Li!
J...

x Y.
amafi
‘ Lfig _ J.
, , .. . ‘ ‘ ‘ dwfiézwf... «Win:
. , ; P; 5 g... mum
4%,.”2. ‘ , . ‘ , ‘ . , 1 4 ‘ . 9.3,, ...
.7; ..

.O
. . .5...
:0...“- . , . . V . ‘ firruu L“
d“.
$6.54
..
r

 

._..m.._...
J A
. as» fum: :I I
‘ 3. . »Wmmu§\4

23!.225:
.i 25:; I. :80.» .
a: .3. v 42.1.3

mum: fiMflflfif

 

.w.

ii...- i‘
{:34 .53.”; to 3‘

‘ skunk v. . Ffiaé‘”
fin; g“.

 

35"-.” a m? u. 335.!
garagfip ,
3‘5iéag if!)
ahnufi. (£1. ‘
.. 5:5»...

 

 

 

 

 

 

‘ ’ ‘-: . .
It-‘fgi’l

A 1.3:. . . .
:53; t, .9:

 

. MHz...” x . ‘ . ‘ 9.. W. . Hugh 4%

 

 

 

THESIS
/

X90 2%

lIBRARY
Michigan State
University

This is to certify that the

dissertation entitled

A Definition of World Class Manufacturing and
an Empirical Analysis of Practice-Performance
Relationships in Manufacturing Plants

presented by
Xinyan Wang

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in Marketing and Supply

 

 

Chain Management

Major professor

Date Izilzll a l

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

 

 

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

 

A

 

AN 0132083

 

meoi

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIFiC/DateDue.p65-p.15

 

A DEFINITION OF WORLD CLASS MANUFACTURING AND AN EMPIRICAL
ANALYSIS OF PRACTICE-PERFOMANCE RELATIONSHIPS IN
MANUFACTURING PLANTS

By

Xinyan Wang

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
Eli Broad Graduate School of Management
Michigan State University

2001

ABSTRACT
A DEFINITION OF WORLD CLASS MANUFACTURING AND AN EMPIRICAL
ANALYSIS OF PRACTICE-PERFORMANCE RELATIONSHIPS IN
MANUFACTURING PLANTS
By
Xinyan Wang

The topic of World Class Manufacturing (WCM) has attracted a lot of attention in
manufacturing industry and operations strategy literature. The wide spread use of
benchmarking which involves copying the “best practices” from “best performing”
plants makes it compelling to define WCM and investigate how “best practices” and
“best performance” are related. Literature in operations strategy has defined WCM based
on either practices or performance each offering their own list of “WCM practices” or
performance. There has been little empirical research concerning the relationship
between the so-called “WCM practices” and performance.

This research is intended to provide a definition of WCM by employing multiple
criteria and an empirical investigation of the relationships between the “WCM practices”
and performance while taking into consideration technical efficiency of each individual
manufacturing plants.

This research first summarized and integrated practice-based and performance-
based definitions of WCM in an effort to provide measurement models for WCM
practices and performance. Methodology and theory from several fields that are related to
manufacturing strategy are employed. Efficiency analysis which is based on traditional
ideas in economics relating outputs to inputs was utilized to identify the “best

performing” plants which are the most technically efficient. It reconciles the notion of

technical efficiency with the notion of world class by verifying the moderating effect of
efficiency on the relationship between practices and performance. Taxonomy analysis,
which is a popular tool in segmenting operations strategy, was performed on efficient
manufacturing plants in the data sample. The results show that based on which strategy
the efficient plants are adopting, the priorities they placed on various “WCM practices”
are different.

This dissertation makes four major contributions to operations strategy literature.
First, this research represents one of the first efforts to synthesize the dimensions of
WCM practices in literature. Comprehensive measurement models, which may provide a
foundation for further empirical testing in this area are proposed and empirically tested.
Second, using technical efficiency in the traditional economic sense overcomes the
weaknesses of existing definitions of WCM in the literature. The research findings
suggested that technical efficiency is a valid measure and is one of the three pre-
requisites of WCM. Third, results of cluster analysis and comparison among the three
clusters within efficient manufacturing plants indicate that these manufacturing plants
have adopted different strategies to become efficient. What is more important is that these
groups have different priorities on WCM practices. This fiirther indicates that the
approach of current WCM literature, which does not realize that differences exist in
priorities on manufacturing practices within best performing plants due to strategy
differences, is inadequate. This research warrants the need to fiu‘ther investigate how the
strategy differences affect the relationship between practices and performance. Finally,
the research findings suggested that to be termed WCM requires all three criteria: engage

in WCM practices, be technically efficient and perform well on all the performance areas.

Copyright by
)CHTVUADJVVAUQC}

2001

DEDICATED TO MY PARENTS, GEPING WANG AND QINGYANG LI

ACKNOWLEGMENTS

I gratefully acknowledge the guidance and advice provided by my dissertation
Chair, Professor Ram Narasimhan, at various stages of my doctoral degree program. He
spent enormous amount of time making sure that I developed the necessary skills to be a
successful researcher and person, even long before the development and completion of
this dissertation. This research would not have been possible without his direction and
care.

I would like to express my deep appreciation for the time and efforts contributed
by other members of my dissertation committee, Dr. Morgan Swink, Dr. Gary Ragatz,
Dr. Srinivas Talluri. They spent inordinate amount of time providing precious comments
that guided this dissertation on the right track. Their serious attitudes toward research
have had great influence on me and on this dissertation as a result. I consider myself
blessed to have been given the opportunity to learn from these scholars. Their quick
feedback on my dissertation drafts also helped tremendously in keeping this research on
schedule.

The managers who contributed their time and thoughts in responding to my
questionnaire should also be thanked. Empirical research would not be possible without
their cooperation. It is my hope that the results of this research and any future research that
may be based on this research will benefit the managers in the long run.

Finally, I would like to reiterate my gratitude to my parents. Without them being a

role model for the attitudes towards work and life since I was very young, I could not

have achieved what I have done today. The sacrifice they made to support me to pursue

vi

graduate education and the encouragement they provided during the entire period of the

doctoral program will always be appreciated and remembered.

vii

TABLE OF CONTENTS

 

 

 

 

 

CHAPTER 1
INTRODUCTION 1
1.1. OVERVIEW OF THE DEFINITION OF WORLD CLASS MANUFACTURING ....................... l
1.2. MISSING LINK IN WORLD CLASS MANUFACTURING LITERATURE ............................. 4
1.3. RESEARCH OBJECTIVES AND STEPS ........................................................................... 5
1.4. STRUCTURE OF THE DISSERTATION ............................................................................ 8
CHAPTER 2
LITERATURE REVIEW 9
2.1. TAXONOMY OF MANUFACTURING STRATEGY ............................................................ 9
2.2. COMPETITIVE PERFORMANCE .................................................................................. 13
2.2.1. Theoretical Components ................................................................................... 14
2.2.2. Measurement Issue ........................................................................................... 15
2.2.3. Trade-off vs. Synergy ........................................................................................ 17
2.3. WORLD CLASS MANUFACTURING ............................................................................ 22
2.4. RELATIONSHIP BETWEEN PRACTICES AND PERFORMANCE ....................................... 28
2.5. SUMMARY OF LITERATURE REVIEW ........................................................................ 30
CHAPTER 3
CONCEPTUAL FRAMEWORK 32
3.1. CONCEPTUAL FRAMEWORK ..................................................................................... 32
3.2. HYPOTHESES DEVELOPMENT ................................................................................... 43
CHAPTER 4
RESEARCH DESIGN 47
4.1. RESEARCH DESIGN .................................................................................................. 47
4.1.1. Scale Validation ................................................................................................ 48
4.1.2. Unit of Analysis and Sample Selection ............................................................. 49
4.1.4. Data Collection ................................................................................................ 50
4.2. METHODOLOGY ....................................................................................................... 52
4.2.1. Establish Measurement Models ........................................................................ 52
4.2.2. Identify Efficient Manufacturing Plants ........................................................... 55
4.2.3. Investigate Moderating Effect of Efficiency ..................................................... 60
4.2.4. Taxonomy of Efiicient Manufacturing Plants ................................................... 63
4.2.5. Compare Strategic Groups within Efficient Manufacturing Plants ................. 64
CHAPTER 5
DATA ANALYSIS 66
5.1. THE PROS AND CONS OF WEB BASED SURVEY ........................................................ 66
5.2. RESPONDENT PROFILE ............................................................................................. 68
5.3. CONFIRMATORY FACTOR ANALYSIS ........................................................................ 71
5.3.1. CFA for WCM Practices .................................................................................. 73
5.3.1.1. Univariate and Multivariate Normality ...................................................... 74
5.3.1.2. Goodness of Fit Indices .............................................................................. 77

