance, the less likely people are to choose
voice option. It supports the part of this study‘s hypotheses that says power distance will have a
negative association with the voice option, since people who have high power distance
orientation believe that employees who have lower power should obey the authority of the

１０３

organization. Therefore, they will be more likely to refuse a confrontational action toward
solving deteriorating situation in the organization they belong to, such as expressing
disagreement or different opinion against the authority
.
Table 8. Regression Outcome on Voice with the total sample
Model 1
Model 2
β

Model 3

β

t

β

t

-2.26

1.80

-2.29

1.86

2.26
1.94
-0.61
-0.19
-0.44
0.13
0.15

*

0.15
0.10
-0.07
-0.21
-0.16
-0.11
0.01
0.06

2.62
1.60
-1.32
-0.60
-0.77
-0.30
0.18
1.33

***

-0.15

-2.99

***

0.20

4.22

Collectivism

0.12

2.70

Face-Saving

0.20

4.34

Conflict-Avoiding

-0.12

-2.45

t
**

(Constant)

-3.61

Korea

0.05

0.74

0.13

Education
Tenure
Managerial
Professional
Laborer
Alternative Quality
Self-Esteem

0.14
-0.03
0.00
-0.04
0.16

2.03
-0.47
0.01
-0.16
0.36

*

0.13
-0.04
-0.07
-0.10
0.05
0.01
0.14

3.07

Power Distance

-0.21

-4.63

LMX

0.27

5.88

²

0.02

R
∆F
***

p < 0.001,

**

2.75

0.17

1.28

21.88

***

**

**

**

***
**

***
*

0.24
12.14

***

*

p < 0.01, p < 0.05, N=471

In contrast, LMX has a positive relationship with voice: people who perceive that they
have a good relationship with their supervisor are more likely to choose the voice option.
Therefore, we can conclude that the finding of this study supports the part of this study‘s
hypothesis that says high LMX promotes employees to choose voice option, since a good
relationship with a supervisor provides employees more opportunities to speak up, exchange

１０４

information or ideas with their supervisors, and use more communication channels compared to
those in low-LMX relationships.
Overall, contrary to the result of regression on the exit option, the regression outcomes on
the voice option revealed distinctive features of the predictors of the voice option. The main
finding is that the voice option has strong culture effects in many ways: 1) country itself has a
strong association with the voice option; 2) all three culture values are strong predictors of
employees‘ propensity toward voice option in a deteriorating situation at work; 3) culture value
is the strongest predictor of the voice options: face-saving was found as the strongest predictor of
the voice option.

5.4.1.3 Regression on the Loyalty Option

² ²

According to Table 9, Model 3 has the highest R (R = 0.20), which means that Model 3
is the best combination of predictors for the loyalty option among the three models. In other
words, the final model significantly improves the ability to predict the loyalty option by adding
three cultural variables –– collectivism, face-saving, and conflict-avoiding.
This table also provides the parameters of the model and their significance levels. We
will look only at the outcome of Model 3, since it was found to be the best combination of
predictors among the three models.
First of all, table 9 shows that country variability (t = 4.14, p<0.000) has a significant
association with the loyalty option; which means that, consistent with the finding in the voice
option, employees‘ propensity for the loyalty option is different depending on the country,
especially if the employees in this study are from Korea. Employees in Korea are found to be
more likely to choose the loyalty option. In particular, the Korea variable is found to be the
１０５

strongest predictor of the loyalty option. Based on this finding, we can expect the outcome of
regression on the loyalty option to show a difference between the Korea sample and the U.S.
sample.

Table 9. Regression Outcome on Loyalty with the total sample
Model 1
Model 2
β

Model 3

β

t

**

-2.24

2.07

***

0.21
0.04
-0.09
0.03
-0.01
0.10

3.55
0.62
-1.55
0.09
-0.04
0.26

Alternative Quality

-0.15

-3.12

Self-Esteem

0.13

2.78

Power Distance
LMX

0.25
0.04

5.59
0.96

(Constant)

-3.42

3.14

Korea
Education
Tenure
Managerial
Professional
Laborer

0.24
0.06
0.02
0.01
-0.04
0.07

3.99
0.91
0.26
0.03
-0.19
0.16

²

0.07

R
∆F
***

p < 0.001,

**

6.16

12.43

***

*

-1.53

1.41

0.25
0.04
-0.09
0.07
0.02
0.13

4.14
0.63
-1.57
0.19
0.07
0.33

**

-0.15

-3.15

**

0.14

2.90

***

0.16

***

t

***

Collectivism
Face-Saving
Conflict-Avoiding

β

0.15
0.05

3.02
1.02

0.10
-0.03
0.17

t

2.26
-0.64

***

**

**
**

*

3.42

**

0.19
5.59**

*

p < 0.01, p < 0.05, N=471

Next, concerning the three culture variables, collectivism (t = 2.26, p<.05) and conflictavoiding (t = 3.42, p<.01) are found to have a significant relationship with the loyalty option. In
more detail, collectivism was found to have a positive relationship with loyalty, which means
that, as the degree of collectivism increases, the incidence of the loyalty option increases. In
other words, this finding supports the hypothesis that employees who highly value collectivism

１０６

will be more likely to choose behavioral options that are considered not to hurt or dissolve their
relationship with the organization. Consequently, employees in the collectivism culture are likely
to be encouraged to choose the loyalty option since it is interpreted as a constructive behavior
within the organization, aimed at maintaining or reviving the relationship with the organization.
Therefore, these employees will continue to fulfill their obligations, rather than hurting or
threatening the relationship with the organization. Thus, we can expect that the part of this
study‘s hypothesis that says collectivism will have a positive association with the loyalty option
will be supported.
However, conflict-avoiding was found to have positive β-values indicating a positive
relationship with loyalty; this can be interpreted that as the degree of conflict-avoiding increases,
the propensity for the loyalty option increases. This finding supports that someone who highly
values conflict-avoiding will pursue integrating behaviors in handling the problematic situation
at work, refusing a confrontational behavior toward that situation. Therefore, they will be more
likely to choose the loyalty option since the loyalty option is regarded as passive response
without explicit expression, in other words, waiting for the situation to get better without their
involvement. Therefore, we can expect that the part of this study‘s hypothesis that says conflictavoiding will have a positive association with the loyalty option, will be supported.
Last, as far as predictors other than the three culture variables, the regression outcome
indicates that alternative quality (t = -3.15, p<.01), self-esteem (t = 2.90, p<.01), and power
distance (t = 3.02, p<.05) have a significant association with the loyalty option. In detail,
alternative quality (β = -0.15) has a negative relationship with the loyalty option, which means
that people who believe that their next job would be better than their current job, or that they
would find a satisfactory job if they had quit the current job, are less likely to choose the loyalty

１０７

option. Therefore, this finding supports the part of the hypothesis of this study that says that
employees having high-quality alternatives to their current job will be encouraged to do
something about work problems; furthermore, high-quality alternative will serve as a source of
power for a change. In contrast, the absence of good alternatives will make the employee wait
quietly for conditions to improve, in other words, remain loyal.
In addition, self-esteem (β = 0.14) was found to have a positive relationship with the
loyalty option. In other words, the more people perceive themselves valued in their organization,
the more they are likely to choose the loyalty option. Therefore, we can conclude that the
findings of this study support the part of this study‘s hypothesis that says that self-esteem will
have a negative association with the loyalty option, since high-level of self-esteem in the
organization that they belong to will encourage them to support the organization, remaining loyal.
Power distance (β = 0.15) was also found to have a positive relationship with the loyalty
option. In other words, the higher the degree of power distance, the more people are likely to
choose the loyalty option. In other words, it supports that employees with higher PD orientations
believe that employees who have lower power should obey the authority of the organization, and
therefore supervisors should have a large amount of power over them. For that reason,
employees in high PD orientation will be afraid to disagree with their superiors; they will be
more likely to refuse a confrontational action aimed at solving a deteriorating situation in the
organization they belong to. This means that even if they disagree with their supervisor, they will
remain loyal, obeying their supervisor‘s opinion and continuing to fulfill their work obligations.
Therefore, we can conclude that the findings of this study support the part of this study‘s
hypothesis that holds that power distance will have a positive association with the loyalty option.

１０８

Overall, consistent with the result of regression on the voice option, the loyalty option is
influenced by culture in many ways. First, the country effect works to predict the employees‘
propensity for the choice of the loyalty option. Furthermore, country is the strongest predictor of
the loyalty option. Second, some of the values in a harmony culture, such as collectivism and
conflict-avoiding, have a significant association with the loyalty option. In addition, conflictavoiding was found to be the strongest predictor of the loyalty option. Along with these culture
values, other variables such as alternative quality, self-esteem, and power distance also affect
employees‘ choice of the loyalty option.

5.4.1.4 Regression on the Neglect Option
Finally, the next table shows the multiple regression analysis summary predicting the

²

neglect option for the total sample. In Table 10, we find that Model 3 has the highest R , which

means that Model 3 is the best combination of predictors out of the three models. The R

² of

Model 3 is 0.19, which means that the linear regression explains 19% of the variance in the data,
even after taking into account the number of predictor variables in this model. In addition, the R

²

²

of Model 3 brought the increase of the R to 0.03, which shows that three cultural variables
explain statistically significant additional variance in the neglect option. It supports that there are
significant relationships between the three cultural variables and the neglect option.
This table also designates the parameters of the model and their significance levels. As in
the regression outcome predicting exit option, we will look only at the outcome of Model 3,
since it was found to be the best combination of predictors among the three models.
１０９

First of all, table 10 shows that country variable (t = 4.83, p<.000), has a significant
association with the neglect option. This means that employees‘ propensity for the neglect option
will be different depending on the country, particularly whether or not the employee is in Korea:
employees in Korea are more likely to choose the neglect option. In particular, the standardized
beta values (β), of these predictors indicated that among these significant predictors, being
Korean was the strongest predictor for the neglect option. Based on this finding, we can expect
the outcome of regression on the neglect option will be different between the Korea sample and
the U.S. sample.

Table 10. Regression Outcome on Neglect with the total sample
Model 1
Model 2

Model 3

β

t

β

(Constant)

-3.28

2.51

-3.46

2.64

Korea

0.28

0.28

Education
Tenure
Managerial
Professional
Laborer

-0.02
0.03
-0.15
-0.07
-0.31

4.80
-0.27
0.55
-0.39
-0.30
-0.76

-0.02
0.02
-0.22
-0.11
-0.34

4.69
-0.31
0.36
-0.61
-0.51
-0.86

Alternative Quality

0.11

2.30

Self-Esteem

-0.13

-2.84

Power Distance

0.20

LMX
Collectivism
Face-Saving

-0.05

4.55
-1.09

***

²
**

**

-2.94

2.24

0.29
-0.02
0.03
-0.19
-0.10
-0.31

4.83
-0.27
0.47
-0.54
-0.47
-0.80

*

0.11

2.18

**

-0.12

-2.51

***

0.13

2.46
-0.19
-1.44
-0.04

0.09
8.06

0.17

***

9.85

*

p < 0.01, p < 0.05, N=471

１１０

***

*

***

*

*

-0.01
-0.07
0.00
0.17

R
∆F
p < 0.001,

t

***

Conflict-Avoiding

***

β

t

3.56
0.19
4.92

**

***

Next, concerning the three culture variables, only conflict-avoiding (t = 3.56, p<.000)
was found to have a statistically significant relationship with the neglect option. In more detail,
table 5.6 indicates that conflict-avoiding has a positive relationship with the neglect option, with
the standardized beta value of 0.17. As hypothesized, people who place a high value on conflictavoiding do not prefer to exhibit confrontational behavior despite problematic situations at work.
They are more likely to choose the neglect option, since neglect is characterized by using passive
response with lax and careless behavior, by sidestepping the situation. Therefore, we can expect
that the part of this study‘s hypothesis that says conflict-avoiding will have a positive association
with the neglect option, will be supported.
Last, as far as predictors other than three culture variables, the regression table 5.6
indicates that self-esteem and (t=-2.51, p<.05) and power distance (t=2.46, p<.05) are associated
with the neglect option. In detail, self-esteem (β = -0.12) was found to have a negative
relationship with the neglect option. In other words, the more people perceive themselves valued
in their organization, the less people are likely to choose the neglect option. Therefore, we can
conclude that the findings of this study support the part of this study‘s hypothesis that concludes
that self-esteem will have a negative association with the neglect option, since in low self-esteem
situations, people think that they are not significant, successful, and worthy in the organization,
which leaves them feeling powerless and leads to passive response, such as neglect, when
confronted with a deteriorating situation at work.
Power distance (β = 0.15) was also found to have a positive relationship with the neglect
option. In other words, the more people believe that employees who have lower power should
obey the authority of the organization, and that therefore supervisors should have a large amount
of power over them, the more they are likely to choose the neglect option. In other words, it

１１１

supports that employees with higher PD orientations believe that employees who have lower
power should obey the authority of the organization; therefore supervisors should have a large
amount of power over them. For that reason, they will be afraid to disagree with their superiors;
they will be more likely to refuse a confrontational action toward solving deteriorating situation
in the organization they belong to. This means even though they disagree with their supervisor or
different opinion against their supervisor, they will set aside instead of trying actively to solve
the situation. Overall, we can conclude that the findings of this study support the part of this
study‘s hypothesis that says power distance will have a positive association with the neglect
option.
In sum, consistent with the result of regression on the voice and loyalty options, the
loyalty option is influenced by culture in many ways. First, the country effect was observed in
the loyalty option indicating that country difference influences employees‘ propensity for the
loyalty option. Furthermore, country is the strongest predictor of the loyalty option. Second, one
of culture values in a harmony culture, conflict-avoiding, was found to have a significant
association with the loyalty option. Furthermore, consistent with the result of regression on the
voice and loyalty options, a culture value, which is conflict-avoiding, was found to be the
strongest predictor of the loyalty option. Additionally, other variables, such as self-esteem and
power distance, also affect employees‘ propensity for the neglect option.

5.4.2 Comparison between the Korean sample and the U.S. sample
The next analysis is a comparison of regression outcomes between the Korean sample
and the U.S. sample to see if there is any difference between these two country samples. There
are two main reasons to conduct this comparison. First, this study hypothesized that collectivism,

１１２

face-saving, and conflict-avoiding are values that affect people‘s behaviors in a harmony culture,
therefore, these values will affect employees‘ behavioral responses to a deteriorating situations at
work. As described earlier, the Korean sample represents a harmony culture and the U.S. sample
is categorized as a non-harmony culture in Kozan‘s culture model (1997). Therefore, by
comparing the regression outcome between the Korean sample and the U.S. sample, we can
examine if these three culture variables are culture-specific values for a harmony culture. The
Chow test outcome will determine if these differences are statistically significant.
Second, previous analysis of the total sample indicates significant association between
dummy country variable and three behavioral outcomes; voice, loyalty, and neglect. This means
that whether or not the respondent is in Korea make a difference in their propensity for voice,
loyalty, and neglect. Comparison of the split regression outcome between the Korean sample and
the U.S. sample will show the details of the difference between the three behavioral outcomes for
each predictor.

5.4.2.1 Regression on the Exit Option
Table 11 presents the multiple regression analysis summary on the exit options with the
Korean sample and the U.S. sample separately, as well as the chow test summary. First, the result
of chow test for the regression model with the Korean sample and the U.S. sample shows that the
regression models are statistically different between the Korean sample and the U.S. sample.
Second, there are two common findings of regression model on exit option between the
Korean sample and the U.S. sample: 1) none of the three culture values has a significant
association with the exit option; 2) alternative quality has a significant and positive association
with the exit option among other predictors tested in these models. Alternative quality was found

１１３

to have a significant and positive association with the exit option both in the Korean sample (t =
8.470, p<.000) and in the U.S. sample (t = 2.58, p<.05). More interestingly, in both samples,
alternative quality was found to be the strongest predictor of the exit option. These two findings
are consistent with the previous finding in the total sample that the three harmony culture values
such as collectivism, face-saving, and conflict-avoiding do not predict the exit option and that
alternative quality is the strongest predictor of exit option. 3) Considering that dummy
organization variable does not have a significant association with the exit option, organization
difference does not exist in each country sample.

Table 11. Regression Outcome on Exit with the Korean sample and the U.S. sample
KOR

US

β

β

-3.21

Dummy Org
Education
Tenure
Managerial
Professional
Laborer
Alternative Quality

0.00
-0.15
0.01

-0.03
-1.88
0.20

0.11
-0.15
0.40

2.26
-1.53
8.47

Self-Esteem
Power Distance
LMX
Collectivism
Face-Saving
Conflict-Avoiding

-0.15
-0.02
0.03
-0.03
0.06
0.06

-3.02
-0.42
0.58
-0.65
1.16
1.20

²

p < 0.001,

**

5.32

2.46

-0.15
0.17
-0.08
-0.49
-0.34
-0.27
0.32

-1.01
1.32
-0.69
-1.34
-0.80
-0.56

*

***
**

9.31

*

*

2.58
-1.95
0.53
-0.91
1.15
0.44
-1.98

-0.24
0.07
-0.11
0.14
0.05
-0.29

0.23

F
Chow Test (F)
Result
***

4.33

***

(Constant)

R

t

t

0.32

***

2.17

*

0.60
Different
*

p < 0.01, p < 0.05, N=398 for the Korean sample, N=73 for the U.S. sample
１１４

Lastly, the difference between the effect of the regression model on exit option in the
Korean sample and the U.S. sample is that the job category of professional (t = 2.26, p<.05) and
self-esteem (t = -0.42, p<.01) are significant predictors of exit option only in the Korean sample.
Looking closers, in Korea, professionals are more likely to choose the exit option. In addition,
someone who has higher self-esteem is less likely to choose the exit option, which supports the
hypothesis that self-esteem will have a negative association on exit option.
Overall, regression outcomes on the exit option in two country samples reconfirmed the
findings in the total sample, by supporting that the three culture variables do not affect people‘s
propensity for exit option, and that alternative quality is the strongest predictor of the exit option
for both countries.

5.4.2.2 Regression on the Voice Option
Table 12 presents the multiple regression analysis summary on the voice options on the
Korean sample and the U.S. sample separately, as well as chow test summary. First, consistent
with the results of the regressions on the exit option, the result of the chow confirms that the
regression models with the Korean sample and the U.S. sample are statistical different each other.
Second, there are two common findings for regression models on voice option for both
countries: 1) among the four predictors, other than the three culture values, only LMX is found to
have a significant and positive association on the voice option, which supports this study‘s
hypothesis that LMX will have a positive association on voice option. 2) As shown in the
regression outcomes on the exit option, dummy organization variable does not have a significant
association with the voice option, which means that there is no significant regression outcome
difference between the two organizations within the same country.

１１５

Table 12. Regression Outcome on Voice with the Korean sample and the U.S. sample
KOR

US

β

β

t

0.25

0.18

0.24
0.29
-0.06
-0.08
-0.34
0.11
0.24
0.07
0.10

1.70
2.25*
-0.53
-0.23
-0.81
0.23
1.93
0.55
0.72

**

0.33

**

0.11

2.80
0.93

***

-0.02

-0.17

*

-0.03

-0.20

t
***

(Constant)

-2.64

Dummy Org
Education
Tenure
Managerial
Professional
Laborer
Alternative Quality
Self-Esteem
Power Distance

-0.01
0.02
-0.07

-0.10
0.31
-1.51

0.02
0.06
0.00
0.06
-0.18

0.41
0.62
-0.07
1.28
-3.43

LMX

0.15

3.05

Collectivism

0.13

2.72

Face-Saving

0.23

Conflict-Avoiding

²

p < 0.001,

**

4.56

-0.11

**

-2.15
0.26

R
F
Chow Test (F)
Result
***

3.92

11.08

**

0.34

***

2.37

*

1.34
Different
*

p < 0.01, p < 0.05, N=398 for the Korean sample, N=73 for the U.S. sample

Last, the difference between the Korean sample and the U.S. sample regression model of
exit option is that power distance (t = 0.13, p<.01), collectivism (t = 0.23, p<.01), face-saving (t
= 4.56, p<.000), and conflict-avoiding (t = -2.15, p<.01) are significant predictors of the exit
option only in the Korean sample. More importantly, as shown in Table 12, face-saving has the
strongest association with the voice option in Korea of any variable. Another interesting finding
is that, contrary to the finding on the exit option in Korea, all three culture values have a
significant association with the voice option in Korea: collectivism (β = 2.72, p<.01), and facesaving (β = 4.56, p<.000) have a significant and positive association with the voice option, while
１１６

conflict-avoiding (β = -0.11, p<.01) has a significant and negative association with the voice
option. Based on this finding, we can suggest that the voice option is heavily influenced by
culture-specific values. In the U.S. sample, only education (t = 2.25, p<.05) is found to be a
significant predictor of the exit option.
In sum, the key findings of regression outcomes of the voice option in two country
samples are as follows: first of all, contrary to the findings in the exit option, the regression
outcome on the voice option in the Korean sample and that in the U.S. sample are different from
each other. Therefore, we can discuss that the mechanism that employees choose the voice
options in Korea is different to that in the U.S. Second, while the exit option does not have a
significant association with three harmony culture values even in Korean sample, the voice
option is influenced by all of these three culture values. Last, the only independent variable
affecting both the Korean sample and in the U.S. sample is LMX. Overall, the regression
outcome on the voice options supports the part of this study‘s hypothesis as well as previous
employees‘ voice behavior literature that a good relationship with a supervisor promotes
employees voice behavior. In addition, the finding of this study confirms that this effect of LMX
on employees‘ voice behavior is universal, across countries.

