XX 7! .

4... u .
.. :fi
5.] . .ll _
‘ nx .Kaaathfia t ,
.18 :33}? . in
.531. g .23: .
.. i mNphifi a. .

a , .
L 3:2. It... lulv
: 3‘5.

s: w.
.
E. 3?»...5, .1. .
n?
u A

. . 1...};
by 1.6.0.an
.. .ut?
é.

3M.

‘ J
hufidfi
5i... . .
5.9.: flu . r .
... I. .3 “MR-"fl. in:
£4.31; 9.
~ I '0’
film”? a“!
la Fir

330...]... r. n
n .. but...”

.
. 4 5...?
. lentil 1,
3. (I 7:.
(1 «Lu. 1. alt-.73.,
!,

I! II
1 ._ r E
{Wining
,

flaming.

 

 

A!

1‘ fl
0 {a :

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

A MULTILEVEL ANALYSIS OF QUALITY MANAGEMENT
PRACTICES, COOPERATIVE CULTURAL VALUES AND WORK
PERFORMANCE

presented by

THOMAS JAMES KULL

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Department of Marketing and
Supply Chain Management

 

 

16614,“ 7k “Ck-7W5“ _

 

Major Professor’s Signature
I 2- / 4 1/ 0?

Date

MSU is an affmnalive-action, equal-opportunity employer

—.—-—.-._,-.—.—._.-.-.~n-.-A-— -

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE
w 1 z o :3
FEB 1 1 2009

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/CIRC/DateDue.indd-p.1

 

 

 

A MULTILEVEL ANALYSIS OF QUALITY MANAGEMENT PRACTICES,
COOPERATIVE CULTURAL VALUES AND WORK PERFORMANCE

By

Thomas James Kull

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Marketing and Supply Chain Management

2007

 

 

 

pare

to (I
ma:
the:
ma;
tha‘

DEr

ABSTRACT

A MULTILEVEL ANALYSIS OF QUALITY MANAGEMENT PRACTICES,
COOPERATIVE CULTURAL VALUES AND WORK PERFORMANCE

By
Thomas James Kull

Unsuccessful quality initiatives often are attributed to an organizational
culture that does not value the cooperative nature of quality management
practices. However, two opposing perspectives exist as to how quality practices
may relate to cooperative cultural values, which is problematic for deciding where
to allocate resources during a quality initiative. Moreover, how the performance
effects of quality practices and cooperative cultural values change over time is
unknown, which has important epistemic implications.

This research finds evidence for a reconceptualization that resolves the
paradox created by the two opposing perspectives. First, secondary data is used
to discriminate between organization-level and workgroup-level quality
management practices. Second, support is found for a sociotechnical system
theory-based explanation as to how cooperative cultural values relate to quality
management within a multilevel model. Third, longitudinal evidence is provided
that shows cooperative cultural values increasing in influence on workgroup
performance while organization-level quality practice decrease in influence.

Based upon the results, quality management practices are seen as
serving the dual role of instilling cooperative values at the organization level and
enabling higher performance at the workgroup level. Also, managerial insights

are provided that recommend a simultaneous approach to changing cultural

valu'

man

their

 

 

values and implementing quality management practices. Finally, quality
managers from several manufacturing facilities were presented these results and

their insights explain why further study is needed on this topic.

Copyright by
THOMAS JAMES KULL

2007

Dedication

This dissertation is dedicated to my son, Adam James Kull. He is my inspiration
for being the best person I can possibly be.

 

 

encI
Cha
thro
beg

alsr

this
ste
DE
0t
IE

IC

ACKNOWLEDGEMENTS

This dissertation would not have come into existence without the
encouragement and dedication of my family and friends. My parents, David and
Charlene Kull, were always there to support me in many different ways
throughout my doctoral years. Also, the courage to end an industry career and
begin an academic one emanated from my good friend Noelle Bowman, who

also aided in editing this manuscript.

The most prominent person to have impacted the strength and quality of
this dissertation is my dissertation chair, Ram Narasimhan. Not only did his
steadfast guidance and wisdom make this research valuable, but his insightful
perspectives on work, family, and life have changed me forever. In addition, the
other members of my dissertation committee have each made a difference in this
research and in my career. Gary Ragatz has from the beginning encouraged me
to explore new topics and to focus on what really matters. Roger Calantone was
instrumental in developing my methodological skills and scientific paradigm.
Morgan Swink taught me how to look critically at my own work so it attains the

highest standards of our discipline.

Finally, I thank the people who were doctoral students with me in the
Marketing and Supply Chain Department at Michigan State University. I
especially acknowledge the people in my entering year; Andrea Prudhomme,

Wenming Chung, and Laird Burns. Each of them supported and challenged me

vi

along the way. They made the character of my doctoral studies to be as special

and unique as they are.

vii

TABLE OF CONTENTS

LIST OF TABLES ................................................................................................. xi
LIST OF FIGURES ..................................................................... xiii
CHAPTER 1: INTRODUCTION ............................................................................ 1
1.1 Research Topic ................................................................................. 1
1 .2 Motivations ........................................................................................ 3
1.2.1 Epistemic Motivations ........................................................................ 3
1.2.2 Practical Motivations ......................................................................... 4
1.3 Meso Paradigm ................................................................................. 6
1 .4 Methodology ..................................................................................... 7
1 .5 Contributions ..................................................................................... 8
1.6 Dissertation Outline ........................................................................ 10
1.7 Summary of Introduction ................................................................. 1 1
CHAPTER 2: LITERATURE REVIEW ................................................................ 12
2.1 Quality Management ....................................................................... 12
2.1.1 Deming’s Framework ...................................................................... 13
2.1.2 The Multilevel Nature of Quality Management ................................ 16
2.1.3 Human Issues in Quality Management ............................................ 18
2.1.4 Performance Implications of Quality Management .......................... 21
2.1.5 Dynamics of Quality Management ................................................... 23
2.2 Organizational Values ..................................................................... 25
2.2.1 Organizational Cultural Values ........................................................ 26
2.2.2 Cooperative Cultural Values ............................................................ 29
2.3 Pertinent Empirical Studies ............................................................. 31
2.3.1 Quality Management and Organizational Culture Studies ............... 32
2.3.2 Practice-Culture Empirical Studies .................................................. 33
2.4 Sociotechnical Systems Theory ...................................................... 36
2.4.1 Sociotechnical System Frameworks ................................................ 37
2.4.2 Reciprocal Influences Between Social and Technical Systems ...... 39
2.4.3 Sociotechnical System Design ........................................................ 41
2.5 Summary of Literature Review ........................................................ 44
CHAPTER 3: RESEARCH FRAMEWORK ......................................................... 46
3.1 Research Perspective ..................................................................... 46
3.2 Development of Hypotheses ........................................................... 48
3.2.1 Organization-level Quality Management Instills Cooperative Values
........................................................................................................ 11S)
3.2.2 Organization-level Quality and Workgroup-level Quality ................. 50
3.2.3 Cooperative Values Encourage Workgroup-level Quality ................ 52

viii

3.2.4 Quality Management at the Organization Level Improves Workgroup
Performance ................................................................................... 54
3.2.5 Cooperative Cultural Values Help Workgroup Performance ........... 55
3.2.6 Workgroup Performance is Improved by Workgroup-level Quality
Management ................................................................................... 55
3.2.7 Diminishing Effects of Organization-level Quality Management ...... 56
3.2.8 Diminishing Effects of Cooperative Values ...................................... 59
3.2.9 Workgroup-Level Quality Management Improves With Time .......... 60
3.3 Research Models ............................................................................ 61
3.4 Summary of Research Framework ................................................. 63
CHAPTER 4: RESEARCH METHODOLOGY ..................................................... 65
4.1 Data Collection ............................................................................... 65
4.1.1 Data Collection Approach ................................................................ 66
4.1.2 The National Partnership for Reinventing Government ................... 68
4.1.3 Sampling ......................................................................................... 71
4.1.4 Assessment of NPR Survey ............................................................ 73
4.2 Measurement Model ....................................................................... 75
4.2.1 Background to Model Development ................................................ 75
4.2.2 Organization-level Quality Practices (00) ........................................ 78
4.2.3 Workgroup-level Quality Practices (QW) .......................................... 79
4.2.4 Organization-level Cooperative Values tS/VO) .................................... 80
4.2.5 Workgroup-level Work Performance (P ) ....................................... 81
4.3 Data Analysis Method ..................................................................... 81
4.3.1 Factor analysis ................................................................................ 83
4.3.2 Data aggregation ............................................................................. 84
4.3.3 Hierarchical Linear Modeling ........................................................... 85
4.3.4 Estimation Procedure ...................................................................... 92
4.4 Testing Diffusion Hypotheses ......................................................... 92
4.5 Summary of Research Methodology ............................................... 93
CHAPTER 5: ANALYSIS RESULTS ................................................................... 95
5.1 Data validation ................................................................................ 95
5.2 Measurement model construction and validation ............................ 98
5.2.1 Confirmatory Factor Analysis .......................................................... 98
5.2.2 Unidimensionality .......................................................................... 100
5.2.3 Discriminant Validity ...................................................................... 101
5.2.4 Predictive Validity .......................................................................... 102
5.3 Common method variance control ................................................ 103
5.3.1 Preliminary assessments .............................................................. 104
5.3.2 Correlated uniqueness .................................................................. 104
5.4 Factor Scoring .............................................................................. 106
5.5 Data aggregation .......................................................................... 109
5.6 Hierarchical Linear Model Results ................................................ 116
5.6.1 Results for 1998 ............................................................................ 117
5.6.2 Results of 1999 ............................................................................. 121

5.6.3 Results of 2000 ............................................................................. 124

5.7 Parameter stability ........................................................................ 125
5.8 Summary of Data Analysis ............................................................ 128
CHAPTER 6: DISCUSSION ............................................................................. 130
6.1 Knowledge of Quality Management .............................................. 130
6.1.1 Single Period Results (1998) ......................................................... 130
6.1.2 Multi-period Results (1998-2000) .................................................. 133

6.2 Knowledge of Sociotechnical System ........................................... 136
6.2.1 Single Period Results (1998) ......................................................... 136
6.2.2 Multi-period Results (1998-2000) .................................................. 138

6.3 Managerial insights ....................................................................... 139
6.3.1 Insights from a Single Period (1998) ............................................. 140
6.3.2 Multi-period Results (98-00) .......................................................... 142

6.4 Summary of Discussion ................................................................ 144
CHAPTER 7: CONCLUSION ............................................................................ 146
7.1 Summary of Research .................................................................. 146
7.2 Reactions from Managers ............................................................. 148
7.3 Limitations of study ....................................................................... 150
7.4 Future research ............................................................................ 151
7.5 Summary ...................................................................................... 153
APPENDIX ........................................................................................................ 154
REFERENCES ................................................................................................. 156

...........

 

 

 

LIST OF TABLES

Table 2-1: Concepts within the Deming Management Model .............................. 15
Table 2-2: Cultural Values Relatedto TQM ........................................................ 20
Table 2-3: Characteristics of Sociotechnical Systems ........................................ 39
Table 2-4: Variations of STS Principles .............................................................. 42
Table 4-1: National Partnership for Reinventing Government Employee Survey,
1998 .................................................................................................. 70
Table 4-2: Federal Agencies Participating in NPR Survey .................................. 72
Table 4-3: Initial Construct Measures ................................................................. 76
Table 4-4: Hypothesis Testing Framework ......................................................... 86
Table 4-5: Expected Changes in Effects ............................................................. 93
Table 5-1: Percent of Responses with IVIissing Values ....................................... 96
Table 5-2: Final Measurement Model from 1998 Data ........................................ 99
Table 5-3: Unidimensionality Test - Monomethod ............................................. 101
Table 5-4: Discriminant Validity ......................................................................... 102
Table 5-5: Assessment of Predictive Validity .................................................... 103
Table 5-6: Effect of Correlated Uniqueness - Common Method Variance (CMV)
Control ............................................................................................. 105
Table 5-7: Unit-Weight Factor Scoring Procedure ............................................ 107
Table 5—8: Correlation Matrices of Factor Scores .............................................. 108
Table 5-9 Overall Descriptive Statistics ........................................................... 109
Table 5-10: Agency Means per Year ................................................................ 111
Table 5—1 1: Results from HLM analysis for NPR - 1998 ................................... 118
Table 5-12: Results from HLM analysis for NPR - 1999 ................................... 122
Table 5—13: Results from HLM analysis for NPR - 2000 ................................... 123

xi

" ‘ s- I I": .

 

 

 

Table 5-14: Structural change from 1998 to 1999 ............................................. 127

Table 5-15: Structural change from 1999 to 2000 ............................................. 127
Table A-1: Difference Statistics Between Actual and Imputed Values .............. 154
Table A-2: Matrices from Factor Scoring Procedure on 1998 Data .................. 155

xii

 

 

 

 

Figt
FigL
FigL
Figt

Fig.

 

LIST OF FIGURES

Figure 2-1: Theoretical Grounding ...................................................................... 44
Figure 3-1: Expected Relationships Within a Single Time Period ....................... 61
Figure 3-2: Qualitative Depiction of Expected Changes Over Time .................... 62
Figure 3—3 : Expected Direction of Change in Relationships Over Multiple Periods

......................................................................................................... (313
Figure 4-1: Flow Chart of Analysis Method ......................................................... 82

xiii

CHAPTER 1: INTRODUCTION

This chapter introduces the research topic for this dissertation and
describes the motivations for pursuing this topic. The meso-paradigm adopted
by this dissertation is presented, followed by the research methodology used.
Contributions to both theory and practice are then discussed. The outline of the

dissertation concludes this chapter.

1.1 Research Topic

This dissertation examines an aspect of the interface between operations
management practice, organizational culture, and performance. Specifically this
dissertation analyzes the relationship between quality management practices,
cooperative cultural values, and work performance. The knowledge that quality
management often has cultural value issues has been well established (Bright
and Cooper, 1993; Huq, 2005; Jabnoun and Sedrani, 2005; Moore and Brown,
2006). Literature has associated successful implementation of quality
management practices with organizational cultures possessing such
characteristics as cooperative values (Detert, Schroeder, and Cudeck, 2003;
Detert, Schroeder, and Mauriel, 2000; Kujala and Lillrank, 2004).

However what is not clear is how quality management practices are
related to cooperative cultural values. Some research has argued that certain

cultural values are prerequisites to successful operations management practices

(Nahm, Vonderembse, and Koufteros, 2004), while other research has argued
that certain cultural values are a consequent of operations management
practices (Naveh and Erez, 2004). This dissertation offers clarity to this apparent
paradox through use of an insight suggested by Juran (1989) who distinguished
between high level quality management practices and lower level quality
management tools. Juran (1989) designated these two levels of quality
management as ““big 0”” and “little O." This dissertation utilized this insight to
provide a richer conceptualization of how quality management practices are
related to cooperative cultural values.

Although this research is restricted to cooperative cultural values and
quality management practices, general inferences can be made to operations
management practices and organizational culture. Tasks and routines embody
how an organization creates value and performs its overall purpose (Thompson,
1967). Improvement initiatives alter the form of these tasks in some way. A
quality improvement initiative is just one of these ways. In addition, an
organization is a social construction (March and Simon, 1958). People within the
organization have to cope with how to work together and how to adapt to external
forces. This is how cultures form, through shared acceptance of certain beliefs,
values and norms (Schein, 2004). Values are a useful proxy for culture as a
whole (Chatman and Barsade, 1995). Hence this study helps the operations
management field understand how an organization’s practices, culture, and

subsequent effectiveness interrelate.

The principal research questions of this dissertation are the following:
What role, if any, do cooperative cultural values play in a quality management
improvement initiative; and do the practice-value-perforrnance relationships
change over time? The next section elucidates the motivations for pursuing

these research questions.

1 .2 Motivations

This section describes both the epistemic and practical reasons for
researching the relationships among quality management practices, cooperative

cultural values and work performance.

1.2.1 Epistemic Motivations

A resolution to the question of how quality management practices are
related to cooperative cultural values will have important theoretical value. At
issue is the nature and effect of both practices and values. The claim that certain
cultural values are antecedent to quality management implies that cultural values
enable quality management to occur (Kujala et al., 2004; Miron, Erez, and
Naveh, 2004). This means that social-cultural aspects of an organization will
prevent or discourage quality management implementation. Such a
characterization offers cultural values to be unchanging and highly influential.
The other claim that certain cultural values are consequent to quality

management implies that cultural values are enabled by quality management

(Boggs, 2004; Naveh et al., 2004). This means that social-cultural aspects of an
organization will change or be encouraged by quality management
implementation. This perspective sees cultural values to be changeable and
influenceable. The current literature on quality management and cultural values
is unable to resolve why these two perspectives coexist. The typical approach is
to adopt one of the two perspectives and leave the other unexamined. This
dissertation is motivated by a need to find evidence for a research framework
that incorporates both perspectives.

Another theoretical issue pertains to how the relationships among quality
management practices, cooperative cultural values and work performance
change over time. The literature on quality management has noted the
importance of the time dimension when examining the effects of quality initiatives
(Narasimhan and Mendez, 2001; Schroeder, Linderrnan, and Zhang, 2005);
however, few empirical analyses incorporate time (Hendricks and Singhal,

2001 b). If a quality initiative can influence and be influenced by cooperative
cultural values, do the relationships remain stable over time? Investigating the
time oriented effects among quality management practices, cooperative cultural

values and work performance are a motivation for this dissertation.

1.2.2 Practical Motivations

Many attempts at quality improvement fail to reap the rewards the
organizational leaders intended to achieve (Hendricks et al., 2001b). A major

reason given for failure is organizational culture (Detert et al., 2000), of which

cultural values are a part. That is, the organization’s employees reject the main
tenets of the improvement initiative because it conflicts with the prevailing ‘way
things are done.’ Because cultural values affect how people perceive the world
(Schein, 2004) and strongly influence norms of behavior (Dewey, 1939), attempts
to improve an operation possessing values that are counter to the initiative is
susceptible to rejection. Organizational culture has been identified as the main
stumbling block in many companies. For instance, General Motors has tried
numerous times to spread lean manufacturing practices via their NUMMI and
Saturn operations, but with limited success (lnkpen, 2005). GM’s culture was
said to be a main problem.

With this in mind, prescriptive advice has been to align cultural values with
the desired practice before attempting to implement the practice (Flanagan,
1995; McDermott and Stock, 1999). The argument is that without the appropriate
values, the initiative at best will achieve mediocre results and at worst speed the
organization toward failure. Embarking on cultural value change initiatives has
therefore been advocated as the prescription for true improvement (Kujala et al.,
2004). Two issues exist with this resource allocation prescription. First, how
much time can the organization afford to take to change cultural values before
embarking on the needed improvement? Culture change takes time, and the
process isn’t exact (Ogbonna and Harris, 1998). The multitude of social
networks and groups within an organization, even relatively small ones, is difficult
to identify and influence. Also, the effects of social networks can have

unexpected consequences, which mean corrective actions will be needed along

 

 

 

 

 

 

me

Va
ma
ml

be

the way (Harris and Ogbonna, 2002). Second, studies have shown that
organizational cultural values can change as a result of improvement initiatives
(Boggs, 2004; Naveh et al., 2004). Cultural values congruent with an excellent
practice may be a result of practice implementation, not an antecedent.
Therefore, a motivation for this dissertation is to provide practical advice as to

where managers should allocate resources before and during a quality initiative.

1.3 Meso Paradigm

In general the practice-culture relationship, and in particular the quality
management-cooperative cultural values relationship, is a paradox (Poole and
Vandeven, 1989) that is unresolved by extant literature. However, the paradox
may be resolvable by reconceptualizing this issue and adopting a multilevel or
meso-paradigm (House, Rousseau, and Thomashunt, 1995). Discriminating
between different levels of quality management is suggested by Juran’s (1989)
“big Q” and “little Q” insight.

On adopting the meso—paradigm, cooperative cultural values are seen to
form and create social norms that exist at the organization level, but their
influence manifests at the workgroup or individual level (Martocchio, 1994;
Oreilly, Chatman, and Caldwell, 1991 ). This recognition of cross-level influence
(i.e., from a higher to lower level or vice versa) is part of adopting a meso-
paradigm. Moreover, recognition of within level influences should occur too.
That is, effects that exist at either the organization level or workgroup level.

Therefore by adopting the meso—paradigm, a conjecture can be made that quality

 

 

 

 

 

man

leve

to it

mar

leve

COD»:

and

 

The
sho
sec

par.

1.4

Sy:
Dl‘é
the

50

SI.

th

management may be antecedent to cooperative values within the organization
level, but these values may in turn be antecedent to quality management across
to the workgroup level. In addition, the effect of organization-level quality
management practices may diminish over time, while the effect of workgroup-
level quality management may increase. These multilevel and cross-level
conceptualizations have been theoretically useful elsewhere (Hofmann, 1997)
and were adopted in this dissertation for epistemic and practical purposes.
Therefore, support for a meso—paradigm has implications for where managers
should devote resources during a quality management initiative. The next
section will review the methodology used to seek support for this meso-

paradigm.

1.4 Methodology

This dissertation tested a multilevel model that utilized sociotechnical
systems (STS) theory to explain the relationships among quality management
practices, cooperative cultural values and work performance. Utilizing STS
theory, hypotheses were made regarding what relationships were expected to be
supported empirically. In addition, diffusion hypotheses were developed for the
multi-year comparisons. The intention of this dissertation is to test the multilevel
model on a secondary data source. This archival data came from a multi-year
survey of US federal government agencies during an enterprise-wide quality
management initiative (O.P.M., 2002). Quality management and organization

theory literatures were used to develop a well grounded measurement model that

 

1.5

mar
leve
was.
effe
mar
cou

imp

am

3L:

 

was subsequently supported. The multilevel model was then tested against each
of the survey years. Results from the HLM data analysis were utilized to draw
conclusions regarding the single period and multi-period hypotheses. The result
of this process was a richer conceptualization of the relationship between quality
management, cooperative cultural values and work performance. The next

section provides an overview of the dissertation findings and contributions.

1.5 Contributions

The first important finding that makes a substantial contribution to quality
management knowledge is the empirical discrimination between organization-
level and workgroup-level quality management practices. This discrimination
was not only validated in the theory-based measurement model, but also in its
effects within the research model. That is, these two levels of quality
management practice influenced work performance in different ways over the
course of the quality initiative. This differentiation in behavior further supports the
importance of acknowledging the multilevel nature of quality management.

The second important finding is the support for the STS theory—based
conjecture for how cooperative cultural values are related to quality
management. That is, it was found that the archival data supported the
hypothesis that organization-level quality management practices are antecedent
to organization-level cooperative cultural values, and that these values are in turn
antecedent to workgroup-level quality management practices. Such a finding

supports the proposed resolution to the aforementioned practice-value relational

 

 

 

 

para
SYSII
ml
coo;
ma l‘

impl

 

 

 

fram

exa'
abil
thrc
ma
coc

Dre

ml

dis

hi

Dr

It:

paradox. Moreover, this result also supports the applicability of sociotechnical
systems theory to human issues in operations management. The implication for
managers is that a quality initiative can be concurrent to the instillation of
cooperative values, rather than post hoc. This finding therefore may help
managers avoid unnecessary time delays in quality management
implementation. Support for the proposed STS theory-based research
framework therefore contributes both to practice and to theory.

A third finding that should be noted is the deeper understanding gained by
examining and comparing the research framework over multiple years. The
ability to assess the dynamics of how quality management practices diffuse
through the organization and affect performance is unique in the operations
management literature. In addition, it was found unexpectedly that the role of
cooperative cultural values becomes more prominent over time; it becomes a key
predictor of workgroup-level performance and quality management practice
implementation — more so than organization-level quality practices. Without this
multiple year examination, such a finding could not have been made. Thus, this
dissertation contributes to the knowledge regarding the dynamics of quality
management practices, cooperative cultural values and work performance.

Finally, when the results of this research were presented to quality
managers, there was unanimous agreement that cooperative cultural values are
highly influential in determining the permanence of quality management
practices. They also concurred with the claim of sociotechnical system theory

that cultural values can both be influenced by management practices. However,

 

 

 

the r

prac

 

qual'
capt
diss
furl?r

out

1.6

COR

 

and
TBSI
exa
inve
COI‘
me
the
dis
8 c

an

the managers found it non-intuitive that organization-level quality management
practices would be less influential over time. In addition, the difference between
quality management and cooperative values was difficult to distinguish. By
capturing the reactions of quality managers to the research findings, this
dissertation makes an important link between theory and practice and motivates
further research. Having described the major findings from this research, the

outline of the dissertation is next presented.

1.6 Dissertation Outline

The next chapter will review the extant literature related to the pertinent
concepts in this dissertation — i.e., quality management, organizational values,
and sociotechnical system theory — as well as empirical studies related to this
research topic. Chapter 4 will develop the multilevel research framework used
examined in this dissertation. Six hypotheses were developed for single-period
investigation while three diffusion hypotheses were developed for year-over-year
comparisons. Chapter 5 provides a detailed description of the research
methodology used. This chapter also provides extensive detail with respect to
the archival data. In chapter 6 the results of the data analysis are presented and
discussed, with the implications for these results described in chapter 7. Chapter
8 concludes the dissertation with a summary, managerial comments, limitations,

and suggestions for future research.

10

1.7 Summary of Introduction

In this chapter the theoretical paradox of how quality management relates
to cooperative cultural values was presented as a motivation for this research.
Also discussed was the method used to find empirical support for a multilevel
model, which suggests a resolution to the paradox. Finally, highlights of the
contributions to knowledge and practice were made explicit. To understand how
these research findings make a contribution, extant literature must be reviewed.

This is accomplished in the next chapter.

11

 

diss'l
SDEC
the l

of qLI

 

 

 

reve
(Mga
\Nhh‘
to<j
INhE
in ft
con

WOr

2.1

'IhE

de\

The

CHAPTER 2: LITERATURE REVIEW

In this chapter, the pertinent literature will be reviewed and related to this
dissertation. The first section will review research on quality management, with
special attention given to the multilevel nature of quality management practices,
the human issues within quality management and the performance implications
of quality management. The second section reviews organizational values to
reveal how extant literature has highlighted these as an important aspect of
organizational culture and to review the pertinence of cooperative cultural values
within this dissertation's context. Empirical studies are presented in section three
to discuss the topics related to this dissertation that have been examined and
where further work is needed. Finally, sociotechnical systems theory is reviewed
in the fourth section. This theory serves as the basis for a richer
conceptualization in this of how quality management, cooperative values and

work performance are related.

2.1 Quality Management

Quality management is a well researched operations management topic.
The roots of quality management trace back to the 1920s when W.A. Shewhart
developed the process control techniques at the Bell Telephone Laboratories.
The success of such tools as SPC were recognized and communicated by W.E.

Deming, J. Juran, P.B. Crosby, and K. lshikawa from the 19503 to the 19805.

12

More general quality management ideals were added by these authors and
others. Historically, quality management has been studied in the manufacturing
environment, but more recently it has been examined in the services context -
for example, in schools (Detert et al., 2003), hospitals (Douglas and Fredendall,
2004), and government (Taveira, James, Karsh, and Sainfort, 2003).

Quality management practices have the goal of improving organizational
effectiveness through the means of high quality processes and outputs. Unlike
other operations management practices, like lean manufacturing and mass
customization, quality management has a longer research history and thus other
business disciplines — such as management and organizational behavior — have

examined the topic as well.

2.1.1 Deming’s Framework

Deming’s management model was a famous example of the
management-level counterpart to the technical-level quality management
practices. This model contained 14 principles to guide management practices
toward the goal of superior quality and operating performance (Deming, 1986).
Deming’s model has implications for an organization’s top-management;
however, it was intended to reach all levels of the organization. A main theme of
the 14 points focuses on how to manage variability, which Deming saw as
inherent in all processes and as essential to organizational survival (Gartner and
Naughton, 1988). The model is applicable to multiple functions in an

organization, and includes customers and suppliers (Gartner et al., 1988).

13

 

 

inte
(19!
sen
und

SUIT

Eacl
colu
The

the

be 5
Que
this
Free:
hos
mar

De.

 

The Deming management model was later organized into seven
interrelated theoretical constructs by Anderson, Rungtusanatham, and Schroeder
(1994), which they said provided a foundation for its use in both products and
services. The framework by Anderson et al. (1994) provides a basis for
understanding the main themes of Deming’s management model and is
summarized by the authors in the following statement:

The effectiveness of the Deming management method arises from

leadership efforts toward the simultaneous creation of a cooperative and

learning organization to facilitate the implementation of process-
management practices, which, when implemented, support customer
satisfaction and organizational survival through sustained employee
fulfillment and continuous improvement of processes, products, and
services.
Each construct is shown in Table 2-1 along with its definition in the second
column. Examples of management practices are provided in the third column.
The meso-paradigm as described in Chapter 1 is applied to this framework and
the resultant organizational levels are shown in column four.

Although the original intent of Deming’s management model was for it to
be applicable to both products and services, the large majority of studies in
quality management have been focused on manufacturing. Only recently has
this quality management framework been applied to services. Douglas and
Fredendall (2004), for instance, tested the Anderson et al. (1994) framework in a
hospital environment. Their results were similar to the earlier studies in

manufacturing environments (Anderson, Rungtusanatham, Schroeder, and

Devaraj, 1995; Rungtusanatham, Forza, Filippini, and Anderson, 1998).