Viii

5.3.1.3. Reliability and Convergent Validity .......................................................... 78

 

 

5.3.1.4. Discriminant Validity and Nomological validity ....................................... 80
5.3.1.5. Common Method Bias ............................................................................... 83
5.3.2. CFA for WCM Performance ............................................................................. 84
5.3.2.1. Univariate and Multivariate Normality ...................................................... 84
5.3.2.2. Goodness of Fit Indices .............................................................................. 86
5.3.2.3. Reliability and Convergent Validity .......................................................... 87
5.3.2.4. Discriminant Validity and Nomological validity ....................................... 88
5.3.2.5. Common Method Bias ............................................................................... 90
5.4. DATA ENVELOPMENT ANALYSIS ............................................................................. 92
5.5. MANCOVA TESTS ................................................................................................. 98
5.5.1. Test of Hypotheses 1 ......................................................................................... 98
5.5.2. Hypothesis 1 Test Result and Analysis ........................................................... 102
5.6. CLUSTER ANALYSIS ............................................................................................... 105
5.6.1. Hierarchical Procedure .................................................................................. 105
5.6.2. K-means Clustering ........................................................................................ 108
5.6.3. Validity and Hypothesis 2 Result .................................................................... 108
5.7. AN OVA AND MULTIPLE RANGE TESTS ................................................................ 110
5.7.1. One Way ANOVA Tests .................................................................................. 110
5.7.2. Multiple Range Tests ...................................................................................... 111
5. 7.3. Hypothesis 3 Test Result and Analysis ........................................................... 11 7
CHAPTER 6
DISCUSSION OF RESULTS - 119
6.1. IMPLICATIONS OF HYPOTHESIS 1 TEST RESULTS ................................................... 1 19
6.2. IMPLICATIONS OF HYPOTHESIS 2 AND 3 TEST RESULTS ......................................... 121
6.3. CONTRIBUTIONS .................................................................................................... 124
6.4. LIMITATIONS AND FUTURE RESEARCH DIRECTIONS .............................................. 125
APPENDIX A
MEASUREMENT INSTRUMENT 128
BIBLIOGRAPHY _ 146

 

ix

LIST OF TABLES

TABLE 2.1 GARVIN'S DECOMPOSED STRATEGIC PRIORITIES .............................................. 16
TABLE 3.1 PRACTICE PROFILE WITH LITERATURE SUPPORT .............................................. 33
TABLE 3.2 PERFORMANCE PROFILE ................................................................................... 38
TABLE 5.1 GROUP 1 RESPONDENTS (N=58) ........................................................................ 69
TABLE 5.2 GROUP 2 RESPONDENTS (N=79) ........................................................................ 70
TABLE 5.3 GROUP 3 RESPONDENTS (N=87) ........................................................................ 70
TABLE 5.4 FACTORS AND VARIABLES IN WCM PRACTICES CFA MODEL ......................... 73
TABLE 5.5 UNIVARIATE STATISTICS: WCM PRACTICES CFA ........................................... 75
TABLE 5.6 MULTIVARIATE KURTOSIS FOR WCM PRACTICES ............................................ 77

TABLE 5.7 GOODNESS-OF-FIT INDICES FOR WCM PRACTICES MEASUREMENT MODEL... 77
TABLE 5.8 MEASUREMENT EQUATIONS WITH STANDARD ERRORS, STANDARDIZED
SOLUTION AND CRONBACH'S ALPHA FOR WCM PRACTICE CFA MODEL ................... 79

TABLE 5.9 VARIANCES FOR EACH FACTOR AND INTER-FACTOR COVARIANCES FOR WCM

PRACTICES CFA .......................................................................................................... 81
TABLE 5.10 MULTIVARIATE LM TEST WITH THETA LAMBDA MATRIX RELEASED ............ 83
TABLE 5.11 FACTORS AND VARIABLES IN WCM PERFORMANCES CFA MODEL ............... 84
TABLE 5.12 UNIVARIATE STATISTICS: WCM PERFORMANCE CFA ................................... 85
TABLE 5.13 MULTIVARIATE KURTOSIS FOR WCM PERFORMANCE ................................... 86

TABLE 5.14 GOODNEss-OF-FIT INDICES FOR WCM PERFORMANCE MEASUREMENT
MODEL ........................................................................................................................ 87
TABLE 5.15 MEASUREMENT EQUATIONS WITH STANDARD ERRORS, STANDARDIZED

SOLUTION AND CRONBACH'S ALPHA FOR WCM PERFORMANCE CFA MODEL ........... 89

TABLE 5.16 VARIANCES FOR EACH FACTOR, INTER-FACTOR COVARIANCES AND SQUARED
CORRELATION FOR WCM PERFORMANCE CFA .......................................................... 90

TABLE 5.17 MULTIVARIATE LM TEST WITH THETA LAMBDA MATRIX RELEASED

FOR WCM PERFORMANCE CFA MODEL .................................................................... 91
TABLE 5.18 INPUT AND OUTPUT DIMENSIONS FOR DEA MODEL ...................................... 92
TABLE 5.19 DISTRIBUTION OF EFFICIENCY SCORES ........................................................... 95

TABLE 5.20 WILK'S LAMBDA FOR NINE MANCOVA MODELS TESTING HYPOTHESIS 1 99

TABLE 5.21 REGRESSION COEFFICIENTS FOR INTERACTION TERMS IN MAN COVA ....... 100
TABLE 5.22 AGGLOMERATION COEFFICIENTS FROM WARD'S METHOD ........................... 107
TABLE 5.23 THE CLUSTER CENTERS FROM WARD'S METHOD ......................................... 107
TABLE 5.24 DISTANCES BETWEEN CLUSTER CENTERS .................................................... 108
TABLE 5.25 ANOVA RESULT FOR K-MEANS CLUSTERING ............................................. 109
TABLE 5.26 ANOVA RESULTS ........................................................................................ 110
TABLE 5.27 TUKEY MULTIPLE RANGE TEST RESULTS ON SEVEN WCM PRACTICES ....... 1 11

TABLE 5.28 DESCRIPTIVE STATISTICS OF WCM PRACTICES FOR THE CLUSTERS AND

MULTIPLE RANGE TEST RESULTS .............................................................................. 1 14

xi

LIST OF FIGURES

FIGURE 3.1 EFFICIENCY AS A MODERATOR BETWEEN WCM PRACTICES AND

PERFORMANCE ............................................................................................................ 41
FIGURE 3.2 CONCEPTUAL FRAMEWORK FOR WORLD CLASS MANUFACTURING ................ 42
FIGURE 4.1 COMPARISON BETWEEN DEA AND REGRESSION ............................................. 57
FIGURE 4.2 ENVELOPMENT SURFACE FOR BCC MODEL .................................................... 58
FIGURE 4.3 ENVELOPMENT SURFACE FOR CCR MODEL .................................................... 58
FIGURE 5.1 DISTRIBUTION OF EFFICIENCY SCORE ............................................................. 96

xii

CHAPTER 1

INTRODUCTION

1.1. Overview of the Definition of World Class Manufacturing

The term "world class manufacturing" was first introduced by Hayes and
Wheelwright in 1984. Since then, various authors have embraced and expanded this
concept. Most of the research pieces are descriptive in nature. Generally, there is no
agreed upon definition about what world class manufacturing is. There have been two
different trends in defining world class manufacturing (W CM): practice-based and
performance-based.