5.4.2.3 Regression on the Loyalty Option
Table 13 presents the multiple regression analysis summary of the loyalty option with the
Korean sample and the U.S. sample separately, as well as chow test summary. First, the result of
the chow test shows that the regression models are statistically different for the Korean sample
and the U.S. sample.

１１７

Second, the only common finding between the countries is that dummy organization
variable does not have a significant association with the loyalty option, meaning that
organization effect does not exist both in Korea and the U.S. sample.

Table 13. Regression Outcome on Loyalty with the Korean sample and the U.S. sample
KOR
US
β
t
β
t
***

(Constant)

-2.20

Dummy Org
Education
Tenure
Managerial
Professional
Laborer
Alternative Quality

-0.05
0.16
-0.08

-0.51
1.97
-1.47

0.00
0.19
-0.15

-0.03
1.80

Self-Esteem
Power Distance
LMX
Collectivism
Face-Saving
Conflict-Avoiding

0.14
0.09
0.02
0.13
-0.03
0.22

²

p < 0.001,

**

-3.07

*

*

2.50
1.58
0.41

*

2.48
-0.61
3.92

***

-2.11

1.12

-0.11
-0.19
0.12
0.08
-0.03
-0.04
-0.08

-0.73
-1.45
0.98
0.22
-0.08
-0.08
-0.63

0.14
0.28
0.16
0.00
-0.11
0.20

1.12
2.01*
1.27
0.00
-0.84
1.33

0.13

R
F
Chow Test (F)
Result
***

4.08

4.86

0.28

***

1.78
0.36
Different

*

p < 0.01, p < 0.05, N=398 for the Korean sample, N=73 for the U.S. sample

Finally, a more interesting finding is that, especially compared to regression outcomes on
other dependent variables, there are no common predictors of the Korean sample and the U.S.
sample that have a significant association with the loyalty option. This means that none of the
predictors tested in this study affect the loyalty option across countries. In particular, four

１１８

predictors including alternative quality, self-esteem, power distance, and LMX did not gain
statistically significant association with the loyalty option; therefore, the parts of hypotheses in
this study that they will have a significant relationship with the loyalty option across countries,
are rejected.
In the Korean sample, alternative quality (t = -3.07, p<.05), self-esteem (t = 2.50, p<.05),
collectivism (t = 2.48, p<.05), and conflict-avoiding (t = 3.92, p<.000) are found to have a
significant association with the loyalty option. More specifically, self-esteem, collectivism, and
conflict-avoiding have significant and positive relationships with the loyalty option. In other
words, the more of these factors found in people, the more likely they are to choose the loyalty
option. Interestingly, conflict-avoiding (β = 0.22) was found to be the strongest predictor of the
loyalty option in the Korean sample. On the other hand, alternative quality was found to have a
significant negative relationship with the loyalty option. This means, the higher the degree of
alternative quality, the lower the propensity for loyalty option.
In the U.S. sample, only power distance (t = 2.01, p<.05) was found to be a significant
predictor of the loyalty option: power distance has a significant and positive relationship with the
loyalty option. In other words, people are more likely to choose the loyalty option when they
have a higher degree of power distance.
Overall, findings in the regression outcomes of the loyalty option in two country‘s
samples are consistent with the regression outcomes of the voice option in some points. First of
all, Chow test result shows that the regression model with the Korean sample is different to that
with the U.S. sample. Secondly, the regression outcomes reveal that three harmony culture
values have partially a significant association with the loyalty option since except face-saving,
two of them such as collectivism and conflict avoiding have a significant association with the

１１９

loyalty option in the Korean sample. Last, contrary to findings in the regression outcomes on the
exit and voice options, no common factor was found in the loyalty option affecting both the
Korean sample and the U.S. sample.

5.4.2.4 Regression on Neglect Option
Table 14 presents the multiple regression analysis summary on the neglect option with
separate Korean and U.S. samples, as well as chow test summary. First, the result of the chow
test shows that the regression models of the neglect option show statistical differences between
the Korean sample and the U.S. sample. This means that in this study, the employees‘ propensity
for the neglect option is influenced by different predictors that depend on their country of origin.
Second, dummy organization variable do not have a significant association with the
neglect option. This means that organization effect does not exist in each country‘s sample.
The last and most interesting finding is that, consistent with the finding of the loyalty
option, there are no common predictors that have a significant association with the neglect option
for either the Korean sample or the U.S. sample. This fact means that none of the predictors
tested in this study have an effect on the neglect option.
According to the regression outcome for the neglect option, in the Korean sample, the job
category of laborer (t = -2.39, p<.05), alternative quality (t = 2.38, p<.05), and conflict-avoiding
(t = 4.04, p<.000) are found to have a significant association with the neglect option: alternative
quality has a significant and positive relationship with the loyalty option. In other words, people
who have a good alternative to their current job are more likely to choose the neglect option.
This finding is the opposite of this study‘s hypothesis that alternative quality will have a negative
association with the neglect option.

１２０

Table 14. Regression Outcome on Neglect with the Korean sample and the U.S. sample
KOR
US
β
t
β
t
***

(Constant)

2.95

Dummy Org
Education
Tenure
Managerial
Professional
Laborer

0.17
-0.07
-0.02

1.84
-0.86
-0.41

-0.01
-0.25

-0.18

Alternative Quality
Self-Esteem
Power Distance
LMX
Collectivism
Face-Saving
Conflict-Avoiding

0.12
-0.09
0.10
0.01
-0.05
0.04
0.23

²

p < 0.001,

**

-2.39

*

*

2.38
-1.67
1.75
0.11
-0.89
0.75
4.04

***

5.31

*

3.81

2.08

0.03
0.11
0.24
-0.22
-0.21
-0.29

0.22
0.83
1.97
-0.60
-0.48
-0.60

0.16
-0.24
0.34
-0.07
-0.04
-0.17
-0.05

1.28
-1.94

0.14

R
F
Chow Test (F)
Result
***

4.29

*

2.50
-0.58
-0.31
-1.37
-0.31
0.32

***

2.17

*

0.75
Different
*

p < 0.01, p < 0.05, N=398 for the Korean sample, N=73 for the U.S. sample

Although it rejects this hypothesis, this finding can be considered plausible after
considering the findings from the turnover studies: according to meta-analysis in turnover
research, withdrawal behaviors such as absenteeism and lateness forecast turnover later (Hom &
Griffeth, 1994).This finding says that when confronting a deteriorating situation at work,
someone who has a good alternative to their current job is likely to become involved in neglect
behavior such as losing motivation to work, lateness, or reducing effort at the job, as a precursor
of quitting later.

１２１

Conflict-avoiding was also found to have a significant and positive association on the
neglect option. This finding supports the interpretation of conflict-avoiding as a style of avoiding
explicit expression that is not consistent with status quo in their mind: since people who put a
high value on conflict-avoiding believe that this confrontational behavior will hurt the
relationship with the organization, they refuse to use confrontational action to solve a
deteriorating situation at work. Furthermore, conflict-avoiding (β = 0.23) is found to be the
strongest predictor of the loyalty option in the Korean sample.
The job category of skilled crafts and laborers was found to have a significant but
negative relationship on the loyalty option. This means that employees who are in the job
category of skilled crafts and laborers are less likely to choose the neglect option. It is possible
that uninvestigated characteristics of the employees who are in the job category of the skilled
crafts, and laborers may affect their propensity for the neglect option.
In the U.S. sample, only power distance (t = 2.50, p<.05) was found to be a significant
predictor of the neglect option: power distance has a significant and positive relationship with
respect to neglect option. In other words, people are more likely to choose the neglect option
when they have a higher degree of power distance.
Overall, there are several interesting findings in the neglect option: 1) the employees‘
intention to choose the neglect option is influenced by different predictors depending on their
home country; 2) there is no organization effect on the regression outcomes for the neglect
option either in the Korean sample or the U.S. sample; 3) there is no common predictor of the
neglect options across countries; 4) in the Korean sample, the job category of laborer, alternative
quality, and conflict-avoiding all have a significant association with the neglect option,
particularly conflict-avoiding, which is the strongest predictor of the neglect option in the Korean

１２２

sample; 5) in the U.S. sample, only power distance has a significant association with the neglect
option.

5.4.3 Additional Analysis
Along with regression analysis on EVLN, additional regression analysis was conducted
for further interpretation of what the data. The exit, voice, loyalty, and neglect option can be
largely interpreted as constructive responses vs. destructive responses, especially considering the
outcome of the behavior on relationship with the organization. There are two reasons to conduct
this additional analysis.
First, the outcome of zero-order correlation matrix indicates interesting findings,
especially for the correlation among the four behavioral responses. According to this correlation
matrix, the exit option has a statistically significant and positive correlation with the neglect
option, and the voice option has a statistically significant and positive correlation with the loyalty
option. Furthermore, the exit option has a statistically significant but negative correlation with
the loyalty option, while the voice option has a statistically significant but negative correlation
with the neglect option. These findings can be interpreted to mean that there is a statistically
significant association among the four behavioral associations.
Based on these findings, we can earn other way of interpreting the relationship among
these four behavioral options: 1) the exit option is related to the neglect option positively,
therefore they can be categorized as one dimension of four behavioral options; 2) the voice
option is related to the loyalty option positively, thus they can be combined as another dimension
of four behavioral options; and 3) the first categorized option and the second categorized option

１２３

have negative association with each other, meaning that these two categories can be interpreted
as the polar opposite option for each other.
Second, this combination of ‗exit and neglect‘ vs. ‗voice and loyalty‘ is consistent with
one of the dimension that previous EVLN researchers have previously suggested through
empirical supports. For instance, when describing the characteristics of each of EVLN option,
Rusbult and colleagues (1982) suggested that these four types of responses would differ from
each other with two evaluative dimensions: one of these dimensions is constructiveness vs.
destructiveness. According to their explanation, as far as the destructive-constructive dimension,
when categorizing these four responses in terms of impact on employee-organization relationship,
voice and loyalty are considered to be constructive responses, since they are aimed at
maintaining or reviving the relationship with the organization, whereas, exit and neglect are
described as destructive responses to this relationship. This categorization of the EVLN construct
has obtained both theoretical and empirical support in a number of studies (Farrell, 1983; Farrell
& Rusbult, 1985; Hagedoorn et al., 1999; Rusbult, Farrell, Rogers, & Mainous, 1988; Withey &
Cooper, 1989).
Overall, based on the finding of the zero-order correlation matrix as well as the
dimension suggested by the empirical support of previous EVLN researchers, this study
conducted additional regression analysis on the constructive vs. destructive responses. In other
words, the exit and neglect options are categorized as ‗destructive response‘ bringing a negative,
confrontational or undesirable outcome, such as quitting, absenteeism, lateness, or reduced effort,
which hurts the relationship between the employees and the organization. On the contrary, the
voice and loyalty options will be categorized as ‗constructive,‘ bringing a positive, cooperative,
or desirable outcome in order to maintain or revive the relationship between the employees and

１２４

the organization, much as making an effort to improve the situation, giving support to the
organization, or practicing good citizenship.
For the analysis of the constructive vs. destructive responses, first, all items indicating the
exit option and the neglect option were summed and averaged to create a single measure of
‗negative‘ response, while all items for the voice option and the loyalty option were summed and
averaged to create a single measure of ‗constructive‘ response. Then, each of these was regressed
upon the following independent variables: collectivism, face-saving, conflict-avoiding,
alternative quality, self-esteem, power distance, and LMX, as well as control variables including
nationality, age, education, tenure in the organization, and occupational type. This regression
analysis was conducted with 1) the total sample, 2) the Korea sample, and 3) the U.S. sample to
see any difference of regression outcomes. As it was in the regressions on EVLN between the
Korean sample and the U.S. sample, the Chow test was conducted to examine if the differences
were statistically significant. In addition, since data was collected in two organizations for each
country sample, dummy organization variable was included as a control variable to see if there
was an organization effect within the country sample.

5.4.3.1 Constructive vs. Destructive Responses with the Total Sample
Face-saving also has a positive relationship with constructive response; in other words,
the more people are concerned with face-saving, the more likely they are to choose constructive
response. This is supported by the fact that people who highly value face-saving will respond to
their deteriorating situation at work based on the criteria of whether or not their behavior will
lead to hurting their own face as well as faces of others in the organization they belong to. They
will, therefore, be more likely to engage in a constructive response, since it will be interpreted as

１２５

a cooperative behavior by other members in the organization, enhancing favorable judgment
from other remembers.

Table 15. Regression Outcome on Constructive vs. Destructive response with the total sample
Constructive
β

Destructive
β

t
*

*

(Constant)

(1.87)

Korea
Education
Tenure
Managerial
Professional
Laborer
Alternative Quality

.24
0.06
-0.09
-0.08
-0.10
0.00
-.14**

4.06
0.94
-1.58
-0.23
-0.48
0.01

Self-Esteem

.15**

Power Distance

-0.04

3.22
-0.86

LMX

.20

Collectivism

.17

***

-2.93

²

**

**

(3.21)

3.43

***

***

0.24
-0.03
0.02
-0.30
-0.13
-0.41
.32

4.04
-0.40
0.39
-0.84
-0.62
-1.05

***

-.20

**

6.73

***

***

-4.19

*

***
*

4.22

***

3.89

0.10
-0.02

1.99
-0.41

***

-0.05

-1.03

**

***

0.04

0.81

***

2.67
0.92
0.24

R
F

**

**

.13
0.04

Conflict-Avoiding

p < 0.001,

***

**

Face-Saving

***

2.14

t

10.92

***

.13

**

2.73

**

0.21
9.10

***

*

p < 0.01, p < 0.05, N=510

However, conflict-avoiding (t = 2.73, p<0.01) was found to have a significant and positive
relationship with a destructive response. When people have a higher degree of conflict-avoiding,
they are more likely to choose destructive response. This finding supports that since people who
place a high value on conflict-avoiding believe that this confrontational behavior will hurt their
relationship with the organization, they refuse a confrontational action even if it means fixing a
negative situation at work.

１２６

Finally, as far as predictors other than the three culture variables, the regression outcome
indicates that alternative quality has a significant negative association with a constructive
response (t = -2.93, p<0.01), while having significant but positive association with a destructive
response (t = 6.73, p<0.001). People who have good job alternatives to their current job will be
less likely to choose constructive response and more likely to choose destructive options. This
finding is consistent with suggestions that good alternatives encourage people to behave
aggressively when confronting a deteriorating situation at work, since it serves as a source of
power for them; in this case, it can be interpreted as a precursor to withdrawing or leaving (exit
and neglect). Therefore, they will not use a constructive response to revive or maintain the
relationship with the organization (voice and loyalty).
Self-esteem works the opposite way from alternative quality: self-esteem has a positive
association with a constructive response (t = 3.22, p<0.01) while negative association with a
destructive response (t = -4.19, p<0.001). This means that the more people perceive themselves
as valued in their organization, the more likely they are to choose constructive responses, instead
of destructive responses. Accordingly, the finding supports that self-esteem indicates the degree
to which individuals believe they are capable, significant, successful and worthy; these positive
beliefs about the self will encourage positive input toward the organization (voice and loyalty)
and discourage controversial behaviors (exit and neglect).
Power distance has a significant and positive association with a destructive response (t =
1.99, p<0.05). This means that the more people have a degree of power distance, the more they
are likely to choose destructive behavior. It supports that power distance will have a negative
association with the voice option, since people who have high power distance orientation believe
that employees who have lower power should obey the authority of the organization, and are

１２７

therefore, more likely to refuse a confrontational action aimed at solving a deteriorating situation
at the organization they belong to, by either expressing disagreement or having a different
opinion from the authority.
LMX was found to have a significant and positive association with a constructive
response (t = 4.22, p<0.05): people who perceive that they have a good relationship with their
supervisor are more likely to choose a constructive response. Therefore, it supports that high
LMX is characterized as mutual trust, professional respect, reciprocal influence, liking, and a
sense of obligation to the leader (Grae & Uhl-Bien, 1995). This favorable context for the
employees encourages them to involve in positive input to the organization, such as voice and
loyalty.
Overall, the key findings are as follows: 1) country itself has a strong association with
both the constructive response and the negative response; 2) people who value collectivism and
face-saving are more likely to choose the constructive response, while people who value conflictavoiding are less likely to choose the destructive response; 3) For other predictors, alternative
quality has a significant and negative association with a constructive response, while having a
significant but positive association with a destructive response; self-esteem has a significant and
positive association with a constructive response, but a significant but positive association with a
negative response. Power distance has a significant and positive association with a destructive
response, and LMX was found to have a significant and positive association with a constructive
response.

5.4.3.2 Constructive vs. Destructive Responses in the Korean Sample and the U.S. Sample
Table 16 presents the multiple regression analysis summary on positive vs. destructive
response to a deteriorating situation at work with the Korean sample and the U.S. sample
１２８

separately, as well as the Chow test summary. First, the result of the Chow test confirms that the
regression model with the Korean sample is statistically different to that with the U.S. sample.
Besides, organization effect was not found in either the constructive or destructive responses for
either country.

Table 16. Coefficient Outcome on Constructive vs. Destructive response with the Korean sample
and the U.S. sample
Constructive
KOR
US
***

(Constant)

-1.03

(2.47)
-0.03
0.07
-0.08

Korea
Education
Tenure
Managerial
Professional
Laborer

0.13
-0.01
0.05
0.06
-0.17
0.10

0.00
0.12

KOR

-.27

**

**

0.16

-.15
0.06

Power Distance

.15
-0.10

LMX

.15

²

**

.31

**

*

-.32

*

*

0.02

.30
-0.12

-0.05

0.05

-0.09

0.06

-0.10

**

***

0.18
0.22

R
F

***

.32
0.09

**

.16
0.03

Conflict-Avoiding

p < 0.001,

0.24

**

Face-Saving

(4.04)
-0.06
0.18
0.13
-0.45
-0.35
-0.37

0.05
.31

Self-Esteem

.19

**

(3.07)
0.12
-0.13
-0.01

0.04

-.14

Collectivism

US

***

**

Alternative Quality

***

Destructive

8.89

***

.20

***

-0.20
0.41
3.09**

*

p < 0.01, p < 0.05, N=398 for the Korean sample, N=73 for the U.S. sample

Second, concerning the three culture variables--collectivism, face-saving, and conflictavoiding--collectivism (β = .19***) and face-saving (β =.16**) have a significant and positive
relationship with the constructive response, while conflict-avoiding (β =.20***) has a significant
and positive relationship with a destructive response. In more detail, consistent with the findings

１２９

of the total sample, the finding supports that people who value collectivism will be more likely to
engage in a constructive response since it will contribute to maintaining or improving harmony
and relationship within the organization they belong to. Face-saving also has a positive
relationship with constructive response, supporting that the more people are concerned with facesaving, the more people are likely to choose constructive response, since it will not hurt their
own face or the face of others in the organization.
However, conflict-avoiding was found to have a significant and positive relationship with
a destructive response. It is also consistent with the finding in the total sample that the more
people have a degree of conflict-avoiding, the more likely to choose destructive response, since
they refuse a confrontational action toward solving a deteriorating situation at work, and prefer
avoiding explicit expression that is not consistent with status quo in their mind.
Last, as far as other predictors than the three culture variables, there are three common
findings of a regression model on constructive response and destructive response across
countries: in both countries, 1) alternative quality has a significant and positive association with
the destructive response; 2) self-esteem has a significant and negative association with the
destructive response; and 3) LMX has a significant and positive association with the constructive
response.
In more detail, alternative quality (β =.31 in the Korean sample, β = .31 in the U.S.) has a
significant and negative association with a destructive response, which means, consistent with
the finding in the total sample, people who have good job alternative quality to the current job
will be more likely to choose the destructive option since good alternative encourages people to
do something aggressive for a deteriorating situation at work, since it serves as s source of power
for them.

１３０

On the contrary, self-esteem was found to have a significant and negative association
with a destructive response (β =.-.15 in the Korean sample, β = -.32 in the U.S.). As supported by
the finding with the total sample, this means that the more people perceive themselves valued in
their organization, the less people are likely to choose destructive response, since self-esteem
will discourage their controversial behaviors (exit and neglect).
As found in the total sample, LMX has a significant and positive association with the
constructive response (β = .15 in the Korean sample, β = .32 in the U.S.). Therefore, people who
perceive that they have a good relationship with their supervisor are more likely to choose a
constructive response since this favorable context for the employee, including mutual trust,
professional respect, reciprocal influence, liking, and a sense of obligation with their leader,
encourages them to engage in positive input, such as voice and loyalty, toward the organization.
Last, the outcomes differ between the Korean sample and the U.S. sample as follows: 1)
alternative quality (β = -.14) has a significant and negative association with constructive
response in Korea, but not in the U.S.; 2) self-esteem (β = .15) has a significant and positive
association with a constructive response in Korea, not in the U.S; and 3) power distance has a
significant and positive association with a destructive response (β = 0.30) only in the U.S.