14

 

Table 2-1: Concepts within the Deming Management Model*

 

Concept

Management practices
exemplifying concept

Related
Deming
Princfliles

Meso-paradigm
level

 

Visionary
Leadership

Internal and
external
cooperation

Learning

Process
Management

Continuous
Improvement

Employee
Fulfillment

Customer
Satisfaction

Clarity of vision, long-range
orientation, coaching
management style, participative

change, employee

empowerment, and planning

and implementing

organizational change

Firm-supplier partnership,
single supplier orientation,
collaborative organization,
teamwork, organization-wide
involvement, systems view of
the organization, trust and

elimination of fear.

Company-wide training,
foundational knowledge,

process knowledge,

educational development,
continuous self-improvement,
and managerial learning.

Management of process,
prevention orientation,
reduction of mass inspection,
design quality, statistical
process control, understanding
variation, elimination of
numerical quotas, elimination of
management by objective,
elimination of merit-reward
systems, understanding
motivation, total cost
accounting and stable

employment.

Continuous improvement

activities

Job satisfaction, job

commitment, and pride of

workmanship

Customer-driven focus

1,2,7,
14

4, 7,8,9

6,12,13

3, 4. 5.
10,11,
12

1,5

10,11,
12

Implicit

Organization level

Organization level
Work-group level

Organization level
Work-group level
Individual level

Work-group level
Individual level

Work—group level
Individual level

Individual level

Organization level
Work-group level

 

 

* Adapted from Anderson et al. (1994)

 

 

Other studies have also found evidence for the efficacy of quality management in
services — for instance Hays and Hill (2001) found support for portions of the
Anderson et al. 1994 framework in hotel settings, while Zhao, Yeung, and Lee
(2004) found the use of quality management systems in wholesaling, retailing,
and trading industries in China.

This dissertation will utilize Deming’s (1986) management model and
Anderson et al.’s (1994) conceptual descriptions to characterize quality
management practices. Although other models are available, such as the
Malcolm Baldridge Quality Award Criteria or the ISO-9000 quality standard,
Deming’s model has a longer history and has been researched in both the
operations management and organizational behavior literatures. Moreover,
Deming’s (1986) model has been used as a basis for relating quality

management to specific organizational cultural values (Detert et al., 2000).

2.1.2 The Multilevel Nature of Quality Management

A concept that has been proposed by Juran is the “big Q and little Q”
distinction in quality management (Juran, 1989). “Big Q” refers to the quality
management practices associated with the overall functioning of the organization
such as continuous improvement in all activities; efficient deployment of
resources; employee, supplier, and customer development and recognition; and
the full participation of employees. In contrast, “little Q” designates those quality

management practices associated with specific tools, techniques, and activities

16

associated with quality control — such as statistical process control, pareto
analysis, and root cause analysis. This “big Q” / “little Q” differentiation highlights
the fact that quality management can exist at different operational and
organizational levels.

Juran’s (1989) insight suggests that there is value in approaching quality
management from a meso-paradigm — that is, examining quality management at
multiple levels in an organization. By taking the multilevel perspective,
researchers can bound the scope of their studies and managers can better
identify the system in which quality management operates. Previous empirical
work in quality management has predominantly been at one level, either the
organization level or the work group (e.g.,, team) level. Examples of studies at
the organization level are: Douglas and Fredendall (2004), Rungtusanatham et
al. (1998), and Hendricks and Singhal (2001a). Examples of research studies in
quality management at the workgroup level or lower include: Waldman (1994)
,who conceptualized quality management issues at the individual-level; Zbaracki
(1998), who used qualitative data to present quality management issues at the
managerial level; and Soltani, Van der Meer, and Williams (2004), who argued
for the importance of the individual in quality management practices. Relatively
few studies have adopted the meso—paradigm to examine cross-level influences
in quality management; Naveh and Erez’s (2004) study dealing with quality
practices in workgroups is one recent example where teams and plants were

examined simultaneously.

17

 

 

 

quai
SUQQ
an 0

and

 

qua.
(fihe
and
Jura
coo

Cros

2.1

to
0F

be

te

Using Juran’s “big Q” / “little Q” concepts, this dissertation investigates
quality management practices at multiple levels in an organization. The literature
suggests that quality management practices are implemented at multiple levels in
an organization — the organization level, the worksite level, the workgroup level
and the individual level. However, what is not well understood is whether or not
quality management practices interact with other organizational aspects
differently at different levels. The conjecture that such interactions might exist
and that they might differ significantly at different levels follows directly from
Juran’s insight regarding “big Q and little O”. For example, the skills and
coordination requirements differ significantly between the “big Q” practice of

cross-functional teams and the “little Q” practice of statistical process control.

2.1.3 Human Issues in Quality Management

According to Schroeder, Linderrnan and Zhang (2005, p. 472) in relation
to quality management, “Research in human resources presents a great
opportunity for further insights, and draws on the vast literature in organizational
behaVior. This literature can be used to develop interesting insights into
implementation, change management, goals, culture and other issues that affect
quality.” Empirical studies examining human issues in quality management can
be divided into two perspectives: studies relating to “enablers” and “inhibitors” of
quality management. For instance, Gupta and Ash (1994) studied how allowing
team self-regulation “enables” higher productivity and safety. Another example is

Rungtusanatham’s (2001) examination of how statistical process control

18

increases employee motivation through its job enriching properties. Stewart and
Chase (1999) examined how human errors “inhibit” service quality. In a related
study, Stewart and Grout (2001) examine how simple “mistake-proofing” devices
achieve marked operational improvements by preventing human errors in
processes. These studies underscore the importance of human issues in quality
management practices.

One dominant human issue in the quality management literature pertains
to organizational culture and, more specifically, organizational values. The topic
of culture became popular with the Peters and Waterman (1982) study that
introduced the strategic importance of organizational culture. Subsequent
studies claimed that quality management makes a number of assumptions about
an organization’s culture (Bright et al., 1993), such as a culture that views conflict
as dysfunctional, that sees management as rational and technical, and that is
unified by the quality imperative. These assumptions were further articulated by
Detert et al. (2000) to consist of eight underlying values and beliefs essential for
a quality management initiative. These were referred to as TQM values and are
shown in Table 2-2. Detert et al. (2000) posited that the values and beliefs that
are closely tied to TQM are 1) factual decision-making, 2) a long-term orientation,
3) problems are systemic in nature, 4) never-ending improvement, 5) stakeholder
satisfaction, 6) cooperation and collaboration, 7) employee involvement, and 8)
customer driven. In a later study, Detert et al. (2003) found empirical support for
these values and beliefs when examining the quality culture of primary and

secondary schools.

19

I Table
mi
1 1. Th
‘ rat

5 2. Th
hc'

3. Ma

 

Imp:

c
Orge

 

 

Table 2-2: Cultural Values Related to TQM
Cultural Dimension

TQM Value

 

 

1.

The basis of truth and
rationality in the organization

The nature of time and time
honzon

Motivation

Stability versus change/

innovation/ personal growth

Orientation to work, task and

coworkers

Isolation versus collaboration

/ cooperation

Control, coordination, and
responsibility

Orientation and focus —
internal and/or external

Decision making should rely on factual
information and the scientific method.

Improvement requires a long-term orientation
and a strategic approach to management.

Quality problems are caused by poor
systems — not the employees. Employees
are intrinsically motivated to do quality work if
the system supports their efforts.

Quality improvement is continuous and
never—ending. Quality can be improved with
existing resources.

The main purpose of the organization is to
achieve results that its stakeholders consider
important. Results are achieved through
internal process improvement, prevention of
defects, and customer focus.

Cooperation and collaboration (internal and
external) are necessary for a successful
organization.

A shared vision and shared goals are
necessary for organizational success. All
employees should be involved in decision
making and in supporting the shared vision.

An organization should be customer driven.
Financial results will follow.

 

According to Cameron and Quinn (1999) organizational values greatly

impact the implementation and efficacy of quality management practices in an

organization -— that is, values influence quality management. For example, an

organization that is concerned for its people would be better prepared for

employee involvement initiatives than an organization with a concern for control.

20

 

 

 

Diet

mm

2.1

sh;
SUI

ch

Also along this theme, Naveh and Erez (2004) found quality management
practices to invoke an organization’s “attention-to-detail” — that is, quality
management influences values. As can be seen, the literature on quality
management has highlighted the importance of the human issue of
organizational culture and values but has yet to develop a full understanding of
the relationship between quality management and culture.

Because this dissertation focuses on the antecedents and consequences
of quality management practice implementation, the preceding discussion of
extant literature suggests that human issues should be of particular interest in
studying these interrelationships. Schroeder et al. (2005) suggest that research
dealing with human resource issues and their relationship to quality management
is especially fruitful. Because organizational culture in general, and cultural
values specifically, have been discussed in the context of quality management, it
follows that “values” should hold promise in relating quality management to
human issues. In addition, a multilevel perspective of quality management

practices should lend additional insights into how social factors influence and are

influenced during a quality management initiative.

2.1.4 Performance Implications of Quality Management

As noted by Schroeder et al. (2005) one of the most common relationships
studied empirically is between quality management and performance. This is not
surprising as the intent of quality management is to improve the quality of

products or services with the final goal to improve organizational performance.

21

Eafiy
andfll
otqul
conhl
Bensl
mant
thl
dhec
lndhj
Fmal

ancfi

nott

 

enst
SUD?
acg
‘whh
ape
Dre

”1a

Early works by Benson, Cunningham and Leachman (1995); Flynn, Schroeder,
and Sankakibara (1995); and Powell (1995) all found partial support for the effect
of quality management on performance, because each study revealed
contingencies where quality management is and is not effective. Specifically,
Benson et al. (1995) found that in semi-conductor manufacturing, quality
management impacts seven unique metrics of quality and productivity differently.
Flynn et al. (1995) found that only quality management “core” practices have a
direct effect on performance, while deeper “infrastructure” practices only
indirectly improve performance through their positive impact on “core” practices.
Finally, Powell (1995) showed that just the “tacit” aspects of TQM, such as such
an open culture and leadership commitment, contribute to competitive advantage
not the “technical” aspects, such as benchmarking and training.

However, these findings led later authors to claim a “TQM paradox”
existed between the popular prescription for TQM and the limited empirical
support for this prescription (Choi and Eboch, 1998). That is, there seemed to be
a contradiction between the supposed benefits of TQM and the dissatisfaction
with TQM’s results. This led many researchers to test explanations for why such
a paradox existed. For instance, Choi and Eboch (1998) found that institutional
pressures to appease customers was most related to TQM and that
manufacturing performance found only weak and non-customer related
improvement. Hendricks and Singhal (1997; 2001b) found that quality
management performance effects occur over a longer period of time than implied

by cross-sectional studies. Their studies suggested that up to five years might

22

 

be It

 

disci
how
Kayl
mus
lists.

direl

2.1.1

patll

effe

 

call

the

exa
seq
cha

be?»

 

De

Uh?

 

be required for improvements to take effect. Kaynak (2003) argued that the
disconnect between TQM and performance lies in scholarly inconsistencies in
how quality management is related to performance. The empirical study by
Kaynak (2003) showed that different levels of performance within an organization
must be considered in order to understand the TQM benefits. These studies are
listed only as representative of a theme in the literature, casting doubt on the

direct and positive effect of quality management on performance.

2.1.5 Dynamics of Quality Management

An important aspect of empirical quality management research is the
paucity of longitudinal studies, given their promise in exploring the temporal
effects of quality management practices on performance. One finds repeated
calls for “future research” to include longitudinal data in order to better explore
the relationships found in cross-sectional data (Kathuria and Davis, 2001).

There are studies in the literature that have explored the dynamic
interactions between quality management practices and performance. For
example, Ahire and Ravichandran (2001) discuss a four-stage diffusion
sequence: adoption — top management commitment and launching; adaptation —
changes in the attitude of employees; acceptance — changes in employee
behavior; and use — quality management actions improve operations. Full
performance gains, according to Ahire and Ravichandran (2001), don’t occur
until the fourth diffusion stage. The authors found empirical support for this

diffusion process within the automotive parts industry.

23

 

 

stud

Nara-

Clual
ISSu.
Imp}
only
she;
Sue»
liter;

rTlar

 

Another stream of research in quality management highlights the
importance of time and dynamic relationships, albeit from an analytical
perspective. Narasimhan, Ghosh and Mendez (1993) proposed a dynamic
diffusion model where time delays between actual and perceived product quality
impact the benefit an organization realizes from quality improvements. This
dynamic was later developed in a study investigating the impact product pricing
has on continuous quality improvements (Narasimhan, Mendez, and Ghosh,
1996). Furthering this research stream was a study by Narasimhan and Mendez
(2001) that investigated the equilibrium properties of the dynamic relationship
between an organization’s quality reputation, marketing decisions, and long-term
profitability. Finally, the temporal cost improvements of experience curves were
studied as an important attribute of quality management dynamics (Mendez and
Narasimhan, 2002).

The underlying themes to these longitudinal perspectives are not only that
quality management practices take time to implement (e.g.,, because of resource
issues or organizational cultural issues), but that quality management also is
implemented in different sequences, delayed for strategic reasons and beneficial
only after delays in customer/market feedback. Dynamics therefore have been
shown to exist both internally, such as with experience curves, and externally,
such as with market diffusion. Therefore, a review of the quality management
literature reveals the criticality of incorporating the temporal effects in quality

management research.

24

 

rdat

mga

com:
pehc
hem
dbsl
impil
efie:
hem

COM

22

emu
mar
Cuh,
Org;
0ng

an{

 

 

The purpose of Implementing quality management practices is improved
organizational effectiveness. Examining performance effects over time and in
relation to other factors, such as human issues, will aid in developing a more
complete understanding of how quality management practices impact
performance than is currently available in extant literature. In addition, it could
help resolve the TQM paradox discussed previously. For these reasons, this
dissertation investigates work performance resulting from quality management
implementation using panel data that enable an examination of performance
effects over time. Having reviewed pertinent aspects of the quality management
literature, the next section will present literature pertaining to an important

concept related to quality management — organizational values.

2.2 Organizational Values

The previous section discussed research that establishes organizational
culture as an important aspect to be considered when implementing quality
management practices. The scope of literature dealing with organizational
culture is vast. The literature relating to organizational values — an aspect of
organizational culture — is of particular relevance to the issues addressed in this
dissertation. The values of an organization are not only a significant feature of
an organization’s culture (Schein, 2004) but also a frequently used proxy for
organizational culture in studies pertaining to quality management (Detert et al.,
2000; Cameron and Quinn, 1999). Moreover, the shared values within an

organization influence employee actions through the norms they create. This is

25

an in
mane

impo

2.2.1

valui
Chal-
The?
“thec
valuT
discfi
SchI
For
indi.
8W
Add
allo,

Ihfc

 

 

 

an informal and implicit force that may exist along with the more formal quality
management initiative that an organization implements, and is, therefore, an

important consideration in this research.

2.2.1 Organizational Cultural Values

Common among the organizational culture studies is the prescription that
values should be studied when considering culture (Martin, 2002; O'Reilly,
Chatman, and Caldwell, 1991; Quinn and Rohrbaugh, 1983; Schein, 2004).
These espoused values can be considered ideological ways of operating or
“theories-in-use” (Argyris and Schon, 1974; Schein, 2004). Although espoused
values may not completely reflect an organization’s culture, they are more readily
discernible than tacit assumptions. The organizational culture framework of
Schein (2004) sees organizational values as well established influential forces.
For instance, if cooperative cultural values exist, they will strongly influence
individuals to act cooperatively. Social penalties and rewards are bestowed upon
employees based upon their adherence to the norms derived from these values.
Additionally, cultural values designate what is of importance to the group and
allow employees to prioritize and select from the various norms of behavior. This
informal encouragement toward a mode of behavior enables the implementation
of formal management systems requiring that behavior mode.

Detert, et al (2000) synthesized twenty years of organizational culture
research to propose the following organizational cultural value dimensions: 1) the

basis of truth and rationality; 2) the nature of time; 3) motivation; 4) stability vs.

26

 

and

relat

 

dime

Kl

al.,

how
and
revii
amt
peo
as a
forr
atte
I[far
be

Cor
Chg

bar

change; 5) orientation to work, task, and coworkers; 6) isolation vs. collaboration,
7) control, coordination, and responsibility; and 8) internal vs. external focus.
Organizational improvement initiatives are said to possess underlying values
matching these dimensions, thereby allowing a comparison between an initiative
and an organization’s culture. To illustrate their conjecture, Detert et al. (2000)
related quality management practices to specific cultural values along the eight
dimensions and later found empirical support for these “quality values” (Detert et
al,2003)

An important issue with organizational values is how they are formed and
how they change. This research topic, in general, is not well developed (Hitlin
and Piliavin, 2004), but some suggestions have been made that are important for
review. Values are thought to form, in part, as a response to an overwhelming
amount of environmental information. That is, because data are available to
people from too many sources, it is often not readily apparent how to use them
as a guide for actions (March et al., 1958). To facilitate decision making, people
form and use values to discern what is important and to what they need to be
attentive (Dewey, 1939). Quinn and Rohrbaugh’s (1983) competing values
framework illustrates this by showing how organizations that value flexibility will
be attentive to diversity and individual initiative, while organizations that value
control will be attentive to structure and coordination. Exactly what values to
choose often come from influential sources — such as leaders, cultural

background, and social groups (Weiss, 1978). When individuals are thrust into

27

new

inforl

withi
prob
infor
cont
the \
and
Woe
anc

imp

0th§
VBIL
192

eV 6

 

 

new situations in which they must make decisions, values aid the selection of
information and choice of action.

Organizational values are said to change as a result of “critical reflection”
within an organizational context (van Woerkom, 2004). As new situations and
problems present themselves, employees must reflect upon the available
information and goals in order to solve the problem. When similar problems
continually occur, or when problems are highly impactful, deeper reflection into
the very premises on which a problem rests is needed. This is “critical reflection”
and it is the situation in which value change is most likely to occur. Van
Woerkom (2004) likens this to organizational learning (Argyris and Schon, 1996)
and states that processes like Deming’s plan-do-check-act (PDCA) continuous
improvement cycle can be catalysts for organizational value change.

An important feature of values is that they are projected onto others. In
other words when one observes the actions of another that contradict one’s
values, one judges the other to be acting poorly or in bad behavior (Dewey,
1939). This essential feature of values is crucial toward understanding how they
eventually influence organizational behavior. As values become shared and
espoused within an organization, social norms of behavior begin to form around
those values (Schein, 2004). Norms can be considered the rules of behavior that
are enforced through such social processes as inclusion, acceptance, and
ridicule. New members of an organization are especially apt to be influenced by
the social norms and values as they make sense of their new and uncertain

situation (Louis, 1980).

28

 

 

also

indi.
selei

ihilti

 

earli

the
org
vali
dis:
Che
infl.

dis:

2.2.

8c:

 

Ins.

 

The above review reveals the multilevel nature of organizational values.
When values are shared they exist at an organizational (i.e., high) level. The
leadership that influences what is valuable in an organization operates at a high
level. Organizational norms that follow from the underlying organizational values
also exist at the organization level. However, norms form in order to influence
individual (i.e., lower-level) action. In addition, values influence information
selection as problems present themselves at these lower levels. This cross-level
influence of organizational values fits well within the mesa-paradigm described

earlier.

This dissertation incorporates these ideas on cultural values. Specifically,
the prominence of values as an influential force in human behavior in
organizations justifies its consideration. However, recent research on how
values change (Naveh et al., 2004; van Woerkom, 2004) supports this
dissertation’s perspective that cultural values are a malleable organizational
characteristic. Therefore it is the perspective that cultural values are both
influenced and influential that underpins the richer conceptualization in this

dissertation of how cultural values interact with a quality management initiative.

2.2.2 Cooperative Cultural Values

Among the many researches that have investigated organizational values,
a common element is the recognition that many types of values exist. For

instance, O’Reilly et al. (1991) has elucidated over fifty different types of values

29

in their organizational culture profile (OCP). Schwartz (1992) also has described
the different values that exist within a culture, such as individualism vs.
collectivism. Quinn (1983) described multiple organizational values as well, but
also showed how some are competing against each other — for instance,
flexibility vs. control. It should be noted, therefore, that values are not only
multifarious but are often incompatible.

The values surrounding collectivist-type values [e.g.,, cooperation
(Chatman et al., 1995)] are of particular interest to this research because of their
(often) assumed relationship to quality management (Detert et al., 2000). In
particular, Detert et al. (2000) highlighted the following quality values that closely
align with what can be termed cooperative-type values: 1) All employees should
be involved in decision making and in supporting the shared vision, 2)
Cooperation and collaboration are necessary for a successful organization, and
3) Employees are intrinsically motivated to do quality work if the system supports
their efforts. Associations have also been made between the success of TQM in
Japanese companies and the collectivist-type values of the Japanese in general.

In addition to the association with quality management, cooperative
cultural traits have been associated with organizational effectiveness and
performance. For instance, Denison and Mishra (1995) found that involvement
and adaptability (as indicated by flexibility, openness, and responsiveness)
helped predict organizational growth, quality, employee satisfaction, and overall
performance. Later Song, Montoya-Weiss and Schmidt (1997) empirically found

a strong, positive relationship between cross-functional cooperation and

30

 

 

2.3

 

performance in new product development. Jones and George (1998) have
argued that interpersonal cooperation and teamwork is an important component
of organizational performance and competitive advantage. Recently, Doolen,
Hacker, and Van Aken (2003) empirically found that an organizational culture
supportive of communication, integration, and cooperation between teams had a
positive relationship with team effectiveness and team satisfaction.

This stream of research then allows this dissertation to focus on a specific
cultural value — the value of cooperation. Because cooperative values have been
prominently associated with quality management, and because values in general
are important to quality management practice implementation, this dissertation
will examine specific relationships between cooperative values and the
implementation of quality management practices at multiple levels of an
organization. Moreover, the performance effects that cooperation may have in
an organization call for the joint examination of cooperative values and quality
management practices. So far, this has not been done in the studies relating to
quality management practices. However, empirical studies have been conducted
relating quality management to other aspects of organizational culture. This

literature is reviewed next.

2.3 Pertinent Empirical Studies

The following section describes empirical research that examined the
relationships between operations management practices — in particular quality

management - and aspects of organizational culture. Definite causal directions

31

withi
reme
certs
char
man

gene

9395

2.3.‘

and
poir
Mor
Sigr
De:

COI‘.

 

inii
Uhg
au‘.

TC

 

within this interface of operations management and organizational behavior
remain unresolved. However, what is strongly supported is that the presence of
certain operations management practices and certain organizational cultural
characteristics are concomitant. This section will review where quality
management research specifically and operations management research
generally have conducted empirical work. Following this, pertinent research

gaps will be presented.

2.3.1 Quality Management and Organizational Culture Studies

In the operations management literature the association between
organizational culture and TQM has been recognized. A likely reason for this is
that TQM is both a set of observable practices (e.g.,, statistical process control
and root cause analysis) and a host of managerial beliefs (e.g., Deming’s 14
points and Taguchi’s quality loss function). Bates, Amundson, Schroeder, and
Morris (1995) found that quality management strategies and culture were
significantly correlated, although they did not suggest directional influence. Later
Detert et al. (2000) offered an interaction hypothesis, where TQM was
considered an element in a more general set of improvement initiatives. Detert
et al. (2000) argued that a unique value structure exists for all improvement
initiatives and that implementation efforts will be less successful if the values that
underpin the organizational culture are incompatible with the initiatives. Other
authors have also dealt with the TQM-culture relationship, some implying that

TQM practices can affect organizational culture (Mead, 1985; Patti, Fok, and

32

 

imp

mai
cult
org;
mar

stu

dim

2.3

reiq

Dra

 

Hartman, 2004) or can be affected by organizational culture (Gallear and
Ghobadian, 2004; Kujala et al., 2004). For instance, Mead (1985) examined a
single case example where quality tools not only improved operating
performance, but also brought new, interactions and relationships among
employees. Patti et al. (2004) validated a survey instrument that revealed
different levels in the organization (i.e., managers and employees) perceive
quality management practices to affect organizational culture differently. Gallear
and Ghobadian (2004) argue, based on empirical evidence, that successful
introduction and practice of quality management often requires culture change.
Kujala and Lillrank (2004) also contend that successful quality management
implementation requires an organizational culture compatible with quality
management practice. A review of these studies reveals that organizational
culture can influence and be influenced by quality management practice. That is,
organizational culture matters to quality management, but how the quality
management-culture relationship should be depicted remains unclear. These
studies also reveal a research gap; no empirical work has incorporated the time

dimension that is implicit in the quality management-culture relationship.

2.3.2 Practice-Culture Empirical Studies

Other operations management researches have investigated the
relationship between organizational culture and operations management
practices. McDermott and Stock (1999) utilized Quinn and Rohrbaugh’s (1983)

competing values framework to investigate if culture influenced the benefits from

33

impl

 

reva
—al
pert .

outt

 

orgl
ope
that
mar
The
of c
am
me
Int

CE

CK

implementing advanced manufacturing technologies (AMT). Their results
revealed statistically significant relationships — some positive and some negative
— among competing-value types, managerial AMT satisfaction, and competitive
performance. Their conclusion was that organizational culture moderates the
outcomes of AMT implementation.

Another study by Yauch and Steudel (2002) investigated how
organizational culture may influence the formation, design, implementation, and
operation of manufacturing cells. Utilizing Schein’s (2004) framework, they found
that cultural factors have little influence on the formation and design of cellular
manufacturing but have wide influence during implementation and operation.
Their conclusion was that organizational culture can impede the implementation
of cellular manufacturing (i.e., implying a culture-as-antecedent hypothesis). The
authors speculate as well that organizational culture can limit cellular
manufacturing’s effectiveness (i.e., inferring a culture-as-moderator hypothesis).
Interestingly Yauch and Steudel mention, but do not explore, the idea that
cellular manufacturing can impact organizational culture (i.e., accepting the
possibility of a culture-as-consequent hypothesis).

Research by Nahm, Vonderembse, and Koufteros (2004) investigated the
relationship between a customer orientation assumption, a set of integrative
beliefs, time-based manufacturing practices (TBMP) and overall performance.
Nahm et al. utilized the cultural framework of Schein (2004). Their proposition
was that a customer orientation assumption leads to integrative beliefs, which

leads to TBMP implementation and then firm performance. Evidence was found

34

 

prar
strc

valt

cu?

def

DE'

TEE

 

 

 

for this sequential relationship in their survey-based study. The implication for
managers was that creating a collaborative, customer-oriented culture should be
done before employing such practice as TBMP (i.e., culture-as-antecedent).

Naveh and Erez (2004) investigated the relationship among quality
improvement practices, the cultural values of innovation and attention-to-detail,
and internal measures of performance. In a longitudinal data set collected from a
single organization, they found support for their hypotheses that 1) quality
improvement practices enable the two cultural values in question and that they
coexist, and 2) performance is affected by the relationship between values and
practices. The implication for managers was that quality improvement practices
strongly communicate organizational goals and therefore elicit certain cultural
values (i.e., culture-as-consequent) that in turn will impact performance.

In the light of these past empirical studies, there is little question that
cultural values relate to quality management in some way. What remains less
definitive is how cultural values should be related to quality management. One
perspective would hold that cultural values are antecedent to quality
management implementation, while another would view cultural values as a
consequent. As was stated in Chapter 1, both views have different epistemic

and practical implications.

This review of empirical studies reveals a deficiency in the literature to
resolve how both perspectives can exist. That is, how can both perspectives
coexist logically and be supported empirically? This dissertation seeks answers

to this epistemic question by approaching the issue from a meso-paradigmatic

35

and
Char
long
rem
pers
diss
IOOI‘

qua

2.4

asi

US

 

and sociotechnical theoretic perspective. The meso—paradigm was covered in
Chapter 1, and sociotechnical system theory will be discussed next. In addition,
longitudinal analysis of the quality management - cultural values relationship
remains to be explored. Researchers have argued from a time oriented
perspective, but have yet to validate these arguments empirically. This
dissertation seeks to close this research gap as well by using longitudinal data to .
look for patterns of change in the relationships among variables to better explain

quality management’s relationship with cultural values.

2.4 Sociotechnical Systems Theory

Although quality management and cooperative values have been
associated in the literature, sociotechnical systems (STS) theory provides a
useful explanation as to how they are related within a meso—paradigm. STS
theory originated in the 19503 to explain how changes in coal-mining techniques
had social ramification and subsequent performance problems (Trist and
Bamforth, 1951 ). Recently, the STS theory has been used to explain Issues in
mass customization (Liu, Shah, and Schroeder, 2006), project management
(Wallace, Keil, and Rai, 2004), and management science (Rosenzweig, 1994).

In this section the salient aspects of STS theory for this research will be
presented. A longer development will be given in Chapter 3 as to how the quality

management - cooperative values relationship can be explained with STS theory.

36

2.4.‘

othJ
met
that
wor
unc‘
des
per
usir
oft
Fre
an
sr.
er
CL

CI“

 

 

2.4.1 Sociotechnical System Frameworks

Sociotechnical system (STS) theory began with the works of Trist and
other colleagues at the Tavistock Institute with initial studies of British coal mining
methods (Trist et al., 1951). An outcome of these studies was the conclusion
that the behavior of organizational employees was so tightly coupled to the way
work was designed that the human system could not be understood without also
understanding the technical system. Moreover, a change in technical system
design would affect the social system and vice-versa (Pasmore, 1988). The STS
perspective views the organization as made up of people who produce outputs
using some technology, and that each affects the operation and appropriateness
of the technology as well as the actions of the people who operate it (Pasmore,
Francis, Haldeman, and Shani, 1982). Influencing both the organization’s social
and technical systems is the environmental system or “totality of systems
surrounding and influencing a focal organization” (Pasmore, 1988). The
environment (e.g., governmental, economic, industrial, transportation, and
cultural systems) is acknowledged to be immensely complex and continually
changing.