Hayes and Wheelwright (1984), Schonberger (1990), and Giffi ct al. (1990) have
discussed world class manufacturing from the perspective of manufacturing practices.

Hayes and Wheelwright (1984), in their classic book "Restoring Our Competitive
Edge: Competing through Manufacturing" (page 375-385), proposed that world class
manufacturing is composed of six dimensions: Workforce skills and capabilities,
management technical competence, competing through quality, workforce participation,
rebuilding manufacturing engineering, incremental improvement approaches. By
comparing the practices of Japanese and German manufacturers with US. manufacturers,
Hayes and Wheelwright claimed that the US. plants must focus on these six broad
categories of practices in order to achieve their world class manufacturing status.

Schonberger (1986, 1996) provided a list of 16 principles of world class
manufacturing Which fall into eight categories: general, design, operations, human

resources, quality and process improvement, information for operations and control,

capacity, promotion and marketing. Schonberger actually asked managers to evaluate
their own plants based on these 16 principles. He warned those plants that scored low on
the 16 principles to identify their problems and make an effort to improve these practices
to keep up with the competition.

Giffi, Roth and Seal (1990) also provided operating principles of world class
manufacturing in the following areas: quality and customer, management approach,
manufacturing strategy, manufacturing capabilities, performance measurement,
organization, human assets and technology. Giffi et al. (1990) considered these practice
dimensions to be interrelated; quality and customer constitue the core of the principles,
all the other dimensions interact with quality dimension. This article was the first one that
brought competitive capabilities into WCM framework. Their View of competitive
capability agrees with both performance-based definition of WCM and cumulative model
of competitive capabilities. Similar to the performance-based definition of WCM, they
argued that achieving multiple capabilities Simultaneously is one of the keys to
manufacturing excellence. They also endorsed F erdows and De Meyer's (1990)
cumulative model in that they believe that the time-based competitive capabilities
(dependable delivery time, delivery speed, and flexibility) are built upon superior quality.

These practice-based definitions of WCM have a lot in common. All of them
imply that adoption of these best practices will lead to superior performance, but none of
them went one step further to discuss the prescriptive properties of these practices (how
strongly each practice is related to performance). If the strength of the relationship
between practice and performance varies, it may be more effective to focus on a subset of

practices selectively.

Still, there are quite a few authors who have adopted the performance-based
definitions of WCM. The performance-based definitions are generally more
straightforward and intuitive.

Todd (1995) defined world—class as being the best in terms of product design and
performance, quality and reliability, least manufacturing cost, the ability to keep
introducing innovative designs more quickly than your competitors, shorter lead times
and more reliable delivery performance, and customer services.

Taninecz (1997 b) asked participants of the Industry Week Census of
Manufacturers to assess their plant's or company's progress toward world-class. Taninecz
found correlations at the plant level between world-class status and best performance, and
identified speed, cost and quality as the best metrics that measured the progress toward
world-class status.

The performance-based definitions generally avoided answering questions that
have been debated in the operations literature; such as whether there are trade-offs among
these performance areas, whether one performance area builds on the others, and whether
synergies exist. The performance-based definitions seem to suggest that WCM is a
homogenous group in terms of their strategic choices, that is, all WCM plants are
working on all the dimensions of competitive performance without placing a priority on
one dimension over the others. This definition poses potential difficulties for
practitioners. Without identifying strategic differences among the best plants, the concept

of WCM is too vague for plants to develop the right benchmarks to use for improvement.

1.2. Missing Link in World Class Manufacturing Literature

Despite the implications of the practice-based definition of WCM that WCM
practices will lead to superior performance, there is a lack of both theoretical and
empirical investigation of the relationship between practices and performance in WCM
literature. The relationship between various manufacturing practices (somewhat
overlapping with WCM practices to some degree) and performance, however, has been
attracting a lot of attention in manufacturing strategy literature. Most of the research that
studied the relationship between manufacturing practices and performance focused only
on a small set of either manufacturing practices or performance. Comprehensive and
systematic study of the relationship between the so called “world class manufacturing
practices” and performance is considered missing in literature, and will be a focus in this
dissertation.

As mentioned earlier, both practice-based definition and performance-based
definition of WCM have their weaknesses, this dissertation intends to overcome these
weaknesses by including multiple criteria and approaches that have been used to identify
“best practices” in strategy literature.

Efficiency analysis is one of the tools that has been utilized for performance
analysis in economics and finance and has seen increased applications in operations
literature. Efficient manufacturing units are defined as those that produce the most
amount of output with the least amount of input. Those efficient manufacturing units, if
plotted on an output vs. input diagram, are placed on efficient frontiers (also called
envelope in data envelopment analysis). From an economic perspective, there is reason to

believe that among all the manufacturing plants, those plants that can best utilize the

inputs are candidates for the title of “WCM” and should be the plants against which the
lower performing plants may want to benchmark themselves.

Taxonomy of strategy is another technique that has been used for identifying the
“best” performing plants and is especially useful when a plant wishes to benchmark its
way to achieve world class status. Porter’s (1980) taxonomy of strategy into cost,
differentiation and focus served as the first argument that there are differences among the
best. He believed that only these strategies (cost, differentiation or focus) will lead to
superior performance, any firm “stuck in the middle” is doomed to low profitability. Yet
authors such as Kotha and Orne (1989) provided taxonomy that segmented
manufacturing strategy types into both attractive and unattractive positions. Despite the
argument that WCM plants can achieve multiple competencies at the same time and
equally well, most empirical evidence in taxonomy analysis suggested that there are
multiple strategy types even Within the best performing manufacturing units. Taxonomy
is crucial for benchmarking since strategy is the fundamental building block for tactical
and operational decisions. If a plant were to benchmark its practices (i.e. operational
decisions) on a plant with totally different strategy, the result can be disastrous.
Therefore, it is essential to identify strategy types within the most efficient manufacturing

units, be it WCM or not.

1.3. Research Objectives and Steps
There are two major objectives in this dissertation. First, a definition of World
Class Manufacturing that overcomes the weaknesses of both practice based and

performance based definitions, establishes linkage between practices and performance

and one that includes multiple criterion will be attempted. Second, an empirical
investigation of the relationship between manufacturing practices and performance will
be carried out in this dissertation. These two goals will be achieved by applying the
procedures discussed below.

Despite the popularity of the WCM topic and ongoing framework that is related to
this topic, there is a lack of theory in WCM literature. As stated by Bacharach (1989), the
primary goal of a theory is to answer the questions of how, when and why rather than to
answer the question of what. In more detailed terms, Bacharach (1989) described theory
as a system of constructs and variables in which the constructs are related to each other
by propositions and the variables are related to each other by hypotheses.

Both the practice—based definition and the performance-based definition are
basically descriptions, which are used to answer the question of what. They did not
answer the questions of how, when and why. Neither did they attempt to establish a
system of variables related to the constructs in discussion. Therefore, in order to establish
a system that leads to propositions and hypotheses, the first step is to establish a system
of measurement for the constructs.

The process of identifying World Class Manufacturing plants start with
identifying “best practice” plants among the peers. Efficiency analysis provides a way to
accomplish this. Efficiency analysis is a non-parametric approach which does not require
the functional form between input and output to be specified. The focus of such analysis
is to find out the “best practice” units that can produce the most amount of outputs with
the least amont of inputs. These characteristics make efficiency analysis an ideal tool for

performance analysis in this scenario. The critical decision in efficiency analysis is to

pick the correct set of inputs and outputs. A reasonable approach is to select a list of
inputs and outputs based on the combination of theoretical foundation from literature and
practical considerations. Afier having established the measurement models for practice
and performance, and selected the variables for efficiency analysis, the next step in this
dissertation was to apply this efficiency analysis and identify the frontier in the data
sample.