１３１

CHAPTER 6
DISCUSSION

This chapter presents the discussion of the results of the analyses conducted in this
dissertation. It includes (1) Summary of Findings, (2) Interpreting Findings, and (3) the Results
of Hypotheses Testing.

6.1 Summary of Findings
So far, the key findings of regression analysis predicting EVLN using the total sample, as
well as a comparison between the Korean sample and the U.S. sample, have been addressed.
Furthermore, additional regression analysis was conducted with the total sample, as well as a
comparison between the Korean sample and the U.S. sample, by categorizing the exit and the
neglect option as ‗destructive‘ response and the voice and the loyalty option as ‗constructive‘
responses. Since this study conveys a complex model with four dependent variables and seven
independent variables with three types of sample, a summary of findings from all of these
analyses is needed to provide one comprehensive picture to answer the question, ―What was
found in this study?‖

6.1.1 Findings from Regressions on EVLN with Total Sample
Overall, several interesting findings were observed in the regression outcome on EVLN
with the total sample, which support or reject this study‘s hypotheses. First of all, the exit option
reveals distinctive features compared to the other three options: that is, the exit option is not
associated with culture at all in that 1) country effect was not found and 2) none of the three

１３２

culture values have a significant association with the exit options. These findings are the
opposite of the findings in the other three options (voice, loyalty, and neglect) in that all of them
have a significant association with country variable and with at least one of three culture values.
As discussed earlier, it is due to the nature of the exit option itself differentiating it to
other option; the exit option brings consideration of possible disadvantage by ‗leaving the
organization‘ while other options assume ‗staying the organization‘ anyhow. Therefore, when
people think of the exit option, they need to think of the realistic losses, especially economic
losses, expected by choosing the exit option, such as loss of seniority, fridge benefits, and costs
to find and transition to a new job. Additional finding that alternative job quality is the strongest
predictor in the exit option supports these possible economic considerations for the exit option.
By association, culture values are too distal to directly affect employees‘ exit-related behaviors.
Findings in turnover literature that proximal predictor, such as quit intention, are the best
predictors of an actual turnover behavior support this lack of the effects of culture on the exit
option.
Second, aside from the exit option, the three other options, which include voice, loyalty,
and neglect, are influenced by culture to some degree. The common effect across these three
options is that they have a significant association with country, which describes whether or not
employees are certain country makes a difference in their behavior of voice, loyalty, and neglect
options. For instance, the regression outcomes show that in this study, Korean respondents are
more likely to choose voice, loyalty, and neglect options. It can be interpreted that, when they
confront a deteriorating situation at work, these employees are more likely to choose behaviors
that assume departure from the organization they belong to.

１３３

Third, for non-exit options, the magnitude of the effects of the three culture values varies
across the three behavioral options. For instance, among the three options, only the voice option
has a significant association with all three culture values: collectivism, face-saving, and conflictavoiding. Based on this finding, we can conclude that the voice options are more heavily
influenced by culture values than other behavioral options. The loyalty option is influenced by
two culture values: collectivism and conflict-avoiding, and the neglect option has a significant
relationship with one culture value only: conflict-avoiding. Therefore, we can conclude that
although all three behavioral options, other than the exit option, are influenced by culture, the
degree of culture varies: the voice option is most heavily influenced by the culture values and the
neglect option is the least affected by the culture values.
Fourth, among three culture values, only conflict avoiding has a significant association
with all three non-exit options. For instance, collectivism is significantly related to the voice and
the neglect options, while face-saving is significantly related to the voice option only. However,
conflict-avoiding is associated with all three non-exit options. Furthermore, conflict-avoiding is
the second strongest predictor of the neglect and loyalty option, following the country variable. It
can be interpreted that conflict-avoiding is the most widely referred and strongest predictor when
employees respond to a deteriorating situation at work. This is possibly due to the nature of
conflict-avoiding itself. Basically, the concept of conflict-avoiding came from research on
conflict management, which defined conflict-avoiding as a style of handling conflict,
characterized by withdrawing from the situation or suppressing the urge to engage in conflict
(Rollinson & Broadfield, 1998). Compared with two other culture values which are general value
applicable to other situations in a culture, conflict-avoiding can be interpreted as a situationspecific value referring to a deteriorating situation at work. This trait of conflict-avoiding will be

１３４

considered widely as a reference for deciding certain types of behavior as a response to a
deteriorating situation at work. It must be observed, however, that the correlation coefficient
between KOR and tenure is .62. It must also be observed that the Korean respondents had, on
average, eleven years more tenure in the organization than the U.S. respondents. Because the
coefficients on KOR were significant in two of the three staying equations (loyalty and neglect),
it may be that some of the effect of KOR may actually be a tenure effect; longer-service
employees are more likely than shorter-service employees to have experienced a negative
situation at work and more likely than shorter-service employees to have determined that loyalty
and neglect are reasonable responses to a negative situation at work.

6.1.2 Findings from Comparison between the Korean Sample and the U.S. Sample
Overall, the comparison of regressions on EVLN between the Korean sample and the U.S.
sample provides several interesting findings to be addressed. First of all, there are common
findings across the EVLN options: 1) Chow test results confirm that the regression outcomes on
EVLN options are statistically different from the Korean sample to the U.S. sample; 2) the result
of regressions on these options shows that there is no organization effect for each country,
meaning that the organization variable does not make any statistically significant variation within
the country; 3) none of the three culture variables are associated with all of the EVLN options for
the U.S. sample, therefore, it supports that the three culture variables valued in a harmony culture
present distinctive traits of their specific culture. Second, the regression outcomes reveal notable
features of each behavioral option. In the exit option, consistent with the findings in the total
sample, regardless of the country, none of the three culture values were found to have a
significant association with the exit option. Therefore, it reconfirms that an employees‘
１３５

propensity for the exit option is not influenced by culture at all. Once again, this supports the
finding that the inherent nature of the exit option differentiates itself from the other three options
by providing alerts to the possible disadvantages of leaving the organization. Therefore, rather
than distal factors, such as personal beliefs or values, proximal predictors such as availability of
alternatives to the current job directly influence the exit option. The dominant effect of
alternative quality on the exit option for both countries supports this realistic consideration of the
exit option.
Third, except for the exit option, the three culture values are associated with three
behavioral options, voice, loyalty, and neglect, in the Korean sample only. All three behavioral
options are significantly associated with at least one culture value in the Korean sample.
However, the degree of effect of the three culture values on employee response to negative
workplace situations varies depends on the respective behavioral options. For instance, the voice
option is most influenced by the harmony culture values since only the voice option has a
significant association with all harmony culture values –– collectivism, face-saving, and
collectivism. The loyalty option is influenced by two culture values, collectivism and conflictavoiding, and the neglect option only has a significant relationship with one culture value:
conflict-avoiding. Therefore, we can conclude that although the three behavioral options, the
voice, loyalty, and neglect option, are influenced by the harmony culture values in the Korean
sample, the degree of effect of the three culture values on these three options varies; the voice
option is most heavily influenced by the harmony culture values and the neglect option is the
least affected by the harmony culture values.
Fourth, another interesting finding was observed in regards to the loyalty and neglect
options. Contrary to the other two behavioral options: exit and voice, there is no common

１３６

predictor found to have a significant association with the loyalty and neglect options across
countries. For instance, alternative quality was found to be a universal predictor of the exit
option both in the Korean sample and the U.S. sample; and LMX has a significant association
with the voice option both in the Korean sample and the U.S. sample. However, there are no
universal predictors found to be associated with the loyalty and neglect option both in the Korean
sample and the U.S. sample.
Last, in the Korean sample, of the three culture values, only conflict-avoiding has a
significant association with all three behavioral options. This is possibly due to the nature of
conflict-avoiding itself. Compared with the two other culture values which are general values
applicable to other situations in a culture, conflict-avoiding can be interpreted as situationspecific proxy of the attitude referring to a deteriorating situation at work. Therefore, conflictavoiding will be widely considered a reference for deciding on a type of behavior as a response
to a deteriorating situation at work.

6.1.3 Findings from Regressions on Constructive vs. Destructive Responses
Rather than providing new findings, the outcomes of regression on constructive vs.
destructive responses within the total sample, the Korean sample, and the U.S. sample are largely
consistent with the outcomes of the EVLN construct; therefore reconfirming most of the findings
of the EVLN construct.
First, the statistically significant country effects were confirmed both in the constructive
and destructive behaviors: 1) the outcome shows that employees in Korea are more likely to
engage in both the constructive and destructive behaviors; 2) Chow test outcomes indicate that

１３７

both regression outcomes of the constructive and destructive behaviors are different between the
Korean sample and the U.S. sample.
Second, three of the culture values have significant association with both the constructive
and destructive behaviors in the Korean sample, and not in the U.S. sample. This finding
reconfirms that the three culture values characterized as the key values in a harmony culture have
a significant impact on the members‘ behaviors, differentiating those in other cultures.
Third, although the significant association between three harmony culture values and the
constructive/destructive behaviors was supported in the Korean sample, the magnitude of
association is different depending on constructive or destructive behaviors. For instance,
collectivism and face-saving have a significant and positive association with the constructive
behaviors, while conflict-avoiding has a significant and positive association with the destructive
behaviors. In other word, there is no harmony culture value associating with all EVLN options.
Fourth, there are common predictors available both in the Korean sample as well as in the
U.S. sample. For instance, for either country, alternative quality has a significant and positive
association with the destructive behaviors, self-esteem has a significant but negative association
with the destructive behaviors, and LMX has a significant and positive association with the
constructive behaviors.
Fifth, among non-harmony culture values, there are also predictors available found either
only in the Korean sample or only in the U.S. sample. For instance, alternative quality has a
significant and negative association with the constructive behaviors for the Korean sample, and
self-esteem also has a significant and positive association only in the Korean sample, not in the
U.S. sample, while power distance has a significant and positive association with the destructive
behavior only in the U.S. sample, not in the Korean sample.

１３８

Sixth, consistent with the outcomes of the EVLN construct the result of regressions show
that no organization effect exists for each country: it means that the organization variable does
not make any statistically significant variation within the country
Finally, with respect to the distinction of constructive-destructive behaviors, it must be
noted that the significant results on Korea must be considered in the context of the high
correlation between Korea and tenure. Both constructs include stayers, thus, it may be that Korea
country variable is picking up some of the effects of tenure.

6.2 Interpreting Findings
So far, the findings of the data analysis have been briefly described as predictions of the
exit-voice-loyalty-neglect option, both separately and with the total sample, along with
additional analyses on the effect of constructive vs. destructive responses. In the final section of
this chapter, more efforts have been made to interpret what these findings tell us by comparing or
contrasting the findings on exit-voice-loyalty-neglect as a response to a negative situation at
work, with findings in the total sample, the Korean sample, and the U.S. sample, respectively.
Reviewing the key findings from all regression analysis conducted in this study, we can observe
that there are overlapping clues across regression analyses that provide insights into the overall
mechanism in order to answer the initial question of this study: ‗How does culture affect
employee response to a deteriorating situation at work?‘

6.2.1 Research Question One: Does Culture Affect Employee Voice Behavior?
According to Rusbult and colleges‘s (1982) EVLN construct, when employees confront a
deteriorating situation at work, they respond to this situation in one of four ways: they can leave

１３９

the organization--exit, express their opinion or idea to make the situation better--voice, continue
to support the organization--loyalty, or reduce their effort--neglect. Despite impressive advances
in the research for a better understanding of EVLN construct, few studies on the effect of culture,
particularly national culture on employees‘ voice behaviors at work, have been conducted
(Botero & Van Dyne, 2009; Edward & Greenberg, 2009; Landau, 2009; Lee & Jablin, 1992;
Price et al., 2001; Xu et al., 2005). Consistent with previous studies on the various cultural
dimensions that influence work-related psychological and behavioral phenomena (Bontempo &
Rivero, 1992; Markus & Kitayama, 1991; Suh, et al., 1998; Triandis, et al., 1995; Wagner, 1995;
Wasti, 2003), this study started with the question: ‗Does culture affect employee response to
negative workplace situations or voice behavior?‘
As described in Hofstede‘s (1980) explanation of culture, which is defined it as ―the
collective programming of the mind that distinguishes the members of one category of people
from those of another,‖ people from the same country share the same culture composed of values,
defined as ―broad tendencies to prefer certain states of affairs over others‖ (Hofstede, 1980, p.
19). Since values in a specific society represent patterns of thinking, feeling, and acting, those
who belong to the society have these values embedded, and would therefore exhibit similar
behaviors.

6.2.1.1 For a ‗Leaver‘, Culture Does Not Matter
One the most consistent and notable findings both in the Korean sample and the U.S.
sample is that culture does not have any significant association with the exit option. There are
two pieces of evidence supported by this study: 1) country effect was not found in the exit
option; 2) none of the three harmony culture values have a significant association with the exit

１４０

option both in the Korean sample and the U.S. sample. These findings are the opposite of the
findings on the other three options (voice loyalty, and neglect), in that all of them have a
significant association with the country variable and with at least one of the three culture values.
First, country effect was not found in the exit option, which is not consistent with
findings from previous cross-cultural research, especially from those supporting Hofstede‘s
(1980) view of culture. According to Hofstede‘s culture study, people from the same country
share values that represent stable patterns of thinking, feeling, and acting over time, therefore
they exhibit similar behaviors. However, the findings of this study reveal that the nationality of
the respondent fails to gain statistical power to predict their choice of the exit option.
Second, the findings of this study do not support any significant association between the
three culture values and the exit option. In this dissertation, it was proposed that, other things
being equal, all three harmony culture values will have a significant association with the exit
option. For instance, it was hypothesized that collectivism-oriented employees, who value
maintenance of harmony within the organization, will be less likely to chose the exit option
because, to them, the exit option will be interpreted as a destructive behavior terminating their
relationship with the organization. Face-saving was also hypothesized to discourage exit option
since exit option would be interpreted as ‗threatening‘ behavior, damaging the employee‘s face
in the organization and leading to other‘s unfavorable judgment of the employee. On the other
hand, conflict-avoiding was proposed to have a positive relationship with the exit option, since
people who have a tendency toward Conflict-Avoiding, prefer the exit option as a way of
escaping from the deteriorating situation, therefore avoiding expressing to others differences in
perceptions, beliefs, and goals in the organization. However, none of these hypotheses were
supported in this study.

１４１

Why did culture fail to affect employees‘ decision to leave? As discussed earlier, this is
possibly due to 1) the nature of the exit option itself and 2) the heavy influence on exit option by
realistic and economic consideration such as alternative quality. First, according to the turnover
studies, turnover itself is a very complex phenomenon (Peterson, 2004), especially considering
these possible negative consequences for the quitter (Mobley, 1982). Turnover researchers have
suggested that turnover can lead to negative consequences for the leaver as well as negative
consequences for the organization. For instance, Mobley (1982) summarized the negative
consequences for the leaver, which include 1) loss of seniority and fringe benefits, 2) transition
stress in a new job, 3) relocation costs, 4) termination of personal and family social networks, 5)
loss of valued community services, and 6) disruption to spouse‘s career. When people choose
the exit option, they need to consider these possible negative consequences to themselves.
However, for the other options such as the voice, loyalty, neglect option, people do not need to
consider these possibilities of these negative consequences so long as they do not intend to exit
the organization.
In this same vein, Mobley (1982) suggested that these warnings can help prospective
leavers to make wise decisions about turnover, as well as potentially deterring their exit.
Therefore, although an employee may confront a deteriorating situation at work, a decision to
leave the organization will be considered carefully in light of all possible advantages and
disadvantages. Of course, some may argue that other behavioral options—voice, loyalty, and
neglect—can bring disadvantage to the person who chooses one of these options. For instance,
the neglect behavior may bring a negative evaluation about the person from the person‘s
managers or coworkers. However, the bottom line is that three other options are in common in
that they will not leave, rather, stay the current organization anyhow; therefore they do not need

１４２

to be concerned about possible negative consequences caused by leaving the organization. For
example, people do not need to worry about the possible loss of seniority and fringe benefits as a
result of exercising voice, being loyal, or neglectful behaviors. Therefore, we can conclude that
the exit option is exclusive in that the other three options are for the ‗stayer‘ and entail continuity
of the employment relationship, while the exit option is for the ‗leaver‘ who needs to consider
these possibly drastic and mainly economically negative consequences.
Second, previous research demonstrates that cultural variables are not significant
predictors of the exit option; the latest meta-analysis showed that there are 45 predictors that
have been examined (Griffeth, Hom, & Gaertner, 2000). It revealed that, in a turnover or
withdrawal process, proximal precursors were among the best predictors. The proximal
predictors include job satisfaction, organizational commitment, job search, comparison of
alternatives, withdrawal cognitions, and quit intentions. It was also found that more distal
predictors such as the characteristics of the work environment, stress, work group cohesion,
autonomy, leadership, and (to a lesser extent) distributive justice and promotional chances, have
small effects on the termination process. By this logic, cultural values themselves are possibly
too distal to directly affect employees‘ exit-related behavior. In addition, it is supported by this
analysis that more realistic and economic considerations, such as job alternative quality, strongly
affect employees‘ propensity for the exit option. For instance, alternative quality was found to be
the strongest predictor of the exit option across cultures.
In sum, we can conclude that the exit option is an exclusive behavioral option
differentiated from the other three behavioral options; therefore, ‗leaver‘ does not care culture.

6.2.1.2 For a ‗Stayer,‘ Culture Does Matter

１４３

Aside from the exit option, the other three options--voice, loyalty, and neglect--are
influenced by culture to some degree. In other words, once employees decide to do something as
a way of response to a negative workplace situation without leaving the organization, then
culture effects activate. There are two pieces of evidence to support this finding. First, the
common finding shared by these three options is that they have a significant association with
country; whether or not employees are from certain country makes a difference in their behavior
of voice, loyalty, and neglect options. For instance, the regression outcomes show that in this
study, Korean respondents are more likely to choose voice, loyalty, and neglect options. It can be
interpreted that, when they confront a deteriorating situation at work, employees in Korea are
more likely to choose behaviors allowing them to stay at their current organization than
employees in other countries. It supports Hofstede (1980)‘s view of culture that people from the
same country exhibit similar behaviors, thus, the nationality of the individual is a strong factor
predicting the behaviors of that individual.
Second, in the Korean sample, all three non-exit options were found to have a significant
association with at least three harmony culture values: collectivism, face-saving, and conflictavoiding. The voice option, for instance, has a significant association with all three of the
harmony culture values; the loyalty option has a significant association with collectivism and
conflict-avoiding; and the neglect option has a significant association with conflict-avoiding.
Last, for all three non-exit options, one of the three culture values was found to have the a
stronger association with these options than any other predictors. For instance, the regression
outcomes presented face-saving as the strongest predictor for the voice option, and conflictavoiding for both the loyalty option and the neglect option, meaning that these values have a
stronger power to predict non-exit options than other predictors tested in this study.

１４４

In sum, we can conclude that once employees decide not to leave their current
organization, culture effects activate, influencing their choice of certain behavioral option to
respond to a negative workplace situation.

6.2.2 Research Question TWO: Which Culture Specific Values Affect Employees‘ Voice
Behavior?
As Hofstede (1980) explained, the core elements in culture are values, and people from
the same country would share the same culture composed of values. In other words, cultural
values are the average priorities attributed to different values by societal members, and reflect
the central thrust of their shared enculturation. In this line, for precise understanding of people‘s
behavior, researchers need to extend their attention to the research on cultural values as
predictors of behaviors.
To investigate cultural values affecting employees‘ responses to a deteriorating situation
at work in a non-western culture, specifically an Asian culture in this study, this dissertation
included Kozan‘s (1997) three cultural models because it provides a holistic theoretical
framework for analyzing how culture affects the whole process of t dealing with a negative
situation. Based on his categorization the Asian countries as the examples of the harmony culture,
this study suggested that there are three key values in the harmony culture which guide
members‘ behavior in response to negative workplace situations: the harmony culture values of
collectivism, face-saving, and conflict-avoiding.

6.2.2.1 Collectivism, Face-saving, and Conflict-Avoiding in a Harmony Culture

１４５

As described above, all three staying options were found to have a significant association
with at least three of the harmony culture values: collectivism, face-saving, and conflict-avoiding.
However, these significant associations between culture values and non-exit options were found
only in the Korean sample, not in the U.S. sample. In the U.S. sample, none of the three culture
values had a significant association with any behavioral options.
These findings support that these three culture values influence employees‘ non-exit
behaviors as a response to a deteriorating situation at work, but only in a harmony culture, since
these values were derived from the key values of a harmony culture that Kozan‘s theoretical
framework predicted for categorizing cultures. Therefore, it can be suggested that these three
culture values are culture-specific values, exclusively applicable to harmony culture.