Most work with STS has been normative (Cherns, 1987; Clegg, 2000), but
the more descriptive aspects (Pasmore, 1988; Seiler, 1967) will be utilized in this
dissertation. According to Pasmore (1988) “the technical system of an
organization consists of the tools, techniques, artifacts, methods, configurations,
procedures, and knowledge used by organizational employees to acquire inputs,

transform inputs into outputs, and provide output or services to clients or

37

CUS

org;

enc
valt
sys
sys

SOC

08!

inc‘

 

co
Be
W.“

Fi'

 

 

customers.” Generally, technical systems are the means to achieve
organizational ends. The social system “is comprised of the people who work in
the organization and all that is human about their presence” (Pasmore, 1988). It
encompasses many things including attitudes, beliefs, group relations, cultures,
values, norms, politics, cooperation, conflict, and emotions. Generally, social
systems are the means to achieve human ends. Therefore, sociotechnical
systems influence both organizational and human ends through technical and
social means.

Numerous STS frameworks have been provided in the literature,
predominantly in ergonomics and human-factors engineering journals (Carayon,
2006). However, examples exist in the organizational behavior literature as well
(Barko and Pasmore, 1986; Pasmore et al., 1982; Seller, 1967). A common
feature of these frameworks, as shown explicitly by Moray (2000) and
Rasmussen (2000), is the multilevel nature of STS. Moray’s (2000) framework in
particular shows that at the center of STS are physical devices, ergonomics, and
individual behavior. The next layer is team and work-group behavior, of which
communication, cooperation, and perceptions of responsibility play major roles.
Beyond the work-group layer is organizational and management behavior, of
which culture, work practices, and chain of command play significant roles.
Finally, beyond the organizational layer is the legal and regulatory environment,
and lastly the societal cultural layer. Such a framework shows how STS exists at

and across many organization levels.

38

Tag?
I Tl: I’
‘ T2: ‘-
‘1 rigid

r3: 1
, entit

T4:
WOT.
‘ efle

. iron

2.4

 

 

Table 2-3: Characteristics of Sociotechnical Systems*

 

Technical features Social features
T1: Input inducement of variation S1: Codependency in work roles

T2: Work environment — stimulation & 82: Output responsibility in work roles
rigidity (directly related to task)

T3: Spatio-temporal distribution of S3: Distribution of resource allocation
entities (e.g., machines, workers) responsibility

T4: Automation level (i.e., amount of S4: Social “diffusionality” of output

work done by humans) moderates affecting variances (i.e., ramifications
effect of other features that matter)

T5: Patterns in operational clusters S5: Possibility for helpful

(i.e., how work is grouped) interdependencies

T6: Flexibility in and human demands $6: How roles affect individual
from operational method experiences, values and identity

T7: Form and effectiveness of process S7: Degree of social interdependencies

maintenance that are task induced
T8: Input supply robustness (i.e., 88: Congruence among individual
reliability with respect to variability) goals, task interdependencies and

formal structures

 

 

* Modified from Fox (1995)

 

2.4.2 Reciprocal Influences Between Social and Technical Systems

To understand how the technical and social systems influence each other,
Emery (1959) listed characteristics of sociotechnical systems — technical features
(T1-T8) and social features (S1-S8) - that were later summarized by Fox (1995)
and are shown in Table 2-3. As can be seen from Table 2-3, technical systems
have a strong influence on how work and employees are arranged and also the
communication effectiveness between work groups. A key component of social

systems is the amount and form of interdependence that exists. Upon viewing

39

 

 

this table, one can see how a technical change in management practice (e.g.,
quality management) can lead to social changes (e.g., cultural values). For
instance, new management methods that bring employees into cross-function
teams (T3 and T5), changes the roles (S1), interdependencies (S7), and
individual experiences (86).

The multi-ordered effects of technological change described by Pasmore
(1982) provide a useful description for how technological changes, in particular,
influence social characteristics over time. The first-order effect of technological
change is the association that employees make between the new work
arrangements and the positive impact on efficiency and effectiveness. A second-
order effect comes from the new work arrangements, which influence how
coordination must be carried out among employees and leaders. Reward
mechanisms (e.g., monetary, political) adapt to reinforce these new coordination
requirements. A third-order effect involves the skills and knowledge required of
employees involved in the new technology. New problem-solving skills require
training, role adjusting, or hiring -— all of which change the skill profile of the
organization. Organizational structure is modified to accommodate the changing
skill profile. New structures, different managerial relations, and modified reward
mechanisms influences how workers are treated — changing employee self-
conceptions and underlying value systems. These various effects take time to
promulgate through the organization and therefore delay the impact of new

management techniques.

40

2.4.

_L\

theo
tech
al.,
org;
can
ana
onl,
IOg:
p. 4
ans
me
prir

an:

J8;

01g

 

2.4.3 Sociotechnical System Design

An important concept that falls under the more normative features of STS
theory is the notion that only through “joint optimization” of both the social and
technical systems can the sociotechnical system attain full potential (Pasmore et
al., 1982). That is, only if the social and technological systems of the
organization are designed to fit the demands of each other and the environment
can performance gains be realized. Although this idea is more conceptual than
analytical, the notion is “that the design and performance of new systems can
only work satisfactorily if the social and the technical (aspects) are brought
together and treated as interdependent aspects of a work system” (Clegg, 2000,
p. 464). With this perspective in mind, Chems (1976, 1987), Pasmore (1988),
and Clegg (2000) have each developed a list of STS work-design principles
meant to guide the process and content of developing operating systems. These
principles have been used in research related to cellular manufacturing (Huber
and Brown, 1991), lean production (Niepce and Molleman, 1998), mass
customization (Liu et al., 2006), and other operational systems (Das and
Jayaram, 2007). The overall support for the STS approach being related to
organizational performance was provided in a review by Van Eijnatten (1992) of

over 1,500 STS studies.

41

 

Table 2-4: Variations of STS Principles

 

 

Chems (1987)

1. Compatibility: Work system
design should involve employees and
should be compatible with
organizational long-term objectives.

2. Minimal critical specifications:
How jobs are to be performed should
be specified as little as possible.

3. Variance control: Variance control
should be performed where variances
originate and should not be exported
across organizational boundaries.

4. Boundary location: Functional
boundaries should not be drawn so as
to impede the sharing of information,
knowledge, and learning.

5. Information/low: Organizational
systems should ensure that
information is provided to those who
require it when they require it.

6. Power and Authority: Those who
need resources to complete their tasks
should have access to, authority over,
responsibility for, and economically
use them.

7. Multifunctionality: Individuals
should be capable of performing a
diverse range of jobs.

8. Support congruence:
Organizational functions that support
operations should reinforce the
behaviors the organization’s structure
is designed to elicit.

9. Transitional Organization: The
design process should embody and
facilitate organizational transitions in
management philosophies.

10. lncompletion: Organization
design should be continual and never
completely stable, with its operating
teams part of redesign via self-
regulation and learning.

Pasmore (1988)

1. Innovation versus
preserving the status quo:
Creatively employ
organizational resources to
enhance organizational design,
products, service, policies,
work design, or technology.

2. Development of human
resources: Encourage
employees to develop their
knowledge through training,
reward systems, demanding
jobs, organizational
commitment, and expanded
vertical responsibilities.

3. Awareness of the external
environment: Regularly
involve of all employees in
scanning the organization’s
external environment to
increase awareness of threats
and opportunities.

4. Maximize cooperative effort:
Ensure organizational units
and actors pursue common
goals in a cooperative fashion
through autonomous
workgroups.

5. Developing commitment and
energy: Create social and
organizational conditions that
will invoke employee
commitment and direct
motivation toward
organizational goals.

6. Utilizing social and
technical resources effectively:
Use variance analysis, select
technologies deliberately and
design operations with STS
thinking to make better use of
people and machines.

Clegg (2000)

Meta-Principles for Design

1. Design is systemic.

2. Values and mindsets are
central to design.

3. Design involves making
choices.

4. Design should reflect the
needs of the business, its users
and their managers.

5. Design is an extended social
process.

6. Design is socially shaped.
7. Design is contingent.

Content Principles of Design

8. Core processes should be
integrated.

9. Design entails multiple task
allocations between and among
humans and machines.

10. System components should
be congruent.

11. Systems should be simple in
design and make problems
visible.

12. Problems should be
controlled at source.

13. The means of undertaking
tasks should be flexibly
specified.

Design Process Principles
14. Design practice is itself a
sociotechnical system.

15. Systems and their design
should be owned by their
managers and users.

16. Evaluation is an essential
aspect of design.

17. Design involves
multidisciplinary education.
18. Resources and support are
required for design.

19. System design involves
political processes.

 

42

 

Van
Che
cha
org.
pro
cha
for

fun.

COP.

 

 

M c

orl
of

Dr

\‘l't . :

 

Variations exist upon the STS design principles as illustrated in Table 2-4. While
Chems and Pasmore focus almost exclusively on the organizational design
characteristics recommended by STS theory, Clegg offers STS principles for how
organizational design should be carried out (see Clegg’s meta-principles and
process principles in Table 2—4). There are at least four common design
characteristics among each listing: 1) employee involvement in and responsibility
for the work system; 2) employee knowledge, skills and activities beyond singular
function; 3) coordination and integration of functions; and 4) deliberate
connections to organizational goals.

The above presentation of STS theory is divisible into two main areas,
basic and applied, both of which will be utilized in this dissertation. Firstly, the
basic theoretical qualities of STS theory are useful for their frameworks and
explanations for change. These frameworks not only give definitions of technical
and social systems that encompass quality management practices and cultural
values, they also match the multilevel quality management perspective of Juran’s
“big Q” and “little Q”. STS’s explanations for change in both technical and social
entities provide plausible descriptions for why organizational leaders choose
quality management and how organizational members change their values.
Moreover, STS theory provides dynamic explanations for cross-level and multi-
ordered change — this is particularly of interest for explaining the delayed effects
of quality management. The application of STS theory involves the design
principles and performance implications. The design principles are congruent

with many elements of Deming’s management model, and in so doing add a

43

theoretical dimension to quality management’s connection to performance. In
addition, STS design argues that performance benefits will be realized with
cooperation within and between autonomous work units. Therefore, STS theory

provides a framework for the research carried out in this dissertation.

 

Figure 2-1: Theoretical Grounding

 

Sociotechnical
Quality Management Literature Systems Theory Organization Theory Literature
Literature

 

(Emery, 1959;
Pasmore, 1988;
Seiler, 1967; Trist

 

(Anderson et al., Quality et al., 1951) . (Detert et al.,
1994; Deming, Management CEITLTrFaTV‘SIiJZS 2000; Wagner,
1986) Practices 1995)

 

 

 

(Douglas et al., (Denison et al.,
2004) Work Performance 1995)

(Juran, 1989) Meso-Paradigm (Schein, 2004)

(Narasimhan et
al,2001)

(van Woerkom,

Dynamics 2004)

 

 

 

 

 

2.5 Summary of Literature Review

Chapter 2 has presented literature pertinent to this dissertation’s study of the
multilevel relationship between quality practices, cooperative cultural values, and
work performance. A summary of the theoretical grounding is provided in Figure
2-1, which depicts how extant literature is related to the topics of interest in this

dissertation. The Deming management model was reviewed in detail because of

44

 

its I
bet
org
cor

Fin

me
lite
aft

ch

 

 

its popularity and relevance to this dissertation. A connection was made
between quality management and cooperative values, an aspect of
organizational culture. The lack of clarity in extant literature as to how these two
concepts might relate was revealed through the review of empirical studies.
Finally, sociotechnical systems theory was presented as a possible perspective
that may provide a richer conceptualization of the relations between quality
management, cooperative value, and work performance. Based upon extant
literature, a research framework was developed to offer a resolution to the
aforementioned theoretical paradox. This framework is presented in the next

chapter.

45

cor
per
jus
qua
mu

the

3.1

 

 

Chapter 3: RESEARCH FRAMEWORK

This chapter will present nine hypotheses on how quality management,
cooperative values and work performance are related. An STS theory
perspective will be given for each hypothesis, as well as context specific
justification from previous studies. In this way, both organizational theory and
quality management literature will be synthesized to provide justifications for the
multiple hypotheses. Concluding this chapter is a multilevel theoretical model

that shows how the hypotheses fit into a research framework.

3.1 Research Perspective

The total quality management (TQM) approach has been termed a
sociotechnical system by organizational behaviorists (Schein, 2004) and
operations management scholars (Ahire et al., 2001). This is likely because of
the similarities between STS design principles (Cherns, 1987) and TQM
recommendations (Deming, 1986). However, to name TQM a sociotechnical
system may be too restrictive, as such a view does not take into consideration
the potential sociotechnical forces operating outside the TQM purview.
Essentially TQM can be seen as a normative set of management principles
focused on organizational goals (Westphal, Gulati, and Shortell, 1997), while
sociotechnical systems exist regardless of purposeful STS design based

management‘methods (Seiler, 1967). TQM acknowledges the existence of

46

sociotechnical systems within the organization and thus attempts to control those
forces at work.

By taking this viewpoint, this dissertation reaches beyond TQM and STS
design literature to incorporate of the literature on STS theory. Such an
approach expands the bounds of what is to be considered when describing how
quality management and cooperative cultural values relate to each other and to
performance. Moreover, STS theory provides useful descriptions for how quality
management and cooperative values act in a multilevel manner (Emery, 1959;
Moray, 2000; Rasmussen, 2000). Finally, STS theory indicates how the time
element may play into the aforementioned relationships (Pasmore et al., 1982).
It is for these reasons that the research framework of this dissertation rests on
STS theory.

From an STS theory perspective, quality management practices are
viewed as a management technology. As Pasmore (1988) described the
technical system of an organization to consist of (among other things)
techniques, methods, configurations, and procedures to provide services to
clients, quality management is understood as the technical end of a
sociotechnical arrangement within an organization. Concurrently, cooperative
cultural values are a dimension of the social system within an organization. STS
theory acknowledges that social values are part of the organizational culture and
that they play a significant role in influencing human decisions and behavior
(Cherns, 1987; Seller, 1967). The particular link between quality management

and cooperative value is studied because of the empirical concomitant variation

47

between the two (Detert et al., 2003) for which STS theory provides an
explanation. That is, STS theory explains how technologies mediate human
experience, how leaders choose technologies for environmental adaptation, how
initiatives change social values, and how social systems influence individual
behavior, of which technological acceptance is a part. Finally, the multi-ordered
effects of technological change described by Pasmore (1982) provide a useful
description for the changes in influence quality management and cooperative
values have over time.

Although past research has studied cultural relations with quality
management (Bates et al., 1995; Boggs, 2004; Gallear et al., 2004; Huq, 2005;
Jabnoun et al., 2005; Kujala et al., 2004; Miron et al., 2004; Moore et al., 2006;
Patti et al., 2004), no research has utilized the meso-paradigm along with the
incorporation of time under a single theoretical explanation. In so doing, this
dissertation sets itself apart from research on quality management. The following
sections develop the research framework of this dissertation based on STS

theory, the quality management literature and research on cooperation.

3.2 Development of Hypotheses

This section presents each hypothesis. All hypotheses are reasoned from
STS theory first in generic form, and then reasoned from literature specifically
pertaining to the hypothesized relationship. The first six hypotheses relate one
concept to another, while the final three hypotheses pertain to expected effects of

a concept over time.

48

3.2.1 Organization-level Quality Management Instills Cooperative

Values

According to STS theory, technologies are often chosen by leaders within
an organization as a response to changing external environmental pressures
(Seller, 1967). Management technologies such as quality management are no
different, as they too can be externally motivated (Westphal et al., 1997;
Zbaracki, 1998). However, technologies possess characteristics with social
ramification, as shown in Table 2-3. Organizational quality management in
particular possesses spacio—temporal characteristics (T3) in its emphasis on
team-work — bringing employees into closer, more regular contact. Such a
technical feature raises the possibility for interdependent help (S5) and new
social interdependencies (S7). Moreover quality management’s prescriptions of
cross-functional decision-making groups work differently (T5) and increase
codependency in work roles (S1) and output responsibility (S2). Couple these
technological arrangements with the ideal of continuous improvement and the
situation can induce a critical reflection that challenges employee preconceptions
of what is good for the organization — that is, organizational values.

Ackoff and Emery (1972) show that when employees understand that the
benefits of joint action outweigh the benefits from individual action, cooperation is
more likely to occur. Moreover, adopting cooperative values solidifies the
decision to act cooperatively; thereby reducing the uncertainty associated with

trying to choose whether if acting collectively or individualistically is most

49

appropriate. In addition, STS theoreticians have argued that technology is
inherently value-laden. Turning again toward quality management, Detert et al.
(2003) found support for their list of underlying values to quality management as
shown in Table 2-2. From this list, many values describable as cooperative
cultural values can be found. In particular the positive attitude toward coworkers
under dimension five, the collaborative preference toward organizing under
dimension six, and the importance of coordination and shared vision described
under dimension seven. The extent to which an organization-wide quality
management implementation occurs communicates the extent to which
cooperation is valued by organizational leaders, which in turn should influence
the employees of the organization. Therefore, the following hypothesis is given:
Hypothesis 1: Organization-level quality management practices (00) have

a positive impact on organization-level cooperative cultural values (V0)

3.2.2 Organization-level Quality and Workgroup-level Quality

An important component of the organization-level sociotechnical system is
the organizational structure (Pasmore, 1988). This component provides the
formal work relationships, such as the organizational chart (for direction and
reporting purposes), as well as the formal reward systems, such as promotion
and compensation. Although these structures exist at the organization level,
their effects are designed to affect workgroups and individuals. Seller (1967)
points out that organizational structure is quite susceptible to change because of

the ease with which such structures can be manipulated. Following from

50

Pasmore’s multi-ordered effect of technological change, the implementation of
organization-level quality management is expected to modify the organizational
structure. For instance, monthly reporting meetings will likely include formal
communication of quality management progress and promotion decisions could
likely be influenced by success with the new quality management approaches.
These formal mechanisms promote the ideals of quality management at the
workgroup level and make implementation of quality management at that lower
level more likely. From the workgroup-level perspective, the organizational
structure embodies much of the workgroup’s environment. As such, the STS
perspective would view the workgroup as adapting to its environment by adopting
quality management practices.

Additionally, the quality management literature has pointed out the
criticality of top management support for its success. Adoption of the quality
management paradigm at first requires a leap-of—faith in the cause-effect
relationships promoted by such a management technology. Workgroups may
see quality management as risky or distracting unless backed up by top
management commitment, patience, and resources. As such, the implementation
of organization-level quality management likely sends a strong message to
workgroups that the leadership is committed to seeing this technology
Implemented. Often new management practices are talked about more than
done - hence leading to the labeling “management fad”. Such a label could be
indicative of managerial lack of support. However, if employees generally see

that organization-wide change is occurring, they are more likely to agree that

51

imp
wor
shc
trai

WO

3.:

8C

06

gr

(7)

 

 

implementation of quality management in their workgroup is both appropriate and
worthwhile. In addition, the presence of organization-level quality management
should indicate the existence of resources available to workgroups to conduct the
training and implement the changes required to adopt quality management at the
workgroup level. Therefore, the following hypothesis is given:

Hypothesis 2: Organization-level quality management practices (00) have

a positive impact on workgroup-level quality management practices (QW).

3.2.3 Cooperative Values Encourage Workgroup-level Quality

As social values are used to judge not only one’s actions but also the
actions of others, it follows that an organization’s cultural values are a key
determinant for cultural norms of behavior (Rokeach, 1973; Seller, 1967).
Cultural norms act as rules that when followed are rewarded socially — that is, via
group inclusion or social status. Workgroup technologies set guidelines for
employee behavior, but may contradict organizational social norms. This
contradiction discourages acceptance and likely encourages only surface
implementation of technologies — that is, tokenism. Moreover, STS theory sees
technology implementation as predominantly a leadership decision based upon
environmental adaptation. From a meso-paradigm viewpoint, a workgroup’s
environment is predominantly internal to an organization (e.g., organizational
climate). Such a multilevel understanding leads to the possibility that
organizational social values and norms create environmental characteristics to

which a workgroup’s leadership must adapt. This process is similar to the

52

dis
wo
inil
co
en
an

Th

gh

 

changes in the organizational environment discussed in section 3.2.2, albeit an
informal rather than formal process. This cross—level influence then would
encourage and discourage certain workgroup technologies to be implemented.

As cooperative-type cultural values are tightly coupled with quality
management practices (Detert et al., 2000), the presence of such cooperative
values should invoke norms of cooperation that encourage quality management
to be practiced by workgroups. In addition, cooperative cultural values represent
part of the mental model within an organization that aids in uncertainty reduction.
When employees are required to act in ways that oppose their mental models,
dissonance and therefore resistance may ensue. Such could be the case with
workgroups being asked to cross-train in other functions as part of a quality
initiative. A lack of cooperative values indicates a mental model discouraging of
collaborative and supportive behaviors. A cooperative act like teaching other
employees one’s function is dissonant with a more individualistic mental model,
and thus the cross-training could be resisted or implemented half-heartedly.
Therefore based upon these sociotechnical influences the following hypothesis is
given:

Hypothesis 3: Organization-level cooperative cultural values (V0) have a

positive impact on workgroup-level quality management practices (QW).

53

3.2.4 Quality Management at the Organization Level Improves

Workgroup Performance

The relationship between organization-level quality management — as in
Juran’s “big Q” — and STS design principles has been observed (Schein, 2004).
For instance the STS principle of technological system compatibility with
organizational long-term objectives (Cherns, 1987; Liu et al., 2006) follows
closely the quality management principle of a shared vision and constancy of
purpose (Deming, 1986; Anderson et al., 1994). In addition, the STS principle of
departmental boundary reduction aligns well with the quality management
principle of cross functionality. Research has shown that STS design principles
are beneficial to organizations. Moreover, they likely influence the organizational
structures that allow workgroups to attain higher performance by removing
internal barriers viewed as counterproductive. STS designs avoid conflicting
goals between workgroups, increase multi-functionality and reduce the
duplication of effort (Pasmore, 1988). Research in quality management has also
found evidence that organizational performance, and by extension workgroup
performance, improve with the implementation of quality management principles
(Choi et al., 1998). Therefore the following hypothesis is given:

Hypothesis 4: Organization-level quality management practices (00) will

have a positive effect on workgroup performance (PW).

54

3.2

lea
cor
sys
cor
flat
infr
or

org

Za

ml

al.

Th

 

 

 

3.2.5 Cooperative Cultural Values Help Workgroup Performance

STS theory contends that social systems in an organization influence
leadership’s choice of organizational structure (Seller, 1967). As the presence of
cooperative cultural values would be a significant part of an organization’s social
system, it would be expected that organizational structures congruent with
cooperation would emerge. Cooperative-type organizational structures could be
flatter organizations, more frequent team oriented actions, many channels for
information sharing, and higher trust between departments. These types of
organizational attributes have been found to improve effectiveness — such as
organizational growth and quality (Denison et al., 1995), or individual satisfaction
and task performance (Denison et al., 1995; Jackson, Colquitt, Wesson, and
Zapata-Phelan, 2006). Moreover, the presence of these types of structures are
congruent with many STS design principles —such as, higher information sharing,
multi-functionality, employee involvement and autonomy (Cherns, 1987; Llu et
al., 2006) — and therefore should be associated with workgroup performance.
Therefore, the following hypothesis is given:

Hypothesis 5: Organization-level cooperative cultural values (V0) will have

a positive effect on workgroup performance (PW).

3.2.6 Workgroup Performance is Improved by Workgroup-level

Quality Management

One of the founding ideals of STS design is the autonomous workgroup

(Pasmore, 1988; Trist et al., 1951). The prescription is to allow workgroups to

55

SB

pit

T8!

prr

10

V8

n-

pr

I0

Q»

self-adjust to uncertainties from their environment. Removing role and
procedural restrictions, it is argued, allows for a better fit between situation,
response, and therefore performance. Furthermore, other STS design principles
predominantly manifest at the workgroup level. Examples include timely access
to information, workers being capable of performing a diverse range of jobs, and
variance control at the point of origin. These principles closely follow Juran’s
“little Q” designation of multiple quality management practices, such as tools for
process control, information for corrective action and employee involvement in
root cause analyses. These “little Q” practices, such as statistical process
control, have been found to have benefits not only for the organization but for
employee job enrichment as well (Rungtusanatham, 2001). Since both quality
management and STS principles have been associated with performance, the
following hypothesis is given strictly at the workgroup level:

W)

Hypothesis 6: Workgroup-level quality management practices (Q will

have a positive effect on workgroup performance (PW).

3.2.7 Diminishing Effects of Organization-level Quality Management

Cultural beliefs and values change as disconfirming evidence is
repeatedly presented to an organization’s employees (Schein, 2004). In the case
of quality management influencing the emergence of cooperative values,
disconfirming evidence would be the observation of successful cross-functionality
where before cross-functionality was deemed hurtful or a waste of time. Initially

such disconfirmations of non-cooperative mental models are explained away as

56

chant
mana
coopr
inknd
mean

Ihecn

aein
dngul
vdue:
popuh
indwk
Oman
whhp
dfius
bdwe
Howe

mane

 

Cigar

 

I'Sr W”
prim;

QuaI

 

chance events or necessary inefficiencies. However if the effort to retain quality
management practices holds over time, the mounting evidence for valuing
cooperation wins over social arguments against cooperation. Increasing
interdependence, norms of cooperative behavior and successful outcomes then
create self-reinforcing feedback, thereby diffusing cooperative values throughout
the organization.

An organization is replete with differing values (O'Reilly et al., 1991) that
are incommensurable and at time competing (Quinn et al., 1983). Observing the
singular cooperative cultural value does not take into consideration the other
values that vie for prominence in an organization. As cooperative values grow in
popularity, they should begin to compete with deep-seated assumptions about
individual freedom and responsibility. This would especially be true for
organizations within a strongly individualistic national culture, like organizations
with predominant operations in the United States (Hofstede, 1998). The above
diffusion pattern should therefore manifest in an initial strong relationship
between organization-level quality management and cooperative cultural values.
However, the relationship strength should diminish in later years of a quality
management initiative.

A similar pattern of decreasing influence is expected between
organization-level quality management and the other constructs of interest - that
is, workgroup-level quality management and workgroup-level performance. The
primary reason for this is the intermediary influence of organizational structure.

Quality initiatives often lead off with top management communications and

57

organization-level changes. Subsequently, training begins and resources are
dedicated to departmental and workgroup-level quality management practice
implementation (Ahire et al., 2001). Following STS theory, formal changes are
made to organizational structures based upon this new management technology.
More organizational structures will come in line with quality management
principles, and should result in increased workgroup-level performance.
However, the number of elements of an organization’s structures is limited.
Seller (1967), for instance, lists 14 dimensions ranging from span-of-control and
levels-of-authority to compensation and skill-levels. As these dimensions are
modified less potential structural changes can occur later in the organizational
quality initiative. In addition, organization-level quality management can only
change workgroup-level activities up to a certain extent. STS theory would
acknowledge that quality management is one of many technologies within an
organization (Pasmore, 1988). It therefore follows that other management
technologies — such as new information systems and human resource practices
— will interact with quality management in unforeseen ways. In addition,
researchers have observed that sustained workgroup-level performance truly
depends upon activities at the workgroup level (Jones et al., 1998; Naveh et al.,
2004). Following these reasons, the hypothesis is given:

Hypothesis 7: The effects of organization-level quality management

practices (QC) on cooperative values (V0), workgroup-level quality

management practices (QW) and workgroup-level performance (PW)

decrease over the years of a quality initiative.

58

3.2.8 Diminishing Effects of Cooperative Values

Changes to organizational cultural values modify cultural norms (Schein,
2004), which in turn influence the organization’s environment — both its structure
and its culture (Seiler, 1967). As an organization embarks on a quality
management initiative, structures and norms based upon older sociotechnical
influences are modified based on the newer influences. For example, a norm of
mono-functionality would yield to a norm of multi-functionality. However, after
sustained effort is given to quality management implementation the number of
organizational elements unaligned with quality management diminishes. This in
turn reduces the impact that-organizational cooperative values have at the
workgroup level. Such a process is similar to the diminishing effect of
organization-level quality management practices, albeit from an informal
structure point of view. Therefore, cooperative cultural values should have the
highest influence on workgroup-level quality management practice
implementation in earlier years of a quality initiative. Because cooperative-type
organizational structures should benefit workgroup-level performance, as less
structures are available to change, then lower performance benefits should be
gained because of cooperative cultural values. Because of these reasons, the
following hypothesis is given:

Hypothesis 8: The effects of organization-level cooperative cultural values
W)

(V0) on workgroup-level quality management practices (Q and

59

workgroup-level performance (PW) decrease over the years of a quality

initiative.

3.2.9 Workgroup-Level Quality Management Improves With Time

A delayed increase in effect is expected for workgroup-level quality
management on workgroup-level performance. This reason for this is two-fold.
First, workgroups are interdependent with other workgroups in the organization.
This is a reality that both STS and quality management principles acknowledge
and leverage. Inter-workgroup formal and informal structural inconsistencies can
cause problems for the individual workgroup. For instance, if workgroup A is
seeking cross-functional knowledge from workgroup B, but workgroup B has not
adopted cooperative values, then the amount of help workgroup A will receive is
likely to be less. In addition, if information sharing in workgroup B is being highly
utilized, but not in workgroup A, the interdependence of the workgroups will make
the information sharing process less helpful for workgroup B.