Taxonomy on the efficient plants may reveal the strategy differences of the most
efficient plants and provide insight for benchmarking effort. Are the plants that are
efficient in producing the most amount of output using the least amount of input
homogenous in terms of strategy or are there inherently different strategic groups among
the efficient plants? The answer to this question has significant implications for
operations strategy literature and makes significant difference for benchmarking effort. In
order to solve this puzzle and provide insights for more interesting topics in WCM
research, a cluster analysis on the efficient plants was performed. If more than one
strategic group is found in data samples, taxonomic analysis will be performed to study
the differences among these groups.

If there is more than one strategic group in efficient plants, one may contemplate
on whether these strategic groups focus on different sets of practices. If this is the case,
what implications do the differences reveal and what should manufacturing managers do
to adjust their actions according to these differences and make the best use of resources?

In summary, this dissertation will try to achieve the two major objectives by the

following procedures:

1. Establish practice and performance measurement models from literature and
empirically test these measurement models.

2. Provide an efficiency analysis of manufacturing plants in the data sample and
identify efficient plants on the data envelope.

3. Empirically investigate the relationship between manufacturing practices and
performance.

4. Provide taxonomy of efficient manufacturing plants.

5. If there is more than one strategic group in efficient manufacturing plants,

study whether these strategic groups emphasize different sets of practices.

1.4. Structure of the Dissertation

Chapter 1 of this dissertation outlines the gaps identified in literature related to
world class manufacturing, major objectives in this dissertation, and procedures that will
be followed to achieve these objectives. Chapter 2 offers a comprehensive review of
literature on several research streams that are related to taxonomy of manufacturing
strategy, competitive performance and world class manufacturing. Chapter 3 develops a
conceptual framework based on literature and the rationale for the research hypotheses.
Chapter 4 describes research design, data collection and data analysis methodology
related issues. Chapter 5 presents the results and validity issues of data analysis. Chapter
6 discusses the implication and contribution of the research results, points out the

limitation of this dissertation and directions for future research.

CHAPTER 2

LITERATURE REVIEW

Chapter 1 provided an overview of literature in the field of world class
manufacturing and theories that may contribute to a better understanding of WCM but
have been overlooked by researchers in WCM area. These areas together with WCM
literature will be examined in detail in this chapter. The existing literature will be
summarized, evaluated and critiqued. Gaps and limitations Will be identified.

Literature review will be organized around four major research streams:
taxonomy of manufacturing strategy, competitive performance, relationship between

practices and performance, and world class manufacturing.

2.1. Taxonomy of Manufacturing Strategy

Porter, in his most cited and debated work "Competitive Strategy", classified
strategy types (1980) into cost, differentiation and focus. His work has influenced a lot of
researchers in the field of manufacturing strategy.

Porter identified a two dimensional, strategic target by strategic advantage matrix.
Both the two dimensions are dichotomy, as strategic target being either industry wide or
particular segment only and strategic advantage being either uniqueness perceived by the
customer or low cost position. The two by two matrix yielded three strategy types:
industry wide low cost as cost strategy, industry Wide uniqueness as differentiation

strategy, particular segment only (either unique or low cost) as focus strategy. Porter

argued that firms "stuck in the middle" -- those that failed to develop their strategies in at
least one of the three directions, are almost guaranteed low profitability.

Porter's taxonomy has been criticized for a lot of reasons. For example, only using
two variables to segment strategies was viewed as too simplified and not comprehensive.
Even for the variables chosen, the dichotomy between uniqueness and low cost does not
seem to capture all the essence of strategic advantage.

Kotha and Orne (1989), upon realizing the criticism faced by Porter, proposed a
three dimensional cube for manufacturing structures. The three dimensions have been
labeled process structure complexity, product line complexity, and organizational scope.
The dichotomy on each of these dimensions is labeled low and high. High process
structure and low product line complexity corresponds with low cost position in Porter's
model. High product line complexity and low process complexity corresponds with
differentiation. High organizational scope corresponds with industry-wide, low
organizational scope corresponds with segmented. This three dimensional cube yielded
eight manufacturing strategies at the corners, among which four strategies which
corresponded with Porter's three generic strategies are identified as highly attractive, two
as reasonably attractive and two as highly unattractive. The authors argued that any other
strategies (represented by the remaining portions of the cube) are also highly unattractive.

Kotha and Orne's taxonomy purports to be an improvement to Porter's model in
the way that these three variables are now independent with each other. However, the
choice of categorical variable to characterize these variables is arbitrary and makes the

model less persuasive. Even the authors described some industries as having medium

10

complexity in these variables. It seems contradictory that the authors then characterized
these positions as highly unattractive.

Ward and Leong (1996)'s taxonomy took into account the linkage among four
constructs: competitive strategy dimensions, environmental dimensions, structural
dimensions, and strategic manufacturing capabilities. They identified sixteen item
measurement on these four construct. Unlike Kotha and Ome, who discussed all possible
combinations of low/high dimensions of the variables, Ward and Leong did not make an
effort to provide such a combination. They only identified four feasible strategies, each of
which has either high or low or insignificant relationship with each of the sixteen items.

Miller and Roth (1994) took a different approach from any of above studies.
Instead of classifying manufacturing strategy theoretically, they did an empirical
classification. They sampled 164 American manufacturers and found three clusters based
on competitive capabilities measures. They identified eleven competitive capabilities that
are relevant to their study: low price, design flexibility, volume flexibility, conformance,
performance, speed, dependability, after sale service, advertising, broad distribution and
broad line. Instead of using categorical scales, they utilized 1 to 7 Likert scales for these
items. The three clusters identified from the sample are caretakers, marketers, and
innovators.

Miller and Roth's approach improved upon the other taxonomies significantly in
several ways:

1. The taxonomy is aimed at manufacturing leading business units. Therefore the

clusters generally served similar purpose as Porter's: provide a benchmark for

the non-leading firms. Kotha et al. '5 taxonomy itself does not provide such

11

functionality. In Ward et al.'s taxonomy, the definition of leaders is based only
upon the authors' judgement.

2. Continuous scale is utilized. This is a step forward in the taxonomy research.

The dimensions are inherently continuous. Using categorical variable to
characterize these continuous variables as is done in other research is a
limitation.

3. It is empirically tested.

However, there are still some weaknesses in Miller and Roth's taxonomy. As
mentioned before, their strength is that their study examines leading manufacturing
business units. However, there was no explanation provided in their paper about what
leading manufacturing business units mean. Nor did they describe how they selected
those leading manufacturing business units. The sample size is small. Therefore, the
generalizability of their research is weak.

Bozarth and McDermott (1998) after reviewing the existing taxonomies in
manufacturing strategy research also called for the empirical testing of theoretical
frameworks.

The literature review of taxonomy of manufacturing strategy gives the direction
for the firture research:

1. A rigorous definition of "leading manufacturing business unit" is needed.

Otherwise the clusters generated from this sample may be biased by the

researchers' opinion as to what is good or bad.

12

2. The taxonomy on the leading manufacturing business units is needed. This
taxonomy, therefore, can provide benchmark for majority of the
manufacturing business units.

3. Performance measure has to be carefully chosen. If a variable is inherently
continuous, continuous measurement Should be chosen over categorical
measurement.

4. Empirical testing is needed to verify the theoretical framework.

2.2. Competitive Performance

Competitive performance is often employed in Operations literature to measure
the outcome of manufacturing strategy. It is some operations capability that is unique and
forms the primary basis for competition. (Anderson, Cleveland, and Schroeder, 1989,
Swink and Hegarty, 1998) It is a bridge that links a company's manufacturing
competence with its product and market competence (Hayes and Wheelwright 1979).
When used to describe the outcome of manufacturing strategy, it is often interchangeable
with competitive capability, comparative advantage, distinctive competence, production
competence etc. The same concept is sometimes used to describe the objective or goals of
a manufacturing unit. When used to describe objective or goals, it is often called
manufacturing objective, competitive priority. The difference between these terms is
largely semantic, although some authors do propose using two constructs to describe
objective and outcome (Vickery, 1991, Swink and Hegarty, 1998).