6.2.2.2 The Effects of these Values Vary Depending on Each Response Option.
Although a ‗stayer‘ is influenced by the three culture values, the magnitude of the effects
of these values varies in each behavioral option. For instance, collectivism and face-saving have
a significant and positive association with the voice option, while conflict-avoiding has a
significant and negative association with the voice option. For the loyalty option, collectivism
has a significant and positive association with the loyalty option and conflict-avoiding has a
significant but negative association with the loyalty option. For the neglect option, conflictavoiding has a significant and positive association with the neglect option.
As described above, in the Korean sample, out of the three non-exit options, only the
voice option has a significant association with all three culture values, the loyalty option is
statistically influence by two of the three culture values, and the neglect option is affected by
only one of them. Based on these findings, we can conclude that, although all three behavioral

１４６

options other than the exit option are influenced by culture values, the degree of effects of
culture values varies depending on the behavioral option: the voice option is most heavily
influenced by the culture values and the neglect option is the least affected by the culture values.

6.2.2.3 Among Them, Conflict-Avoiding is the Most Widely Influencing Predictor in the
Harmony Culture
Among the three culture values, only conflict avoiding has a significant association with
all three non-exit options. For instance, collectivism is significantly related to the voice and the
neglect options, while face-saving is significantly related to the voice option only. However,
conflict-avoiding is associated with all three non-exit options. Furthermore, conflict-avoiding is
the second strongest predictor of the neglect and loyalty option, following the country variable. It
can be interpreted that conflict-avoiding is the most widely influencing predictor when
employees respond to their deteriorating at work, so long as they do not consider ‗leaving the
organization they belong to‘.
This is possibly due to the nature of conflict-avoiding itself. Basically, the concept of
conflict-avoiding came from research on conflict management, characterized by withdrawing
from the situation or suppressing the urge to engage in conflict (Rollinson & Broadfield, 1998).
Compared with two other culture values, which are general values applicable to other situations
in a culture, it can be interpreted as situation-specific, especially problematic situation-specific
value as the proxy of the attitude referring in a deteriorating situation at work. Due to this trait of
the concept of conflict-avoiding, it will be considered widely as a reference to decide certain type
of behavior as a response to a deteriorating situation at work, than other values.

１４７

6.2.3 However, There Are Still Universal Predictors Available Across Cultures
So far, most attention has been directed toward explaining the effects of culture on the
EVLN construct. As discussed, some notable effects of culture on employee voice behavior were
observed in various ways. However, this does not mean culture specific factors are more
important than other factors. Rather, this study is intended to contribute to a further
understanding of employees‘ responses to a deteriorating situation at work by examining underdeveloped factors, such as, culture. For that reason, four non-culture predictors which have been
supported empirically in the previous studies were added in this study as independent variables.
As hypothesized, some, but not all predictors were found to be statistically significant
predictors for certain behavioral options. For instance, as hypothesized, alternative quality was
found to have a significant and positive association with the exit option both in the Korean and
U.S. samples. LMX was also found to have a significant and positive association with the voice
option across countries, which supports this study‘s hypothesis that LMX will have a positive
association with the voice option.
Overall, as proposed in this study‘s model, the findings in this study support that, along
with cultural factors, there are universal predictors affecting behavioral choice.

6.2.4 Dynamics within the EVLN construct with Constructive vs. Destructive Dimension.
On the whole, this study provides further suggestions about predictors of employee
response to negative workplace situations, mainly categorized as 1) culture-specific factors such
as collectivism, face-saving, and conflict–avoiding in a harmony culture and 2) universal factors
such as alternative quality and LMX. However, this is not all; in further steps, the results of data
analysis in this study tell us more about the dynamics of these responses themselves: the zero-

１４８

order correlation matrix provides explanation on the construct of these four classes of responses,
suggesting that these four types of responses would differ from each other or correlate each other.
For instance, according to the Table 6 Correlation Matrix, 1) the exit option is related to the
neglect option positively, and therefore can be categorized as one dimension of four behavioral
options; 2) the voice option is related to the loyalty option positively, thus they can be combined
as another dimension of four behavioral options; and 3) the first categorized option and the
second categorized option have negative association with each other; this means that these two
categories can be interpreted as the opposite option for each other.
Interestingly, these findings are consistent with previous studies on the construct of these
four response options with two dimensions, constructiveness vs. destructiveness and activeness
vs. passiveness, which have been supported by empirical evidence (Rusbult, et. al., 1982; Farrell,
1983; Rusbult & Lowery, 1985). In terms of the consequences of these responses on employeeorganization relationships and immediate sources of dissatisfaction, voice and loyalty are
categorized as constructive responses in which an individual attempts to revive or maintain
satisfactory employment conditions and employee-organization relationships, whereas exit and
neglect are more destructive. In addition, in terms of the impact of an action on a problem, exit
and voice are categorized as active mechanisms through which employees attempt to deal with
dissatisfaction, whereas loyalty and neglect are more passive and diffuse.
Overall, based on the findings from the zero-order correlation matrix, as well as the
dimension that previous EVLN researchers have suggested with empirical support, we can
conclude that these four response options are not totally independent from each other, rather, one
option may imply negative or positive propensity for another option. For instance, the exit option
has a significant and negative correlation with the loyalty option and significant and positive

１４９

correlation with the neglect option. Based on this finding, we can assume that someone who has
a strong exit intention may not have loyalty to the organization and may have a strong tendency
of neglectful behaviors. In this case, the person‘s observable neglectful behavior may be
interpreted as a precursor to leaving. Therefore, if the organization wants to retain that person,
first, directly the organization will provide the person with better benefits or working condition
to make the person‘s current job competitive with other alternatives. However, if that option is
not available, for example, if the organization cannot afford it, the organization can retain the
person by encouraging the person‘s loyalty and discouraging neglectful behaviors. This means
that alternative interpretation of the dynamic existing within the four responses with extend our
understanding of these four response options. Particularly considering greater attention from the
organizational sciences towards these types of destructive employee reactions (e.g. Griffin et al.,
1998 a, 1998b; O‘Leary-Kelly et al., 1996; Van Yperen, et. al., 2000), this alternative way of
interpreting dynamic within the EVLN construct provides the possibility of indirect effects of
factors tested in this study.

6.3 Results of Hypotheses Testing
Based on the summary of data analysis outcomes, Table 17 summarizes the results of
hypotheses testing of the total sample, the Korean sample, and the U.S. sample, respectively.
Regarding harmony culture values, it has been shown that collectivism has a positive
association with the loyalty option and the voice option; it has not been supported that in the
harmony culture, collectivism will have a negative association with the exit option and the
loyalty option. For face-saving, only positive association with the voice option in the harmony

１５０

culture was supported and associations with the other three options were rejected. The
associations with conflict-avoiding and non-exit options in the harmony culture were supported.
Table 17. Summary of Hypotheses Test
Supported
KOR US
H
1

H
2

H
3

H
4

H
5

H
6

H
7

In the harmony culture, collectivism will have
a positive association with loyalty.
a positive association with voice.
a negative association with exit.
a negative association with neglect.
In the harmony culture, face saving will have
a positive association with loyalty.
a positive association with voice.
a negative association with exit.
a negative association with neglect.
In the harmony culture, conflict-avoiding will have
a positive association with loyalty.
a negative association with voice.
a negative association with exit.
a positive association with neglect.
Alternative quality will have a negative association with loyalty.
positive association with voice.
positive association with exit.
negative association with neglect.
Self-esteem will have a positive association with voice.
a positive association with loyalty.
a negative association with neglect.
a negative association with exit.
Power distance will have a negative association with voice.
a positive association with loyalty.
a positive association with neglect.
a positive association with exit.
LMX will have a positive association with voice.
positive association with loyalty.
negative association with neglect.
negative association with exit.

O
O
X
X
X
O
X
X
O
O
X
O
O

X
X
O
X
X

O

X
X

O
O
X
X

X
X
O
O
X
O
X
X
X

Among hypotheses about the four other predictors, as far as alternative quality, the
positive association with the exit option was supported and the positive association with the
voice option and negative association with the neglect option were not supported; the negative
１５１

association with the loyalty option was supported only in the Korean sample, not in the U.S.
sample. For self-esteem, the positive association with the voice option and negative association
with the neglect option were rejected; and the positive association with the loyalty option and
negative association with the exit option were supported only in the Korean sample, not in the
U.S. sample. For power distance, positive association with the exit option was rejected, but
positive association with the loyalty option and the neglect option were supported only in the U.S.
sample, while negative association with the voice option was supported only in the Korean
sample. Finally, only positive association between LMX and the voice option was supported;
while positive association with the loyalty option and negative association with the neglect
option and exit option were rejected.

１５２

CHAPTER 7
CONCLUSIONS

This chapter provides a brief overview of this study and its conclusions, including (1)
Summary of the Dissertation, (2) Theoretical, Methodological, and Practical Contributions, (3)
Limitations and Future Research, and (4) Concluding Remarks.

7.1 Summary of the Dissertation
This dissertation has made significant contributions, both theoretically and practically.
Before the conclusion and the contribution analysis, a summary of the whole study will be
presented.

7.1.1 Summary of the Purpose of the Study and the Theoretical Background
The key purpose of this paper is to investigate how culture affects employees‘ voice
behavior at work. Although numerous studies on employee voice across disciplines have
appeared with new facts and insightful theories, fewer studies have been performed on how
culture affect employees‘ voice studies, particularly in the non-Western context. Considering the
remarkable increase of multi-cultural context in the field of management, culture study of
employee voice has been suggested to be one of the most deserving areas for research toward a
precise understanding of employees‘ voice behavior. In other word, there was a need to test why
and how culture makes a difference in employee voices behavior at work. This was the original
idea of this dissertation.

１５３

More specifically, this study attempted to answer the following questions.
(1) Does culture affect employee response to negative workplace situations?
(2) If so, how does culture affect employee response to negative workplace
situations? In other words, which culture-specific values affect employee
response to negative workplace situations? What key features of a non-western
culture directly make a difference in the application of existing voice-related
models?

To answer these questions, key features of voice literature were reviewed, focusing on
the main concepts, constructs, and predictors of employees‘ voice. In addition, culture research,
especially in employees‘ voice research, was reviewed, focusing on how researchers view and
interpret culture, how they conduct empirical studies on culture, and particularly how they
measure culture. Then, this dissertation proposed a comprehensive theoretical model, which was
built on two theoretical grounds—extended version of Hirschman‘s (1970) exit-voice theory, that
is the EVLN construct developed by Rusbult and colleague (1982) and Kozan‘s (1997) culture
model.
Basically, this proposed model attempts to adapt Kozan‘s culture model to an extended
version of Hirschman‘s exit-voice model, which is the EVLN construct. Then, to explore the
cultural factors to be tested, Kozan‘s culture model was used to suggest key cultural values in a
harmony culture, which will be applied particularly to the Korea culture in this dissertation.
Using Kozan‘s culture model, this dissertation suggests that in a harmony culture, such as an
Asian country like Korea, which places a high value on collectivism, face-saving, and conflictavoiding, people will be guided by these key values when they respond to a negative situation at

１５４

work. Therefore, these three cultural values, that, to my knowledge, have never been tested as the
predictor of employee voice behavior, were proposed as predictors of employee voice behavior
in this dissertation. Besides these, four other voice predictors, which had been supported in the
previous empirical studies—including alternative quality, self-esteem, power distance, and LMX
quality—were also added to the proposed model to examine if they would affect EVLN
behaviors as well.

Corresponding to the model, the following hypotheses were tested:
HP 1. In the harmony culture, while collectivism will have a positive association with voice and
loyalty, it will have a negative association with exit and neglect.
HP 2. In the harmony culture, while face-saving will have a positive association with loyalty and
voice, it will have a negative association with exit and neglect.
HP 3. In the harmony culture, while conflict-avoiding will have a positive association with
loyalty and neglect, it will have a negative association with exit and voice.
HP 4. While alternative quality will have a positive association with exit and voice, it will have a
negative association with loyalty and neglect across cultures.
HP 5. While self-esteem will have a positive association with voice and loyalty, it will have a
negative association with neglect and exit across cultures.
HP 6. While power distance orientation will have a positive association with exit, loyalty, and
neglect, it will have a negative association with voice across cultures.
HP 7. While LMX will have a positive association with voice and loyalty, it will have a negative
association with exit and neglect across cultures.

１５５

7.1.2 Summary of Methodology
In order to test the proposed hypotheses, a field study was conducted in Korea and in the
U.S.—Data was collected from the facilities both in Korea (n = 415) and in the U.S. (n = 95). In
particular, to emphasize cultural effects other than industrial or organizational level factors, data
was collected only in the facilities associated with the Korean automotive industry that employed
local employees. Questionnaires in subjects‘ native languages were administered.
For data analysis, two statistical methods were performed for the analyses of the
proposed model: confirmatory factor analysis and multiple regressions, especially stepwise
regression. First, confirmatory factor analysis was conducted with AMOS 19, to test the
construct validity of three components of the harmony culture, in other words, to test if the items
used in this dissertation measure the values of collectivism, face-saving, and conflict-avoiding in
a harmony culture.
Then, multiple regression, and more specifically, stepwise regression was conducted with
SPSS 19 to examine the relationships between each dependent variable and multiple predictor
variables. The dependent variables in the analysis included exit, voice, loyalty, and neglect
options. Each dependent variable was regressed separately upon the following independent
variables: collectivism, face-saving, conflict-avoiding, alternative quality, self-esteem, power
distance, and LMX. Several control variables that showed statistically significant relationships
with dependent variables (such as nationality, education, tenure, and job category) were also
added in this regression analysis.
Furthermore, in order to assess the contribution of these control variables, other factor
variables, and the three cultural variables, respectively, a series of step-wise regressions was
performed. In detail, control variables that showed statistically significant relationships with

１５６

dependent variables (such as nationality, education, tenure, and job category) were entered first
(Model 1), followed by other independent variables (alternative quality, self-esteem, power
distance, and LMX) (Model 2), then the three main harmony culture variables (such as
collectivism, face-saving, conflict-avoiding) (Model 3). The change in R ² was then assessed. In
addition, these stepwise regressions were conducted with 1) the total sample, 2) the Korean
sample, and 3) the U.S. sample, respectively, to see if any difference of regression outcomes
were influenced by the sample composition. In addition, since data was collected in two
organizations for each country sample, dummy organization variable was added as a control
variable to see if there was an organization effect within the country sample. Last, to examine if
coefficients in linear regressions models are equal between the Korean sample and the U.S.
sample, the Chow test was conducted.

7.1.3 Summary of Results
Utilizing the purified measurement model, which worked well in both the Korean dataset
and the American dataset, the structural model fit was tested for EVLN options. As a result,
several encouraging results were obtained. Overall, the AMOS results confirmed the proposed
model, which supported the significant influence of culture on employees‘ response to
deteriorating conditions at work. In addition, regression analysis provided detailed information
on how culture factors affect employees‘ response to a negative situation at work.
First of all, regardless of the nationality of the sample, none of the three harmony cultural
values predicted exit option. As discussed earlier, this is possibly due to the unique
characteristics of the exit option, rather than those of the three non-exit options. As turnover
researchers have suggested, since turnover can bring huge economic loss to a ‗leaver,‘ for

１５７

example loss of seniority and fringe benefits, people consider primarily economic calculation
such as availability of alternative, rather than distal predictors such as cultural values when
making such a decision. The finding that alternative quality is the strongest predictor of the exit
option across samples supports the importance of economic consideration in the exit option.
Second, in the Korean sample, harmony cultural values were found to be strong
predictors of all three other non-exit options—voice, loyalty, and neglect. In the Korean sample,
face-saving had a significant and positive relationship with the voice option, whereas conflictavoiding had a significant and negative relationship with the voice option. In addition,
collectivism had a significant and negative relationship with the loyalty option, whereas conflictavoiding had a significant but positive relationship with the loyalty option. In addition, facesaving and conflict-avoiding were found to have a significant relationship with the neglect option.
Based on these findings, we can conclude that, as hypothesized in this dissertation, in the
harmony culture, these three values guide and predict peoples‘ behavior, and are strong
predictors of voice behavior within the culture.
Third, contrary to the findings in the Korean sample, none of the three harmony cultural
values were found to have a significant relationship with any of the EVLN options in the U.S.
sample. This means that these three harmony cultural values do not affect employee response to
negative workplace situations in the U.S. sample. Therefore, it confirms that these three cultural
values are unique in the harmony culture, not available in other cultures, such as the U.S. Since
the U.S. is categorized as a confrontational culture by Kozan (1997), the findings in the U.S.
sample lead to the conclusion that the three harmony culture values are the unique values
differentiating a harmony culture from other cultures.

１５８

Finally, among other predictor variables from previous voice research, only LMX was
found to have a significant and positive relationship with the voice option consistently in both
the Korean sample and the U.S. sample. In particular, both in the Korean sample and the U.S.
sample, LMX was found to be the second strongest predictor of the voice option, following facesaving in the Korean sample and power distance in the U.S. sample. Therefore, we can conclude
that, despite whether other predictor variables tested in this dissertation were supported partially
or not at all, only LMX is a universally significant predictor of the voice option regardless of the
nationality of respondents.

7.2 Contributions
According to the results obtained and the conclusions made above, this dissertation has
made significant theoretical and practical contributions. Each of contributions is further
discussed below.

7.2.1 Theoretical Contributions
First of all, this study made significant contributions to the expansion of our
understanding of employees‘ voice behavior by addressing one under-developed but promising
subject that had been suggested by previous employee voice researchers for future research: that
is cross-cultural study of employee voice behavior. For instance, in their recent work, Edwards
and Greenberg (2009) suggested that employee voice is still a significant topic in work-related
research; and cross-cultural difference should be explored further in employee voice research.
Considering consistent findings that culture influences a range of employee and organizational
outcomes, they discussed the significant gap in the voice literature that begs to be addressed by

１５９

future research. In this regard, this study made an attempt to fill the gap by testing the possible
impact of ‗culture‘ on employees‘ voice behavior, with a direct and simple question: ‗Does
culture affect employees‘ voice behaviors?‘
Second, rather than simply showing the difference in employees‘ voice behavior at work
between two countries, this study further answered an in-depth critical question why and how
employees‘ voice behaviors vary across cultures—based on theoretical rationale, by illustrating
the key values in a certain culture. As the researchers suggested, cross-cultural study does not
simply mean showing that country A is different from country B. Addressing cross-cultural study
in employees‘ voice behavior, researchers provided several detailed questions that appear to be
especially interesting for cross-cultural study in employee voice behavior. For example, do
people‘s motives for voice behavior differ across culture? How do various dimensions of
national culture, such as those identified by Hofstede (1980), affect how these behaviors are
perceived, interpreted and used by employees and managers? What are the individual and
organization-level benefits and consequences of employee voice behavior in various cultures? To
what extent does culture influence the tendency of employees to respond to acts of deviance and
ethical breaches by speaking up or being silent?
Although this empirical study does not intend to answer all of these questions, the results
provided several interesting implications for further understanding of employee voice behavior:
It revealed that employees‘ voice behaviors are predicted by different factors depending on
culture. Therefore, some factors which are critical predictors of people‘s behavior in one culture
are not necessarily critical in other culture. For instance, this study showed that in a harmony
culture, Korea in this study, the values of face-saving and conflict-avoiding were critical
predictors of voice behavior, however not in the U.S. sample at all. In addition, the values of

１６０

collectivism and conflict-avoiding were key motives of the loyalty and neglect options in a
harmony culture, Korea in this study, whereas they were found to have no significant impact on
these behaviors in another culture, the U.S. Therefore, rather than simply showing that people in
Korea are different from people in the U.S., this study answered the further question, ‗Which
factor made these differences?‘ In other words, by offering new predictors of employee voice
behavior, which have never been tested in previous studies, this study made significant
contributions to further understanding of employee voice behavior, especially in other cultures.
Third, this study did not find only culture-specific motives for employees‘ voice behavior.
This study also found that, in spite of these culture-specific values, there are universal factors
that people consider across culture when they decide how to respond to a deteriorating situation
at work. For instance, this study revealed that regardless of the nationality of respondents, none
of the three harmony culture values influenced their choosing of the exit option. In addition, this
study found that across culture, LMX and power distance are strong predictors of the voice
option, which is consistent with the findings of previous studies with various samples. These
findings contribute to our understanding that although there are certain culture-specific factors,
there are universal factors influencing employee voice behavior across culture, as well and point
out that for more precise understanding of influencing employee voice behavior, researchers
need to deal with both culture-specific factors and universal factors at the same time.
The research also has theoretical implications for how we should view culture in our
research. Generally speaking, there are two polar opposite views about culture in the
management research field. One extreme is that some researchers exaggerate the importance of
culture, using it to explain almost everything that involves more than one country. The other
extreme is that researchers ignore the impact of culture on management completely, by saying

１６１

that ―business is business and people are people everywhere.‖ However, the findings of this
study tell us that ―both views are right and wrong‖: this study provided empirical evidence of the
uniqueness of a culture as well as the universality of other factors according to sample
phenomena in two countries. Therefore, it must be acknowledged that to understand certain
phenomena of employees in a certain culture, researcher need to closely look at both sides of the
phenomena: generalities of the phenomena caused by universal factors shown in other cultures as
well as uniqueness of the phenomena caused by specific factors embedded in that specific culture.