The second reason a delay in workgroup-level performance is expected
follows from the same empirical observation made at the organization level.
Quality management practices take time to reap benefits (Schroeder et al.,
2005). This is because of learning curves as experienced with other
technologies (Yelle, 1979). Additionally, root cause systemic corrections take
time. For both theoretical and observational reasons, an increase in benefit of
workgroup-level quality management is expected in the later years of

implementation. Therefore the following hypothesis is given:

60

Hypothesis 9: The effect of workgroup-level quality management practices
(QW) on workgroup-level performance (PW) will increase over the years of

a quality initiative.

In the next section, each of the hypotheses discussed are presented
together in order to depict the full network of relationships being examined in this

dissertation.

 

Figure 3-1: Expected Relationships Within a Single Time Period

 

 

Organization-level

 

 

 

 

 

 

 

 

 

 

  
 

 

 

 

 

 

 

Quality
Management H1 Coo erative Cultural
Practices (QC) I\D/alues (V0)
Workgroup-level
Quality Work
Management Performance
Practices (pw)
(QW)

 

 

 

 

 

 

 

 

3.3 Research Models

To position each of the above hypothesized relationships, two research
models are presented. Hypotheses 1 through 6 are shown in Figure 3-1, and
hypotheses 7, 8, and 9 are shown in Figure 3-2. Figure 3-1 is depicted in a

manner congruent with Snidjers and Bosker’s (1999) method of presenting

61

multilevel models. The upper left shows an exogenous organization-level quality
management, which is consistent with STS theory’s perspective that
technologies are chosen as adaptations to external environmental forces. This
depiction is also consistent with the notion of an organization-wide quality
management initiative. All other concepts are endogenous, with only cooperative
cultural values being the other organization- level factor. Both organization-level
quality practices and cooperative values then influence workgroup-level quality
practices. The only fully dependent factor is workgroup-level performance shown

in the lower right comer.

 

Figure 3-2: Qualitative Depiction of Expected Changes Over Time

 

 

H9: Effects of QW

H7: Effects of Q0
H8: Effects of v0

Strength of hypothesized
relationship

 

 

Initiation Maturation

Stage of Quality Initiative

 

 

The qualitative nature of each curve in Figure 3-2 represents hypotheses
7, 8, and 9. The expected changes over time for each relationship are shown in
Figure 3-3. As Figure 3-2 shows, the effects of workgroup-level quality
management practices increase from the beginning stages of a quality initiative

to the maturation stages. The expected rise in effectiveness of workgroup-level

62

quality management practices is shown by the up arrow by the QW-to-PW
relationship shown in Figure 3-3. The shape of the other curve in Figure 3-2
represents qualitatively the expected decrease in strength of the V0 and Q0
effects. These decreasing effects are depicted in Figure 3-3 as down arrows for
each of the relationships captured by hypothesis 7 and 8. Together, Figures 3-1

and 3-3 depict the overall research framework for this dissertation.

 

Figure 3-3 : Expected Direction of Change in Relationships Over Multiple Periods

 

 

Organization-level

 

 

 

Quality
Managemeng W Cooperative Cultural
Practices (Q ) Values (V0)

 

 

 

 

 

 

 

----—----------- ---—------—-—----—-—------------—-——--- ---—----q-----—-—

  
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Workgroup-level
Quality '
Management Perlbvghkance
Practices 7 (PW)
(OW)
3.4 Summary of Research Framework

Chapter 3 discussed the new perspectives possible if quality management
is understood to be a technical component of the sociotechnical systems in an
organization. These new perspectives give insight to the multilevel nature of
quality management, and in so doing provide a plausible explanation for why

cooperative cultural values are concomitant with quality management. Moreover,

63

 

STS theory provides explanations as to how workgroup performance gains |--
emerge within the new network of relationships. Finally, the multi-ordered effects
explained in STS describe what changes to the hypothesized relationship should
be expected over time. To find support for this research framework, an empirical

approach was used and is presented in the next chapter.

64

 

CHAPTER 4: RESEARCH METHODOLOGY

This chapter explains the methodological approach used in this
dissertation. The data requirements and collection method are first presented,
followed by the details of the archival data used in this study. Then, the
measurement model based on dissertation objectives and data availability is
discussed. Finally, the data analysis techniques used in this study are described

and summarized in a flow chart.

4.1 Data Collection

This dissertation seeks empirical support for the hypothesized multilevel,
time-varying relationships among quality practices, cooperative values and work
performance. As such, the data set needs to meet three criteria. The first
criterion is that these concepts need to be measured within the context of a
quality management initiative, as this is a basic premise of the STS theory-based
relationships that were hypothesized in Chapter 2. In addition, this is a practical
motivation for this research — that is, explaining variation in the effectiveness of
quality management programs. The second criterion is that multiple levels within
organizations need to be sampled. This requirement follows from STS theory’s
multilevel description of social effects (Moray, 2000; Rasmussen, 2000; Seller,
1967) and Juran’s (1989) observations highlighting the importance of

distinguishing “big Q” and “little Q" initiatives. The third criterion is the need for

65

longitudinal and specifically, multiyear data. This requirement is caused by prior
research evidence showing that quality management effects do not manifest
immediately and in fact may take several years to unfold (Mitra and Golder,
2007; Schroeder et al., 2005). Each of these requirements was considered in

selecting the data source for this dissertation.

4.1.1 Data Collection Approach

There were two options to collect field data that meet the above
requirements. The first option was to develop an original measurement
instrument that would then be sent to multiple individuals in many organizations
over multiple years. The sample would have to be carefully chosen to include
only those organizations implementing quality management initiatives. This first
option was discarded because of time and cost constraints. In addition, the
nature of a dissertation places constraints upon the amount of time over which
one can sample. Therefore, pursuing this option was deemed prohibitively
“expensive.” The second option was to resort to archival data from a publically
available data source pertaining to quality management practice implementation
in multiple organizations. Following this option required that the data adequately
match the concepts of interest and be readily accessible. This is a common
concern when using secondary data; it is generally resolved by ensuring that a
proper theoretical “network" surrounds the concepts used (Houston, 2004).
Three stages of validation are prescribed: 1) ensuring that a theoretical basis

exists for the secondary measures; 2) assessing the ability of the measures to

66

serve as indicators; and 3) assessing if the measurements reflect how concepts
are expected to relate nomologically (Houston, 2004). Although less control over
measurement instrument design in the use of archival data may pose a threat to
validity, with proper theoretical grounding this option would be a viable approach.
Therefore it was decided to pursue this option for data collection.

Many channels were explored to discover an appropriate data source.
Various online search engines were investigated, as well as financial reports
from large corporations. Major employee surveys were identified, such as
General Electric’s workplace survey1 but access to some of these data was
unavailable. Another option was suggested by a recent special issue of the
Journal of Operations Management, in which operations management issues
were being explored in not-for-profit organizations (Verrna, Youngdahlb,
McLaughlinc, and Johnstond, 2006). This special issue set in motion a search
for data on the well-known initiative by the federal government to implement
quality management practicesz. It was discovered that a survey was conduct by
the United States Office of Personnel Management (O.P.M.) over three years
across many agencies within the US. government. After further investigation
into the content and context of the survey, it appeared this data source may be
appropriate. The fact that this data had been validated by other research was

also attractive (Chun and Rainey, 2005).

 

' see http://www.qe.com/en/citizenship/employees/employee.htm
2 see http://govinfo.library.unt.edu/npr/index.htm

67

 

4.1.2 The National Partnership for Reinventing Government

The National Partnership for Reinventing (NPR) Government conducted a
three-year employee survey from 1998 to 2000 as part of a government-wide
initiative to improve operating practices by installing various quality management
practices (O.P.M., 2002). The survey was intended to gather feedback for
agencies so they could focus on employee satisfaction, increase the
performance of government services, and better serve customers. A desire was
also to assess the progress of organizational changes in agencies - especially in
relation to efforts to reinvent internal operations. These efforts were a major
initiative within the federal government to improve governmental operations
through quality management practices. The initiative was spearheaded by then
Vice President Al Gore, beginning in 1993 and continuing until 2001 (Kamensky,
1996; Qiao and Thai, 2002). Rather than use external consultants, an internal
group of champions was selected from each major agency within the federal
government to develop plans to implement quality management practices and to
execute reinvention projects. Midway through the initiative, an interest arose
regarding organizational climate and culture. This interest partly motivated the
administration of the NPR survey.

The NPR survey was developed by a team of survey experts from the
O.P.M., the Merit Systems Protection Board (MSPB), and the Federal Aviation
Administration (FAA). The 32-item survey (shown in Table 4-1) was
administered by the O.P.M. and was designed to assess employee opinions on

workplace attitudes and the progress of reinvention through the implementation

68

 

of quality management practices in the federal agencies. About half of the items
on the survey were taken from O.P.M.'s organizational assessment survey. Each
year, the NPR survey instrument was reviewed by the O.P.M. to determine areas
of improvement and as such minor modifications were made in wording and
some items were removed or added.

The NPR survey instrument is divided into four sections. The first section
(questions 1 — 21) consists of items measuring the degree to which certain
organization and workgroup traits exist, such as organizational systems for
customer feedback (question 2) and workgroup productivity (question 17). These
were scaled one to five from “strongly disagree” to “strongly agree.” The second
section (questions 22-28) measured the extent to which specific practices could
be characterized, such the extent to which hiring practices are “streamlined”
(question 23). These were scaled one to five from “not at all” to “very great
extent.” The third section (questions 29-33) possessed items seeking the
judgment of the respondent, such as job satisfaction (question 29) and
workgroup quality (question 33). These were also scaled one to five, either on a
“satisfaction” continuum or “goodness” continuum. The final section contained
demographic and other items that were not made available to the public except
for question 42, querying to which organization (agency) the respondent was

employed.

69

 

Table 4-1: National Partnership for Reinventing Government Employee Survey,

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1998
# Description Measurement
Scale
01 There are service goals aimed at meeting customer 1 - strongly
expectations disagree
02 There are well-defined systems for‘linking customers 2 - disagree
feedback/complaints to employees who can act on the 3 - neither
information 4 - agree
03 Managers communicate the organizations mission, vision, 5 - strongly agree
and values
04 My immediate supervisor has organized our work group
effectively to get the work done
05 At the place I work, my opinions seem to count
06 A spirit of cooperation and teamwork exists in my
immediate work unit
07 Teams are used to accomplish organizational goals, when
appropriate
08 Employees are rewarded for working together in teams (for
example, performance ratings, cash awards, certificates,
public recognition)
09 Employees in different work units participate in cross-
functional teams to accomplish work objectives
10 Recognition and rewards are based on merit
11 Creativity and innovation are rewarded
12 Employees receive training and guidance in providing high-
quality customer service
13 Employees receive the training they need to perform their
jobs (for example, on-the-job training, conferences,
workshops)
14 Differences among individuals (for example, gender, race,
national origin, religion, age, cultural background, disability)
are respected and valued
15 Supervisors/team leaders understand and support
employees family/personal life responsibilities
16 My organization has made reinvention a priority (for
example, working smarter and more efficiently)
17 In the past 2 years, the productivity of my work unit has
improved
18 In the past 2 years, I have been given more flexibility in
how I accomplish my work
19 Employees are required to report the hours they work on a
daily basis
20 Corrective actions are taken when employees do not meet
performance standards
21 Management and the union(s) work cooperatively on

 

mutual problems

 

 

70

 

 

Table 4-1: National Partnership for Reinventing Government Employee Survey,
1998 (continued)

 

22

Has your organization implemented simplified travel
regulations?

 

23

Has your organization streamlined the process for hiring
employees?

 

24

Has the use of government credit cards for small office
purchases been implemented in your org?

 

25

Are you clear about how good performance is defined in
your organization?

 

26

Do you have electronic access to information needed to do
your job?

 

27

Is your org working with its regulated community to achieve
better compliance through partnerships?

 

28

Is your org working to streamline its regulatory program to
make it more readable and customer-focused?

1 - not at all

2 - limited extent

3 - moderate extent
4 - great extent

5 — very great
extent

 

29

Considering everything, how satisfied are you with your
job?

 

30

How satisfied are you with your involvement in decisions
that affect your work?

 

1 - very dissatisfied
2 - dissatisfied

3 - neither

4 - satisfied

 

 

 

 

 

 

 

31 How satisfied are you with the recognition you receive for 5 - very satisfied
doing a good job?
32 Overall, how good a job do you feel is being done by your 1 - very poor
immediate supervisor/team leader? 2 - poor
33 How would you rate the overall quality of work being done 3 - fair
in your work group? 4 - good
5 - very good
42 Organization code Various codes
4.1.3 Sampling

Between 1998 and 2000 the US. government averaged civilian

employment of more than 2.7 million employees within the legislative, judicial,

and executive branches. Of these, the executive branch comprised more than

95% of the employees. The NPR survey was administered to about 30% of the

executive agencies at the time of the survey. Table 4-2 presents the list of

agencies sampled in 1998, among which included 12 incidences of “other”

agencies. Section 4.3 describes for how these “other” codes were accounted.

71

 

the ext
federal
public.
restor'
high-i
sun/e
partir
bani

their

A key factor in selecting the agencies to participate in the NPR survey was
the extent to which their services impacted service to the public. Thirty-two
federal agencies had 90 percent of the federal government's contact with the
public. The performance of these high-impact agencies was seen as central to
restoring America's trust in government. Therefore, in 1998 and 1999, 31 of the
high-impact agencies, as well as 17 other agencies, participated in the NPR
survey. In 2000, all high-impact agencies, as well as 17 other agencies,
participated. Individual employees were randomly selected and invited to
participate in the survey. Employees received the survey at either their homes or

their offices, but responses were returned by mail to the address of a consultant

 

 

 

 

 

(O.P.M., 2002).

Table 4-2: Federal Agencies Participating in NPR Survey

Code Description Code Description

100 Dept of Air Force 213 All other Interior

110 Dept of Army 221 Immigration and Natural. Serv
120 Dept of Navy 222 All other Justice

131 Defense Logistics Agency 231 Occup. Safety & Health Admin
132 All other Defense 232 All other Labor

141 Forest Service 241 Bureau of Consular Affairs

142 Food Safety/Inspection Service 242 All other State
143 Animal/Plant Health lnsp Serv 251 Federal Aviation Admin

144 Food & Consumer Services 252 All other Transportation
145 All other Agriculture 261 Internal Revenue Service
Nat Oceanic & Atmospheric

161 Admin 262 US Customs Service

162 Patent & Trademark Office 263 Financial Management Service

163 Bureau of the Census 264 All other Treasury

164 Internal Trade Administration 271 Veterans Health Admin

165 All other Commerce 272 Veterans Benefits Admin

171 Off of Post-Secondary Educ 273 All other Dept of VA

172 All other Education 290 Environmental Protection Agency

180 Dept of Energy 300 Equal Employ. Oppor. Commission
Federal Emergency Management

191 Food 8 Drug Admin 310 Agency

192 Health Care Financing Admin 320 General Service Admin

193 Admin for Children & Families 330 Nat Aeronautics & Space Admin
194 All other HHS 340 Small Business Admin

200 Dept of Housing & Urban Dev 350 Social Security Admin

 

72

 

 

Table 4-2: Federal Agencies Participating in NPR Survey (continued)

 

 

211 National Park Service 360 US Office of Personnel Mgmt
212 Bureau of Land Management 999 Unspecified Agency

 

The 1998 NPR Employee Survey was administered to 34,401 employees
representing 48 government agencies and resulted in a 40% response rate,
which was based on the return of 13,657 completed surveys (O.P.M., 2002). In
September 1999, the NPR Employee Survey was administered to 32,265
employees in the same government organizations, of which 12,755 surveys were
returned for a 40% response rate. The 2000 NPR Employee Survey was
administered to 50,844 employees, with an overall response rate of 42% based
on receiving 21,157 completed surveys. These response rates are higher than
the typical response rate reported in current operations management literature

(Frohlich, 2002).

4.1.4 Assessment of NPR Survey

A review of the NPR survey instrument was conducted for this dissertation
to ascertain its applicability. The first requirement was for the measures to
adequately reflect the concepts of interest. That is, would the responses to the
survey items indicate the use of quality management practices, cooperative
values and work performance? Using extant literature as a guide, it was
determined that the NPR survey would adequately serve this purpose (the next
section will make this explicit). The second requirement was for the data to
possess a multilevel nature. That is, not only should the items be related to

multiple levels of an organization, but also multiple responses had to be nested

73

 

within each organizational unit. The NPR survey satisfies this requirement as
well. Finally, an appropriate sample size over multiple years is needed to pursue
the objective of this research and to test the hypotheses relating to time; the NPR
survey satisfies this requirement. Therefore, after extensive review, the NPR
survey was deemed appropriate for use in the research study and measurement
model development.

The data were made publicly available in 2002 through the Inter-university
Consortium for Political and Social Research (ICPSR) — a part of the Institute for
Social Research (ISR) of the University of Michigan3. ICPSR maintains and
provides member institutions access to an archive of social science data for
research and instruction. Michigan State University is a member of the ICPSR
Direct program. This program allows access to ICPSR’s data holdings for all
member institution students, faculty, and staff; and the data can be directly
downloaded from the ICPSR website. In order to comply with human—subjects
research issues that concern the many Institutional Review Boards (IRBs) of
universities, data acquired by ICPSR undergoes confidentiality reviews to
discover information to identify respondents. If information is found, it is altered
after consultation with the principal investigator to create files that limit the risk of
disclosure. In the case of the NPR data files, no respondent information was
made available except for a unique identifier and an organizational code with
which the respondent was associated. Communications with the ICPSR

revealed that no issues exist with use of this data for the purposes of this

 

3 See http://www.icpsr.umich.edu/indexhtml

74

dissertation. The above assessment established the promising nature of this

data set. As such, a measurement model was developed and is presented next.

4.2 Measurement Model

This section describes in detail the background to and the justification for

the initial selection of construct measurement items.

4.2.1 Background to Model Development

An initial set of items to construct relationships was needed in order to
begin the process of measurement model development. In order to do this, first
the various concepts of relevance needed to be identified from literature. Using
the frameworks of Anderson et al. (1994) and Detert et al. (2000), the items
corresponding to quality management and cooperative values were identified.
The items for work performance were identified based on service operations
literature (Douglas et al., 2004; Parasuraman, Zeithaml, and Berry, 1988;
Schmenner, 2004). The second task was to make the multilevel differentiation
between the organization and workgroup levels. This was accomplished similar
to the approach of Meade and Eby (2007). Use of the terms “organization,
employees, or workplace” designated the item as reflective of an organization-
level construct. Use of the terms “you, your work unit, or your supervisor”
designated an item as reflective of workgroup-level phenomena. Items without

such explicit designations were assessed for their implicit levels. Based on this

75

approach, the initial measurement model was created as shown in Table 4-3. A

discussion of each concept follows.

 

Table 4-3: Initial Construct Measures

 

NPR Questions

Quality practice for which
measure is reflective

 

Quality Management: Organization level (0°)

1. There are service goals aimed at meeting
customer expectations.

2. There are well-defined systems for linking
customers” feedback and complaints to employees
who can act on the information.

3. Managers communicate the organization’s
mission, vision, and values.

7. Teams are used to accomplish organizational
goals, when appr0priate.

8. Employees are rewarded for working together in
teams (for example, performance ratings, cash
awards, certificates, public recognition).

9. Employees in different work units participate in
cross-functional teams to accomplish work
objectives.

12. Employees receive training and guidance in
providing high-quality customer service

13. Employees receive the training they need to
perform their jobs (for example, on-the- job training,
conferences, workshops).

20. Corrective actions are taken when employees
do not meet performance standards.

Customer Service (Anderson
et al., 1994; Cronin et al.,
1992)

Customer Service (Anderson
et al., 1994; Cronin et al.,
1992; Zhao et al., 2004)

Visionary leadership (Douglas
et al., 2004; Zhao et al., 2004)

Support internal cooperation
(Anderson et al., 1994)

Human Resource Focus
(Zhao et al., 2004)

Support internal cooperation
(Anderson et al., 1994)

Focus on Human Resources
(Zhao et al., 2004) Training
(Douglas et al., 2004; Zhao et
al., 2004)

Focus on Human Resources
(Zhao et al., 2004) Training
(Douglas et al., 2004; Zhao et
al.2004)

Continuous improvement
(Deming, 1986; (Douglas et
al., 2004)

 

 

Quality Management: Workgroup-level (0")

10. Recognition and rewards are based on merit.

Human Resource Focus
(Zhao et al., 2004)

 

76

 

 

Table 4-3: Initial Construct Measures (continued)

 

11. Creativity and innovation are rewarded.

18. In the past two years, I have been given more
flexibility in how I accomplish my work.

26. Do you have electronic access to information
needed to do your job?

29. Considering everything, how satisfied are you
with yourjob?

30. How satisfied are you with your involvement in
decisions that affect your work?

31. How satisfied are you with the recognition you
receive for doing a good job?

Continuous improvement
(Deming, 1986; Douglas et al.,
2004; Amabile, 1996)

Empowerment (Anderson et
al,1994)

Process Management
(Anderson et al., 1994),
Information (Zhao et al.,
2004), Manage by Fact
(Douglas et al., 2004)

Understanding motivation &
employee fulfillment
(Anderson et al., 1994)

Employee empowerment
(Anderson et al., 1994),
Elimination of fear (Deming,
1986)

Understanding motivation &
employee fulfillment
(Anderson et al., 1994)

 

 

Cooperative Values: Organization level (V0)

5. At the place | work, my opinions seem to count.

14. Differences among individuals (for example,
gender, race, national origin, religion, age, cultural
background, disability) are respected and valued.

15. Supervisors/team leaders understand and
support employees’ family/personal life
responsibilities.

16. My organization has made reinvention a priority

(for example, working smarter and more efficiently).

21. Management and the union(s) work
cooperatively on mutual problems. (“If you don’t
know leave this item blank” was added in 1999.)

25. Are you clear about how “good performance” is
defined in your organization?

Employee involvement in
decision making (Detert et al.,
2000)

Willingness to work together
(Wagner, 1995), Collaboration
(Detert et al., 2000)

Concern for others (Chatman
et al., 1995), Achieving results
that stakeholders consider
important (Detert et al., 2000)

Shared vision is necessary,
improvement is valued (Detert
etaL,2000)

Cooperation helps everyone
(Chatman et al., 1995),
Cooperation and collaboration
(Detert et al., 2000)

Shared vision is necessary
(Detert et al., 2000)

 

77

 

 

Table 4-3: Initial Construct Measures (continued)

 

 

Work Performance: Workgroup-level (PW)

4. My immediate supervisor has organized our work Effectiveness of teamwork
group effectively to get the work done. (Douglas et al., 2004)

17. In the past 2 years, the productivity of my work Work flow (Schmenner, 2004)
unit has improved.

 

 

32. Overall, how good a job do you feel is being Effectiveness of teamwork
done by your immediate supervisor/team leader? (Douglas et al., 2004)

33. How would you rate the overall quality of work Service quality (Parasuraman
being done in your work group? et al., 1988)

4.2.2 Organization-level Quality Practices (0°)

As shown in Table 4-3, organization-level quality management practices
were associated with nine items. These nine items were related to five quality
management practices: customer service, visionary leadership, supportive of
internal cooperation, continuous improvement, and a focus on human resources
(Anderson et al., 1994; Zhao et al., 2004). Items 1 and 2 relate to customer
service goals and feedback, which are indicative of a customer-service focus
(Anderson et al., 1994) and implicitly exist at the organization level. Item 3
reflects Anderson et al.’s (1994) visionary leadership category, while items 8, 12
and 13 reflect Zhao et al.’s (2004) focus on human resources — rewarding or
developing employees. Anderson et al.’s (1994) category of quality management
practices that support internal collaboration / cooperation are shown in items 7
and 9 related to the use of teams and cross-functionality. Finally, the item 20

relating to corrective actions is expected to indicate the presence of continuous

78

improvement practices (Anderson et al., 1994). Many of these items refer to

employees and are deemed to relate to organization-level initiatives.

4.2.3 7 Workgroup-level Quality Practices (0")

Seven items were identified as likely measures for workgroup-level quality
practices. These items were deemed to relate to workgroup-level initiatives
because they explicitly ask about the respondent’s immediate environment (e.g.,
flexibility in their work, their involvement in decisions) or ask about initiatives
implicitly local within the organization [e.g., creative or meritorious work (Amabile,
1996; Brown, 2001; Drazin, Glynn, and Kazanjian, 1999)]. Specifically, item 10
measures the presence of a human resource focus (Zhao et al., 2004), while
item 11 assesses continuous improvement (Anderson et al., 1994) because of its
emphasis on innovation. Items 18 and 30 ascertain the level of employee
empowerment (Anderson et al., 1994) by asking respondents about flexibility and
involvement, respectively. The question as to whether employees have
electronic access to information in item 26 is reflective of process management
(Anderson et al., 1994) and emphasizing information (Zhao et al., 2004). Finally,
item 29 is concerned with satisfaction while item 31 relates to recognition; both
are indicative of the workgroup’s understanding of motivation and employee
fulfillment (Anderson et al., 1994). Together these items were expected to reflect

workgroup-level quality practices.

79

4.2.4 Organization-level Cooperative Values (V0)

Organizational cultural values are an aspect of organizational culture and
are understood to exist at the organization level (Schein, 2004). Many of the
values described by Detert et al. (2000) associated with quality practices can be
described as reflective of a cooperative environment and perceptual measures
reflecting those values were identified from the NPR survey. For instance with
respect to item 5, if many respondents agree that their “opinions seem to count,”
this could be construed to reflect an organization that values employee
involvement in decision making (Detert et al., 2000). Item 14 asks if “differences
are respected,” which captures a willingness to work together (Wagner, 1995)
and valuing collaboration (Detert et al., 2000). Item 15, which asks a question
regarding the support for employee personal issues, could be indicative of a
concern for others (Chatman et al., 1995), especially when employees are
considered stakeholders (Detert et al., 2000). When employees recognize that
their organization has made “reinvention a priority” (item 16) and are clear as to
what is “good” performance (item 25), these reflect valuing a shared vision of
improvement (Detert et al., 2000). Finally, an organization that believes
cooperation helps everyone (Chatman et al., 1995; Detert et al., 2000) would be
indicated by a high degree of mutual problem solving between management and
an employee union (item 21 ). These six measures are expected to collectively

indicate the presence of cooperative cultural values in an organization.

80

4.2.5 Workgroup-level Work Performance (PW)
Performance is often measured in terms of efficiency and effectiveness
(Douglas et al., 2004; Miron et al., 2004; Schmenner, 2004) and three of the four
work performance items seek to measure efficiency and effectiveness.
Specifically, items 4 and 32 measure the degree of effective workgroup
organization and management similar to Douglas et al.’s (2004) idea of
effectiveness of teamwork in a service setting. Work unit productivity is
measured in item 17, which parallels Schmenner’s (2004) work flow performance
concept. The fourth performance item (item 33) is a perceptual measure of the
quality of work being done. In a service setting, this can be equated to service
quality (Parasuraman et al., 1988). It should be noted that each item relates to
either the respondent’s immediate supervisor or the workgroup. As such, each of
these four perceptual measures is reflective of performance at the workgroup

level. While this section discussed the conceptual validity of the measurement

items, the next section presents the data analysis method used in this research.

4.3 Data Analysis Method

The data analysis approach used in this dissertation is presented in Figure
4-1. The data analysis avoided unnecessary aggregation, averaging or
dichotomizing of the data. In addition, efforts were made to avoid eliminating
data to maximize the use of the available information. Although the number of

respondents in the data set is quite large, a relatively small number of

81

 

Figure 4-1: Flow Chart of Analysis Method

 

 

 

 

 

 

 

   

 

 
    

Analysis Performed Data Used
Raw NPR Data
All items
Missing Value Analysis
EM algorithm imputation 3
/ Semi-imputed NPR
/ Data
CFA with validity checks
CMV assessment and 5
control

    

CMV~adjusted
structure matrix

 
   

 

 

 

Unit-weighted factor scoring
Indeterminacy assessment

 

  

Factor scores
All respondents

 

  
   

a i

Aggregation assessment
Organizational averaging
Remove invalid org codes

 

 

 

 

 

  
 

Level 1 Data
Level 2 Data

Parameter Estimates ,
Hypothesis Testing

 

   

    

 

 

organizations exist within the sample. The organizations coded as other were
problematic because multiple organizations could be present within each other
category. Therefore, responses with other designation were utilized for data

analysis until a specific organization assignment was required (i.e., item 8 in

82

Figure 4-1). In addition, respondents with missing items were retained. The
expectation-maximization (EM) algorithm available in SPSS 15.0 was used for
dealing with missing values (Dempster, Laird, and Rubin, 1977). Although some
issues have been raised with earlier SPSS versions of missing value analysis
(von Hippel, 2004), all respondents and all items were utilized for imputing the

missing values in order to increase imputation effectiveness (Acock, 2005).