It has been viewed as a multi-dimensional construct from the time the concept

was formed. Its theoretical components are evolving over time. Its empirical realization

l3

varied from study to study. The trade-off vs. synergy View of its various dimensions is
becoming a hot topic. The following literature review on competitive performance is

divided into three parts: theoretical components, empirical realization, trade-off vs.

synergy.

2. 2. 1. Theoretical Components

Most writers have focused on four components of competitive performance: cost,
quality, dependability (delivery), flexibility (Hayes and Wheelwright 1984, Swamidass
and Newell 1987, Hill 1989, New 1992, Krajewski and Ritzman 1993, Schemnner 1993,
Boyer 1998). Recent literature has included reliability, customer satisfaction, innovation,
service etc (Nobel 1995, Garvin 1993). Some authors have tried to break some items
down into smaller sub—components.

The concept was first proposed by Hayes and Schmenner (1978) as competitive
priority. It was composed of five dimensions: price, quality, dependability, product
flexibility, and volume flexibility. They argued that manufacturing functions best when
its facilities, technology, and policies are consistent with recognized priorities.

Seven key competitive edge criteria was first presented in New’s (1979) Cranfield
Manufacturing Strategy Model and later modified by New and Sweeney (1984). These
seven key competitive edge criteria included delivery lead time, delivery reliability,
quality capability, quality consistency, flexibility design, flexibility volume and price.

Schroeder, Anderson and Cleveland (1986) asked managers to rate the importance
of manufacturing objectives on similar dimensions. Those items, from highly ranked to

lowly ranked, are quality, delivery performance, unit cost, flexibility to change volume,

14

 

 

 

I/‘t

flexibility to change product, employee relations, inventory turnover, and equipment
utilization.

Garvin (1993) in proposing strategic manufacturing initiatives (SMI), argued that
strategic priorities are too highly aggregated to direct decision making. He decomposed
five strategic priorities (cost, quality, delivery, flexibility, and service) into thirty-six
items. Garvin's list of decomposed strategic priorities is the one of the most

comprehensive lists in literature.

2. 2. 2. Measurement Issue

The empirical realization of competitive performance varied from study to study.
Generally speaking, empirical research included more detailed items to measure and
describe the construct. The actual items included often varied by the authors' interest of
study.

For example, Cleveland et al. (1989) offered a single, overall measure of
production competence based on individual firm’s performance in “nine key areas.
Strength or weakness in these key areas could mean the difference between success or
failure of the business plan”. These nine key areas are: adaptive manufacturing, cost-
effective of labor, delivery performance, logistics, production economies of scale,
production technology, quality performance, throughput and lead-time, vertical
integration. For six midwestem firms, a company representative first rank-ordered the
nine key areas according to their relative importance to the firm. Each respondent then

rated his or her firm’s strength or weakness (S/W) in each area using a (-1, 0, or +1)

15

scale. Logarithms of the reversed ranking values were multiplied by their respective S/W

ratings and summed together to derive the measure of production competence.

Table 2.1 Garvin's Decomposed Strategic Priorities

 

Strategic
Priorities

Decomposed Strategic Priorities

 

Cost

Initial cost

 

Operating cost

 

Maintenance cost

 

Quality

Performance

 

Features

 

Reliability

 

Conformance

 

Durability

 

Serviceability

 

Aesthetics

 

Perceived quality

 

Delivery

Accuracy

 

Completeness

 

Availability

 

Speed

 

Information Accessibility

 

Quality

 

Ease of Orderirbg

 

Ordering Flexibility

 

Shipment Flexibility

 

Ease of Return

 

Flexibility
(Product
Flexibility)

New Products

 

Customization

 

Modification

 

(Volume
Flexibility)

Uncertain Forecasts

 

Ramp-ups

 

(Process
Flexibility)

Mix Flexibility

 

Changeover Flexibility

 

Rerouting Flexibility

 

Material or Factor Flexibility

 

Sequencing Flexibility

 

 

Service

Customer Support

 

Sales Smort

 

Problem Solving

 

 

Information

 

16

 

 

Co:
mu

lIaC

Vickery et al. (1993) tested a refined measure of production competence, which
they define as “the degree to which manufacturing performance supports the strategic
objectives of the firm”. The new measure centered on 31 components of production
competence, such as delivery speed and process flexibility. In a study of 65 firms, a
representative from each firm evaluated the importance of each component to the
business strategy (7-point scale), assessed manufacturing’s relative performance with
regard to each component (--3 to +3), and estimated manufacturing responsibility for the
component (%). These three values were then multiplied together for each component
and summed to derive the production componence score. A stated objective of Vickery et
al. (1993) is to test a model linking production competence, business strategy, and the
interaction between these two constructs and business performance. As such, these
measures needed to be broad-based, generalizable, and easy to apply across large
samples.

Pannesi (1989) offered a detailed list of characteristics of every dimension of
manufacturing capability. His proposal was to pick several key characteristics for a
specific industry, then have managers to rate company performance on these key

characteristics.

2.2.3. T rade-ojf vs. Synergy

Skinner’s (1969) conceptual framework was the origin of the trade—off theory. He
contended that the facility can not be expected, or be required, to perform well on
multiple manufacturing tasks simultaneously, and that some manufacturing tasks must be

traded off for others. The task of manufacturing strategy is, therefore, to configure

l7

 

 

 

 

In

Ch

production systems to reflect the priorities and tradeoffs inherent in an organization's
competitive situation and strategy.

Porter (1980) applied the idea of tradeoffs among competitive priorities to his
taxonomy of general strategy. He divided competitive priorities into cost and
differentiation, which included quality, dependability, flexibility etc. He warned that
those firms attempting to simultaneously pursue both competitive priorities are firms
"stuck in the middle" and will be guaranteed low profitability.

Hayes and Wheelwright (1984) defined the competitive priorities as the ways in
which a firm chooses to compete in the marketplace and the types of markets it pursues.
They explicitly advise against the pursuit of multiple competitive priorities, stating,

"It is difficult (if not impossible), and potentially dangerous, for a company to try
to compete by offering superior performance along all of these dimensions
Simultaneously, since it will probably end up second best on each dimension to some
other company that devotes more of its resources to developing that competitive
advantage."

The work of more recent authors, including Hill (1989), Anderson et al. (1989),
Hayes and Pisano (1994), Miller and Roth (1994) continues to support the tradeoff
model. Most common examples of tradeoffs included cost versus quality and short
delivery times versus low inventory investments.

There have been increasing amount of research that questioned the trade-off
model in the field of operations management. There are two research streams that
challenged the tradeoff model. Some authors including Nakane, De Meyer, Ferdows,

Hall, and Nobel proposed or empirically tested a cumulative model that argued

18

manufacturers should build one capability upon another sequentially. The other authors
confronted the trade-off model conceptually, contending that the world class firms seem
to compete on the basis of quality, dependability, cost, delivery and flexibility, all at the
same time.

While studying Japanese manufacturers, Nakane (1986) suggested a cumulative
model with quality improvement as the basis of all other improvement, followed by
dependability. One should only improve on dependability if the quality level in the
company has reached a critical level. His sequence continued by asserting that quality
and dependability improvements are pre-conditions to cost efficiency improvements; cost
efficiency becomes almost a consequence of quality and dependability improvements.
Finally, flexibility improvements can only be obtained if a company has its quality,
dependability and cost efficiency under control.

Hall (1987) echoed Nakane in stating that manufacturers should follow a stepwise
progression through the capabilities in order to maintain a unity of purpose. The typical
goal progression according to Hall is (1) quality improvement, (2) dependability or lead
time, (3) cost reduction and (4) flexibility.