7.2.2 Methodological Contributions
This study also made significant contributions to cross-cultural research in employees‘
voice behaviors, particularly in the methodology. As discussed earlier, much empirical research
for different countries has been criticized for its limited of methodology. The most common
criticism is that they rely on convenience sampling. Especially in cross-cultural research, data is
often collected by conducting surveys with international students in one or a couple of
universities in a country because of the convenience for the researchers. This dissertation
avoided these shortcomings by pursuing a purposive sample, keeping a balance between
theoretical grounds and practical grounds, as well as by following suggestions for sampling
presented by researchers. Based on this guidance, data was collected from local employees of the
Korean automotive industry facilities in Korea and in the U.S.
First of all, in planning data collecting, this dissertation designed a sampling strategy
based on Lyons and Chryssochoou‘s (2000) guidance on sampling for cross-cultural research.
They argue that although sampling is at the core of data collection, many researchers frequently
select particular cultures because they are familiar with these cultures or they know somebody in

１６２

that particular country. Arguing the limitations of this convenience sampling, they suggested that
purposive sampling is more appropriate. They suggest that 1) selection of subject countries
should be based on theoretical grounds; 2) in selecting the subgroups within each culture, the
researcher should try to match the subgroup to the research purpose across the selected cultures
in the first stage of the sample.
Accordingly, in this dissertation, Korea and the U.S. were selected based on Kozan ‗s
(1997) categorization of culture model. Kozan stated that the Asian countries are examples of the
harmony model. Therefore, this dissertation selected Korea as a main culture group which is
highly compatible with the Kozan harmony model. In addition, in this study the U.S. was
selected as an example of a non-harmony model since Kozan (1997) categorized the U.S. as an
example of a confrontational culture.
Then, employees in Korea and in the U.S. were selected as subgroups, since this
dissertation focuses on certain employee behaviors at work. In detail, employees, not students
were selected as subjects, despite the fact that previous cross-cultural research in management
has often used student samples. As researchers argued (e.g. Miceli & Near, 2005) using college
student samples is not appropriate to examining people‘ voice behavior at work, given the highly
contextual nature and the complex decision-making processes involved in employee behavior. In
particular, considering that the way employees handle negative or problematic situations at work
is heavily influenced by various contextual factors—such as the organization‘s climate, including
policies, relationships with supervisor, unionization setting—it is questionable that a student
sample could capture many of the features that employees show in decision of certain behavior at
work. Accordingly, this dissertation used an employee sample.

１６３

Furthermore, the subjects were restricted to local employees in each country. Therefore,
the American employees working in Korea, Korean employees working at the U.S. sites, and
Korean Americans were not considered as subjects in this dissertation, due to the powerful
process of culturalization. In the process of culturalization, people are continuously losing their
sensitivity toward their native cultures and their responses can wash out the magnitude of culture
effects. For that reason, cross-cultural research will become much stronger when subjects are
recruited from their native cultures. The subjects were restricted to local employees at the
Korean facilities in the manufacturing industry, in particular the automotive industry, in the U.S.
and Korea, to make the samples as compatible with each other as possible.
Overall, this study attempted to overcome the limitations of methodology pervasive in
previous research of cross-cultural effects on voice by 1) sampling from real employees 2) from
the same company, or at least the same industry, in both the U.S. and Korea. By doing so, the
sampling strategy of this dissertation yielded theoretical grounds, providing more plausible and
realistic evidences representing the effects of real-world settings at the same time to see the
cultural differences in the employee response to negative workplace situations.

7.2.3 Empirical Contributions
As expected, this study made significant contributions to cross-cultural research in
employee voice behavior by providing strong empirical evidences about how employees respond
to negative workplace situations across culture. With an employee sample from the automotive
manufacturing industry in Korea and in the U.S., the findings of this study comprehensively
show whether or not there is a difference of the effects of cultural, individual, and organizational
factors between in the Korean sample and the U.S. sample.

１６４

First of all, Chow test results confirm that the regression outcomes on EVLN options are
statistically different between the Korean sample and the U.S. sample; this means the way that
employees respond to their negative workplace situation is different between the Korean sample
and the U.S. sample. Therefore, we can conclude that in Korea, the EVLN construct is exercised
by different mechanism than it is in the U.S. In addition, the results of regressions on these
options show that there is no organization effect within the country. It means that the
organization variable does not make any statistically significant variation within the country.
Second, as expected, key differences of regression model on the EVLN construct were
found between the Korean sample and the U.S. sample. In the Korean sample, except for the exit
option, the three harmony culture values, collectivism, face-saving, and conflict, are associated
with three behavioral options: voice, loyalty, and neglect. In other words, all three behavioral
options are significantly associated with at least one culture value in the Korean sample. For
instance, the voice option has a significant association with all three harmony culture values; the
loyalty option is influenced by two culture values: collectivism and conflict-avoiding; and the
neglect option only has a significant relationship with one of the harmony culture values:
conflict-avoiding. Therefore, we can conclude that, although all three behavioral options other
than the exit option are influenced by culture, the degree of effects of culture varies; the voice
option is most heavily influenced by the culture values and the neglect option is the least affected
by the harmony culture values. On the contrary, none of the three harmony culture values are
associated with all of the EVLN options in the U.S. sample, therefore, the findings of this study
support that the three harmony culture values represent distinctive traits of a harmony culture.
Last, despite these differences, some interesting common outcomes between the two
countries were found. For instance, regardless of the country, none of the three culture values

１６５

was found to have a significant association with the exit option. This shows that the exit option is
not influenced by culture at all. As discussed earlier, this is possibly due to the inherent nature of
the exit option as differentiated from the other three options; only the exit option provides alerts
to possible disadvantages to ‗leaving the organization.‘ Therefore, rather than distal factors such
as personal beliefs or values, proximal predictors, such as availability of alternatives to the
current job, directly influence an employees‘ exit options. In this line, the significant effect of
alternative quality on the exit option both in the Korean sample and the U.S. sample supports
employees‘ realistic consideration for the exit option.
In addition, as hypothesized, alternative quality was found to have a significant and
positive association with the exit option both in the Korean and U.S. samples. LMX was also
found to have a significant and positive association with the voice option across countries, which
supports this study‘s hypothesis that LMX will have a positive association with the voice option.
Therefore, as proposed in this study‘s model, the findings in this study support that, as well as
cultural factors, there are universal predictors affecting certain behavioral option across cultures.
Overall, the empirical evidence of this study supports the difference of significant factors
on the EVLN in Korea and in the U.S., as well as commonality of the factors at the same time.
By providing both culture-specific and universal factors of the EVLN construct, this study
contributes a more comprehensive and precise understanding of employees‘ response to negative
workplace situations across cultures.

7.2.4 Practical Contributions
This study also provides practical implications for practitioners, mainly by providing
empirical evidences of the importance of cultural differences, especially in the increasing multi-

１６６

cultural setting of business. Over time, organizations are getting more exposed to multi-cultural
setting as a result of the changing business environment, characterized by increasing
globalization and international business. Many companies attempted to enter into markets abroad
and poured their resources aboard; they move their facilities abroad or build new facilities or
business units in other countries where they are not familiar before. In this process, it may be
relatively easy to manage tangible resources such as buying land, or building new plants.
However, the real obstacles for many organizations are often caused by lack of
understanding of other cultures, since understanding another culture does not simply mean
learning their language or the related regulations, or analyzing their market. As Hofstede (1999)
illustrated, management is always about people, the essence of management in organization is
dealing with people and all their daily issues at the lowest possible level within the organization.
Since people‘s behaviors in a society are affected by the values governing part of the collective
programming of people's minds in that society, management is subject to cultural values of the
society in which it takes place.
First and foremost, this study showed that employees‘ cognitions and behaviors operate
differently in different cultures, influenced by different factors valued in their country. This
study showed that three harmony culture values—collectivism, face-saving, and conflictavoiding—are critical predictors of employees‘ responses to their deteriorating situation in the
workplace, as long as they do not leave, since exit decision is not influenced by these cultural
values. At the same time, this study showed that these three values do not predict any employee
behaviors in the U.S. This differing result after predicting the same behavior between Korea and
the U.S. tells us that for successful business abroad, organizations need to open their minds,
carefully investigate foreign culture, and adjust their strategies, especially HRM strategies. They

１６７

need to acknowledge that assuming ―business is business everywhere‖ and ―employee is the
same everywhere‖ is very dangerous. The results of this study revealed that this view is flawed
in that it fails to recognize that fundamental cultural differences exist among people around the
world.
Second, this study also provides organizations with suggestions on an operational level,
because this study specifically tested some of the relationships among cultural factors and
employee behaviors, in this case response to deteriorating situation at work. In a global business
setting, when working with local employees in other cultures, the findings from this study
suggest applicable lessons for organizations, especially how to promote employee voice and to
facilitate communication with employees. For instance, by developing the related policies such
as new communication policies and training for both local and expatriate employees,
organizations will be able to enhance their understanding about cultural differences. In particular,
as discussed earlier, effective communication is essential for the healthy functioning of an
organization, contributing to organizational innovation (Ryan & Oestreich, 1991), organizational
learning (Argyris, 1976), team processes (Edmondson, 2003), perceptions of fair treatment
(Thibaut & Walker, 1975), employee well-being (Cortina & Magley, 2003), creating and
maintaining an ethical workplace (Near & Miceli, 1985), extra-role work behavior (Van Dyne &
LePine, 1998), and job performance (LePine & Van Dyne, 2001). For this reason, facilitating
employee voice behavior is key issue for an organization.
However, suppose an organization transplants same employee voice practices that
worked successfully in their home country to new facilities in other countries. Or suppose that
the organization ignores some voice practices commonly used in other countries simply because
these practices do not fit their organizational policies or they have experienced success without

１６８

these practices in their home country. They may not be able to understand why their local
employees leave their organization or why turnover is high in the local facilities, or what does
silence of local employees mean exactly. Even worse, they may confront unexpected conflicts
with local employees. For instance, this study revealed that face-saving had a positive
relationship with the voice option in Korean respondents. This means that in Korea, employees
are more likely to exercise their voice when they believe that their face-saving interest is
protected. Therefore, they may not want to speak out in front of other since they may be
concerned that their face will be hurt by others who observe this behavior. For that reason, they
may prefer not to be identified, expressing their opinion anonymously or indirectly— such as
with a hot line or appeal to union—rather than openly speaking-upin a department meeting or a
lunch with the CEO. In this way, this research may help develop employee voice practices that
are customized for employees in their culture.
The last contribution of this research to practice is that, as well as exploring cultural
difference in employees‘ voice behavior, this study reconfirmed the universally important role of
supervisors for their subordinate‘s voice behaviors: previous voice research suggested that
organizational factors mainly deal with an organizational mechanism that is conductive to
employee voice provided by the organization, both formally–for instance, providing grievance
procedures, suggestion boxes, open door policies, and lunch with the CEO--as well as
informally–for instance, talking to supervisors on a causal basis (Gorden et al., 1988; Kassing,
2000; Landau, 2009). In particular, researchers have emphasized the importance of supervisors
in employees‘ voice (Botero & Van Dyne, 2009; Brown et al., 2005; Chia et al., 2000; Detert &
Burris, 2007; Janssen et al., 1998; Kassing, 2000; Landau, 2009; Saunders et al., 1992; Van
Dyne et al., 2008; Walumbwa & Schaubroeck, 2009). Consistent with the findings of previous

１６９

research, this study also demonstrated that the LMX quality is a strong predictor of voice
behavior, both in Korea and in the U.S. This means that, across cultures, LMX, referred to as the
quality of supervisor–subordinate relationships, is an important contextual climate affecting
employees‘ voice behaviors by providing employees more opportunities to speak up, exchange
information or ideas with their supervisors, and use more communication channels. This implies
that training supervisors in good voice managing will promote communication within the
organization.
In addition, this study found that self esteem is a universal factor preventing employee
exit. Both in the Korean sample and the U.S. sample, self esteem was found to have a significant
and negative relationship with the exit option. The importance of self-esteem in employee
response to negative workplace situations provides organization with practical implications on
how to retain their employees. According to the findings of this study, in deteriorating situations
in the workplace, employees will be more likely to consider leaving when they have low selfesteem, in other words, when they think that they are not valued in the organization. The results
of this study show that such a negative belief will discourage them from acting positively within
the organization, instead, it lead to them leaving the organization to find an alternative
organization that would make them feel capable, significant, successful, and worthy.
In particular, considering that turnover is not desirable behavior, causing cost to an
organization or causing a struggle to retain their valued employee, this study suggests one
practical implication for how to reduce employee exit. That is to develop HRM practice,
enhancing the employees‘ self-worth in the organization. For instance, the compensation or
reward programs and actively praising employees, such as ‗employee of the month or of the
year,‘ will be able to make employees believe that they are successful and valuable in the

１７０

organization. HR practices promoting employees‘ involvement in the decision-making process
will also be helpful to raise employees‘ self-worth and belief that they are capable, significant,
and worthy. These practices will encourage employees to take risks and exercise their behavior
with confidence so that they can contribute to a meaningful change at work and will discourage
consideration of the exit option, when handling a deteriorating situation in the workplace.

7.3 Limitations and Future Research
To my knowledge, this is the first study to consider three cultural values—collectivism,
face-saving, and conflict-avoiding—differentiating harmony culture from other cultures, as
predictors of voice behavior, particularly with employee samples in Korea and in the U.S. Thus,
a key strength of this research is its unique focus on specific substantive relationships in two
different cultural settings. Although the findings of this study represent a significant
advancement in the systematic examination of employee voice behavior, especially by providing
new strong predictors of employee voice behavior, the results must be interpreted with a degree
of caution, as methodological limitations qualify the inferences which can be derived. There are
a number of limitations which need to be pointed out; hopefully, they will inspire more ideas on
employee voice study and lead to more fruitful cross-cultural research in the future.
The first limitation to be overcome in future research is about the generalizability of this
study. This study is the first study to suggest three key harmony culture values such as
collectivism, face-saving, and conflict-avoiding as predictors of employee voice behavior. In this
dissertation, these three harmony culture values are treated as significant predictors guiding
employee voice behavior, particularly in a harmony culture. The regression analysis results show
that these three harmony culture values affect the way that employee respond to negative

１７１

workplace situations in the Korean sample, which is categorized as a harmony culture country by
Kozan (1980), but not in the U.S. sample, which is categorized as confrontational culture by
Kozan (1980). However, since data was only collected from two countries, Korea and the U.S.,
the results will have much stronger generalizability if data from more cultures is collected.
Therefore, for future research, it will be important to see whether results are unique to our
samples or if they generalize to other samples, especially with another Asian country sample,
such as China, Japan, or Singapore, since Kozan (1980) categorized the Asian countries as
examples of harmony culture.
The second limitation of this study is that only direct correlations between the harmony
culture values and the employee voice behaviors were tested, since this is the very initial study
examining these cultural values as predictors of employee voice behaviors. It may be argued that
this model is too simple to capture the complicated mechanism existing between these predictors
and the outcomes.
In this vein, it can be argued that this study does not take the interaction effects of
predictors on employee voice behavior into account. However, as prior studies have shown with
their predictors, future research can investigate the possible interaction effects of employee voice
predictors. For instance, Botero and Van Dyne (2009) empirically tested LMX and PD as
predictors of employee voice behavior. In addition, they tested an interaction between LMX and
PD to predict employee voice behavior, saying that PD makes more of a difference in voice
when employees have high-quality LMX relationships with their supervisors. They revealed this
interaction effect, based on trait activation theory. The key notion of trait activation theory is that
individual characteristics (traits) make a difference in behavior only when they are theoretically
relevant to specific outcomes that are being considered. Consistent with this theory, they

１７２

demonstrated that individual differences, such as personality (or traits such as the cultural value
of PD in their study), are salient in some situations and not salient in other situations; High LMX
allows employees to feel they have the choice to act in accordance with their fundamental beliefs
about appropriate roles and behaviors when interacting with someone in a position of power,
such as their supervisor.
Similarly, the question can be raised that individual‘s cultural value can affect the
relationship with other predictors and voice behavior. For instance, it can be argued that an
individual‘s collectivism orientation affects the relationship between LMX and voice behavior.
Since a key feature of collectivism is putting a high value on ‗relationships‘ between oneself and
others, there may be more influence from LMX to choose voice behavior when there is a cultural
trait of collectivism. In this line, the interaction effect between collectivism and LMX to predict
voice behavior is an issue to be tested by future research.
Furthermore, a more thorough explanation of mechanism existing between the predictors
and outcomes tested in this study will be possible by adding other predictors not tested in this
study. In other words, it can be argued that these relationships may be affected by other factors
which have not been taken into consideration in this dissertation. The most likely argument is
that organizational factors influence these relationships. As described in sampling strategies, in
this dissertation, data was collected only in the facilities of the Korean automotive industry in
Korea and in the U.S. to investigate if respondents respond to negative workplace situations
differently between the Korean sample and the U.S. sample, mainly focusing on the possible
impacts of cultural factors other than industrial or organizational level factors. Therefore, it is
possible that some industrial or organizational level factors affect the relationship between the
three harmony cultural values and employees response to negative workplace situations.

１７３

As discussed earlier, organizational factors such as characteristics of the organizations,
practices of the organizations, and leadership style are important determinants of people‘s
thoughts and behaviors. Therefore, researchers have emphasized the importance of
organizational factors in employee voice behavior (Rao, Landau & Chia, 2001; Gorden et al.,
1988; Kassing, 2000; Peirce et al., 1989). In this vein, some researchers provided empirical
evidence on organizational factors affecting employee voice behavior. For instance, Morrison
and Milliken (2000) described that organizational climate which leads employees decision
whether to speak includes 1) patterns of organizational policies and structures, such as open-door
policies, grievance systems, regular team meetings, suggestion boxes, and hotlines (Landau,
2009), 2) belief structures of top management teams, and 3) the process of collective
sensemaking and communication. Similarly, Spencer (1986) empirically found that the number
of voice mechanisms was positively related to employees‘ expectations for problem resolution,
because that the number of voice mechanisms can be a signal to employees that management
wants them to speak up and will take their concerns seriously. Considering these effects of
organization factors on employee voice behavior, in the future research, we can explore
additional organizational factors that were not included in this study, such as the number of voice
mechanism. For instance, in the future research we can test 1) if the number of voice mechanism
affect employee voice behavior across cultures; 2) if the number of voice mechanism moderate
the effects of three harmony cultural values on employee voice behavior. These additional tests
would provide additional insights into factors that influence employee response to negative
workplace situations.
Furthermore, other factors at the country-level may be as important as culture. For
instance, it is evident that the political context in China influences Chinese employees‘ values,

１７４

thinking, and behaviors (Hofstede, 1999). This political context has resulted in the establishment
in China of an economy based on low labor costs. In such a situation, organizational success may
be determined primarily by minimizing labor costs and organizations can adapt to cultural
differences once they have relocated in the new country.
Another important limitation of this study is about methodological issues: as described
earlier, this study uses the cross-sectional self-report survey, which has sometimes been the
target of criticism. The key notion of the criticism is that usually self-report surveys inflate
correlations between variables measured with, due to common method variance (CMV), since
the method variance component would be shared or would be common across variables assessed
with a given method (e.g. Podsakoff, MacKenzie, Lee, & Podsakoff, 2003).
Although this criticism has become widely accepted, this study uses the cross-sectional
self-report method for several reasons. First of all, as other researchers have argued theoretically
and empirically, automatic criticism of the cross-sectional self-report is not desirable since it
becomes distorted and exaggerated over time (e.g. Spector, 1994, 2006). With empirical
evidence, for instance, Spector (1994) has cast doubt that the method itself produces systematic
variance in observations that inflates correlations to any significant degree; even measures that
are considered valid indicators of an intended trait can be influenced by many other variables.
Therefore, he argues that CMV is not about certain type of measurement inherently.
Furthermore, Spector (1994, 2006) argues that the reasonableness of using self-reports
depends on the purpose of the study itself: a more important issue is which constructs will be
measured and which method can be considered to represent these constructs, rather than
automatically following popular criticism on cross-sectional self-report. Accordingly, automatic
criticism of the methodology of this study simply because it uses cross-sectional self-report may