4.3.1 Factor analysis

To refine the initial measurement model, a confirmatory factor analysis
(CFA) was conducted at the respondent level on the 1998 data. After
modifications, the resultant measurement model was then validated using the
1999 data. An important goal of the CFA process was to end with at least three if
not four items per construct, provided that the model attained acceptable fit
indices (Anderson and Gerbing, 1988; Bentler, 1995). The measurement model
was modified iteratively using the Lagrange Multiplier (LM) test for item deletion,
while retaining the proper theoretical content in the model. Once a preliminary
acceptable model was identified, a series of composite reliability and discriminant
validity tests were conducted (Fornell and Larcker, 1981; Venkatraman, 1989).
Any issues identified during these tests were resolved through further model
modifications. Following these validity checks, an assessment was made for
common method variance (CMV), which is likely to occur with perceptual
measures with social desirability (Malhotra, Kim, and Patil, 2006; Podsakoff,

MacKenzie, Lee, and Podsakoff, 2003; Song and Zahedi, 2005). Based upon

83

the constructs of interest, an intermixing of construct items was observed in the
NPR survey instrument, which aids in CMV control (Podsakoff et al., 2003).
However, a significant presence of CMV would further be controllable via the
correlated uniqueness approach within the CFA. As shown in Figure 4-1, at the

end of this series of tests, a valid measurement model was developed.

4.3.2 Data aggregation

The multilevel analysis method used in this dissertation is termed
hierarchical linear modeling (HLM). In order to prepare the data for the HLM
analysis, some degree of aggregation was necessary. That is, in the context of
this dissertation, HLM was not to be used for covariance analysis like structural
equations modeling (SEM). Instead, estimated values for each concept (i.e.,
factor values) were required for each respondent. For example, a single value
for workgroup-level performance (PW) was required for the each respondent.
These values were then used as workgroup responses within each organization.
When needed, the values within each organization were aggregated via an
average. For example, the workgroup response values for organization-level
quality management practices (QO) within each organization were averaged.
Before aggregating, interclass correlation (ICC) values and within group
correlations (rwg) values were assessed (de Jong, de Ruyter, and Wetzels, 2005).

One approach to estimating factor values for each respondent is to sum or

average the item values. Although this approach is common, it not only ignores

CFA informatidn (i.e., factor loadings), it also removes the CMV controls present

84

within the CFA. Another approach would be to use factor scores; however, many
methods assume factors to be orthogonal (Gorsuch, 1983), which in the
nomological context of this dissertation is incorrect. Therefore a unit-weighting
approach was chosen, which is recommended by Grice (2001) for computing
factors assumed to be non-orthogonal — i.e., oblique. This method utilizes the
structure matrix computed from the CFA, which in this case is CMV controlled, as
will be demonstrated in Chapter 5. Under the unit-weighting procedure,
individual items estimated to correlate over a certain threshold with any factor are
used to compute that factor score. A unit-weight is used, not the structure matrix
value, but the sign of the correlation is preserved. In this approach, one item
may positively and/or negatively affect multiple factors. The resultant set of
values (i.e., -1, 0 or 1) resides in a weight matrix that can be used in subsequent
data sets. Grice (2001) found this approach to reproduce factor correlations
favorably and to be stable over multiple periods. The unit-weight factor scoring
approach was chosen for this reason. Finally, a concern with factor scores is the
issue of indeterminacy (Wilson, 1928). Therefore, the approach for

indeterminacy evaluation provided by Gorsuch (1983) was utilized.

4.3.3 Hierarchical Linear Modeling

Two sets of data were generated for each year. The first set of data,
termed level-1, contained the responses for workgroup iwithin organization j for
each factor (i.e., Qoij, V0,), Qwij, and PW”). The second data set was termed level-

2 and it contained the aggregated (i.e., averaged) values for the theory-based

85

organization-level factors (i.e., 00,,- and V9,). As was stated, multilevel modeling
is necessary for testing the hypotheses and the postulated relationships among
these variables (constructs). However, multilevel modeling has difficulties with
endogenous relationships like those shown in Figure 3-1 because of the resultant
correlated effects (Kim and Frees, 2006). Unlike SEM and other simultaneous
equation methods, HLM requires that one dependent variable to be specified
(Snijders et al., 1999). One approach to accommodate mediating factors is the
use of instrumental variables (Kim et al., 2006). Another approach has been to
test models with mediation in multiple stages (de Jong et al., 2005; Naveh et al.,
2004). Because of the ability to isolate specific relationships for examination, the
multi-stage approach was chosen for this dissertation. This approach will test
multiple models in a sequential manner as shown in Table 4-4. The first analysis
will test the relationship between Q0 and V0 (Model 1) to draw conclusions with
respect to hypothesis 1. The second analysis will test the relationships among
00, V0 and QW (Model 2) to make conclusions regarding hypotheses 2 and 3.
Lastly, the relationships among 0", v0, QW and PW (Model 3) will be analyzed to

make conclusions regarding hypotheses 4, 5, and 6.

 

Table 4-4: Hypothesis Testing Framework
Dependent Factor
Model 1 Model 2 Model 3
Independent I7a QH PW

 

 

 

meter
0 H1 (7.") H20?) H4051)
V“ Hair?) Hsrn’zi
0“ Harris)

 

86

4.3.3.1 Model1 Formulation
The first stage of HLM analysis tests the hypothesized relationship
between Q0 and V0 through a type of random-intercept model known as means-

as-outcomes model (Raudenbush and Bryk, 2002).

Level 1: V0.0 = fig, +5," (1)
Level 2: AI, = 73’ + r." (Qf’ - Q”) + at," (2)
Where,

lg": Cooperative cultural values in organization j measured by workgroup i

3;”, = Parameter estimate of aggregate of V in organization j

r,’ = Variation of Vfor workgroup i in organization j, assuming r; ~ N(0,oz)

75" = Parameter estimate of overall aggregate of V

y,” = Parameter estimate of effect of Q.) on aggregate of V in organization j

Q? = Organizational quality management practices in organization j = ZQf/n,
Q": Grand mean of Q0 for all organizations

at = Organization j unique variation, assuming “i ~ N(0,r0,,)

In this two-level model, the variance of level-1 dependent variable shown

in (1) is explained by two parameters: the random intercept ,ng that varies by

organization and errorigf’ that varies by workgroup. The intercept is interpreted as

87

the organizational mean V0. This intercept is the level-2 dependent coefficient in
(2) and is explained by a grand V0 mean y: , a grand centered mean organization
value of 00, and a random effect associated with each organization uj. The

organization-level predictor 00 is grand-mean centered to enable interpretation
of y: to be the grand mean of V0 (Raudenbush et al., 2002). An assumption is

that 00 influences the workgroup-level response through the organization-level

mean of V0. Thus, HLM will fit this system of nested models and test for the
significance of y,“ thereby providing a test of hypothesis 1 that posits Q0 to

positively influence V0.

4.3.3.2 Model 2 Formulation

The second stage of HLM analysis will test the hypothesized relationship

between 00, V0 and QW also through a means-as-outcomes model.

Level 1: Q',’ = p09, + r? (3)
Level 2: 4?,- = 7.? + 719(Q? - Q") + 729W,“ - 17.0) + “i (4)
Where,

Q,',.” = Quality management implemented by workgroup i in organization j
,6,” = Parameter estimate of aggregate of QW in organization j
r19: Variation of QW for workgroup i in organization j, assuming 27,9 ~ N(0,az)

V, = Cooperative cultural values in organization j = 2V”- /n,

88

179: Grand mean of V0 for all organizations

)4? = Parameter estimate of overall aggregate of QW

7,9 = Parameter estimate of organizational Q19 effect on QW in organization j
y? = Parameter estimate of organizational Vj effect on QI in organization j

14?: Organization j unique variation, assuming uf.’ ~ N(0,r00)

The variance of level-1 dependent variable 0'” shown in (3) is explained

by two parameters: the random intercept ,609, that varies by organization and

error n19 that varies by workgroup. The intercept is interpreted as the
organizational mean 0‘”. This intercept is the level-2 dependent variable shown
in (4) and is predicted by an estimated grand 0‘” mean yOQ, two grand-mean
centered organization values 00 and V0, and a random effect associated with

each organization u]? The organization-level predictors are grand-mean

centered to enable interpretation of 7,? to be the grand mean of 0‘”. A special

note is that V0 is the average cooperative value factor score within organization j.

Thus, HLM will fit this system of nested models and test for the significance
of 7,9 and 7? thereby providing a test of hypotheses 2 and 3 that posit QO and V0

positively influence 0‘”.

89

4.3.3.3 Model 3 Formulation
The third stage of HLM analysis will test the hypothesized relationships
between QO, V0, QW and PW through a random-intercept with level-1 covariate

model (Raudenbush et al., 2002).

Level 1: PHI” 2 ,ij + ,B,P(Q;V -Q,”")+ r0.” (5)

Level 2: flu”,- = ml; + mire? — Q") + 70W,“ -— 7.0) + at, (6)

Level 2: A” = 75, + u,” (7)

Where,

R,” = Workgroup-level work performance for workgroup i in organization j
,BJ’, = Parameter estimate of average PW in organization j

pi; = Parameter estimate of workgroup Q?” effect on PW in organization j
Q,” = Mean of 0'” for organization j

n,” = Variation of PW for workgroup i in organization j, assuming 2;,” ~ N(0,02)
mi, = Parameter estimate of overall average of PW

ml]: Parameter estimate of organizational Q? effect on PW in organization j
Vii; = Parameter estimate of organizational V_, effect on PW in organization j
75,: Parameter estimate of overall effect of QW on PW

ug’, = Organization j unique variation in aggregate of PW, assuming ug’, ~ N(0, r00)

uf: = Error in estimated effect of PW, assuming u,’,. ~ N(0,0)

90

The variance of level-1 dependent variable PW shown in (5) is explained

by an intercept ,60’; that varies by organization, a workgroup-level group-centered

0'”, and errornj.” that varies by workgroup. The intercept is interpreted as the

organizational mean PW.
ln model 3, two level-2 equations exist. In the first, the level-1 intercept is

the level-2 dependent variable shown in (6) and is predicted by an estimated

grand PW mean 70%, two grand-mean centered organization values 00 and V0,
and a random effect associated with each organization uf,. The organization-

Ievel predictors are grand-mean centered to enable interpretation of 7.1;, to be the

grand mean of PW. In the second level-2 equation (7) the slope ,6,” representing

the workgroup-level effect of 0‘” is the dependent variable. This slope is
assumed non-randomly varying for statistical efficiency and computational

stability (Raudenbush et al., 2002). Thus, it is predicted solely by a grand

effect 7,’,’, with no error (i.e., u,” is assumed to have a mean and variance of zero, it
is retained in the model for post-hoc analysis). Thus, HLM will fit this system of
nested models and test for the significance of 751,753, and 7,’,’,thereby providing a

test of hypotheses 4, 5, and 6 that posit 00, V0 and QW positively influence PW

respectively.

91

4.3.4 Estimation Procedure

To estimate level-2 parameters, HLM uses a Generalized Least Squares
(GLS) estimation procedure. This provides a weighted level-2 regression such
that the groups with more precise level-1 estimates of the dependent variable
receive more weight in the level-2 regression equation (Hofmann, 1997; Snijders
et al., 1999). For each individual, two predicted intercept and slope values can
be estimated: one from the level-1 model, and the other from the level-2 model.
To combine both estimates, HLM computes an empirical Bayes estimate of the
level-I intercepts and slopes for each individual that optimally weights the
Ordinary Least Squares (OLS) level-I estimates and the level-2 predicted values
for these same estimates. These empirical Bayes estimates are contained in the
residual file generated by the HLM software. Hypothesis testing will be
conducted using the point estimates and significance levels for the parameters
corresponding to the relationships hypothesized above. See Table 4-4 for the
explicit hypothesis-parameter correspondence. The next section discusses the

approach for testing the three diffusion hypotheses.

4.4 Testing Diffusion Hypotheses

In order to test the diffusion hypotheses 7, 8, and 9; year-over-year
comparisons were made for each of the parameters shown in Table 4-4. The
expected parameter changes suggested by the diffusion hypotheses are shown

in Table 4-5. Hypothesis 7 consists of three relationships that are represented by

parameters 7,” , 7,9 and 75. A significant decrease in strength is expected for each

92

from 1998 to 1999 and from 1999 to 2000. Hypothesis 8 consists of two
relationships represented by parameters 729 and 75;. A significant year-over-year
decrease in strength is also expected for these two parameters. Hypothesis 9
consists of one relationship represented by parameter 7,7,, of which a significant

increase in strength from year to year is expected. An equality of coefficients t-
test utilizing parameter and standard error estimates was used to detect changes
from year-to-year. The test results were compared to the hypothesized changes

and conclusions were drawn.

 

 

 

 

 

 

Table 4-5: Expected Changes in Effects

Hypothesis Relationship Parameter Expected Pattern

7: EffeCts Of QO 00 —r V0 71V 7151993 > 713999 > 712000
QO -> QW 71Q 78993 > 73999 > 732000
0° 4 PW 75’. ring. > 751...... > 7512......

8: EffeCtS Of VO V0 —-+ QW 72g 7291993 > 7291999 > 729.2000
V0 —) PW 7:2 7(1):,i998 > 75.1999 > 75.2000

93 EffeCts 0f QW QW —+ PW 71,0 781993 < 7110,1999 < 7110,2000

 

 

4.5 Summary of Research Methodology

In this chapter the research methodology used was presented in detail.
The data requirements were first described which led to the use of secondary,
archival data from the National Partnership for Reinventing Government. This
data was assessed based upon extant literature to see if the concepts of
interests could be ascertained from the survey instrument. After this was

confirmed, the item-to-construct relationships were presented. The remainder of

93

the chapter described the data analysis approach to test the hypotheses
developed in chapter 3. This approach included a CMV-controlled CFA, a unit-
weighting factor score approach, and a multi-stage HLM process. Having
described the methodological approach for testing the nine research hypotheses,

the next chapter will present the results of this analysis and testing.

94

CHAPTER 5: ANALYSIS RESULTS

This chapter will review the process of data preparation, the results of the
hierarchical linear modeling (HLM) and the year-over—year test results. In
addition, the data validation, factor analysis and measurement model validation
are presented. Conclusions are drawn with respect to the empirical support of

the nine research hypotheses.

5.1 Data validation

Upon initial examination of the NPR data it was discovered that a majority
of cases had some degree of missing data. That is, either the respondent had
left an item blank or had responded with a don't know answer. As shown in
Table 5-1, although generally each item is missing, less than 5% of values, a list-
wise deletion of respondents would result in only 31% to 46% of usable cases,
depending upon the year. As this is a substantial loss of information, replacing
missing values was deemed essential. There was a general increase in the
percent of missing values within each item in later years. As was stated in
Chapter 4, the EM algorithm available in SPSS version 15.0 was used to impute

these missing values using all respondents and all 32-33 items in the survey.

95

 

Table 5-1: Percent of Responses with
Missing Values

 

1998 1999 2000
Sample Size 13,689 18,154 31,975

 

 

Number valid if 4,188 11,543
had list-wise - (31%) 8,384 (36%)
deleted (46%)
NPR Questions
Quality Management: Organization-level (0°)
1 0.8 3.6 6.5
2 0.9 4.5 9.8
3 0.5 1.4 2.5
7 0.4 2.2 6.3
8 0.4 3.4 7.0
9 0.6 4.0 9.0
12 0.5 1.5 2.8
13 0.3 0.5 0.7
20 1.5 8.1 9.0
Quality Management: Workgroup-level (0")
10 0.6 2.7 3.2
11 0.8 2.0 3.4
18 1.5 2.8 5.1
26 1.8 2.3 15.8
29 (28*) 0.9 1.1 0.5
30 (29*) 1.0 1.1 0.6
31 (30*) 1.1 1.1 0.8
Cooperative Values: Organization-level (V0)
5 0.3 1.9 1.7
14 1.3 2.2 1.8
15 1.2 2.6 3.2
16 1.6 3.6 7.1
21 4.3 43.5 42.1
25 1.9 4.0 6.9

 

96

 

 

Table 5-1: Percent of Responses with
Missing Values (continued)

 

Work Performance: Workgroup-level (PW)

4 0.4 1.6 3.0

17 1.5 5.2 9.1
32 (31*) 1.1 1.2 0.9
33 (32*) 1.2 1.4 1.0

 

* Question number for 1999 & 2000

 

 

 

An important consideration with missing data is the assessment of
whether the data are missing completely at random (MCAR). When missing
values are considered MCAR they are considered to be randomly distributed
across all observations. In such instances missing case may be ignored through
listwise deletion, otherwise imputation is warranted (Little and Rubin, 1989).
Little’s MCAR chi-square test was performed for each year, where an
insignificant result means the data are MCAR. For each year the MCAR
hypothesis was rejected: For 1998 f = 18152 (17,538), p<.01; for 1999 f:
71,108 (62,436) p<.001; for 2000 72 = 169,361 (155,329) p<.001. The
assumption was therefore made that the data were missing at random (MAR),
which means the missing data depends upon the non-missing data. Using all the
available response data, missing values were then imputed using the EM
algorithm. An assessment of the impact this imputation had on the means and
correlations of the variables was done (see Appendix A.1). These differences
were deemed not substantial to warrant concern and therefore analysis

progressed to factor analysis.

97

5.2 Measurement model construction and validation

Upon establishing the data set as acceptable to conduct analysis, the
process of refining and validating the measurement model ensued. This section

presents the steps taken during this process.

5.2.1 Confirmatory Factor Analysis

Subsequent to replacement of missing values, a CFA was performed
using EQS 6.1 (Bentler, 1995) on the 1998 data with the initial set of target items
(as shown in Table 4-3 in Chapter 4). The result was 72 (df) = 33,913 (372),
RMSEA = 0.081; NFI = 0.84; CFI = 0.84. For the alternative fit statistics — such
as NFI and CFI — to be considered acceptable, their values should be at least
0.90 and approaching 0.95 (Bentler and Bonett, 1980; Hu and Bentler, 1999).
Also, the RMSEA value should be below 0.8 (Curran, Bollen, Chen, Paxton, and
Kirby, 2003). The fit indices for this initial model were therefore deemed
unacceptable. Model modification was subsequently performed utilizing the
lagrangian multiplier (LM) test to detect where significant 12 changes may result
from model changes. Items cross-loading were eliminated unless they
possessed important theoretical content. For instance, question 9 regarding
cross-training has important sociotechnical implications and was retained. Also,
the motivation to have three or more items per construct led to the retention of
question 4.

The result of the model modification process for the 1998 data is shown in
Table 5-2. A four-item-per-factor model was produced with fit statistics as

98

follows: 23 (df) = 5,821 (98), RMSEA = 0.065; NFI = 0.94; CFI = 0.94. The CFI
and NF I fit statistics are well above the 0.90 criterion and the RMSEA value is
well below the 0.8 criterion. All loadings were significant (p<.01) and above 0.5
except for question 26, which was 0.44. Although this was below a 0.5 ideal
value, retaining four items for QW was deemed more crucial for measurement

validity.

 

Table 5-2: Final Measurement Model from 1998 Data

 

 

Measures Loadings t-value

Fit Indices x2 (df) = 5821 (98); RMSEA = 0.065; NFI = 0.94; CFI = 0.94

Quality Management: Organization-level, 00 (n = 4; CR = 0.74 )

2. There are well-defined systems for linking customers’ 0.61 73.70
feedback and complaints to employees who can act on the
information.

3. Managers communicate the organization’s mission, vision, 0.75 96.84
and values.

9. Employees in different work units participate In cross- 0.61 73.54
functional teams to accomplish work objectives.

13. Employees receive the training they need to perform their 0.62 74.99
jobs (for example, on-the-job training, conferences,
workshops).

Quality Management: Workgroup-level, QW (n = 4; CR = 0.82 )

11. Creativity and innovation are rewarded. 0.78 105.18

26. Do you have electronic access to information needed to do 0.44 51.04
your job?

30. How satisfied are you with your involvement in decisions 0.80 109.27
that affect your work?

31. How satisfied are you with the recognition you receive for 0.80 109.29
doing a good job?

Cooperative Values: Organization-level, V0 (n = 4; CR = 0.72 )
14. Differences among individuals (for example, gender, race, 0.59 72.38
national origin, religion, age, cultural background, disability)
are respected and valued.

16. My organization has made reinvention a priority (for 0.66 81.52
example, working smarter and more efficiently).
21. Management and the union(s) work cooperatively on 0.50 59.53

mutual problems. (“If you don’t know leave this item blank”
was added in 1999.)

25. Are you clear about how "good performance” is defined in 0.70 88.10
your organization?

 

99

 

Table 5-2: Final Measurement Model from 1998 Data (continued)

 

Work Performance: Workgroup-level, PW (n = 4; CR = 0.85 )

4. My immediate supervisor has organized our work group 0.84 116.55
effectively to get the work done.

17. In the past two years, the productivity of my work unit has 0.60 74.26
improved.

32. Overall, how good a job do you feel is being done by your 0.86 120.40
immediate supervisor/team leader?

33. How would you rate the overall quality of work being 0.66 83.25

done in your work group?

 

 

Note: All t-values are significant at p<.05, CR = composite
reliability

 

 

In order to assess if the model adequately represents the proportion of
measured variance attributable to the latent variable, composite reliability p0
values were computed (Venkatraman, 1989). A pc over 0.5 implies that the
variance captured by the model is more than the error components. These
values for 1998 ranged from 0.72 to 0.85 and were therefore deemed reliable.
Further testing was done on 1999 data for validation purposes. The CFA for
1999 produced the following fit statistics: f (df) = 4,115 (98), RMSEA = 0.064;
NFI = 0.94; CF I = 0.94. All loadings followed acceptable patterns. These results
revealed that the measurement model was acceptable. Next, tests for
unidimensionality and other conventional validation test were carried out, which

are discussed in the following sections.

5.2.2 Unidimensionality

To assess the unidimensionality and convergent validity of each construct,
the approach suggested by Venkatraman (1989) was utilized. In this approach

the goodnesS-of-fit for each item-construct set is assessed independently

100

through a CFA. The 7? value and incremental fit index A (Bentler and Bonett,
1980) are computed for each CFA. The incremental fit index indicates the
practical significance of the model in explaining the data and is computed with
the equation (F0 - Fk)/F0, where F0. is the null model and Fk is the specific model
for construct k. These results are presented in Table 5-3. Although all ,1? values
are significant (p<.001), this is not surprising because of the large sample size.
Alternatively each A is above the 0.95 recommended threshold (Bearden,
Sharma, and Teel, 1982). Based on this information, it was concluded that each
construct passed the tests for unidimensionality and convergent validity at the
monomethod level of analysis. This established confidence that the measures
reflected their expected constructs. However, quality management and
cooperative values are closely associated; therefore tests were conducted to

assess discrimination between the constructs, which are discussed next.

 

 

 

 

 

Table 5-3: Unidimensionality Test - Monomethod
CFA Results
Factor No. of 2; df p-Ievel A
indicators
0° 4 96.776 2 <.oo1 0.992
0'” 4 71.335 2 <.oo1 0.996
v0 4 11.848 2 <.001 0.999
PW 4 998.199 2 <.oo1 0.956
5.2.3 Discriminant Validity

After establishing reliability and unidimensionality, the discriminant validity
of the constructs was tested. The results are shown in Table 5-4. Pair-wise tests

between constructs were made to assess if construct correlations are

101

significantly different from unity. To do this, pair-wise CFAs with correlations
constrained to be one were compared with unconstrained pair-wise CFAs.
Significantly lower X2 values for the unconstrained model supports discriminant
validity. Six tests were performed for each unique pair with each A72 assessed.
Significantly lower values were found, thereby providing support for discriminant
validity. The relationship between Q0 and V0 was noted to be highly correlated;

however, more discrimination occurred later with the CMV modifications.

 

Table 5-4: Discriminant Validity

 

 

 

ML est Constrained Unconstrained
(pairwised) model model
Test# Pair Unconstrained 7’ df 2r df Af’ Sig.
Correlation
(t-value)
Q0 to
1 W 0.865 (176.1) 2,412.39 20 1,569.02 19 843.37 <.001
2 V" 0.975 (197.2) 442.51 20 415.62 19 26.89 <.001
3 PW 0.779(137.8) 4,722.94 20 2,301.32 19 2,421.62 <.001
tho
4 V" 0.921 (199.9) 1,176.05 20 879.15 19 296.90 <.001
5 P” 0.801 (173.0) 5,561.11 20 2,236.16 19 3,324.96 <.001

 

 

6 P 0.796 (136.3) 3,692.50 20 2,102.81 19 1,589.70 <.001

 

5.2.4 Predictive Validity

A final validity assessment was made that assesses if the constructs
correspond to each other in a nomological network (Gerbing and Anderson,
1988). This was done in a structural equations model (SEM) with the entirely
dependent construct PW being predicted by the other three constructs QO, QW

and V0. Note that the true theoretical test is in the HLM context — that is, Q0 and

102

VO operate at a level higher than at the CFA respondent level. Therefore this
SEM was used solely to assess for construct (i.e., predictive) validity. The result
of this test is shown in Table 5—5 with the strength of each structural relationship
represented by 7. All relationships were found to be significant, thereby
bolstering the claim of predictive validity. Due to nature of the constructs,
however, the correlations observed could be explained by social desirability and
use of a single measurement instrument. To assess and control for this,

common method variance analysis was conducted and is discussed next.

 

 

 

Table 5-5: Assessment of Predictive Validity'r
Fully Deegndent Factor

Explanatory t-value

Factors Y

0° 0.318 34.776 (***)

0‘” 0.510 47.165 (**)

PW 0.249 28.219 ("*1

*All respondents from all organizational codes

(**)—p < 0.05.

(***)—p < 0.01.

 

 

 

5.3 Common method variance control

This next section reviews the process for how common method variance
(CMV) was assessed and controlled. First preliminary tests were done to assess
the degree to which CMV was present. A moderate amount of CMV was
detected and therefore without controls some erroneous inferences might have
been made. The last portion of this section details the method used to control for

CMV.

103

5.3.1 Preliminary assessments

A basic concern regarding CMV is erroneous inferences about the effect
of exogenous constructs on endogenous constructs caused by the common
survey instrument. Multiple CMV. assessment techniques were used to detect
the degree to which this is a possibility. One approach used was the marker
variable technique (Malhotra et al., 2006) where the lowest element in 2 matrix
from the CFA indicates a percent of method bias. For the 1998 data this value
was 0.197 indicating a possible 19.7% inflation. Another approach is Harmon’s
one-factor assessment using exploratory factor analysis (Podsakoff, 2003). This
approach estimated that 45% of the 1998 data variance can be explained by one
factor. The problem with this technique is that it includes true variance as well as

method variance.

5.3.2 Correlated uniqueness

The final test conducted has been termed the correlated uniqueness test,
where construct item errors between constructs of concern are allowed to covary
in a CFA (Podsakoff, 2003). The effect of free error covariance is then assessed
and, if necessary, controlled. In a CFA this equates to certain free off-diagonal
elements in the (95 matrix. Because the construct Q0 was the only true
exogenous factor, its four item errors were free to correlate with the other 12
endogenous construct item errors. Constraints were then placed upon each free
path to equate to zero but were allowed free if the LM test revealed a significant
impact. Comparisons were made with and without correlated uniqueness

104

 

 

 

 

 

Table 5-6: Effect of Correlated Uniqueness - Common Method Variance
CMV) Control
Before CMV control After CMV control
Fit indices X2 = 5822 (98); NFI=.941; X2 = 3762 (61); NFI=.962;
CFI=.942; RMSEA=.065 CFl=.963; RMSEA=.067
Error Error
Question Loading Variance R2 Loading Variance R2

Organization-level Quality Practices
Q2 0.608 0.794 0.370 0.634 0.773 0.403
Q3 0.753 0.658 0.567 0.770 0.639 0.592
Q9 0.606 0.796 0.367 0.568 0.823 0.322
Q13 0.615 0.788 0.379 0.587 0.809 0.345

Workgroup-level Quality Practices
Q11 0.782 0.623 0.612 0.773 0.635 0.597
Q26 0.437 0.900 0.191 0.429 0.903 0.184
Q30 0.804 0.594 0.647 0.807 0.590 0.652
Q31 0.806 0.592 0.649 0.814 0.581 0.662

Organization-level Cooperative Values
Q14 0.594 0.804 0.353 0.608 0.794 0.369
Q16 0.656 0.755 0.430 0.637 0.771 0.406
021 0.503 0.864 0.253 0.497 0.868 0.247
Q25 0.697 0.717 0.486 0.705 0.709 0.497

Workgroup-level Work Performance
Q4 0.841 0.541 0.707 0.839 0.545 0.703
Q17 0.604 0.797 0.365 0.595 0.804 0.354
Q32 0.859 0.511 0.739 0.870 0.493 0.757
Q33 0.661 0.750 0.437 0.656 0.754 0.431

 

controls. Without error covariances f (df) was 5,821 (98) with a consistent

 

Akaiki information criteria (CAIC) of 4790. The LM test suggested 37 of the 48
possible Game” elements should be freed. Upon doing so the result was a ,1]
(df) of 3762 (61) with a CAIC of 3120. This represented about a 35% model
improvement — large enough to warrant CMV control. Such a control was
accomplished through retention of the 37 free error covariances when computing
the structure matrix used for calculating factor scores. For illustrative purposes
Table 5-6 reveals the change in factor loadings as a result of CMV-control. As
shown, some loadings increase while others decrease. An important note is in
relation to the discriminant validity concern between 00 and V0 noted earlier.

105

Prior to CMV control the A12 was 26.89 (see Table 5-4, test #2), indicating a
slight but significant discrimination. However, with CMV control the Ag; value
increased to 105.39, indicating an improvement in discrimination. This process
controlled for method variance, thereby giving higher confidence towards
inferring correlations to be of causal and not methodological origins. Because
singular values for each construct are required for multilevel analysis, a factor

scoring process was used and is discussed in the next section.