De Meyer et al. (1989) surveyed 574 Japanese, North American, and European
manufacturers. Their result suggested that Japanese manufacturers were focusing on the
trade-off between cost efficiency and flexibility. De Meyer et al. reasoned that Japanese
manufacturers could do so because they had already reached prerequisite levels of
quality, dependability, and cost efficiency sequentially over time. Their data showed that
North American manufacturers were focusing on quality, with flexibility being low on

their list of competitive priorities. Therefore they suggested that North American

19

manufacturers were still on a lower level of competitive priority sequence. The findings
for the European manufacturers were less clear.

Hall and Nakane (1990) expanded the preceding goal progression by adding new
goals as follows: (1) a company—developed culture, (2) quality, (3) dependability, (4)
waste reduction, (5) flexibility, and (6) innovation. They suggested that Japanese
manufacturers had begun their goal progression in the 19505 and 1960S by starting with
the quality capability, had progressed through the other high level goals, and by the mid-
19805 were concentrating on the flexibility capability. In contrast, the US. manufacturers
were focusing equally on quality and cost.

Ferdows and De Meyer (1990), in the light of increasing evidence of goal
progression, formally proposed a cumulative model, which they called "sand cone
model". They argued that many of the excellent manufacturers in Europe, North America,
and especially Japan followed a distinct sequence of improvement programs which aim at
building one capability upon, and not instead of, another. Their purpose was not to defy
trade-off theory completely. Instead, they suggested that the trade-off theory is not valid
under all contingencies and certain approaches change the trade-off into a cumulative
one. They suggested that the approach which avoids trade-offs and ensures cumulative
buildup of manufacturing capabilities in the long run is one which focuses on quality
first, then quality and dependability, then quality, dependability, and flexibility, and
finally on all three plus cost efficiency.

They used data from 1988 European Manufacturing Futures survey (167
respondents) to test and illustrate their model. Testing a sequential model on a static data

set is arguable and they only claimed to find partial support to their model. They found in

20

their data that 25% on respondents improved on only one capability, a large majority
(62%) of the respondents improved on more than one capability simultaneously. They
reasoned that trade-off theory would have predicted that the group of companies which
have improved only one of the four measures would be the most numerous, therefore
trade-off theory is refuted! Another support for cumulative model came from the
observation that the "better than average" performers in quality conformance, in on—time
delivery, in speed of new product development, and in inventory turnover have all
emphasized zero-defect programs significantly more than "worse-than-before" group.
They argued that this is the evidence that quality capability is at the bottom of the sand-
cone -- the other capabilities all build upon quality capability. They attributed the lack of
additional evidence to the use of a pre-existing database not designed for this particular
study.

Ferdows and De Meyer's testing of cumulative model is far less than rigorous.
First, the evidence of more than one capability is improved does not necessarily refute
trade-off theory since trade-off theory does not make the assumption that only one
capability can be improved over some period of time. Second, the association between
zero-defect program and other capabilities is hardly enough to prove that quality is at the
bottom of the sand-cone. They acknowledged this by saying that their purpose was not to
"prove" the cumulative model but to provide some evidence for it.

Nobel ( 1995) extended previous studies by statistically testing the cumulative
model in a multi-country context. She compared and contrasted the manufacturing
strategies of 265 North American, 129 European, and 167 Korean factories by region.

Similarly to F erdows and De Meyer's finding, her result Showed that competing on the

21

basis of multiple capabilities simultaneously is associated with higher performance and
better performing plants ofien compete with multiple capabilities-- usually with quality in
the group.

The other stream of conceptual research that confronted trade-off model included
Schonberger (1986), Roth and Miller (1990, 1992), Collins and Schmenner (1993),
Schmenner (1993). They argued that the trade-off theory has been outdated by world
class manufacturing.

New (1992) represented another opinion in this debate: he suggested that under
the impact of world class manufacturing model, traditional trade-offs have changed over
time: some disappeared, some still exists. For example, the trade-off between lead time
and delivery reliability can be eliminated by high process repeatability and low buffers.
However, the trade-off between quality and price still exists and may not be eliminated.
New believed that a firm can not be the best in the world at all seven criteria
simultaneously from the same manufacturing mix, which defies the belief of world class

manufacturing model and is one of the major points this study wants to address.

2. 3. World Class Manufacturing
As discussed in section 1, there have been two different trends in describing world
class manufacturing: practice-based and performance-based. A majority of literature has
been practice-based. The researchers have been engrossed in describing what a
manufacturer needs to do step by step in order to achieve world class status.
Hayes and Wheelwright (1984) were the first to use the term "world class

manufacturing". Through an in-depth analysis and comparison of the practices

22

implemented by Japanese, German and US. firm, they developed six practices that world
class manufacturers must adopt.

The first practice Hayes and Wheelwright proposed is to "build the Skills and
capabilities of your workforce". They found that German and Japanese firms do much
more ongoing training than their US. counterpart. They criticized the assumption made
by American manufacturers that motivation and work habits were developed elsewhere
or that they would develop naturally as the worker matured on the job. Therefore, training
must focus on both skills and attitude. They recommended that US. manufacturers take a
proactive stance on developing workforce skills and capabilities through apprenticeships,
internal training institute, and cooperative arrangements with vocational technical
institutes.

They also found that "build technical competence throughout management" is a
necessity for world class manufacturing and is where Japanese and German performed
better than Americans. They suggested providing technical training for managers,
developing more managers with engineering or technical degrees and rotating managers
through technical functions in their organizations.

Hayes and Wheelwright's third practice "competing through quality" recognized
that quality involves all parts of the organization as well as the customer's organization.
They emphasized the coordination among marketing, manufacturing, and design
engineering in achieving customer orientation.

In terms of "real worker participation", they went beyond recognizing the value of
teamwork and of close bonds between firnctions and organizational levels. They pointed

out that a long period of preparation and confidence building is required: workers and

23

managers have to develop a common vocabulary, knowledge base and set of
competences; production workers, engineers, and mangers have to be brought together so
that interaction is natural and unavoidable.

Hayes and Wheelwright's fifth practice, "rebuild manufacturing engineering",
describes the internal development of equipment with unique characteristics, which is
difficult for competitors to copy.

They proposed "incremental progress" in their last topic "tortoise and hare
approaches to industrial competition". While US. firms have traditionally pursued
strategic leaps (referred to as the hare approach) as a means of manufacturing
improvement, they suggest that the path of the race that leads to world class
manufacturing be by incremental improvement (referred to as the tortoise approach).
They also stressed the importance of measurement systems that can track this kind of
progress, compensation and promotion systems that encourage it and resource allocation
systems that support it for firms that adopt the incremental approach.

Other authors (Schonberger 1986, 1990, 1996, Giffi et a1. 1990) have been more
detailed in describing the attributes or practices of world class manufacturing. They
added more dimensions and did not fully agree with what Hayes and Wheelwright
proposed.

Schonberger (1996) proposed 16 principles for world class manufacturing. He
claims that these principles apply in nearly every case. The 16 principles fall into eight
categories. These 16 principles are:

General
1. Team up with customers; organize by families of customers or products (what

customers buy/use).

24

2. Capture and apply customer, competitive, and best-practice information

3. Dedicate to continual, rapid improvement in quality, response time, flexibility, and
value.

4. Frontline employees involved in change and strategic planning -- to achieve unified
purpose.

Design

5. Cut to the few best components, operation and suppliers.

Operations

6. Cut flow time, distance, start-up, and changeover times all along the chain of
customers.

7. Operate close to customers' rate of use or demand.

Human Resources

8. Continually enhance human resources through cross-training, job and career-path
rotation, and improvements in health, safety, and security.

9. Expand the variety of rewards, recognition, pay, and celebration -- to match the
expanded variety of employee contributions.

Quality and Process Improvement

10. Continually reduce variation and mishaps.

11. F rontline teams record and own process data at the workplace.

Information for Operations and Control

12. Control root causes of cost and performance, thereby reducing internal transactions
and reporting; simplify external communications.