１７５

not be appropriate. In Particular, he argues that there seems to have been relatively little criticism
in the use of self-reports when measuring people‘s feeling about and perceptions of work, and
the criticism has focused mainly on instances where it measures the objective job environment
(Spector, 1994). Considering that this study is mainly about employees‘ own beliefs, values,
attitudes, and behavioral propensity, the results of this study should not be discounted merely
because this study uses cross-sectional self-report; rather, self-report is reasonable in this study.
Despite of it, another remaining issue in the limitation of cross-sectional design is that all
data are collected at one point in time, and, therefore, the variables measured by cross-sectional
design cannot obtain the evidence about the causality. In this line, a longitudinal study would
allow for more confident conclusions about causal relations, which are difficult with crosssectional designs, regardless of measurement method such as self-report survey. However, even
longitudinal designs and multiple methods cannot control for all possible factors that can
produce observed results, because there may be many feasible, and perhaps likely, alternative
explanations for the observed results of a study (Spector, 1994). Particularly, in this study, as
described earlier, there were many practical restrictions to accessing the organization, such as
limitations in terms of time, space to conduct a survey. For those reasons, even though
longitudinal design might have been better than cross-sectional design, it was not available in
this study.
Overall, despite the weakness of the cross-sectional self-report methodology, this design
is one of the major ones used in most studies of organizational behavior (Spector, 1994), because
it is useful in providing a picture of how people feel about and view their jobs, showing us the
intercorrelations among various feelings and perceptions, bringing important insights, and can be
useful for deriving hypotheses about how people react to jobs. Especially cross-sectional design

１７６

is useful because it can provide a relatively easy first step in studying phenomena of interest to
many research topics of organizational behavior studies (Spector, 1994). As discussed earlier,
this study is the initial study to suggest new factors such as face-saving and conflict-avoiding
that has not been tested yet in employee voice behavior studies. Therefore, it is reasonable to use
the cross-sectional self-report methodology to examine the association between these newly
suggested cultural values and employee voice behavior in this study. For future research,
additional methodologies, such as longitudinal design, will be needed to fully test the findings of
this study, and show the causality between the three culture values and employee voice behavior
more precisely.
Another possible limitation in this study is that, as discussed earlier, this study does not
measure ‗actual behavior‘ of EVLN construct, rather it measures ‗would you‘; meaning the
intention or propensity of these behaviors. Although it is true that this study does not measure the
actual behavior, it is consistent with previous EVLN studies. This is partly because measuring
the actual behaviors is unreasonable due to the inherent boundary condition of the EVLN
construct. As described earlier, the precondition of this construct is that there is a situation
leading to the actor‘s behavioral choice, particularly a deteriorating situation at work; therefore,
we assume that employees start to think of a certain way of response once one has experienced
this situation. This means that there is a certain point at which the mechanism of EVLN becomes
operative. Therefore, as discussed earlier, in EVLN studies, it is hard to measure actual behavior
so long as there is an actual situation leading to all respondents‘ EVLN behavior. For that reason,
most studies of EVLN provided respondents with a scenario, such as, ―Assume that there is a
situation…. How would you….?‖
In addition, even though actual behavior is measured, for example, actual turnover, it is

１７７

hard to be sure if it is exactly what is intended to be measured. As discussed by turnover
researchers (e.g. Hom & Griffeth, 1995), for instance, most turnover research relies on personal
files provided by the organization. Even these so-called objective records are typically deficient
in measuring voluntary turnover, since they fail to capture and classify all reasons. Measuring the
intention should not be automatically discounted, because researchers support that the intention
is not far from the actual behavior; rather, it is the strongest precursor of the actual behavior. For
instance, in turnover research, a fundamental tenet of modern thought on turnover is that a
decision to withdraw from the workplace best portends subsequent withdrawal (Hulin,
Roznowski, & Hachiya, 1985; Mobely et al., 1979; Price & Mueller, 1986; Susbult & Farrell,
1983). The meta-analysis outcomes also revealed that intention to quit best predicted actual
departure (e.g. Hom & Griffeth, 1995). Overall, when measuring actual behaviors of the EVLN,
timing does matter, due to the nature of EVLN construct itself. A possible alternative way to
examine the link between the intention and the actual behavior is to measure an employee‘s
experience retrospectively when he or she has suffered a recent negative situation at work, as
well as intention in the given situation.
Another possible limitation in methodology is about the sample used in this study:
although the data collection plan has been designed based on thorough consideration of
theoretical suggestions, rather than simply relying on a convenient sample such as a student
sample, practical restrictions occurred in the process of data collection in the organizations
involved in this study, which may possibly lead to bias in the analysis of the results of this study.
The biggest obstacle to determining subjects for the study was the difficulty of getting
approval from the organizations that satisfied the criteria of sampling strategy in this study,
which was 1) Korean organizations; 2) operating a the U.S. facilities as well as one in Korea; 3)

１７８

with local employees in both sites, enough to perform cross-cultural comparison; 4) in the same
industry. However, most organizations on the initial contact list refused to join the survey,
pointing out that this survey conveys many sensitive and uncomfortable issues to the
organization, such as employees‘ voice behavior and relationship with their supervisor.
Even after granting permission to survey subjects, the organizations gave very strict
limits for the access of the facilities. For that reason, the details of the accessible place and time
in each site depend on the conditions that each organization required for the permission, due to
their high concern and therefore very high restrictions. For instance, one organization in Korea
allowed me access to only one cafeteria among a couple of cafeterias on the site, and only during
lunch break.
These practical limitations on data collection in this study may cause possible bias in this
sample. For instance, as described in the explanation of sample, most of the respondents are male
and a skilled crafts or laborer; this is possibly due to the characteristics of a labor force in the
manufacturing industry. Second, the Korean sample shows relatively longer tenure than in the
U.S. sample: that is possibly because among organizations that participated in this study, the
organizations for the U.S. sample were established much later than the organizations for the
Korean sample. In addition, in Korea life-long employment remained somewhat, especially than
in the U.S. Finally, as for job categories, organization 2 in Korea shows a high composition of
the job category of 1) Executive/Management, Sales/Marketing, and Administrative/Clerical
(70.3%). This is possibly due to restriction of access. As described above, one organization
allowed me to access only one cafeteria among a couple of cafeteria in the site. It is possible that
this available cafeteria is located close to a managerial or clerical department, and for that reason
a relatively larger number of managerial employees in the plant participated in the survey than in

１７９

other sites. Furthermore, it can be argued that the cafeteria in which the organization permitted
the distribution of questionnaires possibly serves a greater number of managerial or clerical
employees than the other cafeteria.
Despite these obstacles to data collection, the bottom line is that the overall procedure of
data collection, from deciding sampling strategy to actual data collection, was guided by the
consideration of theoretical suggestions, rather than simply convenient sampling. First of all,
sampling strategy was suggested based on thorough consideration of researchers‘ suggestions for
cross-cultural research, such as Lyons and Chryssochoou‘s (2000) guidance for cross cultural
research. They suggest that deciding which cultures/nations are to be used should be based on
theoretical grounds: researchers should decide which variables/aspects of a culture are likely to
have an effect on the target behavior and then choose cultures which are likely to vary among
those particular dimensions. Since Kozan (1997) stated that the Asian countries are examples of
a harmony model, this dissertation takes Korea as a main culture group. The U.S. is used as an
example of non-harmony model since Kozan mentions the U.S. as an example of non-harmony
culture.
Then as for which subgroup within each culture will be involved in the research, the
researcher should try to match the subgroup to the research purpose, across the selected cultures
in the first stage of the sample (Lyons & Chryssochoou, 2000). For instance, if the research
interest is in cultural difference in consumer behavior, it will not make sense to compare middle
class adults in one culture with working class adults in another. The research focus of this
dissertation is to explore certain employees‘ behaviors at work, particularly the effects of culture
on employee voice behavior. Therefore, this dissertation uses employee samples in the U.S. and
Korea. Furthermore, by selecting a sample in the same industry with sample country origin of

１８０

organization, this dissertation attempts to make these two countries‘ samples as compatible with
each other as possible. Overall, the subject organization has been selected based on all of these
consideration, in detail, 1) Korean organizations; 2) operating the U.S. facilities as well as in
Korea; 3) with local employees in both sites, enough to cross-cultural comparison; 4) in the same
industry.
Second, even in the data collection procedure, despite a variety of detailed restrictions
given by the organizations, the bottom line ensured across sites are that 1) all participants in the
research were totally voluntary, 2) data collection was conducted by myself alone, without any
involvement of HR managers or other managers in the organization, and 3) the survey does not
allow individuals to be identified in any case. The survey does not label respondents by name or
by other information which might allow participants to identify themselves or others, and no data
that would allow respondents to be identified as individuals will be shared with anyone at the
organization.
Last, and most importantly, the results of data analysis support that these restriction given
by the organizations did not affect the results of analysis. The demographic information for
respondents, such as education, tenure, and job category, were added as control variables in each
regression on EVLN; however, none of them were found to have a significant association with
the dependent variable.
One suggestion for this possible limitation of data in the study is to use other types of
samples to examine the consistency of the findings of this study. For instance, larger samples, or
at least sample size similar to the Korean site, will strengthen the analytic power of the results. It
will be interesting to test if the findings in this study are applicable to other industries that do not
share the demographic characteristics of the sample in this study: such as using a sample from a

１８１

service industry, female-dominant industry, or high-technology industry. By using other types of
samples, we will be able to confirm the findings of this study or enhance the analytic power of
these findings.

7.4 Concluding Remarks
In conclusion, this study represents the initial attempt to empirically provide an answer to
the question, ―Does culture matter in employee response to negative workplace situations?‖ With
the strong empirical evidence gathered about a not fully investigated value, culture, this study
attempts to present a new way of looking at employee voice-related behaviors, increasing our
understanding of employee response to negative workplace situations across cultures. Although,
more research needs to be done in the future, it appears that this research has effectively set the
stage for further empirical examination, and theoretical development. By providing an improved
comprehensiveness and significant parsimony in employee voice literature, this study takes a
step toward stimulating future research in employee voice literature, and expanding our
understanding employee response to negative workplace situations across cultures for both
researchers and practitioners.

１８２

APPENDICES

１８３

Appendix A: Survey in English
Instructions
Thank you for taking the time to complete this survey. This survey is to understand how
employees respond to dissatisfaction at work differently across culture. This survey will take
approximately 10 minutes to complete and it is divided into 4 parts:
1.
2.
3.
4.

Describing perception of yourself and your behaviors at work
Describing your relationships with your supervisor
Describing your response to conflict or dissatisfactory situation at work
Demographic information

Please read each question carefully and answer each question honestly. There are no right or
wrong answers, we are interested in your personal experience. Remember that all your responses
will be kept confidential.
Part I. Describing a perception of yourself at work
1. Please indicate your level of agreement with each of the following statements.
Not at all----------------------Moderate Amount---------------------Great Extent
1----------------2-------------3------------4-------------5---------------6----------------7
(1)
(2)
(3)
(4)
(5)
(6)
(7)

If a team is slowing me down, it is better to leave it and work alone.
To be superior, a person must stand alone.
One performs a better job working alone than in a group.
Only those who depend upon themselves get ahead in life.
Individuals should sacrifice self-interest for the group that they belong to.
Group success is more important than individual success.
Individuals should pursue their goals after considering the welfare of the group.

2. Please indicate your level of agreement with each of the following statements.
Not at all----------------------Moderate Amount---------------------Great Extent
1----------------2-------------3------------4-------------5---------------6----------------7
(1)
(2)
(3)
(4)

It is important to be sensitive to another person‘s self-worth.
I am concerned with helping the other person to maintain his/her credibility.
I am concerned with not bringing shame to myself.
I am concerned with protecting my self-image.

１８４

3. Please indicate your level of agreement with each of the following statements about yourself
in your workplace.
Not at all----------------------Moderate Amount---------------------Great Extent
1----------------2-------------3------------4-------------5---------------6----------------7
(1)
(2)
(3)
(4)
(5)

I am important around here
I am valuable around here
I am trusted around here
I am efficient around here
I am cooperative around here

4. Please indicate your level of agreement with each of the following statements.
Not at all----------------------Moderate Amount---------------------Great Extent
1----------------2-------------3------------4-------------5---------------6----------------7
(1) If I left this job, my next job would probably better than the job you have now.
(2) I am confident that I would find a satisfactory job if I had quit this job

Part II. Describing your relationships with your supervisor
1. Please indicate the extent of your agreement with the following statements concerning your
supervisor.
Not at all----------------------Moderate Amount---------------------Great Extent
1----------------2-------------3------------4-------------5---------------6----------------7
(1) My supervisor is investing a great deal in my career.
(2) I often share my good ideas with my supervisor.
(3) I am open to my supervisor‘s suggestions regarding my work.
(4) My working relationship with my supervisor is better than average.
(5) My working relationship with my supervisor is very effective.
(6) My boss handles my concerns promptly.
(7) My boss listens carefully to what I say when I bring in a concern.
(8) My boss is fair when I take a concern to him/her.
(9) My boss takes action to correct the concerns that I speak to him or her about.
(10) My boss gives high priority to handling employee concerns.
(11) My boss listens to what I say when I bring in a concern.

１８５

Part III. Describing your response to conflict or dissatisfactory situation at work
1. Please indicate your level of agreement with each of the following statements, in case there is
a conflict in your workplace.
Not at all----------------------Moderate Amount---------------------Great Extent
1----------------2-------------3------------4-------------5---------------6----------------7
(1)
(2)
(3)
(4)
(5)
(6)

I believe it is better to keep negative opinion to ourselves rather than create hard feelings.
I try to smooth over conflicts by trying to ignore them.
I avoid being put "on the spot" by keeping conflict to ourselves.
I try to stay away from disagreements.
I usually avoid open discussion of my differences with the other person.
I generally avoid an argument.

2. Please indicate your level of agreement with each of the following statements, in case there is
a conflict in your workplace.
Not at all----------------------Moderate Amount---------------------Great Extent
1----------------2-------------3------------4-------------5---------------6----------------7
(1) It is better not to disagree with management decisions.
(2) When my supervisor makes a decision with which I disagree I prefer to accept the decision
rather than question it.
(3) I believe that it is not right to disagree with my boss.
(4) People in higher positions should make most decisions without consulting people in lower
positions.
(5) People in higher positions should not ask opinions of people in lower positions too frequently.
(6) People in lower positions should not disagree with decisions made by people in higher
positions

3. Now consider a situation in which you think something important in your job needs to be
changed or improved. For instance, consider a situation in which you think that you are unfairly
treated at work. What would you do about it?
Not at all----------------------Moderate Amount---------------------Great Extent
1----------------2-------------3------------4-------------5---------------6----------------7
(1)
(2)
(3)
(4)
(5)
(6)

I would think about quitting my job.
I would let my supervisor know that I was thinking about quitting.
I would accept another job offer.
I would quit my current job even without a job offer.
I would go to my immediate supervisor to discuss the problem.
I would ask my co-workers for advice about what to do.

１８６

(7) I would talk to another member of management about how I felt about the situation.
(8) I would try to solve the problem by suggesting changes in the way work was supervised in
the office.
(9) I would hang in there and wait for the problem to go away.
(10) I would stick with my job.
(11) I would still think that my job was as good as most.
(12) I would wait patiently for the problem to disappear.
(13) I would lose motivation to do my job as well as I might otherwise.
(14) I would show up late because I wasn't in the mood for work.
(15) I would call in sick occasionally because I didn't feel like working.
(16) I would put less effort into my job.
(17) I would take a lot of breaks.

Part IV. Demographic information
Please answer the following questions about yourself
1. What is your biological sex?
(1) Female
(2) Male
2. Indicate your age (in years).
3. What is the highest level of education you have completed?
(1) Some high school
(2) High school diploma or GED
(3) Some college
(4) College degree
(5) Masters degree or higher
4. How long have you worked for your current organization (in years)?
5. Which of the following categories best describes your current job?
(1) Executive/Management, Sales/Marketing, and Administrative/Clerical
(2) R&D and Technical/Professional
(3) Skilled Crafts and Laborer
6. Which of the following best describes your ethnic background?
(1) Caucasian
(2) Asian
(3) African/black
(4) Hispanic/ Latino
(5) Others

１８７

Appendix B: Survey (In Korean)

본 설문참여를 위해 시간을 응해주셔서 감사합니다. 본 설문은 서로 다른 문화에서
근로자들이 직장 내 불만사항에 대해서 어떻게 서로 다르게 대응하는 지를 이해하고자
기획되었습니다. 본 설문을 작성하는 데에는 약 10 분이 소요될 것이며, 설문은 다음
4 부분으로 이루어져 있습니다.

1.

직장 생활에 대한 귀하의 생각 및 행동

2.

귀하의 직속 상사와의 관계

3.

직장 내 귀하의 의견개진 및 의사표현 관련 행동

4.

귀하의 인적 사항

각 문항을 읽고 솔직하게 답변해 주시기 바랍니다. 문항에 대해 정해진 정답은 없으며
귀하의 경험에 대해서 알아보고자 합니다. 귀하의 모든 답변은 공개되지 않을 것입니다.

I. 먼저 직장생활에 대한 귀하의 평소 생각에 대한 질문입니다

1. 다음 문항에 어느 정도 동의하시는지 표시해주시기바랍니다.
전혀 그렇지 않다 -------------그저 그렇다 --------------------- 매우 그렇다
1----------------2-------------3------------4-------------5---------------6----------------7
1) 집단 때문에 내 일이 지체된다면, 집단을 떠나 혼자 일하는 것이 낫다
(2) 남들보다 뛰어나기 위해서는 홀로서야 한다
(3) 집단보다는 혼자 일하는 것이 낫다
(4) 자신에게 의지하는 사람만이 인생에서 성공한다
(5) 개인은 자신이 속한 그룹을 위해 자기 자신의 이익을 희생해야 한다
(6) 조직의 성공이 개인의 성공보다 중요하다
(7) 개인은 조직의 안녕을 고려한 후 자신의 목표를 성취해야 한다

2. 직장내 인간관계과 관련하여 다음의 문항에 어느정도 동의하시는지 표시해주시기
바랍니다.

１８８

전혀 그렇지 않다 -------------그저 그렇다 --------------------- 매우 그렇다
1----------------2-------------3------------4-------------5---------------6----------------7
(1) 다른 사람의 자부심을 배려하는 것은 중요하다
(2) 다른 사람이 그들의 신뢰를 유지할 수 있도록 돕는데 신경 쓴다
(3) 나는 비난 받지 않기 위해 신경을 쓴다
(4) 나는 나의 이미지를 지키는데 신경을 쓴다

3. 직장내 귀하 자신에 대한 다음 문항에 어느 정도 동의하시는지 표시해주시기바랍니다.
전혀 그렇지 않다 -------------그저 그렇다 --------------------- 매우 그렇다
1----------------2-------------3------------4-------------5---------------6----------------7
(1) 나는 이곳에서 중요한 사람이다
(2) 나는 이곳에서 가치있는 사람이다
(3) 나는 이곳에서 신뢰를 받고있는 사람이다
(4) 나는 이곳에서 유능한 사람이다
(5) 나는 이곳에서 협조적인 사람이다

4. 다음 문항에 어느 정도 동의하시는지 표시해주시기바랍니다.
전혀 그렇지 않다 -------------그저 그렇다 --------------------- 매우 그렇다
1----------------2-------------3------------4-------------5---------------6----------------7
(1) 나는 이곳을 그만두더라도 다음 직장은 이곳보다 나을 것이다
(2) 나는 이곳을 그만둔다면 더 만족스러운 직장을 찾을 수 있다고 확신한다

II. 다음은 귀하의 직장 상사와의 관계에 대한 질문입니다
1. 귀하의 상사와의 관계에 대하여 다음의 문항에 어느정도 동의하시는지 표시해주시기
바랍니다.
전혀 그렇지 않다 -------------그저 그렇다 --------------------- 매우 그렇다
1----------------2-------------3------------4-------------5---------------6----------------7
(1) 나의 상사는 나의 커리어에 상당히 투자한다
(2) 나는 나의 상사와 나의 좋은 아이디어를 자주 나눈다
(3) 나는 나의 업무와 관련한 나의 상사의 제안에 귀를 기울인다

１８９

(4) 나와 상사의 관계는 평균이상으로 좋다
(5) 상사와 나의 업무관계는 매우 효율적이다
(6) 나의 상사는 내가 염려하는 점에 대해 즉각적으로 조치를 취한다
(7) 나의 상사는 내가 염려하는 점을 얘기하면 이를 경청한다
(8) 나의 상사는 내가 자신에 대한 문제점을 지적할때 공정히 대한다
(9) 나의 상사는 내가 염려하는 점을 얘기하면 이를 수정하기 위해 필요한 행동을 취한다
(10) 나의 상사는 직원들의 고충처리에 우선순위를 둔다
(11) 나의 상사는 내가 우려사항에 대해 얘기하면 이를 경청한다

III. 다음은 직장내 의견개진 및 의사표현 관련 행동에 대한 질문입니다
1. 직장내 갈등이 생겼을 경우, 다음의 문항에 어느정도 동의하시는지 표시해주시기
바랍니다.

전혀 그렇지 않다 -------------그저 그렇다 --------------------- 매우 그렇다
1----------------2-------------3------------4-------------5---------------6----------------7
(1)
(2)
(3)
(4)
(5)
(6)

나는 불편한 감정을 일으키느니 그냥 속으로 담아두는 것이 낫다고 믿는다
나는 갈등을 외면함으로써 갈등을 무마시키려고 노력한다
나는 갈등을 드러내기를 꺼린다
나는 의견충돌에서 거리를 두려고 노력한다
나는 대체적으로 논쟁을 꺼린다
나는 남들과 부딪치는 것을 꺼린다

2. 직장내 갈등이 생겼을 경우, 다음의 문항에 어느정도 동의하시는지 표시해주시기
바랍니다.