5.4 Factor Scoring

With CMV controlled, a unit weighting factor scoring procedure was used
to derive factor scores for each respondent (Grice, 2001). The procedure is
shown in Table 5-7. In step one; the 1998 data was used to derive a full
structure weight matrix, which is the estimated degree to which each item
correlates with an underlying construct. Correlations exist between each item-
construct pair. In steps two and three, the most salient items were parsed out
based upon the threshold value or. The outcome is a unit weight matrix WU
based upon the 1998 analysis. In step four, the same WU is used to compute
factor scores for 1998, 1999 and 2000. The approach used to determine or and

the unit-weight matrix are detailed in the Appendix section A2.

106

 

Table 5-7: Unit-Weight Factor Scoring Procedure

 

 

1“: Compute Full Structure Weight Matrix, WFS = Rw‘18vf
RW‘1 = Inverse of actual correlation matrix of v variables
SW = Structure matrix (variable to factor correlations) from CMV-controlled CFA

2nd : Create absolute valued WFS, determine max value per factor (MAXf)

3rd : Create Unit Weight Matrix (wuvf)
For each element of WFS, if element leSIIMAXf > a, then wU = 1 or -1 depending
on sign of original wFs, where a = salience value

4th : Compute factor scores Funf = ZnVWUVf, where va=standardized n responses to v
variables

 

 

Before accepting the computed factor scores as valid, the degree of
indeterminacy was assessed by observing the validity coefficients in the diagonal
of R,,. The issue of indeterminacy, long noted as a problem in factor analysis
(Wilson, 1928), pertains to the fact that an “an infinite number of ways for scoring
the individuals on the factors could be derived that would be consistent with the
same factor loadings” (Grice, 2001, 430). Therefore, an assessment of the
computed factor scores’ indeterminacy should be conducted to determine their
reliability. The validity coefficients represent the correlations between the
theoretical factors and the factor score estimates, and a value above .90 is

desirable (Gorsuch, 1983). The Rf, correlation matrix is derived from the

following expression: R f, = S}..vaS: .Where W.) is the factor score unit-

weight matrix, S’ ,. is the factor structure matrix, and S;,‘ is the diagonal matrix of

factor score standard deviations (Gorsuch, 1983). Validity coefficients for QO,
0‘”, V0 and PW were 0.931, 0.938, 0.9442 and 0.916 respectively. Therefore,

factor indeterminacy was not deemed problematic.

107

Following these validity checks, factor scores were derived for 1998, 1999
and 2000. These scores were subsequently standardized because in later HLM
analysis the meaning of a zero value becomes important. That is, without
standardization, a zero value is meaningless because on the original survey
instrument a zero was not a valid response. With standardization, a zero
represents the average value for a respondent, with below and above average
values having salience within the context of cooperative values, work
performance and the like. The factor correlations for each year are shown in
Table 5-8 and their revealed consistency supports Grice’s (2001) claim to the

stability of the unit weight scoring procedure.

 

Table 5-8: Correlation Matrices of Factor Scores

 

 

1998 1999 2000
0° 0°” v° PW 0° 0W v° PW 0° 0W v° PW
QO 1 .941 .962 .749 1 .938 .963 .753 1 .935 .962 .743

QW .941 1 .961 .815 .938 1 .963 .816 .935 1 .960 .807
V0 .962 .961 1 .790 .963 .963 1 .799 .962 .960 1 .790
PW .749 .815 .790 1 .753 .816 .799 1 .743 .807 .790 1
N 13,689 18,154 31,975

 

 

 

The overall descriptive statistics are shown in Table 5-9 for each factor, for
each year, assessed at the respondent (i.e., workgroup) level. All “other”
organizations have been removed, which explains why the means for each factor
are not zero. The normality assumption that is required within HLM can be
problematic with data derived from Likert-type responses. However this
assumption was supported upon viewing that the skewness and kurtosis values
were within +/-2 and +/-4 respectively. Not only was this true overall but for each
organization also. With the basic assumptions of normality verified, the next step

108

was to aggregate the responses for the organizational-level constructs.
However, an assessment of workgroup consistency within each organization is

needed to validate such a step. This assessment is presented next.

 

 

 

5.5

 

Table 5-9: Overall Descriptive Statistics
1998 1999 2000

N 1 0220 1 0343 27702
Organization-level Quality Management (0")

Mean -0.001 0.027 -0.008

Std. Deviation 1.003 0.997 0.992

Skewness -0.241 -0.260 -0.284

Kurtosis -0.565 -0.528 -0.371
Workgroup-level Quality Management (0“)

Mean -0.003 0.036 -0.005

Std. Deviation 1.003 1.004 0.988

Skewness -0.276 -0.302 -0.31 7

Kurtosis -0.682 -0.658 -0.485
Organization-level Cooperative Values (V0)

Mean 0.005 0.033 -0.005

Std. Deviation 1.007 0.995 0.991

Skewness -0.236 -0.246 -0.259

Kurtosis -0.565 -0.589 -0.432
Workgroup-level Work Performance (PW)

Mean 0.000 0.013 -0.004

Std. Deviation 1.002 1.009 0.996

Skewness -0.521 -0.514 -0.625

Kurtosis -0.446 -0.431 -0. 1 56

 

Data aggregation

 

In order to develop the values for the organization-level factors, data
aggregation is required. However, before averaging the workgroup responses
for the organization-level factors an assessment of workgroup consistency within

each organiZation was required. If the workgroups were not deemed consistent,

109

then aggregation would not be empirically justified. Three indices were used to
assess for consistency: within-group reliability rwga, (James, Demaree, and Wolf,
1984), intraclass correlation (lCC)(1) and a reliability of means index denoted
lCC(2) (Bliese, 2000). An rwcti) estimate was made for each organization. The
values were found to range from 0.798 to 0.939 for Q0 and 0.815 to 0.939 for V0.
This showed a high consistency among workgroups within organizations (James,
Demaree, and Wolf, 1993). Next, the ICC (1) coefficient with unequal sample
sizes were computed based upon ANOVA mean squares. The lCC(1) value
represents the proportion of variance that is accounted for at the organization
level and represents the degree of reliability with a single workgroup’s
assessment of an organization-level factor. The results of 0.023 for 00 and
0.021 for V0 revealed that a relatively small proportion of Q0 and V0 variance
exists between organizations. These values show the importance of gathering
many respondents within each organization. For the reliability of means ICC (2)
calculation high values are preferred. According to James (1982, p. 222), lCC(2)
may be interpreted as follows: “If another sample of 71; individuals were sampled
randomly from each of the same j organizations, then the correlation between the
two sets of means would be approximately equal to lCC(2).” Computing ICC (2)
with unequal sample sizes (Bliese, 2000) resulted in values of 0.867 for 00 and
0.855 for V0. These results justified data aggregation (i.e., averaging) of OD and
V0 for each organization when it was deemed appropriate in the subsequent

HLM process. Table 5-10 depicts the means for Q0 and V0 for each organization

110

over the three years. The mean values for the workgroup-level factors QW and

PW are also shown for informative purposes.

 

Table 5-2: Agency Means per Year

 

 

 

 

 

 

 

 

 

Agency 1998 1999 2000
National N 32
Partnership for 00 0.601
Reinventing 0‘” 0.653 No data
Government v0 0,797
PW 0.381
Department of N 167 189 541
the Air Force 0° 0.089 0.048 0.103
0‘” 0.094 0.067 0.104
v° 0.053 0.050 0.101
PW -0.064 -0.079 0.055
Department of N 218 234 326
the Army 0" 0.014 -0.064 0.060
0‘” 0037 -0.101 0.036
V" -0.036 -0.102 0.023
PW —0.078 -0.119 0096
Department of N 184 193 331
the Navy 0° -0.045 -0.165 0.167
0‘” -0.063 —0.185 0.130
V" 0090 -0.159 0.138
PW 0102 -0.111 0.091
Defense N 243 289 270
Logistics 0° 0.015 0.018 -0.081
Agency 0W -0.017 0.019 -0.098
v° 0.016 0.050 0090
PW 0016 -0.085 0204
Forest Service N 273 274 329
00 0.045 0.056 0034
CW 0027 0.020 -0.066
v° -0.012 -0.018 -0.046
PW 0033 -0.029 0079
Food Safety and N 295 286 5730
inspection 0° -0.134 -0.173 0309
Service 0‘” -0.113 -0.111 -0.242
v° -0.085 -0.137 0.240
PW 0214 -0.215 -0.304

 

111

 

 

Table 5-3: Aggicy Means per Year (continued)

 

Animal and
Plant Health
Inspection
Service

 

Food and
Consumer
Services

 

Nat Oceanic &
Atmospheric
Admin

 

Patent and
Trademark
Office

 

Bureau of the
Census

 

International
Trade
Administration

 

Student
Financial
Assistance

 

 

Office of Post
Secondary
Education

N

E

259%

002
o

S

b§<o

E

ggQZE$0%2

‘D
E

002
o

seezsseezess

b
E

002
o

2

‘6E <6

318 290
0.053 0.028
0.033 0.015
0.048 0.033
0.002 0.002

326 236

-0.115 0.017
-0.077 0.082
-0.067 0.098
-0.087 0.051

341 310
0.110 0.274
0.124 0.347
0.102 0.319
0.139 0.212

203 195
0.008 0.215
0.043 0.197
0.063 0.210
0.059 0.128

342 333
0.181 0.363
0.150 0.315
0.143 0.292
0.183 0.269

301 249
0.055 0.065
0.035 0.067
0.024 0.077
0.123 0.089

No data
312 108
-0.100 -0.115
-0.089 -0.060
-0.060 -0.063
-0.100 -0.088

302
0.049
0.041
0.030

-0.110

193
-0.165
—0.070
-0.079
-0.057

424
0.076
0.122
0.131
0.072

3889
0.089
0.106
0.062
0.171

5458
0.117
0.061
0.082
0.171

288
0.009
0.012
0.032
0.067

363
0.099
0.115
0.142
0.127

33
-0.348
-0.170
-0.360
-0.190

 

112

 

 

Table 5-4: fiency Means per Year Gontinued)

 

Department of
Energy

 

Food and Drug
Administration

 

Health Care
Financing
Administration

 

Administration
for Children
and Families

 

Department of
Housing &
Urban
Development

 

Dept of Interior
Bureau of Land
Management

 

National Park
Service

 

 

Immigration
and
Naturalization
Service

N

00
o

E

8.5%

E

002
0

Es

E

E

‘D
05002 ~°§<00E002 7°$<°0§002 bE<°D 002

2%

E

002
o

76§<b

002
o

E

B
E

364 262
0.071 0.029
0.041 -0.029
0.025 -0.056
0.036 —0.042

296 263
0.067 0.188
0.042 0.149
0.037 0.166
0.008 0.049

308 382

-0.335 -0.105
—0.311 -0.075
-0.330 -0.105
-0.178 -0.073
305 266
-0.108 -0.063
-0.083 -0.019
-0.067 -0.029
-0.054 -0.009
204 357
-0.039 0.076
0.002 0.188
0.082 0.163
-0.020 0.100

No data

358 306
-0.019 0.003
-0.038 0.003
-0.030 -0.037
-0.049 -0.072
212 310
-0.415 -0.326
-0.408 -0.288
-0.441 -0.319
-0.219 -0.201

430
0.095
0.044
0.035

-0.016

468
0.215
0.180
0.184
0.082

319

-0.118
-0.086
-0.113
-0.083

534
0.012
0.002
0.015

-0.037
401
-0.031
0.034
0.021
0.024
289
-0.038
-0.093
-0.123
-0. 182
319
-0.045
-0.088
-0.103
-0.167
204
-0.407
-0.350
-0.406
-0. 159

 

113

 

 

Table 5-5: Aggcy Means per Year (continued)

 

Occupational
Safety 81 Health
Administration

 

Bureau of
Consular Affairs

 

Federal
Aviation
Administration

 

Internal
Revenue
Service

 

US. Customs
Service

 

Financial
Management
Service

 

Veterans Health
Administration

 

 

Veterans
Benefits
Administration

N

00
o

E

*6S 3,

E

DEDOZ V$<O0§002 133330002

‘0S 3)

g%%%22%%%2

002
o

E

2%

E

002
o

2%

314
-0.132
-0.147
-0.111
-0.172

243

0.174
0.176
0.181
0.142

276
-0.288
-0.352
-0.348
-0.269

266
-0.195
-0.151
-0.107
-0.006

260
-0.143
-0.128
-0.140
-0.146

219
-0.026
-0.042

0.000
-0.053

187
-0.042
-0.078
-0.026
-0.030

318
-0.106
-0.064
-0.018
-0.035

864
-0.204
-0.227
-0.210
-0.214

224

0.299
0.282
0.306
0.294

284
-0.226
-0.216
-0.274
-0.072

327

0.002
0.068
0.050
0.033

258
-0.306
-0.282
-0.328
-0.321

226

0.071
0.016
0.071
-0.012

169
-0.140
-0.158
-0.194
-0.123

237
-0.088
-0.028
-0.045

0.011

468
-0.088
-0.104
-0.113
-0.200

260

0.268
0.213
0.276
0.174

359
-0.351
-0.400
-0.397
-0.321

414
-0.051

0.005
-0.020
-0.068

334
-0.177
-O.166
—0.196
-0.205

440

0.180
0.131
0.217
0.129

319
-0.065
-0.148
-0.125
-0.139

440

0.031
0.057
0.087
0.010

 

114

 

 

Table 5-6: Agency Means per Year (continued)

 

Environmental
Protection
Agency

 

Equal

Employment
Opportunity
Commission

 

Federal
Emergency
Management
Agency

 

General Service
Administration

 

National
Aeronautics &
Space
Administration

 

Small Business
Administration

 

Social Security
Administration

 

 

US. Office of
Personnel
Management

N

00
o

259%225.

002
o

‘D ‘D
Egggzsgggzggz

002
o

E

2%

E

002
o

c
E 5<°

002
o

2%

309
0.224
0.218
0.204
0.190

277

-0.177
-0.192
-0.173
-0.011

379

-0.074
-0.042
-0.037
-0.058

257
0.156
0.181
0.170
0.110

385
0.518
0.476
0.474
0.435

358
0.035
0.107
0.121
0.059

273
0.053
0.047
0.025

-0.045

297
0.219
0.184
0.2 12
0.177

308
0.289
0.297
0.281
0.335

235

-0.055
-0.090
-0.062
0.050
300
-0.046
-0.007
-0.020
-0.079

256
0.257
0.311
0.293
0.209

316
0.632
0.571
0.583
0.404

397
0.089
0.182
0.227
0.190

282

—0.086
-0.108
-0.075
-0.146

328
0.306
0.266
0.292
0.259

407
0.143
0.110
0.092
0.102

519

-0.004

0.042
0.020
0.180

346

-0.028

0.012
0.023

-0.041

456
0.305
0.307
0.332
0.204

443
0.576
0.507
0.491
0.317

350
0.015
0.104
0.148
0.136

327

-0.061
-0.076
-0.039
-0.145

379
0.230
0.206
0.252
0.181

 

115

 

The information in Table 5-10 demonstrates the variability in sample size
between organizations and year over year. For example, the Patent and
Trademark Office ranged in sample size from 203 in 1998 to 3,889 in 2000.
There was also variation between organizations as demonstrated in 2000
between the Forest Service (n=329) and the Food Safety and Inspection Service
(n=5,730). Such widely unbalanced data provides anotherjustification for using
HLM because it does not assume sample size equality. Once singular
organization-level factor values were computed for each organization, HLM data

analysis was possible and is presented next.

5.6 Hierarchical Linear Model Results

In order to prepare the data for HLM analysis, two databases were
constructed for each year. The first database is termed level-one data. This
data set included workgroup-level values Qwij and PWI’j , as well as V0,,- responses
for workgroup iin organization j. The second database included the aggregated
organization-level values 00,- and V0]. Both databases were related through the
organization code and all “other” organization codes were removed. Multivariate
data matrices (MDM) were created from these databases in HLM 6.02 software
(Raudenbush, Bryk, Cheong, Congdon, and du Toit, 2004). To assess the
specified models 1, 2 and 3, first an empty model was tested, and then the
hypothesized model was tested. An empty model, also known as a random

effects analysis of variance model, contains only random groups and random

116

variation within groups (Snijders et al., 1999). An empty model is used to create
a base case to which the hypothesized model can be compared. Two r2 values

(one for each level) were computed based upon the change in 0’2 and 12 values

from the empty model (Snijders et al., 1999).

The HLM analysis also included a test for homogeneity of level-1 variance
among the organizations. The assumption of homogenous level-1 variance is
often made in multilevel modeling; however, because of the endogenous V0 and
QW factors some degree of heterogeneity was expected (Raudenbush et al.,
2002). The Bartlett and Kendall (1946) test statistic used by the HLM software
did detected heterogeneous variance among organizations. In order to control
for these, the main effects in each model were specified to explain the log of
level-one error variance. For example, in model 1 the log of c2 was specified to
be predicted by 0". This is the default approach used in HLM and assures the
estimates are consistent with only positive values of 0'2. Besides aiding with
model fit, specifying sources of heterogeneity added insights — this will be
discussed in subsequent sections. Because of this heterogeneity and the
underlying likert-based responses, robust estimation procedures standard in

HLM 6.02 were used for hypothesis testing.

5.6.1 Results for 1998

The results from 1998 are shown in Table 5-11. Model 1 is depicted in the

first two columns with first the “empty” model and then the hypothesized model.

117

The estimated organization-level effect of Q0 on V0. 71V. was 0.930 (p<.001).

Since only a level—2 main effect is considered for model 1, the level-2 :2 is of

 

 

 

 

 

 

 

 

Table 5-7: Results from HLM analysis for NPR - 1998
Mode] 1 Model 2 Model 3
Outcome V0 W PW
Variable ‘7 Q‘ "
Empty £110 1 8 met! E901 415m t M1
Parameters Est. Est. Est. Est. Est. Est.
(std.error) (std.error) (std.error) (std.error) (std.error) (std.error)
[t-value] [t-value] [t—value] [t-valueL [t-value] [t-value]
7‘, 0.007 0.011
0 (0.029) (.006)
[0.245] [1.632]
i" 0.930 ***
H1
7‘ ( ) (.035)
[26.232]
70 -0.003 0.005
0 (.028) (.002)
{-0.119] [1.944]
Q 0.392 ***
H2
7' ( ) (.066)
[5.890]
Q 0.594 ***
, H3
7- ( ) (.074)
[7.982]
}/P -0.004 0.007 (.009)
0" (.023) [0.741]
{-0.188}
P 0.466 *
2’01 (H4) (.192)
[2.427]
P 0.244 (.183)
702 (”5) [1.330]
” ' 0.793 ***
7'0 (H6) (.006)
“i111 [126.543]
Deviance 28959 28837 28874 28745 28926 17046
5-1-mgr) tit 122(3)11 129 (5)11 11880 (7) 11‘
Level 1 R2 0.030 0.030 0.664
Level 2 3’ 0.884 0.886 0.812
* p < .05
* "‘ p < .01
"‘ * * p < .001
i Robust estimation results, 11' Includes heterogenous sigma-sq explanatory variables , Hi" (If is
the difference in no. of parameters, 11'1“] ,6, held constant across organizations — post hoc relaxation
Ewlow reliability (.08) meaning little organizational variation

 

118

interest and was computed to be 0.844. These results provided support for
hypothesis 1, which posited that organization-level quality management practices
would influence the emergence of organization-level cooperative cultural values.

Model 2 is depicted in the third and fourth column, with the estimated

cross-level effects of Q0 and V0 on QW (i.e., 719 and 7? ) being 0.392 (p<.001)
and 0.592 (p<.001) respectively. Again the level-2 r2 was of interest and it was
computed to be 0.866. Therefore it was concluded that hypothesis 2 was
supported, which posited that organization-level quality management practices
would influence workgroup-level quality management practices. In addition,
hypothesis 3, which posited that organization-level cooperative cultural values
would influence workgroup-level quality management practices, was also
supported. It should be noted that the effect of V0 was estimated to be higher

than the effect of 00.

Finally, model 3 is shown in the last two columns. The organization-level

effects of Q0 and V0 on PW (i.e., 7:, and 14:; ) are estimated at 0.466 (p<.05) and

0.244 (not significant), while the workgroup-level effect of QW on PW ( 7(5) is

estimated at 0.793 (p<.001). The level-1 r2 0.664 while the level-2 r2 was 0.812.
These results support the hypothesized effects of both organization-level and
workgroup-level quality management practices (i.e., H4 and H6), but fail to
SUpport the hypothesized effects of organization-level cooperative cultural values
On workgroup-level performance (i.e., H5).

As was stated, variance controls were included for each model. In each
model the logarithm of level-1 variance is predicted by an intercept and the main

119

effects in the model (e.g.,, Iog(62)=ao+aiQO)-). Of which the a1 value is of interest.
A A78 is computed to assess if variance control is warranted. For model 1, 01

had a significant value of -0.346 and a significant 7138 of 17.9, revealing that
higher QO values are associated with less VO error variance. For model 2, Q0
and V0 had insignificant alpha values of 0.192 and -0.469 respectively, but a
significant sz of 11.4, revealing only marginal variance control. For model 3,
00, V0 and QW had alpha values of -0.503, 0.039 and -0.441 respectively, of
which only Qw’s alpha was significant. The ME for model 3 variance control was
significant at 774.7. For all three models the variance controls in the HLM
analyses revealed insights that are developed in Chapter 6.

As was noted in Chapter 4, the level-1 coefficient 81 was held constant
across organizations. This assumed that the effect of workgroup-level quality
management practices was the same across organizations. Organizations likely
implement quality management practices differently, and as such, this
assumption could be questioned. Therefore, to assess this assumption, the error
term Uj in (7) was unconstrained and the model was reanalyzed. A reliability
estimate (Raudenbush et al., 2004) of 0.08 was computed indicating nearly no
organizational variation in the effect of 0'” on PW. This inferred a high
consistency in the effectiveness of workgroup-level quality management

practices.

120

5.6.2 Results of 1999

The results from 1999 are shown in Table 5-12. The estimated

organization-level effect 71V is 0.975 (p<.001), with a level-2 :2 of 0.919. Both of

these are increases from 1998 and provide further support for hypothesis 1.

Model 2 had estimated cross-level effects y? and y? to be 0.166 (insignificant)

and 0.802 (p<.001) respectively, with a level-2 r2 of 0.912. Interestingly, H2 was

not supported for 1999 (different than 1998) but H3 was. Finally, model 3’s
organization-level effects 1(5), and y; were estimated at 0.188 (insignificant) and
0.596 (p<.001), showing a reversal in hypothesis support from 1998 — that is H5
is accepted for 1999 while H4 is not. The workgroup-level effect 71’; is estimated

at 0.803 (p<.001), thereby further supporting H6. In this final model, the level-1 r2
was 0.670 while the level-2 r2 was 0.866.

In regard to variance control, model 1’s alpha had a significant value of
-0.321 and a significant sz of 22.1, again revealing that higher Q0 values are
associated with less VO error variance. The model 2 variance control had similar
results to 1998 with Q0 and V0 having insignificant alpha values of -0.243 and
-0.023 respectively, and a small but significant sz of 15.4. For model 3, QO, V0
and QW had alpha values of 0.155 (insignificant), -0.698 (p<.05) and -0.433
(p<.001) respectively. These results are different than 1998 in that V0 had
become significant and negatively related to PW error variance. The sz for
model 3 variance control was significant at 770.0. The post-hoc test for [31

reliability was 0.24, which is an increase from 1998 but still a low value.

121

 

 

 

 

 

 

Table 5-8: Results from HLM analysis for NPR - 1999
Model 1 Model 2 Model 3
Outcome V0 W w
Variable ’7 Q" P"
_p_yEm t M i _MEm t M 1 mm M 1
Parameters Est. Est. Est. Est. Est. Est. (std.error)
(std.error) (std.error) (std.error) (std.error) (std.error) [t-value]
[t-value] [t-value] [t-value] [t-value] [t-value] '
y!" 0.037 0.037 ***
0 (.033) (.008)
[1.112] [4.454]
1’ 0.975 ***
H 1
7' ( ) (.031)
[30.502]
70 0.041 0.040 ***
0 (.032) (.004)
[1.288] [8.744]
Q 0.166
H2
7' ( ) (.082)
[2.007]
Q 0.802 ***
, H3
7_ ( ) (.089)
[8.958]
7P 0.017 0.021 * (.010)
"0 (.028) [2.112]
[0.615]
P 0.188 (.136)
4
70‘ (H ) [1.381]
P 0.596 ***
702 (H5) (.145)
[4.112]
P 0.803 ***
H6
7'0 ( ) (.006)
1111 [124.666]
Deviance 28916 28788 29123 28987 29327 17201
dX‘ia‘f} iii 128(3) ii 136(5) ii 12126(7)11
Level 1 R2 0.039 0.037 0.670
Level 2 R2 0.919 0.912 0.866
* p < .05
* "‘ p < .01
* ** p < .001
"l“ Robust estimation results, it Includes heterogenous sigma-sq explanatory variables , ‘H‘T df is
the difference in no. of parameters, 111“] ,6, held constant across organizations — post hoc relaxation
fids low reliability Q4) meaning little organizational variation

 

 

122

 

 

Table 5-9: Results from HLM analysis for NPR - 2000

 

 

 

 

Model 1 Model 2 Model 3
Outcome 0 W W
Variable Vi" Q‘ R’
_E_m. Mt MT 411301 t MT .1215”! t 111.101
Parameters Est. Est. ‘ Est. Est. Est. Est.
(std.error) (std.error) (std.error) (std.error) (std.error) (std.error)
[t-value] [t—value] [t-value] [t-value] [t-value] [t-value]
71' 0.018 0.014
0 (.029) (.007)
[0.624] [1.897]
V 0.888 ***
H1
7' ( ) (.044)
[20.133]
70 0.017 0.012
0 (.027) (.007)
[0.643] [1.504]
Q 0221 "'
H2 ’
7' ( ) (.106)
[2.072]
Q 0.715 ***
, H3
7" ( ) (.111)
[6.399]
7P -0.009 -0.004
00 (.025) (.015)
{-0.373] {-0.265]
P -0 044
H4 '
701 ( ) (.166)
{-0.269]
P 0.761 ***
H5
702 ( ) (.154)
[4.938]
P 0.769 ***
H6
7’10 ( ) (.013)
“lift [58.317]
Deviance 77420 77317 77318 77198 77413 46779
aflfifj 1’11 103 (3) 11 120(5) 11 30634 (7)
11
Level 1 R2 .031 0.027 0.644
Level 2 R2 .943 0.945 0.765
* p < .05
** p < .01
*** p < .001

T Robust estimation results, 11' Includes heterogenous sigma-sq explanatory variables , HT (1 ‘is
the difference in no. of parameters, 111‘] ,8, held constant across organizations —— post hoc relaxation
lields moderate reliability (.49) showing some true organizational variation

 

123

 

5.6.3 Results of 2000

The results from 2000 are shown in Table 5-13. The estimated

organization-level effect 71V is 0.888 (p<.001), which is a decrease from 1999 but

with an increased level-2 I2 of 0.943. Therefore for 2000 H1 remains supported.

Model 2 had estimated an increased 719 cross-level effect of 0.221 (p<.05), but a

decreased 30 of 0.715 (p<.001). The level-2 r2 increased to 0.945, therefore

both H2 and H3 are supported for 2000 (similar to 1998). Finally, model 3’s
organization-level effects 7:, and 7,]: were estimated at -0.044 (insignificant) and
0.761 (p<.001), furthering the trend started in 1999 that did not support H4 but
did support H5. The workgroup-level effect 71’; decreased from 1999 to 0.769

(p<.001) but still supported H6. Model 3 had a computed level-1 :2 of 0.644 and
a level-2 r2 of 0.866 — both a decrease from 1999.

Regarding the year 2000 variance control, for the first time model 1’s alpha
was not significant with an estimate of just -0.045 and a non-significant sz of
1.1, which revealed that higher 00 values no longer lessen V0 error variance.
The model 2 variance control was similarly unsuccessful in predicting error
variance with 00 and V0 having insignificant alpha values of -0.026 and] 0.078
respectively, and an insignificant 7178 of 0.9. However model 3’s variance control
was slightly more successful. The main effects Q0, V0 and QW had alpha values
of -0.374 (insignificant), -0.210 (insignificant) and -0.444 (p<.001) respectively.
These results show a diminished ability of the organization-level effects to predict

PW error variance. The A78 for model 3 variance control was much higher than

124

both 1998 and 1999 at 2195.5, revealing this final control to be the most
successful. The post-hoc test for 2000 B1 reliability was 0.49, which continues
the increasing trend from 1999 and shows that organizations were differentiating
on the effect of QW on PW. Reviewing the results from each year reveals
substantial support for the proposed multilevel model. However, as was noted
earlier, the sociotechnical effects of quality management practices are dynamic.

Therefore, year-over-year comparisons were made and are presented next.

5.7 Parameter stability

The effects of a quality initiative were expected to change over time. Not
only is this caused by the diffusion process noted in the literature (Ahire et al.,
2001), but also because of the organizational and environmental dynamics
associated with quality management (Narasimhan et al., 2001). Moreover, the
sociotechnical effects of introducing a new management technology such as
quality management should effect changes in the theorized model. This is
particularly expected in the case of cooperative cultural values. Being able to
detect significant changes in the strength of relationships over time will add
insight into the process of quality initiatives and, more generally, of sociotechnical
systems.