13. Align performance measures with universal customer wants: quality, speed,
flexibility, and value (QSFV).

Capacity

14. Improve present equipment and human work before considering new equipment and
automation.

15. Seek simple, flexible, movable, low-cost, readily available equipment and work

facilities - in multiples, one for each product/customer family.

25

Promotion and Marketing
16. Promote, market, and sell your organization's increasing capability and competence --

every improvement (the results of the other fifteen principles).

Schonberger's l6 principles have slightly different focus on the same aspects from
Hayes and Wheelwright's definition of world class manufacturing. Schonberger has
incorporated more aspects and sometimes disagrees with Hayes and Wheelwright in
defining the key practices of world class manufacturing. While Hayes and Wheelwright
emphasize primarily on worker (skill and capability of workforce, real worker
participation), manager (management technology competence), customer (competing
through quality), and technology (rebuild manufacturing engineering, incremental
progress), Schonberger's l6 principles can be described in a more concise way as
customer focused, employee driven, and data (fact) based. Customer is Viewed as the core
of the principles (principles 1,2,7), supplier is brought into the picture (principle 5), how
to reward and get employee involved is detailed (principles 4, 8, 9, 11). Design (principle
5), operations (principle 6, 7), performance measurement and reward systems (principle
9, l3), marketing (principle 16) which were ignored by Hayes and Wheelwright are key
aspects perceived by Schonberger. In some aspect, they even disagree: while Hayes and
Wheelwright suggest proprietary equipment is the key to lasting capability, Schonberger
proposes improving present equipment and seeking simple, readily available equipment.

Giffi, Roth and Sea] (1990) have taken a similar approach to Schonberger's. They
described world class manufacturing in terms of principles in the following way: quality
and customer dimension is the core of the principles, the other dimensions such as

management approach, manufacturing strategy, manufacturing capabilities, performance

26

measurement, human assets, organization and technology aspects all interact with quality
and customer dimension. Giffi et al. were the first to add manufacturing strategy into the
framework. They also provided some insight about trade-off vs. synergy about
manufacturing capabilities. They argued that achieving multiple capabilities (quality,
dependability, cost and flexibility) simultaneously is one of the key success factors for
manufacturing excellence. On the other hand, they stressed that quality is current measure
of success. They predict that time-based competitive capabilities (such as dependable
delivery time, delivery speed, and flexibility) will be the key success factor for the future.
They also emphasized that it does not mean quality will no longer be important, but
superior quality will be assumed. This line of thinking echoes Ferdows and De Meyer
(l990)'s cumulative model on building competitive capabilities. Schmenner (1993) also
found agreement with Giffi et al. in that world class manufacturing outdated tradeoff

theory.

27

2. 4. Relationship between Practices and Performance

The relationship between practices and performance has been attracting a lot of
attention in manufacturing strategy literature. There are two main reasons why the practice-
perforrnance relationship has been the focus of attention for some time. First, the wide-
spread use of benchmarking (copying the best practices) to achieve competitive advantage
implicitly assumes a strong relationship between manufacturing practices and performance.
Second, unrelenting pressure on firms from stakeholders to increase asset productivity has
been forcing companies to carefully evaluate and justify investments in manufacturing
practices. Despite this topic's practical significance and importance, there has not been a
great deal of literature that has examined the relationship between practice and performance
comprehensively.

Manufacturing practices typically include shop floor practices (e. g. application of
Total Quality Management), human resource management practices, supply chain
management practices, to name a few. The definition of manufacturing performance
typically varies based on the objective of the individual study.

Studies in the literature have mostly focused on the relationship between a subset of
practice with a subset of performance.

For example, Flynn et al. (1995, a) investigated the relationship between quality
management practices to quality performance. Flynn et al. (1995, b) confirmed that the use
of unique JIT practices was positively related to JIT performance, given use of common
infrastructure practices. Flynn et al. (1995, b) also verified that the use of common

infrastructure practices was positively related to JIT performance, and the use of unique

28

TQM practices was positively related to JIT performance, given use of unique JIT and
common infrastructure practices.

The relationship between some specific practices and composite manufacturing
performance also provoked researcher's great interest. Young (1992) has proposed direct
relationship between performance and some aspects of practice. These aspects include rate
of adoption of JIT, high quality delivery, the rate of adoption of a secondary control system
etc. Yet these relationships remain to be examined empirically. Sakakibara et al. (1997)
studied the impact of JIT and its infrastructure on manufacturing performance and suggested
that JIT practices have value only when they are used to build infrastructure.

Voss et al. (1995) and Collins et al. (1996) have asserted that the adoption of best
practices leads to high performance. Both authors found this relationship to hold at the
aggregate level through an examination of scatter plots of data relating to a performance
index versus an aggregate index of manufacturing practices. These two studies represent the
first two studies to systematically examine the relationship between manufacturing practices
and performance and they have defined a new area for research Within the rubric of
manufacturing strategy. Beaumont et al. (1997) found that the use of AMT which includes
direct technologies (NC and laser cutting), indirect technologies (CAD and MRP), and
philosophies (JIT and TQM) in general is not associated with manufacturing or business
success, which is an apparent contradiction in light of the conclusions reached by Voss et
al., and Collins et al. However, the authors also found that specific technologies are

associated with some aspects of performance.

29

The review of literature in the relationship between practice and performance
reveals that an empirical testing between the relationship of a comprehensive set of
manufacturing practices and a comprehensive set of manufacturing performance was
lacking. The link between specific manufacturing practices and specific manufacturing
performance was not well understood. The relationship between quality and cost is a
good example. Increasingly, quality is now recognized as a major contributor to cost
reduction. Thus, in a cost competitive environment, quality programs may be the most
appropriate response rather than cost reduction programs. These specific relationships are
of great interest to operations managers. Decisions on adoption of specific practices are
usually made afier manufacturing strategies have been made. Managers are especially

interested in deciding which practices have an effect on their manufacturing goals.

2. 5. Summary of Literature Review

The review of literature in the area of world class manufacturing, relationship
between practices and performance, taxonomy of manufacturing strategy, and
competitive performance revealed several issues.

First, literature in world class manufacturing has been going into two directions:
practice-based and performance-based. Despite recognizing that “best practices” lead to
superior performance, there was no systematic approach to study this relationship in the
context of world class manufacturing.

Second, measurement issue in world class manufacturing, taxonomy of
manufacturing strategy, and competitive performance needs to be solved. Empirical

studies in these fields will help advance theory in these areas.

30

Third, the knowledge bases of taxonomy, relationship between practice and
performance, competitive performance and world class manufacturing are intertwined. It
is not surprising that the advance of one field will help to solve the questions in another
field. The trade-off theory VS. cumulative model is one example. It originated as a debate
in the area of competitive performance. The experiences coming from world class
manufacturing offered a great deal of explanation in this debate.

There is increasing number of research pieces that utilized the methodology and
knowledge fiom multiple areas in strategy literature. It can be argued that taxonomy of
manufacturing strategy, relationship between practices and performance, and competitive
performance are all relevant topics within world class manufacturing. An effort to apply
the techniques and theory from these areas to the context of world class manufacturing
will lead to the advance of knowledge base and a better understanding of strategy in
world class manufacturing and manufacturing in general. A framework that integrates

knowledge bases from multiple areas will be presented in chapter 3.

31

CHAPTER 3

CONCEPTUAL FRAMEWORK

Chapter 2 identified the limitations and gaps in several research streams that are
related to world class manufacturing. The gaps and limitations in literature helped
identifying research questions in this dissertation. The review of literature, in return, also
offered the potential solutions to these questions. Chapter 3 will develop a conceptual
framework based on literature review, propose research hypotheses and provide rationale
for research hypotheses.

3.1. Conceptual Framework

This dissertation aims to provide a definition of world class manufacturing and an
empirical investigation of the relationship between manufacturing practices and
performance.

The relationship between practices and performance has been assumed by
practice-based definition of WCM and many other research pieces in manufacturing
strategy. However, it has not been studied in a systematic way. To study the relationship
between practices and performance, the first and foremost task is to establish
measurement models for practice and performance.