전혀 그렇지 않다 -------------그저 그렇다 --------------------- 매우 그렇다
1----------------2-------------3------------4-------------5---------------6----------------7
(1) 경영진의 결정에 이의를 제기하지 않는 게 좋다
(2) 상사가 내 의견과 상충되는 결정을 할때, 나는 이의를 제기하기 보다는 그 결정을
수용하는 편이다
(3) 상사의 견해에 동의하지 않는 것은 옳지 않다고 생각한다
(4) 상급자는 부하직원과 상의하지 않고 대부분의 결정을 내려야 한다
(5) 상급자는 부하직원의 의견을 너무 자주 묻지 말아야 한다
(6) 부하직원은 상급자의 결정에 반대해서는 안된다

１９０

3. 회사에 불만사항이 있어 무언가 개선하기 위해 바꾸어야할 부분이 있다고 생각하고
있을 때를 떠올리십시오. 가령 회사에서 귀하께서 부당대우를 받거나 회사의 비윤리적인
행위를 알게되었을 경우를 떠올리십시오. 위의 경우와 관련하여 다음의 항목에 대해
구체적으로 어떻게 행동하시겠습니까?
전혀 그렇지 않다 -------------그저 그렇다 --------------------- 매우 그렇다
1----------------2-------------3------------4-------------5---------------6----------------7
(1) 이직을 고려할것이다
(2) 이직을 고려하고 있다는 것을 상사에게 알릴 것이다
(3) 다른 일자리 제안을 받으면 수락할 것이다
(4) 다른 일자리 제안이 없더라도 회사를 그만둘 것이다
(5) 직속상사를 찾아가 문제에 대해 상의할 것이다
(6) 어떻게 해야할지에 대해 직장동료에게 조언을 구할것이다
(7) 이 상황에 대해 내가 느끼고 있는 바를 다른 관리자에게 얘기할 것이다
(8) 개선방안을 제안하여 문제해결을 위해 노력할 것이다
(9) 일단 문제점이 어떻게 진행되어가는지 지켜볼 것이다
(10) 상황이 좋든 안좋든 내 업무에 집중할 것이다
(11) 여전히 내 직장을 최상이라고 생각할 것이다
(12) 문제점이 사라질때까지 참고 기다릴 것이다
(13) 그전과 달리 일할 의욕이 상실될 것이다
(14) 일할 기분이 아니기 때문에 태만할 것이다
(15) 일하고 싶은 생각이 들지 않을 것이기 때문에 종종 병가(휴가?)를 낼 것이다
(16) 내 업무에 대한 노력을 줄일것이다
(17) 자주 쉬거나 열심히 일하지 않을 것이다

IV. 다음은 귀하의 인적 사항에 대한 질문입니다.

1. 귀하의 성별은 어떻게 되십니까?
(1) 여성
(2) 남성
2. 귀하의 만나이는 어떻게 되십니까? (만나이 기준: 우리나라 나이 -1) _____ 세
3. 귀하의 최종 학력은 어떻게 되십니까?
(1) 고졸미만
(2) 고졸
(3) 2 년제 대학

１９１

(4) 4 년제 대학
(5) 대학원 이상 (석사, 박사 등)
4. 귀하의 근속년수는 얼마나 되셨습니까?

______ 년

5. 귀하께서 하시는 일은 다음 중 어디에 해당됩니까?
(1) 관리·사무·영업직
(2) 연구개발·기술 전문직
(3) 생산·기능직

１９２

REFERENCES

１９３

REFERENCES
Addison, J. T., & Belfield, C. (2004). Union voice. Journal of Labor Research, 25(4), 583–596.
Adler, P. S., & Kwon, S. (2002). Social capital: Prospects for a new concept. Academy of
Management Review, 27, 17−40.
Allen, D. B., Miller, E. D., & Nath, R. (1988). North America. In R. Nath (Ed.), Comparative
management (pp. 23–54). Cambridge, MA: Ballinger.
Angle, H. L., & Perry, J . L. (1981). An empirical assessment of organizational effectiveness.
Administrative Science Quarterly, 26, 1-14.
Ashford, S. J., Rothbard, N. P., Piderit, S. K., & Dutton, J. E. (1998). Out on a limb: The role of
context and impression management in selling gender-equity issues. Administrative
Science Quarterly, 43, 23-57.
Ashforth, B. E., & Humphrey, R. H. (1995). Labeling processes in the organization:
Constructing the individual. In L. L. Cummings & B.W. Staw (Eds.), Research in
Organizational Behavior (pp. 413−461). Greenwich, CT: JAI Press.
Avery, D. (2003). Personality as a predictor of the value of voice. The Journal of Psychology,
137(5), 435-446.
Avery, D. R., & Quinones, M. A. (2002). Disentangling the effects of voice: the incremental
roles of opportunity, behavior, and instrumentality in predicting procedural fairness.
Journal of Applied Psychology, 87, 81–86.
Batt, R., Colvin, A. J. S., & Keefe, J. (2002). Employee voice, human resource practices, and
quit rates: evidence from the telecommunications industry. Industrial and Labor
Relations Review, 55(4), 573–594.
Bentler, P. M. (1990). Comparative fit indexes in structural equation models. Psychological
Bulletin, 107, 238–246.
Block, R. N., Berg, P., & Belman, D. (2004), ―The Economic Dimension of the Employment
Relationship,‖ in The Employment Relationship: Examining Psychological and
Contextual Perspectives, Jackie Coyle-Shapiro, Lynn Shore, Susan Taylor, Lois Tetrick
eds., Oxford, UK: Oxford University Press, 2004, pp. 94-118.
Boehnke, K. Lietz, P., Schreier, M., & Wilhelm, A. (2011). Sampling: the selection of cases for
culturally comparative psychological research. In D. Matsumoto & F. J. R. Van De
Vijver (Eds.). Cross-cultural research methods in psychology (pp.101-129). New York:
Cambridge University Press.

１９４

Bond, M. H., & Wang, S. H. (1983). Aggressive behavior in Chinese society: The problem of
maintaining order and harmony. In A. P. Goldstein & M. Segal 1 (Eds.), Global
perspective on aggression (pp. 58-74). New York: Pergamon Press.
Bontempo, R., & Rivero, J. C. (1992). Cultural variation in cognition: The role of self-concept in
the attitude behavior link. Paper presented at the meeting of the American Academy of
Management, Las Vegas, Nevada, August.
Boroff, K. E., & Lewin, D. (1997). Loyalty, Voice, and Intent to Exit a Union Firm: A
Conceptual and Empirical Analysis. Industrial and Labor Relations Review, 51(1), 50–63.
Botero, I. C. (2005). Getting one‘s way vs. maintaining the relationship: effects of leadermember exchange (LMX) quality on upward influence message production across two
cultures. Unpublished doctoral dissertation, Michigan State University.
Botero, I. C. & Van Dyne, L. (2009). Employee voice behavior: interactive effects of LMX and
power distance in the United States and Columbia. Management Communication
Quarterly, 23(1), 84-104.
Brockner, J. (1985). The relation of trait self-esteem and positive inequity to productivity.
Journal of Personality, 53, 517-529.
Brockner, J. (1988). Self-esteem at work: Research, theory, and practice. Lexington, MA:
Lexington Books.
Brockner, J., Siegel, P. A., Martin, C., Reed, T., Heuer, L., Wiesenfeld, B., Grover, S., &
Bjorgvinsson, S. (1998). The moderating effect of self-esteem in reaction to voice :
Converging evidence from five studies. Journal of personality and social
psychology, 75(2), 394-407.
Brockner, J., Ackerman, G., Greenberg, J., Gelfand, M. J., Francesco, A. M., Chen, Z. X., Leung,
K., Bierbrauer, G., Gómez, C., Kirkman, B. L. & Shapiro, D. L. (2001) Culture and
Procedural Justice: The Moderating Influence of Power Distance on Reactions to Voice,
Journal of Experimental Social Psychology, 37, 300–315.
Brown, M. E., Trevin˜o, L. K., & Harrison, D. A. (2005). Ethical leadership: A social learning
perspective for construct development and testing. Organizational Behavior and Human
Decision Processes, 97, 117–134.
Breukelen, W. V., Vlist, R. V., & Steensma, H. (2004). Voluntary employee turnover: Combining
variables from the „traditional‟ turnover literature with the theory of planned behavior.
Journal of Organizational Behavior, 25, 893–914.
Budd, J. W. (2004). Employment with a Human Face: Balancing Efficiency, Equity, and Voice.
ILR Press, Ithaca, NY.

１９５

Byrne, B. M. (2001). Structural equation modeling with AMOS. Mahwah, NJ: Lawrence
Erlbaum.
Chan, A. (1996). Confucianism and Development in East Asia. Journal of Contemporary Asia.
26(1), 28-45.
Chia, H., Landau, J., & Ong, J. (2000). Predictors of voice: The moderating effect of the general
economic climate. Research and Practice in Human Resource Management, 8(2), 3-28.
Chinese Culture Connection. (1987). Chinese values and the search for culture-free dimensions
of culture. Journal of Cross-Cultural Psychology, 18, 143-164.
Chiu, R. K., & Kosinski, F. A. Jr. (1994). Is Chinese conflict-handling behavior influenced by
Chinese values? Social Behavior & Personality, 22(1), 81-90.
Chow, G. C. (1960). Tests of Equality between Sets of Coefficients in Two Linear Regressions.
Econometrica, 28(3), 591–605.
Coopersmith, S. (1967). The antecedents of self-esteem. San Francisco.
Costigan, R. D., Insinga, R. C., Berman, J. J., Ilter, S. S., Kranas, G., & Kureshov, V. (2006).
The effect of employee trust of the supervisor on enterprising behavior: A cross-cultural
comparison. Journal of Business and Psychology, 21(2), 273-291.
Cotton, J. L., & Tuttle, J. M. (1986). Employee Turnover: A Meta-Analysis and Review with
Implications for Research. Academy of Management Review, 11(1), 55-70.
Dansereau, F., Jr., Graen, G., & Haga, W. J. (1975). A vertical dyad linkage approach to
leadership within formal organizations: A longitudinal investigation of the role making
process. Organizational Behavior and Human Performance, 13, 46–78.
Detert, J. R. & Burris, E. R. (2007). Leadership behavior and employee voice: Is the door really
open? The Academy of Management Journal, 50(4), 869-884.
Dorfman, P. W., & Howell, J. P. (1988). Dimensions of national culture & effective leadership
patterns: Hofstede revisited. In E. G. McGoun (Ed.), Advances in international
comparative management (Vol. 3, pp. 127-149). Greenwich, CT: JAI.
Dowding, E., John, P., Mergoupis, T., & Van Vugt, M. (2000). Exit, voice and loyalty: Analytic
and empirical developments. European Journal of Political Research, 37, 469–495.
Dutton, J. E., & Ashford, S. J. (1993). Selling Issues to Top Management. The Academy of
Management Review, 18(3), 397-428.

１９６

Dyer, B., & Song, X. M. (1997). The Impact of Strategy on Conflict: A Cross-National
Comparative Study of U.S. and Japanese Firms. Journal of International Business Studies,
28(3), 467-493.
Earley, P. C. (1993). East Meets West Meets Mideast: Further Explorations of Collectivistic and
Individualistic Work Groups. The Academy of Management Journal, 36(2), 319-348.
Edwards, M. S., & Greenberg, J. (2009). Sounding Off on Voice and Silence. In Greenberg, J., &
Edwards, M. S. (Eds.). Voice and Silence in Organizations. Bingley, UK: Emerald Press.
Erez, M. (1994). Toward a model of cross-cultural industrial and organizational psychology. In
Triandis HC, Dunnette MD, Hough LM (Eds.), Handbook of industrial and
organizational psychology. (Vol. 4, pp. 559-608). Palo Alto, CA: Consulting
Psychologists Press.
Falbo, T., & Peplau, L. A. (1980). Power strategies in intimate relationships. Journal of
Personality and Social Psychology, 38, 618-628.
Farrell, D. (1983). Exit, voice, loyalty, and neglect as responses to job dissatisfaction: A
multidimensional scaling study. Academy of Management Journal, 26, 596-607.
Farrell, D., & Rusbult, C. E. (1981). Exchange variables as predictors of job satisfaction, job
commitment, and turnover: The impact of rewards, costs, alternatives, and investments.
Organizational Behavior and Human Performance, 27/28, 78-95.
Farrell, D., & Rusbult, C. E. (1985). Understanding the retention function: A model of the causes
of exit, voice, loyalty, and neglect behaviors. Personnel Administrator, 30, 129-140.
Farrell, D., & Rusbult, C. E. (1992). Exploring the exit, voice, loyalty, and neglect typology: the
influence of job satisfaction, quality of alternatives, and investment size. Employee
Responsibilities and Rights Journal, 5, 201–218.
Farrell, D., Rusbult, C. E., Lin, Y. H., & Bernthal, P. (1990). Impact of job satisfaction,
investment size, and quality of alternatives on exit, voice, loyalty, and neglect responses
to job dissatisfaction: A cross-lagged panel study. In L. R. Jauch «& J. L. Wall (Eds.),
Best Papers Proceedings of the Fiftieth Annual Meeting of the Academy of Management,
211-215. San Francisco: Academy of Management.
Fishbein, M., & Ajzen, I. (1975). Belief Attitude, Intention and Behavior: An Introduction to
Theory and Research. Reading, MA: Addison Wesley.
Fleishman, E. A., & Edwin, F. H. (1962). Patterns of leadership behavior related to employee
grievances and turnover. Personnel Psychology, 15, 43-56.
Folger, R., & Cropanzano, R. (1998). Organizational justice and human resource management.
London: Sage.

１９７

Forbes, J. D. (1985). Organizational and political dimensions of consumer pressure groups.
Journal of Consumer Policy, 8, 105-131.
Freedman, R. B., & Medoff, J. L. (1984). What Do Unions Do? New York: Basic Books
Freeman, R. B. (1976). Individual Mobility and Union Voice in the Labor Market. American
Economic Review, Papers and Proceedings, 66, 361-368.
___ __ __ __ _ (1980). The Exit-Voice Tradeoff in the Labor Market: Unionism, Job Tenure,
Quits, and Separations. Quarterly Journal of Economics, 94, 643-673.
Folger, R. (1977). Distributive and procedural justice: Combined impact of ―voice‖ and
improvement of experienced inequity. Journal of Personality and Social Psychology, 35,
108–119.
Frese, M., Teng, E., & Wijnen, C. J. D. (1999). Helping to Improve Suggestion Systems:
Predictors of Making Suggestions in Companies, Journal of Organizational Behavior,
20(7), 1139-1155.
Friedman, R., Chi, S., & Liu, L. A. (2006). An expectancy model of Chinese-American
differences in conflict-avoiding. Journal of International Business Studies, 37(1), 76-91.
Gelfand, M. J., Erez, M., & Aycan, Z. (2007). Cross-Cultural Organizational Behavior. Annual
Review of Psychology, 58, 479-514.
George, J. M., & Brief, A. P. (1992). Feeling good-doing good: A conceptual analysis of the
mood at work-organizational spontaneity relationship. Psychological Bulletin, 112, 310329.
Gordon, W. I. (1988). Range of employee voice. Employee Responsibilities and Rights Journal,
1(4), 283-299.
Graen, G. B., & Uhl-Bien, M. (1995). Relationship-based approach to leadership: Development
of leader-member exchange (LMX) theory of leadership over 25 years: Applying a
multilevel multi-domain perspective. Leadership Quarterly, 6(2), 219-247.
Graham, J. W. (1991). An essay on organizational citizenship behavior. Employee
Responsibilities and Rights Journal, 4, 249–270.
Graham, J. W., & Keeley, M. (1989). Critical issues of exit, voice, and loyalty. Unpublished
manuscript, Loyola University of Chicago.
Griffeth, R. W., Hom, P. W., & Gaertner, S. (2000). A meta-analysis of antecedents and correlates
of employee turnover: Update moderator tests, and research implications for the next
millennium. Journal of Management, 26, 463–488.

１９８

Griffin, R.W., O‘Leary-Kelly, A., & Collins, J. M. (1998a). Dysfunctional Behavior in
Organizations: Violent and Deviant Behavior: Part A. JAI Press, Stamford.
Griffin, R.W., O‘Leary-Kelly, A., & Collins, J. M. (1998b). Dysfunctional Behavior in
Organizations: Violent and Deviant Behavior: Part B, JAI Press, Stamford.
Hagedoorn, M., Van Yperen, N. W., Van de Vliert, E., & Buunk, B. P. (1999). Employees‘
reactions to problematic events: A circumplex structure of five categories of responses,
and the role of job satisfaction. Journal of Organizational Behavior, 20, 309–321.
Harlos, K. P. (1999). Cultures of injustice: Organizational correlates of unjust employment
relationships. Paper presented at the Third Australian Industrial and Organizational
Psychology Conference, Brisbane, QLD.
Hirschman, A. O. (1970). Exit, Voice, and Loyalty: Responses to Decline in Firms,
Organizations, and States. Cambridge, MA: Harvard University Press.
Hofstede, G. (1980). Culture‘s consequences: International differences in work related values.
Beverly Hills, CA: Sage.
Hofstede, G. (1999). Problems Remain, But Theories Will Change: The Universal and the
Specific in 21st Century Global Management. Organizational Dynamics, 27(1), 34-44.
Hofstede, G., & Bond, M. (1988). The Confucius Connection: From Cultural Roots to Economic
Growth. Organizational Dynamics, 16(4), 5-21.
Hogarty, K. Y., Hines, C. V., Kromrey, J. D., Ferron, J. M., & Mumford, K. R. (2005). The
quality of factor solutions in exploratory factor analysis: The influence of sample size,
communality, and overdetermination. Educational and Psychological Measurement, 65,
202-226.
Holt, J. L., & DeVore, C. J. (2005). Culture, gender, organizational role, and styles of conflict
resolution: a meta-analysis. International Journal of Intercultural Relations, 29, 165–196.
Hom, P. W., & Griffeth, R. W. (1995). Employee turnover. Cincinnati, OH: South-Westem
Howell, J., & Higgins, C. (1990). Champions of technological innovation. Administrative
Science Quarterly, 35, 317–341.
Hui, C. H. (1984). Individualism-collectivism: Theory, measurement, and its relations to rward
allocation. Unpublished doctoral dissertation, University of Illinois.
Hui, C. H., & Triandis, H. C. (1986). Individualism-collectivism: A study of cross-cultural
researchers. Journal of Cross-Cultural Psychology, 17, 225-248.