The assessment of parameter stability was accomplished through year-
over—year t-tests of coefficient equivalence. Because no unique respondent
information was available and the surveys were randomly administered, each

year was treated as an independent sample. Also, because it was discovered

125

that error variances were susceptible to main effect changes (see section 5.6)
and that organizational main effect values shifted year over year (see Table 5-10)
it was assumed that variances were not equal between years. Because of these
assumptions of independence and unequal variances, a one-tailed Welch t-test
based comparison was made between years (Welch, 1947). Equation 8 shows
the test statistic calculation while equation 9 shows the computed degrees of
freedom that is rounded down to give a bias to the null hypothesis (Aczel and
Sounderpandian, 2005). A 0.10 threshold was chosen for significance because

of the relatively small number of organizations.

k k
0y+1—6(J’l)2

(while:

[8; 0i] 2
fly+1ia/(”y+1‘1)+(ek)4/("y'1)

 

(8)

 

(9)

 

df =

 

 

Where, 6;; represents the coefficient k in year y and ef, is the standard error of
195‘, which is determined by 67y N};

The results of the test between 1998 and 1999 are shown in Table 5-14
and the results between 1999 and 2000 are presented in 5-15. With respect to
H7, it was expected that the effects of Q0 would negatively change from 1998 to

1999 and from 1999 to 2000. As can be seen only one significant change was

observed from 1998 to 1999 — that is Ar? was -0.226, which was in the direction

126

 

expected. Subsequently, from 1999 to 2000 only one significant change

occurred in the effect of Q0 — that is Ay,” was -0.087, which again was in the

direction expected. Interestingly, the general trend of the other changes was

also negative, albeit only significant when 1998 was compared to 2000, such as

with 7;. Therefore, it was concluded that H7 was supported by the multiyear

results.

 

Table 5-10: Structural change from 1998 to 1999

 

 

 

 

 

 

 

_ 1998 1999
Number of level 1 units 10220 10343
Number of level 2 units 37 36
Estimate Std .Err Estimate Std.Err A t-value df
H7 71" (Q0 __, V0) 0.930 0.035 0.975 0.031 0.045 0.96 70
719 (Qo _, QW) 0.392 0.066 0.166 0.082 -0.226 ** 2.15 67
761 (00 _, PW) 0.466 0.192 0.188 0.136 -0.278 1.18 64
H8 720 (V0 __, QW) 0.594 0.074 0.802 0.089 0.208 ** 1.80 68
71:2 (V0—+ PW) 0.244 0.183 0.596 0.145 0.352 1.51 67
H9 711) (QW—e PW) 0.793 0.006 0.803 0.006 0.010 1.18 70
*p<0.10, ** p<0.05
Table 5-11: Structural chagge from 1999 to 2000
1999 2000
Number of level 1 units 10343 27702
Number of level 2 units 36 38
Estimate Std.Err Estimate Std.Err A t-value Df
H7 7:! (00 _+ V0) 0.975 0.031 0.888 0.044 -0.087 1.62 65
7'9 (QO 6 QW) 0.166 0.082 0.221 0.106 0.055 0.41 68
7:] (Qo _’ PW) 0.188 0.136 -0.044 0.166 -0.232 1.08 70
H8 7? (V0 _+ QW) 0.802 0.089 0.715 0.111 -0.087 0.61 69
7:2 (VG—i PW) 0.596 0.145 0.761 0.154 0.165 0.78 71
H9 )3: (QW—s PW) 0.803 0.006 0.769 0.013 -0.034 ** 2.37 51

 

 

* p<0.10, ** p<0.05

 

127

 

 

With respect to H8 regarding the effects of V0, it was expected that

negative changes would be observed between 1998 and 1999 and between

1999 and 2000. As can be seen in Table 5-14, A729 and A7,]; were 0.208 (p<.05)
and 0.352 (p<.10), which was unexpected. In Table 5-15 it is shown that A73

and Ayg’2 were insignificant, which also was unexpected. Therefore, it was

concluded that H8 was not supported because the direction of change was in the
opposite expected direction. The implications of this finding are detailed more
extensively in Chapter 6, but the results implied that the causal mechanisms
differ substantially between organization-level quality management practices and
cooperative cultural values.

Finally, regarding the effects of Q”, it was also expected that positive

changes would be observed between 1998 and 1999 and between 1999 and
2000. Table 5-14 shows that 3):; was insignificant between 1998 and 1999, but

was an unexpected negative value (i.e., -0.034) between 1999 and 2000. These
results give no support for H9, meaning that although the strength of workgroup-
Ievel quality management practices remained high, they changed very little

throughout the years of the initiative.

5.8 Summary of Data Analysis

This chapter reviewed the data preparation and factor analysis processes

necessary to find support for the proposed research framework. Results were

128

presented that validated the measurement model and factor scoring procedure
utilized to construct an appropriate data set from the secondary data source.
The results of the HLM analysis were presented in which general support for the
research framework was found. In addition, the stability of parameter estimates
were assessed, which will provide an understanding of the dynamics of quality
management, cooperative values and workgroup performance. These results
make epistemic and practical knowledge contributions, which are discussed in

the next chapter.

129

CHAPTER 6: DISCUSSION

In this chapter, both the epistemic and practical insights gained from this
research are discussed. The first two sections will review the contributions to
quality management knowledge and sociotechnical systems theory. The third
section discusses managerial insights pertaining to the relationships among
quality management practices, cooperative cultural values and work
performance. Contributions made by this dissertation are highlighted throughout

each section.

6.1 Knowledge of Quality Management

This subsection reviews the insights to quality management knowledge.
The first subsection will discuss the insights drawn from the single 1998 period,
and the second subsection will discuss additions to knowledge based upon the

year-over-year results.

6.1.1 Single Period Results (1998)

One of the fundamental premises of this dissertation was the conjecture
that a difference exists between quality management practices at the
organization level and the workgroup level. This was based in part on Juran’s
(1989) insight regarding the “big Q” and “little Q" differences in quality
management. Through use of established concepts in the quality management

130

literature (Anderson et al., 1994; Douglas et al., 2004), discrimination was found
in the 1998 archival data between organization-level and workgroup-level quality
management practices. Moreover, these two discriminant concepts had different
relationships with workgroup-level performance and cooperative cultural values,
a concept known to be concomitant with quality management (Detert et al.,
2003). The epistemic implication of this empirical confirmation of Juran’s insight
is that recognition of the multilevel nature of quality management practices
should exist. That is, much of past research has treated quality management as
a single level concept (Rungtusanatham et al., 1998; Zbaracki, 1998). Doing so
could ignore important phenomena, such as the antecedent role organization-
Ievel quality management practices have to cooperative cultural values.
Moreover, recognizing workgroup-level quality management practice as
significantly influenced by both technical and social aspects of an organization
will aid in understand variations in quality management implementation.
Therefore, an important contribution of this dissertation is the empirical
discrimination of the different levels of quality management practices in an
organization.

Another issue underpinning this dissertation was the problem of explaining
the relationship between quality management and cooperative values. Based
upon Detert et al.’s (2000) research, organization-level cooperative cultural
values were measured using the 1998 archival data and related to the different
levels of quality management practices. Empirical evidence was found for

cooperative cultural values to be a consequence of organization-level quality

131

management practices, but also an antecedent to workgroup-level quality
management practices. Such an interrelationship explains why these cultural
values are deemed essential to a quality initiative (Detert et al., 2000; Kujala et
al., 2004). Moreover, the results imply organization-level quality management
practices reduce the variability of cooperative cultural values. As such, these
findings imply that more extensive quality management initiatives stabilize the
level of cooperation in organizations. Such a discovery may not have been
possible without the different relationships proposed in this research. Therefore,
another important contribution of this dissertation is the empirical support for a
richer conceptualization of how quality management relates to cooperative
values.

An important component of a quality management initiative is its
implementation at the workgroup level (Naveh et al., 2004; Soltani et al., 2004).
This research conjectured that both organization-level quality management
practices and cooperative cultural values should influence the occurrence of
workgroup-level quality management practices. Evidence was found in the 1998
archival data that both organization-level concepts had the conjectured influence.
However, the cross-level influence of cooperative values on workgroup-level
quality management practices was significantly higher than the influence of
organization-level quality management practices. This finding implies that
informal processes are more influential at the workgroup level than formal
processes. That is, the cooperative norms of behavior that likely emerge from an

organization valuing cooperation become a strong enabler of workgroup-level

132

quality management practices. This does not remove the importance of
organization-level quality management practices, as these were found to be a
strong influence on cooperative values. This mediating role of cooperative
values can be likened to Argyrisand Shon’s (1996) conception of double-loop
learning, where only after organizational members change their mental models
will the full effect of new technologies occur. The elucidation of the differential
effects of the two organization-level concepts is another contribution of this

dissertation.

6.1.2 Multi-period Results (1998-2000)

The literature on quality management has emphasized the importance of
time in quality management initiatives (Narasimhan et al., 2001; Schroeder et al.,
2005). This is true not only for the time to implement the practices but also for
the effects of these practices to manifest. Two hypotheses related to the
changing effects of quality management practices were given, of which only one
was empirically supported -— hypothesis 7. This hypothesis posited that the
effects of organization-level quality management practices come from changes to
organizational characteristics and structures. As these organizational aspects
are changed, the potential for further influence diminishes and therefore the
strength of effect should decrease. This hypothesis was supported by the
comparison of year-over—year results. The strength of each relationship
diminishedwithin either one or two years. Such a finding implies that the

immediate effects of organization-level quality management practices may not be

133

sustainable. In other words, practices like cross-functional teams and
communicating customer needs have a more immediate than long-term impact.
The long-term effects may emerge more from sustaining the cooperative values
and workgroup-level practices than from organization-level quality management
practices. It should be noted that diminished correlation between organization-
level and workgroup-level quality management practices further supported the
discrimination of these two constructs.

The other hypothesis relating to the changing effects of quality
management was hypothesis 9, which posited an increasing impact of
workgroup-level quality management practices. This hypothesis was based upon
the idea that as more workgroups in an organization learn and adopt quality
management practices, a stronger relationship with workgroup-level performance
should emerge. This conjecture was not supported by the data analysis. An
explanation for this is that the other significant predictor of work performance
(i.e., cooperative values) began to differentiate high and low performing
workgroups better than workgroup-level quality management practices. This
would imply that informal processes (e.g.,, cooperative management-union
relations and valuing differences among individuals) have a cross-level impact
equivalent to the workgroup-level effects of job recognition and information
access. Another explanation for the lack of empirical support for H9 comes from
the recognition that although the strength of workgroup-level quality management
practices decreased, the change was very small. This implies that once quality

management practices are implemented at the workgroup level, the effects

134

remain relatively stable over time. Although more investigation is needed, this
dissertation has made insights into the effects of quality management through the
multiyear analysis used.

This research also posited that the effects of organization-level
cooperative cultural values would diminish over time. Similar to hypothesis 7,
hypothesis 8 was based on a conjecture that once organizational changes were
made, the potential for more change diminishes. This hypothesis, however, was
not supported, and, in fact, the strength of cooperative cultural values increased
over time. This could be explained two ways. First, it is likely that the amount of
organizational structures influenced by organizational values is much greater
than the amount of structures influenced by management practices. That is,
although quality management practices influence how problems are solved and
direction is communicated, it may not influence the hiring practices, interpersonal
relations and bargaining agreements that cooperative values may influence.
These other processes are part of the routines of an organization (Feldman and
Pentland, 2003; Ray, Barney, and Muhanna, 2004). Organizational values likely
have a much stronger impact on these than quality management practices. This
would imply that the amount of change possible caused by organizational value
change is much larger than expected.

A second explanation for why organization-level cooperative values
increased in effect year over year may be in the characterization of
organizational structures as barriers. That is, it may be that quality management

removes those structures that are barriers to organizational effectiveness while

135

cooperative values add structures that are enablers to effectiveness. This
differentiation is most evident in the changing influence on workgroup-level
quality management practices. While organization-level quality management
decreased, organization-level cooperative values increased. If workgroup-level
practices like rewarding creativity and innovation are indicative of other
effectiveness enablers, then it follows that a sustained presence of cooperative
values may further differentiate high and low performing workgroups. Although
more research is needed to support these conjectures, the use of multi-year data
to explore the dynamics of the hypothesized model was critical to discovery. Not
only did this research make epistemic contributions to quality management, but
also towards sociotechnical systems theory, which is presented in the next

secfion.

6.2 Knowledge of Sociotechnical System

This section will discuss the added epistemic insights to sociotechnical
system theory resulting from this dissertation. The first subsection will discuss
the insights drawn from the single 1998 period, and the second subsection will

discuss additions to knowledge based upon the year-over—year results.

6.2.1 Single Period Results (1998)

The theoretical grounding for much of this dissertation has been based

upon sociOtechnical systems (STS) theory (Emery, 1959; Pasmore, 1988;

136

Pasmore et al., 1982; Seiler, 1967; Trist et al., 1951). This theory was utilized to
explain how quality management and cooperative values were related from a
meso—paradigm perspective and a dynamic perspective. Quality management
practices were considered to be part of the technological systems used by the
different organizational levels for environmental adaptation. Cooperative cultural
values were considered to be part of the social systems in an organization
functioning to satisfy the human needs for stability and priority. As STS theory
views organizations as sociotechnical systems, a reciprocal relationship was
expected between quality management practices and cooperative cultural
values.

The first period results from this research provide support for the
applicability of STS theory to the study of quality management practices and
cooperative cultural values. Significant relationships existed among
organization-level quality management practices, organization-level cooperative
values and workgroup-level quality management practices. This support for the
applicability of STS theory to quality management may be useful for future
examinations into the antecedents and consequences of this management
practice. For instance, the growth of Six Sigma as an important quality
management tool could have an impact on the social values of an organization,
such as attention to detail or a reluctance to take uninformed action. Moreover,
STS theory has predominantly been applied to physical manufacturing
technologies such as work cells and mass customization (Huber et al., 1991; Liu

et al., 2006). However, this research found support for STS theory in a nonprofit

137

service context. As such, this dissertation could provide the justification to seek
applicability of STS theory in other service setting with other management
technologies, such as with warehouse management systems (VVMS) within a
logistics context.

A few empirical results, however, were not accounted for with STS theory
and may be areas where the theory needs further development. One issue was
in the lack of support in the first period for the relationship between cooperative
cultural values and workgroup-level performance. STS theory implied that
cooperative values should induce an organizational structure conducive to
workgroup effectiveness, but because this was not supported, the theory should
be questioned. It may be that cooperative cultural values do not change the
structure as expected, or that these changes take longer than can be detected in
a single period. These issues, along with other insights from the multiple period

data analysis are developed in the next subsection.

6.2.2 Multi-period Results (1998-2000)

The insights for STS theory were enhanced from fitting the theoretical
model across consecutive time periods. For instance, the lack of support in 1998
for hypothesis 5 relating cooperative values to workgroup-level performance was
changed in 1999 and 2000 to full support. Such a result not only provides more
confidence in STS theory, it also gives insight into the delayed influence of social
values. It should be noted that each STS theory-based hypothesis was

supported in at least one of the years, and that three (i.e., H1, H3 and H6) were

138

supported in all the years. Overall these results demonstrate the potential for
STS theory in future studies.

However, STS theory was unsuccessful in predicting the year-over-year
effects. This was evident in thelack of support for two of the three diffusion
hypotheses. The STS literature has observed the biased preoccupation STS
theorist have toward understanding the effects of technology on social systems
(Fox, 1995). This bias can be explained by the field’s motivation to find means to
improve quality of work life (QWL) for employees (Griffith and Dougherty, 2001).
Such a focus has served STS theory well with understanding the consequences
of technical system change, but not in the consequences of social system
change. As such, the lack of empirical support for most of the diffusion
hypotheses may motivate future research into the dynamics of STS theory. This
section has presented the epistemic contributions of this research, but practical

insights have been made as well and these are presented next.

6.3 Managerial insights

In this section, suggestions for managers on how to use these research
findings are presented. Insights are drawn from the single 1998 period results,
where the multilevel nature of quality management was validated. This is
followed by practical ideas gained from analyzing the multiple years of archival

data and comparing year-over-year results.

139

6.3.1 Insights from a Single Period (1998)

The operations management literature has found that quality management
is beneficial for both manufacturing and service operations (Douglas et al., 2004;
Kathuria et al., 2001; Zhao et al., 2004). This dissertation, therefore, expected to
find support for quality management influencing work performance. The
uniqueness of this dissertation, however, is that quality management practices at
two distinct organizational levels were tested in a nonprofit service context.
Finding support for quality management in this context would broaden the
applicability of such a management approach. Analyzing archival data from 1998,
support was found for both organization-level and workgroup-level quality
management practices influencing workgroup-level performance. This result
supports the use of both organization-level practices (e.g.,, cross-functional
teams, customer feedback systems, and vision communication) and workgroup-
level practices (e.g.,, innovation rewards, employee involvement, and information
access) to improve workgroup-level quality and productivity. Moreover, this
result supports quality management even in a nonprofit environment where the
“voice of the customer” may not be as strong.

Another important point is the high strength and significance of workgroup-
level quality management practices. This research did not have an expectation
as to which level of quality management would affect workgroup performance
more. Because much of the literature on quality management extols the
importance of top management support (Brah, Wong, and Rao, 2000; Flynn et

al., 1995; NaVeh et al., 2004), an argument could be made for organization-level

140

quality management practices to be most influential for work performance.
However, the 1998 archival data show that workgroup—level practices were more
influential. This result underscores the necessity for managers to implement
change at the lower levels of the organization and to not be satisfied with
organization-level change. In fact, subsection 6.3.2 reveals that the work
performance impact of organization-level practices diminishes completely by the
year 2000. This result provides further support for the importance of
distinguishing and emphasizing both “big Q" and “little Q” practices in a quality
initiative.

It should be noted that organization-level performance was not a construct
measured in this dissertation. Because of this, the importance of organization-
Ievel quality management practice should not be diminished. Moreover, because
of the cascading influence that organization-level quality management practices
have throughout the organization (i.e., inducing cooperative cultural values and
workgroup-level quality management practices), its work performance impact
may still be strong but indirect. For these reasons, organization-level practices
are likely still necessary for the successful quality initiatives but not sufficient.

A crucial decision that managers must make when embarking on a quality
initiative is where to allocate resources. Should managers devote personnel time
to training in quality practices, or should an organization-wide communication
campaign begin the quality initiative? A common theme within the quality
management literature is the importance of having a “quality” organizational

culture prior to implementing quality management practices (Detert et al., 2000;

141

Kujala et al., 2004). This dissertation measured a significant portion of what
characterizes a “quality” culture through the cooperative cultural values construct
(V0). This dissertation also found support for the hypothesis that organization-
level quality management practices are antecedent to cooperative values. A
conclusion could be made that resources should be devoted to organization-wide
initiatives first and then the right “quality” culture will follow. Such a conclusion is
premature. There are well known problems with definitive inferences regarding
the direction of causality with cross-sectional data (Mitchell and James, 2001).
Moreover, the theoretical model is by no means deterministic as Table 5-13
shows. An insight from this dissertation is the possibility that a simultaneous
approach may be the most beneficial. That is, rather than embarking solely on
an organization-wide culture change initiative that could have uncertain
consequences (Harris et al., 2002), this research implies implementation of
organization-level quality management practices simultaneously complements

and aids the culture change process.

6.3.2 Multi-period Results (98-00)

An important question faced by managers is the stability of effects that can
be expected throughout the years of a quality initiative. That is, the impact of
organization-level and workgroup-level quality management practices may
diminish from their initial effects. If this is so, managers must be aware of
changes that may ensue. This is one of the motivations for the longitudinal

design employed by this dissertation.

142

A review of the year-over-year comparisons made in Tables 5-16 and 5-17
revealed two relationships that appeared relatively unchanged from 1998 to
2000. These relationships were between 1) organization-level quality
management practices (0") and organization-level cooperative cultural values
(V0); and 2) workgroup-level quality management (QW) practices and workgroup-
level performance (PW). It should be noted that these were the only same-level
relationships posited in the theoretical model. The relative stability of the QO-VO
relationship implies that these two concepts are tightly coupled. In addition, the
managerial insight is that a sustained effort at implementation of organization-
level quality management practices should have a marked effect on the values of
the organization. The other stable relationship (i.e., between QW and PW) has
quite similar implications and should lend support to managerial convictions that
workgroup-level quality management practices will make a difference.

The unexpected result found in the rising influence of organization-level
cooperative cultural values should be brought to the attention of managers who
are narrowly focused on the technical aspects of a quality initiative. From the
initial non-significant relationship between cooperative values and work
performance in 1998 to highly influential impact in 2000, the growing influence of
this organization-level construct is remarkable. The managerial insight is that
highly performing workgroups will be differentiated based in part by the level at
which that workgroup’s organization values cooperation. The expectation was
that organizations with both formal QO characteristics and informal VO

characteristics will aid in workgroup-level performance. What is instead implied

143

is that the formal characteristics have only short-ten'n workgroup benefits and
that the truly long-term gains are realized by the more informal characteristics.
Therefore, attention to organization-level cooperative cultural values and their ilk
is likely necessary.

A final unstable relationship that managers should take note of is the
decreasing impact of organization-level quality management practices on
workgroup-level performance. In a general sense, while organization-level
cooperative values rose in prominence, the organization-level quality
management practice declined. It should be noted that a decline in influence
was expected because of the diminishing number of potential removable barriers
to work performance. However, an important causal chain from organization-
Ievel quality management practices to workgroup-level performance remained in
effect throughout the quality initiative. That is the following process:

QO—+ VO—>QW—>PW. In each year this connection remained strong and in place.
The managerial insight is that perhaps this should be the mental model of why
organization-level quality management practices matter to an organization. That
is, although a direct QO-PW relationship is temporary, a stable indirect relationship

should be maintained in order to reap the full benefits of a quality initiative.

6.4 Summary of Discussion

This chapter discussed how the findings from this research have made
substantial contributions to study of quality management and its interaction with

organizational values and workgroup performance. The validated multilevel

144

perspective of quality management practices helped explain a theoretical
paradox unresolved in the current literature. Moreover, the longitudinal analysis,
which is uncommon in the quality management literature, helped develop insights
into the sociotechnical changes during a quality initiative. In addition, managerial
suggestions were provided regarding why to take a simultaneous approach
towards instituting both cooperative cultural values and quality management
practices. The next chapter will draw final conclusions regarding this dissertation
research, present insights from quality managers regarding the results, discuss

research limitations and propose directions for future research.

145

CHAPTER 7: CONCLUSION

In this chapter, a summary of this dissertation is given that highlights the
background, propositions, methods, findings and implications of this research.
Following this summary is a review of the limitations of this study and potential

avenues for future research.

7.1 Summary of Research

The lack of success of many quality management initiatives has perplexed
researchers and managers alike (Choi et al., 1998). Extant literature has
examined the various possible reasons for this lack of success (Schroeder et al.,
2005). One prominent finding has been the association of quality management
practices with cooperative cultural values (Detert et al., 2003; Detert et al., 2000).
However, the question of how these values are associated with quality
management has been unresolved; researches have adopted either an
antecedent or consequent position but have not incorporated both. This
research proposed a meso-paradigm model that sought resolution for the
questions: what role do cooperative cultural values play in a quality initiative and
what changes should be expected over time?

In order to develop the meso-paradigm model, literatures from different
disciplines were required (see Figure 2-1). First, the quality management

literature was utilized to distinguish between the multilevel and dynamic nature of

146

quality management practices (Juran, 1989; Narasimhan et al., 2001). Next the
organization theory literature was examined to identify the types of cultural
values that associate with quality management, as well as how these values may
change over time (van Woerkom, 2004; Wagner, 1995). A review of empirical
studies that investigated similar issues was conducted to determine where further
research was required. Finally, a sociotechnical system theory-based set of
single and multi-period hypotheses (see Figure 3-1) were proposed to seek
resolution for the apparent epistemic paradox.

To find support for posited research framework the use of archival data
was required and this was accomplished from a publicly available data source. A
multiyear employee survey administered within the federal government during a
portion of its seven year quality initiative was utilized (O.P.M., 2002). A
measurement model based on extant literature in quality management and
organization theory was tested and supported by the archival data. To avoid
threats to validity, extensive measures were taken to maximize the use of
information, control for underlying biases, and check for measurement adequacy.
The multilevel model was then tested over three consecutive years and
conclusions were drawn regarding the six single period hypotheses. In addition,
year-over-year comparisons were made in order to find support for the three
diffusion hypotheses.

The findings from this data analysis were original and valuable. First, it
was confirmed that an empirical and causal discrimination exists between

organization-level and workgroup-level quality management practice. The

147

implication was that researchers and managers should account for this difference
when considering a quality initiative. Second, support was found for the posited
relationship between cooperative cultural values and quality management
practices. This result presented a possible resolution to the existing paradox in
the literature and gave insight to managers for where to allocate resources in a
quality initiative. Third, the multi-period benefits of quality management and
cooperative cultural values in a governmental context were insightful. Such a
finding supports the claim that quality management, and more generally
operations management, has applicability in nonprofit service situations. Finally,
the unexpected rise in prominence of cooperative cultural values during the
quality initiative was remarkable. This finding motivated a reexamination of some
STS theory premises and also called for managers to assure attention is given to

cooperative values in a quality initiative.

7.2 Reactions from Managers

An important component of this research was the presentation of the
findings to quality managers from several manufacturing facilities that had
recently undergone quality initiatives. Three plant-level managers and one
corporate-level manager were interviewed. Their titles are listed in the appendix
(A3). The managers were forwarded summary material in advance and then
interviewed collectively via teleconference. This conference was recorded with

the managers’ consent and later transcribed for analysis. The managers

148

expressed agreement with some of the major findings but disagreement with
others. In addition, the managers suggested areas of future research.

The quality managers unanimously agreed that cooperative cultural
values should be highly influential in determining how permanent newly
implemented quality management practices would be. One plant-level manager
mentioned that “if (quality management) becomes part of the value system, then
it’ll maintain... managers will no longer need to be involved.” The corporate-level
manager interpreted the results to mean that “values replace practices because
we believe in it, not because we say it.” This same manager commented that he
has observed the following: “a culture develops that won’t let (quality
management) die. If a leader changed the quality push, the people would revoltl”
These comments show that one of the findings from this research resonates with
quality managers.

There was also concurrence with the STS theory claim that cultural values
can be influenced by management practice implementation. A plant-level
manager told a story about multiple plants within his business unit;

“Each had its own identity (and would not cooperate with each other). To

counter act this, we changed the profit and loss reports to be at the

business unit level, not the plant. This had a huge change, plants stopped
being so individualistic. Another way was through process improvements

— evaluating best practices at each plant and then duplicating efforts in

other plants."

These examples show how management practices can influence cultural values.
The corporate—level manager summarized the process as follows: “as groups

install quality practices, and once things get to some level of success, people

transform into being believers.”

149

Some lack of support was given by two plant-level managers as to the
discrimination between organization-level quality management practices and
organization-level cooperative values. "T here isn’t a distinction,” was there
primary comment. This shows how tightly coupled cooperative cultural values
can be with quality management. In addition, one plant-level manager could not
understand why organization-level quality practices diminished in their influence
on workgroup-level practices. “I’d think the effects of organization-level quality
would be increasing over time. Unless (organization-level quality) is continually
reinforced, then workgroup-level performance dies.” The effect this manager
described could still occur through the QO—> VO—>QW—>PW process discussed in
section 6.3.2, but at least one plant-level manager firmly believed there was a
direct effect.

The discussion of the research findings with quality managers provided
richer insights into the multilevel relationships examined herein. Moreover, they
suggested future research should account for the amount of leadership
involvement and measure the depth of deployment of practices. As one
manager put it, “it takes a while before the teams buy in, but with management

support, they really start believing and doing things we wouldn’t have thought of.”

7.3 Limitations of study

There are some important limitations of this dissertation. The lack of
detailed information on respondent characteristics prevented some issues to be

examined. That is, a contrast could not be made between managers and non-

150

managers, nor between unionized and non-unionized employees. In addition,
context-specific factors such as whether the workgroup served a front-line or
support function were not accessible. Moreover, the tracking of unique
workgroups within each organization from year to year prevented a more detailed
dynamic analysis. Each of the above queries was prevented because of data
inaccessibility and would have added more clarity to the research.

Another limitation to this research was the use of perceptual measures for
the quality management and work performance constructs. This is a known
issue for much survey research and the potential problems have been well
documented elsewhere (Ketokivi and Schroeder, 2004). As such, the findings of
this dissertation should be understood within the context of those limitations. It
should be noted that direct measures of observable quality management
practices at the organization and workgroup level were not available for
consideration in the archival data. Moreover, objective measures of workgroup-
level performance would have been difficult to attain given the lack of unique
workgroup identification. This lack of data availability does not discount the
research results given here but motivates the need for future research, discussed

next.

7.4 Future research

Multiple avenues for future research should be pursued to further
elucidate relationships among quality management practices, cooperative

cultural values and work performance. Specifically, factors that may potentially

151

moderate the various relationships should be examined. For example, the
effects of organizational size and employee turnover could significantly moderate
the influence of organization-level quality management practices and cooperative
cultural values. Moreover, the benefits of both organization and workgroup-level
quality management practices may have changed depending upon the proximity
each workgroup had to customers. Another moderator could be the variance in
respondent measures of cultural values. Such variance could serve as a proxy
for cultural strength; a strong culture could be harder to influence but at the same
time be more influential. These and other moderators should be examined in
future research.