１９９

Janssen, O., T., de Vries, & Cozijnse, A. (1998). Voicing by adapting and innovating employees:
An empirical study on how personality and environment interact to affect voice behavior.
Human Relations, 51(7), 945-968.
Jun, J. S., & Muto, H. (1995). The Hidden Dimensions of Japanese Administration: Culture and
Its Impact. Public Administration Review, 55(2), 125-134.
Kassing, J. W. (2000). Investigating the relationship between superior-subordinate relationship
quality and employee dissent. Communication Research Reports, 17(1), 58-70.
Dowding, K., John, P., Mergoupis, T., & Van Vugt, M. (2000). Exit, voice and loyalty: Analytic
and empirical developments. European Journal of Political Research, 37, 469–495.
Kim, A. E., & Park, G. S. (2003). Nationalism, Confucianism, work ethic and industrialization in
South Korea. Journal of Contemporary Asia, 33(1), 37-49.
Kim, D. K., & Kim, C. W. (1989). Korean value systems and managerial practices. In K. H.
Chung & H. C. Lee (Eds.), Korean managerial dynamics (pp. 207-216). New York:
Praeger.
Kim, D. K., & Kim, S. S. (2003). Globalization, Financial Crisis, and Industrial Relations: The
Case of South Korea. Industrial Relations, 42(3), 341-367.
Kim, Y. C. (1998). Industrial Reform and Labor Backlash in South Korea: Genesis, Escalation,
and Termination of the 1997 General Strike, Asian Survey, 38(12), 1142-1160.
Kipnis, D., Schmidt, S. M., & Wilkinson, I. (1980). Intraorganizational influence tactics:
Explorations in getting one's way. Journal of Applied Psychology, 65, 440-452.
Kirkbride, P. S., Tang, Sara F. Y., & Westwood, R. I. (1991). Chinese conflict preferences and
negotiation behavior: cultural and psychological influences. Organization Studies, 12, 3,
365-386.
Kline, R.B. (2005), Principles and Practice of Structural Equation Modeling (2nd Edition ed.).
New York: The Guilford Press.
Kozan M. K. (1997). Culture and conflict management: a theoretical framework. International
Journal of Conflict Management. 8, 338–60
Kristof, A. L. (1996). Person-organization fit: an interactive review of its conceptualizations,
measurement, and implications. Personnel Psychology, 49, 1–49.
Krone, K. J. (1991). Effects of leader member exchange on subordinates‘ upward influence
attempts. Management Communication Quarterly, 8, 9, 18. 20

２００

Krone, K. J. (1992). Achieving communication goals in superior subordinate relationships:
Upward influence maintenance tactics. Communication Quarterly, 40, 1, 1-15.
Kweit, M. G. (1986). Ideological congruence of party switchers and nonswitchers: The case of
party activists. American Journal of Political Science, 30, 184-196.
Lammers, C. J. (1969). Strikes and mutinies: A comparative study of organizational conflicts
between rulers and ruled. Administrative Science Quarterly, 14, 558-572.
Landau, J. (2009). To Speak or Not to Speak: Predictors of Voice Propensity. Journal of
Organizational Culture, Communication and Conflict, 13(1), 35-54.
Larson, E. W., & Fukami, C. V. (1985). Employee absenteeism: The role of ease of movement.
Academy of Management Journal, 28, 464-471.
Leck, J. D., & Saunders, D. M. (1992). Hirschman's loyalty: Attitude or behavior? Employee
Responsibilities and Rights Journal. 5, 219-230.
Lee, J., & Jablin, F. M. (1992), A cross-cultural investigation of exit, voice, loyalty, and neglect
as responses to dissatisfying job conditions. The Journal of Business Communication,
29(3), 203-228.
LePine, J. A., & Van Dyne, L. (1998). Predicting voice behavior in work groups. Journal of
Applied Psychology, 83, 853–868.
LePine, J. A., & Van Dyne, L. (2001). Voice and cooperative behavior as contrasting forms of
contextual performance: Evidence of differential relationships with big five personality
characteristics and cognitive ability. Journal of Applied Psychology, 86, 326-336.
Leung, K. (1988). Some Determinants of Conflict Avoidance, Journal of Cross-cultural
Psychology, 19(1), 125-136.
Lewin, D., & Mitchell, D. (1992). Systems of employee voice: theoretical and empirical
perspectives. California Management Review, 34(3), 95–111.
Liljegren, M., Nordlund, A. & Ekberg, K. (2006). Employee behavioural responses and selfrated health and burnout: a two-year follow-up study. Manuscript submitted for
publication.
Liljegren, M., Nordlund, A., & Ekbergs, K. (2008). Psychometric evaluation and further
validation of the Hagedoorn et al. modified EVLN measure. Scandinavian Journal of
Psychology, 49, 169–177.
Lincoln, J. R., & Kalleberg, A. L. (1996). Commitment, Quits, and Work Organization in
Japanese and U.S. Plants. Industrial and Labor Relations Review, 50(1), 39–59.

２０１

Lind, E. A., Kanfer, R., & Earley, P. C. (1990). Voice, control, and procedural justice:
instrumental and noninstrumental concerns in fairness judgments. Journal of Personality
and Social Psychology, 59, 952–9.
Lind, E. A., & Kulik, C. T. (2009). ―Gear Me Out: Voice and Justice,‖ in Voice and Silence in
Organizations, Jerald Greenberg and Marissa S. Edwards eds., Bingley, UK: Emerald
Press, 135-156.
Lind, E. A., & Tyler, T. R. (1988). The social psychology of procedural justice. New York, NY:
Plenum Press.
Liu, W., Zhu, R., & Yang, Y. (2010). I warn you because I like you: Voice behavior, employee
identifications, and transformational leadership. The Leadership Quarterly, 21, 189-202.
Lyons E., & Chryssochoou, X. (2000). Cross-cultural research methods. In Breakwell, G. M.,
Hammond, S. & Fife-Schaw, C. (Eds.). Research methods in psychology. (2nd ed.) (134146). London: Sage Publications.
Lyons, W. E., & Lowery, D. (1986). The organization of political space and citizen responses to
dissatisfaction in urban communities: An integrative model. Journal of Politics, 48, 321346.
MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996), Power Analysis and
Determination of Sample Size for Covariance Structure Modeling, Psychological
Methods, 1(2), 130-149.
Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion,
and motivation. Psychological Review, 98, 224–253.
March, H. W., Balla, J. R., & Hau, K. T. (1996). An evaluation of Incremental Fit Indices: A
clarification of mathematical and empirical properties. In G. A. Marcoulides & R. E.
Schumacker (Eds.), Advanced structural modeling, issues and techniques (pp. 315–353).
Mahwah, NJ: Erlbaum.
McCabe, D., & Lewin. D, (1992) "Employee Voice: A Human Resource Management
Perspective". California Management Review, 34(3): 112-23.
McLaughlin, G. W., & Butler, R. P. (1974). Use of weighting and data on alternative careers to
predict retention and commitment: Journal of Applied Psychology, 59, 87-89.
Miceli, M. P., & Near, J. P. (1989). The incidence of wrongdoing, whistle-blowing, and
retaliation: Results of a naturally occurring field experiment. Employee Responsibilities
and Rights Journal, 2, 91-108.
Milburn, T. W., Schuler, R. S., & Watman, K. H. (1983). Organizational crisis, part II: Strategies
and responses. Human Relations, 36, 1161-1180.

２０２

Miller, P. & Charles, M. (1991). Australian Evidence on the Exit/Voice Model of the Labor
Market. Industrial and Labor Relations Review, 45, 1, 44–57.
Milliken, F. J., Morrison, E. W., & Hewlin, P. F. (2003). An exploratory study of employee
silence: Issues that employees don't communicate upward and why. Journal of
Management Studies, 40, 1453−1476.
Millward, N., Bryson, A., & Forth, J. (2000), All Change at Work? Routledge, London.
Minton, J. W. (1988). Justice, satisfaction, and loyalty: Employee withdrawal and voice in the
din of inequity. Unpublished dissertation, Duke University.
Mobley, W. H. (1982), Employee turnover: Causes, Consequence, and Control. Reading, Mass.:
Addison-Wesley.
Molseed, M. J. (1989). Gender, silence, and group structure. Greenwich, CT: JAI Press.
Moore, C. A. (1967). The Chinese Mind. Honolulu: University of Hawaii Press.
Morris, M. W., Williams, K.Y., Leung, K., Larrick, R., Mendoza, M. T., Bhatnagar, D., Li, J.,
Kondo, M., Luo, J. L. & Hu, J. C. (1998). Conflict management style: accounting for
crossnational differences, Journal of International Business Studies, 29(4), 729–748.
Morrison, E. W., & Milliken, F. J. (2000). Organizational silence: A barrier to change and
development in a pluralistic world. Academy of Management Review, 25, 706-725.
Morrison, E. W., & Phelps, C. C. (1999). Taking charge at work: extra-role efforts to initiate
workplace change. Academy of Management Journal, 42, 403–419.
Muchinsky, P. M., & Maassarani, M. A. (1980). Work environment effects on public sector
grievances. Personnel Psychology, 33, 403-414.
Nunnally, J. C. (1978). Psychometric theory (2nd ed.). New York: McGraw-Hill.
Oetzel, J. G., & Ting-Toomey, S. (2003). Face Concerns in Interpersonal Conflict: A CrossCultural Empirical Test of the Face Negotiation Theory. Communication Research, 30,
599-624.
Ohbuchi K, Osamu F, & Tedeschi J. T. (1999). Cultural values in conflict management: goal
orientation, goal attainment, and tactical decision. Journal of Cross-Cultural Psychology,
30, 51–71
O‘Leary-Kelly, A. M., Griffin, R. W., & Glew, D. J. (1996). Organization-motivated aggression:
A research framework. Academy of Management Review, 21, 922–924.

２０３

Olson-Buchanan, J. B. & Boswell, W. R. (2002). The role of employee loyalty and formality in
voicing discontent. Journal of Applied Psychology, 87(6), 1167-1174.
Organ, D. W. (1988). Organizational citizenship behavior: The good soldier syndrome.
Lexington, MA: Lexington Books.
Parker, L. E. (1993). When to fix it and when to leave: relationships among perceived control,
self efficacy, dissent, and exit. Journal of Applied Psychology, 78, 949–59.
Peirce, E., Smolinski, C. A., & Rosen, B. (1998). Why sexual harassment complaints fall on deaf
ears. Academy of Management Executive, 12, 41-54.
Pfeffer, J., & Lawler, J. (1980). Effects of job alternatives, extrinsic rewards, and behavioral
commitment on attitude toward the organization: A field test of the insufficient
justification paradigm. Administrative Science Quarterly, 25, 38-56.
Pierce, J. L., Gardner, D. G., Cummings, L. L., & Dunham, R. B. (1989). Organization-based
self-esteem: Construct definition, measurement, and validation. Academy of Management
Journal, 32, 622- 648.
Pinder, C. C., & Harlos, K. P. (2001). Employee silence: Quiescence and acquiescence as
response to perceived injustice. In G. R. Ferris (Ed.), Research in personnel and human
resources management (Vol. 20, pp. 331-369). Greenwich, CT: JAI Press.
Podsakoff, P. M., MacKenzie, S. B., Lee, J. Y., & Podsakoff, N. P. (2003). Common method
biases in behavioral research: A critical review of the literature and recommended
remedies. Journal of Applied Psychology, 88, 879-903.
Price, K. H., Hall, T. W., Van den Bos., K., Hunton, J. E., Lovett, S., & Tippett, M. J. (2001).
Features of the value function for voice and their consistency across participants from
four countries: Great Britain, Mexico, the Netherlands, and the United States.
Organizational Behavior and Human Decision Processes, 84(1), 95-121.
Raykov, T., & Marcoulides, G. A. (2006). A first course in structural equation modeling (2nd ed.).
Mahwah, NJ: Lawrence Erlbaum Associates Inc.
Redding, S. G., Norman, A., & Schlander, A. (1994). The nature of individual attachment to the
organization: A review of East Asian variations. In H. C. Triandis, M. Dunnette, & L.
Hough (Eds.), Handbook of industrial and organizational psychology (2nd ed., Vol. 4, pp.
557– 607). Palo Alto, CA: Consulting Psychologists Press.
Robinson, S. L. (1996). Trust and breach of the psychological contract. Administrative Science
Quarterly, 41, 574–599.

２０４

Robinson, S. L. & Morrison, E. W. (1995). Psychological contracts and OCB: the effect of
unfulfilled obligations on civic virtue behavior. Journal of Organizational Behavior, 16,
289–298.
Rollinson, D., & Broadfield, A., (2002). Organizational Behavior and Analysis—An Integrated
Approach, second ed. Pearson Education.
Rusbult, C. E., Johnson, D. J., & Morrow, G. D. (1986). Determinants and consequences of exit,
voice, loyalty and neglect: Responses to dissatisfaction in adult romantic involvements,
Human Relations, 39, 45–63.
Rusbult, C. E., & Farrell, D. (1983). A longitudinal test of the investment model: The impact on
job satisfaction, job commitment, and turnover of variations in rewards, costs,
alternatives, and investments. Journal of Applied Psychology, 68(3), 429-438.
Rusbult, C. E., Farrell, D., Rogers, G., & Mainous, A. G. III. (1988). Impact of exchange
variables on exit, voice, loyalty, and neglect: an integrative model of responses to
declining job satisfaction. Academy of Management Journal, 31, 599–627.
Rusbult, C. E., & Zembrodt, I. M. (1983). Responses to dissatisfaction in romantic involvements:
A multidimensional scaling analysis. Journal of Experimental Social Psychology, 19(3),
274-293.
Rusbult, C. E., Zembrodt, I. M., & Gunn, L. K. (1982). Exit, voice, loyalty, and neglect:
Responses to dissatisfaction in romantic involvements. Journal of Personality and Social
Psychology, 43, 1230-1242.
Rusbult, C. E., & Lowery, D. (1985). When bureaucrats get the blues: Responses to
dissatisfaction among federal employees, Journal of Applied Social Psychology, 15(1),
80-103.
Saunders, D. M. (1992). Introduction to research on Hirschman‘s exit, voice, and loyalty model,
Employee Responsibilities and Rights Journal, 5(3), 188-190.
Saunders, D. M., Sheppard, B. H, Knight, V., & Warshaw, P. R. (1992). Employee voice to
supervisors, Employee Responsibilities and Rights Journal, 5(3), 241-257.
Schaffer, B. S., & Riordan, C. M. (2003) A review of cross-cultural methodologies for
organizational research: A best-practices approach. Organizational Research Methods,
6(2), 169–215.
Schumacker, R. E., & Lomax, R. G. (2004). A beginner's guide to structural equation modeling
(2nd ed.). Mahwah, NJ: Lawrence Erlbaum Associates.
Sheppard, B. H., Lewicki, R. J., & Minton, J. W. (1992). Organizational Justice. The Search for
Fairness in the Workplace, Lexington Books, New York.

２０５

Singh, J. (1990). Voice, exit, and negative word of mouth behaviors: An investigation across
three service categories. Journal of the Academy of Marketing Sciences, 18, 1-15.
Singleton, R., Straits, B. C., Straits, M. M., & McAllister, R. J. (1988). Approaches to social
research. Oxford, UK: Oxford University Press.
Smith, P. B., Dugan, S., Peterson, M. F., & Leung, K. (1998). Individualism-collectivism and the
handling of disagreement: a 23 country study. International Journal of Intercultural
Relations, 22, 351–367.
Spector, P. E. (1987). Method variance as an artifact in self-reported affect and perceptions at
work: Myth or significant problem? Journal of Applied Psychology, 72, 438-443.
Spector, P. E. (1994). Using self-report questionnaires in OB research: A comment on the use of
controversial method. Journal of Organizational Behavior, 15, 385-392.
Spector, P. E. (1997). Job satisfaction: Application, assessment, causes, and consequences.
Thousand Oaks, CA: Sage.
Spector, P. E. (2006). Method variance in organizational research: Truth or urban legend?
Organizational Research Methods, 9(2), 221-232.
Spencer, D. G. (1986). Employee voice and employee retention. Academy of management
Journal, 29, 488-502.
Stroh, K., Northcraft, G., & Neale, M. (2002). Organizational behavior: A management hallenge.
Mahwah, NJ.
Suh, E., Diener, E., Oishi, S., & Triandis, H. C. (1998). The shifting basis of life satisfaction
judgments across cultures: Emotions versus norms. Journal of Personality and Social
Psychology, 74, 482–493.
Tang, F. Y., & Kirkbride, P. S. (1986). Developing conflict management skills in Hong Kong: an
analysis of some cross-cultural implications. Management Education and Development,
17(3), 287-301.
Tannen, D. (1990). You just don't understand: Women and men in conversation. New York:
Ballantine Books.
Thibaut, J., & Walker, L. (1975). Procedural justice: A psychological analysis. Hillsdale, NJ:
Erlbaum.
Thomas, K. W. (1992). Conflict and negotiation processes in organizations. In M. Dunette (Ed.),
Handbook of industrial and organizational psychology (2nd ed., Vol. 3, pp. 651-717).
Palo Alto, CA: Consulting Psychologists Press.

２０６

Ting-Toomey, S., & Oetzel, J. G. (2001). Managing intercultural conflict effectively. Thousand
Oaks, CA: Sage.
Tinsley, C. H. (1998). Models of conflict resolution in Japanese, German, and American cultures.
Journal of Applied Psychology, 83, 316–323.
Tinsley, C. H. (2001). How negotiators get to yes: predicting the constellation of strategies used
across cultures to negotiate conflict. Journal of Applied Psychology, 86, 583–593.
Triandis, H., C., Chan, D. K. S., Bhawuk, D. P. S., Iwao, S., & Sinha, J. B. P. (1995).
Multimethod probes of allocentrism and idiocentrism. International Journal of
Psychology, 30, 461–480.
Trochim, W. M. (2001). The Research Methods Knowledge Base. 2nd., Cornell University,
Cincinnati, OH.
Trubisky, P., Ting-Toomey, S., & Lin, S. L. (1991). The influence of individualism-collectivism
and self-monitoring on conflict styles. International Journal of Intercultural Relations,
15, 65-84.
Tse, D. K., Belk, R. W., & Zhan, N. (1988). Learning to Consume: A Longitudinal and CrossCultural Content Analysis of Print Advertisements from Hong Kong, People's Republic
of China and Taiwan. Journal of Consumer Research, 52(4), 81-95.
Turnley, W. H., & Feldman, D. C. (1998). Psychological contract violations during
organizational restructuring. Human Resource Management, 37, 71-83.
Tyler, T. R., Rasinski, K. A., & Spodick, N. (1985). Influence of voice on satisfaction with
leaders: Exploring the meaning of process control. Journal of Personality and Social
Psychology, 48, 72–81.
Vakola, M., & Bouradas, D. (2005). Antecedents and consequences of organizational silence: An
empirical investigation. Employee Relations, 27(4-5), 441-459.
Van Buren, H. J. III, & Greenwood, M. (2008). Enhancing Employee Voice: Are Voluntary
Employer-Employee Partnerships Enough? Journal of Business Ethics, 81, 209-221.
Van den Bos, K. (1999). What are we talking about when we talk about no-voice procedures? On
the psychology of the fair outcome effect. Journal of Experimental Social Psychology, 35,
560–577.
Van den Bos, K. (2005). What is responsible for the fair process effect? In J. Greenberg & J. A.
Colquitt (Eds.), Handbook of organizational justice: Fundamental questions about
fairness in the workplace (pp. 273–300). Mahwah, NJ: Erlbaum.

２０７

Van den Bos, K., & Spruijt, N. (2002). Appropriateness of decisions as a moderator of the
psychology of voice. European Journal of Social Psychology, 32, 57–72.
Van der Vegt, G. S., & Van de Vliert, E. (2005). Effects of skill dissimilarity and task
interdependence on helping in work teams. Journal of Management, 31, 73-89.
Van Dyne, L., Ang, S., & Botero, I. C. (2003). Conceptualizing employee silence and employee
voice as multidimensional constructs. Journal of Management Studies, 40(6), 1359-1392.
Van Dyne, L., & LePine, J. A. (1998). Helping and voice extra-role behavior: evidence of
construct and predictive validity. Academy of Management Journal, 41, 108–19.
Van Dyne, L., Cummings, L. L., & McLean Parks, J. (1995). ‗Extra-role behaviors: in pursuit of
construct and definitional clarity (a bridge over muddied waters)‘. In Cummings, L. L.
and Staw, B. M. (Eds), Research in Organizational Behavior, Vol. 17. Greenwich, CT:
JAI Press, 215–85.
Van Dyne, L., Graham, J. W., & Dienesch, R. M. (1994). Organizational citizenship behavior:
construct redefinition, operationalization, and validation. Academy of Management
Journal, 37, 765–802.
Van Dyne, L., Kamdar, D. A., & Joireman, J. (2008). In-role perceptions buffer the negative
impact of low LMX on helping and enhance positive impact of LMX on voice. Journal of
Applied Psychology, 93(6), 1195-1207.
Van Yperen, N. W., Hagedoorn, M., Zweers, M., & Postma, S. (2000). Injustice and employees‘
destructive responses: The mediating role of state negative affect. Social Justice Research,
13, 291–312.
Wagner, J. A., III (1995). Studies of individualism-collectivism: Effects on cooperation in
groups. Academy of Management Journal, 38, 152–172.
Walumbwa, F. O., & Schaubroeck, J. (2009). Leader Personality Traits and Employee Voice
Behavior: Mediating Roles of Ethical Leadership and Work Group Psychological Safety.
Journal of Applied Psychology, 94(5), 1275–1286.
Wasti, S. A. (2002). Affective and continuance commitment to the organization: Test of an
integrated model in the Turkish context. International Journal of Intercultural Relations,
26, 525–550.
Watson, C. J. (1981). An evaluation of some aspects of the Steers and Rhodes model of
employee attendance. Journal of Applied Psychology, 66, 385-389.
Wilson, N., & Peel. M. J. (1991). The Impact on Absenteeism and Quits of Profit-Sharing and
Other Forms of Employee Participation. Industrial and Labor Relations Review, 44(3),
454–468.

２０８

Withey, M. J., & Cooper, W. H. (1989). Predicting exit, voice, loyalty, and neglect.
Administrative Science Quarterly, 34, 521–39.
Xu, H., Van de Vliert, E., & Van der Vegt, G. S. (2005). Breaking the silence culture:
Stimulation of participation and employee opinion withholding cross-nationally.
Management and Organization Review, 1(3), 459-482.
Zald, M. N., & Berger, M. (1978). Social movements in organizations: Coup d'etat, insurgency,
and mass movements. American Journal of Sociology, 83, 823-861.
Zhou, J., & George, J. M. (2001).When job dissatisfaction leads to creativity: encouraging the
expression of voice'. Academy of Management Journal, 44, 682–96.

２０９