The intent of this research was to assess the possibility of cross-level
influences within multiple single periods. However, future research can be done
investigating the viability of cross—period influences within a multilevel model.
That is, further support for the direction of influence from organization-level
quality management practices to organization-level cooperative cultural values if
prior periods can predict future period outcomes (Mitchell et al., 2001). A
dynamic econometric methodology is a choice for approaching issues of this
manner (Hendry, 1995). This approach could also include objective
organization-level measures obtainable from known public sources. Valuable
insights may be gained from this dynamic examination of such a sociotechnical
system.

An important finding of this dissertation is the stable causal sequence

discussed in‘section 6.3.2. That is, the sequence characterized as

152

Qo—i VO—tQW—ePW. This process of influence could well be explored through a
case study methodology using the approach of narratives (Pentland, 1999). The
case studies could utilize retrospective interviews (Cohen, Kasen, Bifulco,
Andrews, and Gordon, 2005) to create generic narratives— i.e., a fabula (Langley,
1999) — that may explain the empirical observation of the aforementioned causal
sequence. The outcome from such an investigation would be an even richer

conceptualization of the relationships among quality management practices,

cooperative cultural values and work performance.

7.5 Summary

This dissertation has suggested a resolution to a theoretical paradox
previously unresolved by extant literature. Such a resolution was possible
through a reconceputalization. That is, utilizing Juran’s “big Q” and ”little Q”
quality management insight, support was found for a multilevel model explaining
how quality management can be both an antecedent and a consequent of
cooperative cultural values. However, this study does not conclude the search
for clearer understandings on this issue. Rather, it inspires further attempts to
clarify why quality management, cooperative values and workgroup performance
relate as they do. Moreover, the benefits of acknowledging the multilevel nature
of quality management should motivate researchers to explore how other
practices may have similar effects. Only through improving our understanding of
the management practices we use can we suggest new and better management

methods in the future.

153

APPENDIX

A.1 Imputation

The overall difference between actual and imputed values is shown in
Table A-1. The means demonstrated a median difference of less than 0.003 for
each year, while the median correlation difference was less than 0.06 for each
year. These differences were deemed not substantial to warrant concern and

therefore analysis progressed to factor analysis.

 

Table A-1: Difference Statistics Between Actual and Imputed Values

 

 

1998 1999 2000

Median absolute difference in mean 0.0007 0.0021 0.0029
(0.02%) (0.06%) (0.09%)

Median absolute difference in correlation 0.0005 0.0201 0.0507
(0.13%) (4.98%) (13.02%)

 

 

A.2 Factor Scoring

Grice (2001) assigned a value of 1/3 but described this as somewhat
arbitrary. In this dissertation the goal was for the factor scores to closely
replicate the CFA factor correlations. As such multiple iterations were conducted
to adjust a until the score-based factor correlations were within .05 of the CFA
factor correlations. The result of this procedure was a a of 1/4 and the
subsequent weight matrix is shown in Table A-2. An important feature in

recognizing factor obliqueness is that many items have multiple factors in which

1. 54

they reflect. In addition, one item (Q26) is not used at all and one item (Q17)

serves to decrease three factors.

 

 

 

Table A-2: Matrices from Factor ScoriLcLProcedure on 1998 Data

8,, : Structure Matrix WU.” : Unit Weight Matrix

0° 0““ v° PW 0° 0”“ v° PW
Q2 0.634 0.562 0.589 0.494 1 1 1 0
Q3 0.770 0.682 0.714 0.600 1 1 1 0
Q4 0.653 0.675 0.668 0.839 0 1 1 1
Q9 0.568 0.503 0.527 0.442 1 0 0 0
Q11 0.685 0.773 0.710 0.622 0 1 1 0
Q13 0.587 0.521 0.545 0.458 1 0 0 0
Q14 0.564 0.558 0.608 0.484 1 0 1 0
Q16 0.592 0.586 0.637 0.508 0 0 1 0
Q17 0.463 0.479 0.474 0.595 -1 -1 -1 0
Q21 0.461 0.457 0.497 0.396 0 0 1 0
Q25 0.654 0.648 0.705 0.562 1 0 1 0
Q26 0.380 0.429 0.394 0.345 0 0 0 0
Q30 0.716 0.807 0.742 0.650 1 1 1 0
Q31 0.721 0.814 0.748 0.655 1 1 1 0
Q32 0.678 0.700 0.693 0.870 1 1 1 1
Q33 0.511 0.528 0.523 0.656 0 0 0 1

 

 

 

A.3 Titles of quality managers interviewed

Four quality managers were interviewed for their reaction to the findings of
this research (as described in section 7.2). There were three plant-level
managers and one corporate-level manager. The plant-level managers’ titles
were as follows: Business Excellence & Six-Sigma Manager, Director of
Research and Development (with Malcohm Baldridge Award involvement since
1990), and Total Quality Manager. The corporate-level manager’s title was

Senior Consultant on Business Excellence.

155

REFERENCES

Ackoff, R. L. and Emery, F. E. 1972. On purposeful systems. Chicago, IL: Aldine-
Atherton, Inc.

Acock, A. C. 2005. Working with missing values. Journal of Marriage and the
Family, 67(4): 1012-1028.

Aczel, A. D. and Sounderpandian, J. 2005. Complete business statistics (6th
ed.): McGraw-HilI/Irwin.

Ahire, S. L. and Ravichandran, T. 2001. An innovation diffusion model of tqm
implementation. leee Transactions on Erlqineering Manafiment, 48(4): 445-464.

Amabile, T. M., R. Conti, H. Coon, J. Lazenby, M. Herron. 1996. Assessing the
work environment for creativity. Acad. Management J., 39: 1154-1184.

Anderson, J. C. and Gerbing, D. W. 1988. Structural equation modeling in
practice: A review and recommended two-step approach. Psvchologicafiulletin,
103(3): 411-423.

Anderson, J. C., Rungtusanatham, M., and Schroeder, R. G. 1994. A theory of
quality management underlying the deming management method. Academv of
Management Review, 19(3): 472-509.

Anderson, J. C., Rungtusanatham, M., Schroeder, R. G., and Devaraj, S. 1995. A
path analytic model of a theory of quality management underlying the deming

management method: Preliminary empirical findings. Decision Sciences, 26(5):
637-658.

Argyris, C. and Schdn, D. A. 1974. Theorv in practice : Incrtgsing professional
effectiveness. San Francisco: Jossey—Bass Publishers.

Argyris, C. and Schon, D. A. 1996. Qgganizational learning ii: Theorv, method
and practice. Reading, MA: Addison-Wesley.

Barko, W. and Pasmore, W. 1986. Introductory statement to the special issue on
sociotechnical systems - innovations in designing high-performing systems.
Journal Of Applied Behavioral Science, 22(3): 195-199.

Bartlett, M. S. and Kendall, D. G. 1946. The statistical analysis of variance-
heterogeneity and the logarithmic transformation. Journal of the Roval Statistical
Society Series iStatistical Methodrology, 8(1): 128-138.

156

Bates, K. A., Amundson, S. D., Schroeder, R. G., and Morris, W. T. 1995. The
crucial interelationship between manufacturing strategy and organizational
culture. Management Science, 41(10): 1565.

Bearden, W. O., Shanna, S., and Teel, J. E. 1982. Sample-size effects on chi-
square and other statistics used in evaluating causal-models. Journal of
Marketing Research, 19(4): 425-430.

Benson, R. F., Cunningham, S. P., and Leachman, R. C. 1995. Benchmarking
manufacturing performance in the semiconductor industry. Prodpction and
Operations Management, 4(3): 201-216.

Bentler, P. M. and Bonett, D. G. 1980. Significance tests and goodness of fit in
the analysis of covariance-structures. Psychological Bulletin, 88(3): 588-606.

Bentler, P. M. 1995. Egs structural equations program manual. Encino, CA:
Multivariate Software, Inc.

Bliese, P. D. 2000. Within-group agreement, non-independence, and reliability:
Implications for data aggregation and analysis. In S. W. J. K. K. J. Klein (Ed.),
Multilevel theory, research. and methods in organizations: 349-381. San
Francisco, CA: Jossey—Bass.

Boggs, W. B. 2004. Tqm and organizational culture: A case study. The Quality
Management Journal, 11(2): 42.

Brah, S. A., Wong, J. L., and Rao, B. M. 2000. Tqm and business performance in
the service sector: A singapore study. lntemational Journal of Operations &
Production Mamement, 20(11-12): 1293-1312.

Bright, K. and Cooper, C. L. 1993. Organizational culture and the management of
quality. Journal of Managerial Psychology, 8(6): 21.

Brown, M. 2001. Merit pay preferences among public sector employees. Human
Resource Management Journal, 11(4): 38.

Cameron, K. S. and Quinn, R. E. 1999. Dimosing and changing organizational
culture. Reading, MA: Addison-Wesley.

Carayon, P. 2006. Human factors of complex sociotechnical systems. Applied
Ergonomics, 37(4): 525-535.

Chatman, J. A. and Barsade, S. G. 1995. Personality, organizational culture, and
cooperation - evidence from a business simulation. Administrative Science
Quarterly, 40(3): 423-443.

Chems, A. 1987. Principles of sociotechnical design revisited. Human Relations,
40(3): 153.

157

Choi, T. Y. and Eboch, K. 1998. The tqm paradox: Relations among tqm
practices, plant performance, and customer satisfaction. Journal of Operations
Management, 17(1): 59.

Chun, Y. H. and Rainey, H. G. 2005. Goal ambiguity and organizational
performance in US. Federal agencies. Journal of Public Administration Research
and Theonr, 15(4): 529.

Clegg, C. W. 2000. Sociotechnical principles for system design. Applied
Ergonomics, 31(5): 463.

Cohen, R, Kasen, S., Bifulco, A., Andrews, H., and Gordon, K. 2005. The
accuracy of adult narrative reports of developmental trajectories. lntemational
Journal of Behavioral Development, 29(5): 345-355.

 

Curran, P. J., Bollen, K. A., Chen, F., Paxton, P., and Kirby, J. B. 2003. Finite
sampling properties of the point estimates and confidence intervals of the rmsea.
Sociological Methods and Research, 32(2): 208.

Das, A. and Jayaram, J. 2007. Socio-technical perspective on manufacturing
system synergies. International Journal of Production Research, 45(1): 169.

de Jong, A., de Ruyter, K., and Wetzels, M. 2005. Antecedents and
consequences of group potency: A study of self-managing service teams.
Management Science, 51(11): 1610-1625.

Deming, W. E. 1986. Out of crisis. Cambridge: Massachusetts Institute of
Technology, Center for Advanced Engineering Studies.

Dempster, A. P., Laird, N. M., and Rubin, D. B. 1977. Maximum likelihood from
incomplete data via em algorithm. Journal of the Royal Statistical Society Series
B-Methodolggical, 39(1): 1-38.

Denison, D. R. and Mishra, A. K. 1995. Toward a theory of organizational culture
and effectiveness. Organization Science, 6(2): 204.

Detert, J. R., Schroeder, R. G., and Mauriel, J. J. 2000. A framework for linking
culture and improvement initiatives in organizations. Academy of Management
Review, 25(4): 850.

Detert, J. R., Schroeder, R. G., and Cudeck, R. 2003. The measurement of
quality management culture in schools: Development and validation of the
sqmcs. Journal of Operations Management, 21(3): 307-328.

Dewey, J. 1939. Theory of valuation. Chicago: The University of Chicago Press.

158

Doolen, T. L., Hacker, M. E., and Van Aken, E. M. 2003. The impact of
organizational context on work team effectiveness: A study of production team.
leee Transactions on Engineering Management, 50(3): 285-296.

Douglas, T. J. and Fredendall, L. D. 2004. Evaluating the deming management
model of total quality in services. Decision Sciences, 35(3): 393-422.

Drazin, R., Glynn, M. A., and Kazanjian, R. K. 1999. Multilevel theorizing about
creativity in organizations: A sensemaking perspective. Acagemv of
Management. The Academy of Management Review, 24(2): 286.

Emery, F. 1959. Characteristics of socio-technical systems. In F. Emery (Ed.),
The emergence of a new paradm of work. Canberra: Centre for Continuing
Education.

Feldman, M. S. and Pentland, B. T. 2003. Reconceptualizing organizational
routines as a source of flexibility and change. Administrative Science Quarterly.
48(1): 94-118.

Flanagan, P. 1995. The abcs of changing corporate culture. Management
Review, 84(7): 57.

Flynn, B. B., Schroeder, R. G., and Sakakibara, S. 1995. The impact of quality
management practices on performance and competitive advantage. Decision
Sciences, 26(5): 659-691.

Fornell, C. and Larcker, D. F. 1981. Evaluating structural equation models with
unobservable variables and measurement error. Journal of Marketing Research,
18(1): 39-50.

Fox, W. M. 1995. Sociotechnical system principles and guidelines: Past and
present. Journal of Applied Behavioral Science, 31(1): 95-105.

Frohlich, M. T. 2002. Techniques for improving response rates in om survey
research. Journal of Operations Management, 20(1): 53.

Gallear, D. and Ghobadian, A. 2004. An empirical investigation of the channels
that facilitate a total quality culture. Total QualityManagement 8 Business
Excellence, 15(8): 1043-1067.

Gartner, W. B. and Naughton, M. J. 1988. The deming theory of management.
Academy of Management Review, 13: 138-142.

Gerbing, D. W. and Anderson, J. C. 1988. An updated paradigm for scale
development incorporating unidimensionality and its assessment. Journal of
Marketing Research, 25(2): 186-192.

159

Gorsuch, R. 1983. Factor anaILsis (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum
Associates.

Grice, J. W. 2001. A comparison of factor scores under conditions of factor
obliquity. Psychological Methods, 6(1): 67-83.

Griffith, T. L. and Dougherty, D. J. 2001. Beyond socio-technical systems:
Introduction to the special issue. Journal of Engineering Technology
Management, 18: 2007-2218.

Gupta, Y. P. and Ash, D. 1994. Excellence at rohm and haas kentuckyz'A case
study of work-team introduction in manufacturing. Production and Operations
Management, 3(3): 186-200.

Harris, L. C. and Ogbonna, E. 2002. The unintended consequences of culture
interventions: A study of unexpected outcomes. British Journal of Management,
13(1): 31.

Hays, J. M. and Hill, A. V. 2001. A preliminary investigation of the relationships
between employee motivation/vision, service learning, and perceived service
quality Journal of Operations Management, 19(3): 335.

Hendricks, K. B. and Singhal, V. R. 1997. Does implementing an effective tqm
program actually improve operating performance? Empirical evidence from firms
that have won quality awards. Management Science, 43(9): 1258-1274.

Hendricks, K. B. and Singhal, V. R. 2001a. Firm characteristics, total quality
management, and financial performance. Journal of Operations Management,
19(3): 269-285.

Hendricks, K. B. and Singhal, V. R. 2001b. The long-run stock price performance
of firms with effective tqm programs. Management Science, 47(3): 359-368.

Hendry, D. F. 1995. Dynamic econometrics. New York: Oxford University Press.

Hitlin, S. and Piliavin, J. A. 2004. Values: Reviving a dormant concept. Annual
Review of Sociolggy, 30: 359-393.

. Hofmann, D. A. 1997. An overview of the logic and rationale of hierarchical linear
models. Journal of Management, 23(6): 723-744.

Hofstede, G. 1998. Attitudes, values and organizational culture: Disentangling
the concepts. Organization Studies, 19(3): 477.

House, R., Rousseau, D. M., and Thomashunt, M. 1995. The meso paradigm - a
framework for the integration of micro and macro organizational-behavior,
Research in organizational behavior: An annual series of analytical essays and
critical reviews, vol 17, 1995, Vol. 17: 71-114.

160

Houston, M. B. 2004. Assessing the validity of secondary data proxies for
marketing constructs. Journal of Business Research, 57(2): 154.

Hu, L.-t. and Bentler, P. 1999. Cutoff criteria for fit indexes in covariance
structure analysis: Conventional criteria versus new alternatives. Structural
Equation Modeligg: A Mgltidisciplinary Journal. 6(1): 1-55.

Huber, V. L. and Brown, K. A. 1991. Human resource issues in cellular
manufacturing: A sociotechnical analysis. Journal of Operations Management,
10(1): 138.

Huq, Z. 2005. Managing change: A barrier to tqm implementation in service
industries. Managing Service Qualigr, 15(5): 452.

Inkpen, A. C. 2005. Learning through alliances: General motors and nummi.
California Management Review, 47(4): 114.

Jabnoun, N. and Sedrani, K. 2005. Tqm, culture, and performance in uae
manufacturing firms. The Quality Management Jogrnal, 12(4): 8.

Jackson, C. L., Colquitt, J. A., Wesson, M. J., and Zapata-Phelan, C. P. 2006.
Psychological collectivism: A measurement validation and linkage to group
member performance. Journal of Applied Psychology, 91(4): 884-899.

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

James, L. R., Demaree, R. G., and Wolf, G. 1993. R(wg) - an assessment of
within-group interrater agreement. Journal of Applied Psychology, 78(2): 306-
309.

Jones, G. R. and George, J. M. 1998. The experience and evolution of trust:
Implications for cooperation and teamwork. Academy of Management Review,
23(3): 531-546.

Juran, J. M. 1989. Juran on leadership for quality. an executive handbook. New
York, NY: The Free Press.

Kamensky, J. M. 1996. Role of the "Reinventing government“ Movement in
federal management reform. Public Administration Review, 56(3): 247.

Kathuria, R. and Davis, E. B. 2001. Quality and work force management
practices: The managerial performance implication. Production and Operations
Management, 10(4): 460.

161

Kaynak, H. 2003. The relationship between total quality management practices
and their effects on firm performance. Journal of Operations Management, 21(4):
405.

Ketokivi, M. A. and Schroeder, R. G. 2004. Perceptual measures of performance:
Fact or fiction? Journal of Operations Management, 22(3): 247.

Kim, J. S. and Frees, E. W. 2006‘. Multilevel modeling with correlated effects.
Psychometrika(Accepted).

Kujala, J. and Lillrank, P. 2004. Total quality management as a cultural
phenomenon. The Quality Management Journal, 11(4): 43.

Langley, A. 1999. Strategies for theorizing from process data. Academy of
Management Review, 24(4): 691-710.

Little, R. J. A. and Rubin, D. B. 1989. The analysis of social-science data with
missing values. Sociological Methods 8 Research, 18(2-3): 292-326.

Liu, G., Shah, R., and Schroeder, R. G. 2006. Linking work design to mass
customization: A sociotechnical systems perspective*. Decision Sciences, 37(4):
519.

Louis, M. R. 1980. Surprise and sense making - what newcomers experience in
entering unfamiliar organizational settings. Administrative Science Quarterly.
25(2): 226-251.

Malhotra, N. K., Kim, S. S., and Patil, A. 2006. Common method variance in is
research: A comparison of alternative approaches and a reanalysis of past
research. Manggement Science, 52(12): 1865-1883.

March, J. G. and Simon, H. A. 1958. Organizations. New York: John Wiley 8
Sons, Inc.

Martin, J. 2002. Organizational culture: Mapping the terrain. Thousand Oaks, CA:
Sage.

Martocchio, J. J. 1994. The effects of absence culture on individual absence.
Human Relations, 47(3): 243-262.

McDermott, C. M. and Stock, G. N. 1999. Organizational culture and advanced
manufacturing technology implementation. Journal of Operations Manggement,
17(5): 521.

Mead, E. F. 1985. Building a corporate quality culture: A test case. Quality
Progress, 18(3): 10.

162

Meade, A. W. and Eby, L. T. 2007. Using indices of group agreement in
multilevel construct validation. Organizational Research Methods, 10(1): 75.

Mendez, D. and Narasimhan, R. 2002. Examining market oriented aspects of
cost of quality. IEEE Transactions on Engineering Management, 49(2): 131.

Miron, E., Erez, M., and Naveh, E. 2004. Do personal characteristics and cultural
values that promote innovation, quality, and efficiency compete or complement
each other? Journal of Organizational Behavior, 25(2): 175-199.

Mitchell, T. R. and James, L. R. 2001. Building better theory: Time and the
specification of when things happen. Academy of Management Review, 26(4):
530-547.

Mitra, D. and Golder, P. N. 2007. Quality is in the eye of the beholder. Harvard
Business Review, 85(4): 26.

Moore, B. and Brown, A. 2006. The application of tqm: Organic or mechanistic?
The International Journal of Q_uality & Reliability Management, 23(7): 721.

Moray, N. 2000. Culture, politics and ergonomics. Ergonomics, 43(7): 858-868.

Nahm, A. Y., Vonderembse, M. A., and Koufteros, X. A. 2004. The impact of
organizational culture on time-based manufacturing and performance. Decision
Sciences, 35(4): 579.

Narasimhan, R., Ghosh, S., and Mendez, D. 1993. A dynamic model of product
quality and pricing decisions on. Decision Sciences, 24(5): 893.

Narasimhan, R., Mendez, D., and Ghosh, S. 1996. An examination of the effect
of continuous quality improvements on optimal pricing for durable goods.
Decision Sciences, 27(3): 389.

Narasimhan, R. and Mendez, D. 2001. Strategic aspects of quality: A theoretical
analysis. Production and Operatjons Management, 10(4): 514.

Naveh, E. and Erez, M. 2004. Innovation and attention to detail in the quality
improvement paradigm. ManagemenLScience, 50(11): 1576-1586.

Niepce, W. and Molleman, E. 1998. Work design issues in lean production from a
sociotechnical systems perspective: Neo-taylorism or the next step in
sociotechnical design? Human Relations, 51(3): 259.

O'Reilly, C. A., Chatman, J., and Caldwell, D. F. 1991. People and organizational
culture - a profile comparison approach to assessing person-organization fit.
Academy of Management Journal, 34(3): 487-516.

163

O.P.M. 2002. National partnership for reinventing government series. US.
Federal Government: Inter-University Consortium for Political and Social
Research: 3419 - 3421.

Ogbonna, E. and Harris, L. C. 1998. Managing organizational culture:
Compliance or genuine change? British Journal of Management, 9(4): 273.

Oreilly, C. A., Chatman, J., and Caldwell, D. F. 1991. People and organizational
culture - a profile comparison approach to assessing person-organization fit.
Acadermr of Management Journal, 34(3): 487-516.

Parasuraman, A., Zeithaml, V. A., and Berry, L. L. 1988. Servqual - a multiple-
item scale for measuring consumer perceptions of service quality. Journal of
Retailing, 64(1): 12-40.

Pasmore, W., Francis, C., Haldeman, J., and Shani, A. 1982. Sociotechnical
systems: A north american reflection on empirical studies of the seventies.
Human Relations, 35(12): 1179-1204.

Pasmore, W. 1988. Designing effective organizations: The socio-technical
systemflerspective. New York: Wiley.

Patti, A. L., Fok, L. Y., and Hartman, S. J. 2004. Differences between managers
and line employees in a quality management environment. The lntemational
Journal of Quality 8 Reliability Management, 21 (2/3): 214.

Pentland, B. T. 1999. Building process theory with narrative: From description to
explanation. Academy of Management Review, 24(4): 711-724.

Peters, T. J. and Waterman, R. 1982. In search of excellence: Lessons from
americas best run companies. New York, NY: Warner Books.

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

Poole, M. S. and Vandeven, A. H. 1989. Using paradox to build management
and organization theories. Academy of Management Review, 14(4): 562-578.

Powell, T. C. 1995. Total quality management as competitive advantage - a
review and empirical-study. Strategic Management Journal, 16(1): 15-37.

Qiao, Y. and Thai, K. V. 2002. Reinventing government at the federal level: The
implementations and the prospects. Public Administration Quarterly, 26(1/2): 89.

Quinn, R. E. and Rohrbaugh, J. 1983. A spatial model of effectiveness criteria:
Towards a Competing values approach to organizational analysis. Marfigement
Science, 29(3): 363.

164

Rasmussen, J. 2000. Human factors in a dynamic information society: Where are
we heading? Ergonomics, 43(7): 869-879.

Raudenbush, S. W. and Bryk, A. S. 2002. Hierarchical linear models (2nd ed.).
Thousand Oaks: Sage Publications.

Raudenbush, S. W., Bryk, A. S., Cheong, Y. F., Congdon, R., and du Toit, M.
2004. Him 6: Hierarchical linear and nonlinear modeling. Lincolnwood, IL:
Scientific Software International, Inc.

Ray, G., Barney, J. B., and Muhanna, W. A. 2004. Capabilities, business
processes, and competitive advantage: Choosing the dependent variable in

empirical tests of the resource-based view. Strategic Management Journal,
25(1): 23-37.

Rokeach, M. 1973. The nature of human values.

Rosenzweig, P. M. 1994. When can management science research be
generalized internationally? Management Science, 40(1): 28.

Rungtusanatham, M., Forza, C., Filippini, R., and Anderson, J. C. 1998. A
replication study of a theory of quality management underlying the deming
management method: Insights from an italian context. Journal of Operations
Management, 17: 77-95.

Rungtusanatham, M. 2001. Beyond improved quality: The motivational effects of
statistical process control. Journal of Operations Managgment, 19(6): 653-673.

Schein, E. H. 2004. Organizational culture and leadersmp (3rd ed.). San
Fransisco, CA: Jossey-Bass.

Schmenner, R. W. 2004. Service businesses and productivity. Decision
Sciences, 35(3): 333-347.

Schroeder, R. G., Linderman, K., and Zhang, D. L. 2005. Evolution of quality:
First fifty issues of production and operations management. Production and
Operations Management, 14(4): 468-481.

Schwartz, 8. H. 1992. Universals in the content and structure of values -
theoretical advances and empirical tests in 20 countries. Advances in
Experimental Social Psycholggy, 25: 1-65.

Seiler, J. A. 1967. Systems analysis in organizational behavior. Homewood, III.:
R. D. Irwin.

Snijders, T. and Bosker, R. 1999. Multilevel analysis. London: Sage Publications.

165

Soltani, E., Van der Meer, R. B., and Williams, T. M. 2004. Challenges posed to
performance management by tqm gurus: Contributions of individual employees

versus systems-level features. TotaLanlity Management 8 Business
Excellence, 15(8): 1069-1091.

Song, J. and Zahedi, F. 2005. A theoretical approach to web design in e-
commerce: A belief reinforcement model. Management Science, 51 (8): 1219-
1235. '

Song, X. M., Montoya-Weiss, M. M., and Schmidt, J. B. 1997. Antecedents and
consequences of cross-functional cooperation: A comparison of r8d.
manufacturing, and marketing perspectives. Journal of Progpct Innovation
Management, 14(1): 35-47.

Stewart, D. M. and Chase, R. B. 1999. The impact of human error on delivering
service quality. Production and Op_e_rations Management, 8(3): 240-263.

Stewart, D. M. and Grout, J. R. 2001. The human side of mistake-proofing.
Progpction and Operations Management, 10(4): 440-459.

Taveira, A. D., James, C. A., Karsh, B. T., and Sainfort, F. 2003. Quality
management and the work environment: An empirical investigation In a public
sector organization. Applied Ergonomics, 34(4): 281-291.

Thompson, J. 1967. Organizations in action. New York: McGraw Hill.

Trist, E. L. and Bamforth, K. W. 1951. Some social and psychological
consequences of the longwall method of coal-getting. Human Relations, 1: 3-38.

Van Eijnatten, F. M. 1992. Aggnthology of the socio-technical system design
paradigm. Eindhoven: Eindhoven University of Technology Press.

van Woerkom, M. 2004. The concept of critical reflection and its Implications for

human resource development. Advances in Developing Human Resources, 6(2):
178.

Venkatraman, N. 1989. Strategic orientation of business enterprises - the
construct, dimensionality, and measurement. Management Science, 35(8): 942-
962.

Verma, R., Youngdahlb, W., McLaughlinc, C., and Johnstond, R. 2006.
Innovative operations management applications in not-for-profit, public and
government services. Journal of Operations Management 24(4).

von Hippel, P. T. 2004. Biases in spss 12.0 missing value analysis. flag
Amerjggn Statistician, 58(2): 160.

166

Wagner, J. A. 1995. Studies of individualism-collectivism - effects on cooperation
in groups. Academy of Management Journal, 38(1): 152-172.

Waldman, D. A. 1994. The contributions of total quality management to a theory
of work performance. Acagggnr of Management Review, 19(3): 510-536.

Wallace, L., Keil, M., and Rai, A. 2004. How software project risk affects project
performance: An investigation of the dimensions of risk and an exploratory
model". Decision Sciences, 35(2): 289.

Weiss, H. M. 1978. Social-leaming of work values in organizations. Journal of
Applied Psychology, 63(6): 71 1-718.

Welch, B. L. 1947. The generalization of students problem when several different
population variances are involved. Bipmetrika, 34(1-2): 28-35.

Westphal, J. D., Gulati, R., and Shortell, S. M. 1997. Customization or
conformity? An institutional and network perspective on the content and

consequences of tqm adoption. Administrative Science Quarterly, 42(2): 366-
394.

Yauch, C. A. and Steudel, H. J. 2002. Cellular manufacturing for small
businesses: Key cultural factors that impact the conversion process. Jgumal of
Operations Management, 20(5): 593.

Yelle, L. E. 1979. The learning curve: Historical review and comprehensive
survey. Decision Sciences, 10(2): 302.

Zbaracki, M. J. 1998. The rhetoric and reality of total quality management.
Administrative Science Quarterly. 43(3): 602-636.

Zhao, X. D., Yeung, A. C. L., and Lee, T. S. 2004. Quality management and

organizational context in selected service industries of china. Jggrnal of
Operations Management, 22(6): 575-587.

167

       
  
    

IA GA

llllll 111191111

312

U

llllllll

029

3 BR

[lililllllil

5 420