In order to do so, efforts were made to synthesize the dimensions of practice and
performance from the most classical literature on WCM. Nine practice constructs were
identified based on Hayes and Wheelwright (1984), Schonberger (1990, 1996), Giffi,

Roth and Sea] (1990).

32

The practice profile is listed in Table 3.1. The descriptions on these dimensions
from previous researchers are also listed. A symbol (a), (b), or (c) for each practice
construct indicates the literature where the description comes from. (a) indicates it comes
from Hayes and Wheelwright (1984), (b) indicates it comes from Schonberger (1990,

1996), (c) indicates it comes from Giffi, Roth and Seal (1990).

Table 3.1 Practice Profile with Literature Support

 

1. Quality Management

(a) Competing though quality

(b) Principle 3: Dedicate to continual, rapid improvement in quality, response time,
flexibility, and value.
Principle 10: Continually reduce variation and mishaps.
Principle 11: Frontline teams record and own process data at the workplace.

(c) Define quality in terms of the customers' needs. Make customer closeness the number
one priority.
View quality fiom a global perspective. Achieving quality of product should be no

more and no less important than achieving quality of process and service.

2. Customer Orientation

(a) Principle 1. Team up with customers; organize by families of customers or products
(what customers buy/use).

Principle 2: Capture and apply customer, competitive, and best practice information
Principle 7: Operate close to customer's rule of use or demand.

(b) Integrate the concept of customer closeness into the organization so that everyone in
the organization has a customer, and everyone's goal is to provide quality product and
service to his or her customer.

(c) Define quality in terms of the customers' needs. Make customer closeness the number

one priority.

33

3. Labor Force Management

(a) Workforce skills and capabilities
Workforce participation

(b) Principle 4: Frontline employees involved in change and strategic planning -- to
achieve unified purpose.
Principle 11: Frontline teams record and own process data at the workplace
Principle 8: Continually enhance human resources through cross-training, job and
career-path rotation and improvements in health, safety and security.

(0) Invest in people; develop a pattern for updating workforce skills and capabilities
consistent with the evolution of technology within the organization.
Empower teams of workers to carry out the mission of the organization. Seek ways to
liberate the teams from traditional organizational controls, and reward and motivate,
based upon ability to achieve meaningful goals.
Eliminate the terms supervisors and supervision. Develop leaders who can create and
execute the strategic vision through the teams.
Evaluate the success of your human assets on the basis of their ability to learn, adopt
to change, and improve performance within their area of responsibility.

Develop accelerated and integrative learning programs.

4. Operations Flow

(a) Not mentioned

(b) Principle 5: Cut to the few best components, operations, and suppliers
Principle 6: Cut flow time, distance, start-up, and changeover times all along the
chain of customers.
Principle 7: Operate close to customers' rate of use or demand

(c) Make dependability and consistency in quality, delivery, and service to your
customers the goal of all operations.
Develop manufacturing operations that are flexible and able to respond rapidly to
changes in products and markets. Value leanness of operation above all other

manufacturing performance indicators.

34

Restructure engineering operations to reduce waste and inefficiency and improve
quality. Develop engineering capabilities characterized by adherence to concurrent
development techniques that integrate customer quality requirements.

Consider the environmental impact of all products and processes, making
environmental considerations an integral component of the design and production

pI'OCCSSCS.

5. Operations Capability

(a) Incremental improvement approach

(b) Principle 3: Dedicate to continual, rapid improvement in quality, response time,
flexibility and value
Principle 5: Cut to the few best components, operations and suppliers
Principle 10: Continually reduce variations and mishaps

(c) Develop manufacturing operations that are flexible and able to respond rapidly to
changes in products and markets.

Develop measurement systems that encourage continual learning.

6. Strategy Formulation and Measurement

(a) not mentioned

(b) Principle 9: Expand the variety of rewards, recognition, pay and celebration -- to
match the expanded variety of employee contributions.

(0) Establish a clearly defined strategic intent; define success in terms of winning for the
long term.
Establish a strategy consistent with the potential for developing needed
manufacturing capabilities. Base the strategy on a realistic assessment of capabilities
and priorities, but always look to push beyond the present.
Develop a global perspective on competition, responding to international competition
with at least the same intensity as to domestic challenges.
Make the strategy more than a formal statement of policy. Make it a blueprint for
action, a pattern of decision to be executed over time. Deal with structural elements,

infrastructural issues, and integration elements.

35

Put the strategy in writing. Manufacturers that have written strategies have been more
forceful in the change process than those that have not written strategy.

Develop the strategy through a participative approach and freely share it with all
employees in the organization.

Develop a strategy that is flexible and adaptive over time as the competitive
enviromnent changes. Review the strategy on a scheduled, periodic basis to ensure
congruence with current and future goals and capabilities. Make modifications and
enhancements as required.

Implement the strategy to effect change as a gradual process of small continuous
steps over time. Allow your strategic intentions to govern the size of your steps.
Focus on the competitive variables the customer requires. Allow customer success to
drive financial success.

Promote and measure knowledge and skill development. Develop measurement
systems that encourage continual learning.

Increase the Vitality of the entire business by focusing attention on integrated business
management. Measure the effectiveness of the boundary management that occurs.
Tailor the performance measurement system to the company's strategic action
programs. Manufacturing strategy is defined explicitly in terms of performance

measurement. As strategy changes, so too should the critical measures of success.

7. Technology

(a) Management technical competence, Rebuilding manufacturing engineering

(a) Principle 14: Improve present equipment and human work before considering new
equipment and automation

(b) Principle 15: Seek simple, flexible, movable, low-cost, readily available equipment
and work facilities -- in multiples, one for each production / customer family.
Develop an investment for the continual enhancement of technology throughout the
organization, based on a clearly defined vision of future competitive requirements.
Identify the competitive advantage of the knowledge base that advanced technology
can create; simultaneously implement new technology and develop the new

knowledge base.

36

Carefully plan technological upgrades to be consistent with infrastructural upgrades,
Benefits can be achieved only when the infrastructure is capable of integrating and

exploiting the technology advantage offered.

8. Supplier Relationship
(a) Not mentioned
(b) Principle 5: Cut to the few best components, operations, and suppliers.

(0) Not mentioned

9. New Product Development

(a) Not mentioned

(b) Principle 1: Team up with customers; organize by families of customers or products.
Principle 2: Capture and apply customer, competitive, and best practice information
Principle 3: Dedicate to continual, rapid improvement in quality, response time,
flexibility and value.
Principle 5: Cut to the few best components, operations and suppliers.

(0) Not mentioned

 

It is Shown from Table 3.1 that the first seven practice dimensions in this practice
profile received greater attention from literature. The last two dimensions: supplier
relationship and new product development are relatively new and received less attention.
These two dimensions, however, are perceived to impact the manufacturing capability by
many recent researchers. Their impact on performance is somewhat less determined.
They are added into practice profile due to the consideration that the three classic works
on WCM practices are somewhat less recent and may not be complete in including all the
dimensions that emerged in recent practices. It is the goal of this dissertation to either

confirm or disconfinn the relationship between these practices with performance.

37

CO
St“.

rh-i

 

 

The performance profile measures will take the most conventional constructs:
cost, quality, delivery and flexibility. The detailed breakdown for each construct is
selected and modified from Garvin's decomposed strategic priorities. Table 3.2 lists all

the items in performance profile.

Table 3.2 Performance Profile

 

1. Cost

(1) initial purchase cost

(2) manufacturing overhead cost

(3) maintenance cost

2. Quality

(1) Product overall quality performance

(2) Product feature

(3) Product reliability

(4) Product conformance

(5) product durability

(6) Product serviceability (speed, courtesy and competence of repair)

(7) Customer satisfaction

(8) Impact of brand name

3. Delivery

(1) Delivery accuracy (correct items were delivered)

(2) Delivery availability (the probability that items will be in stock at the time of order)

(3) Delivery dependability (delivered on the agreed upon date)

(4) Delivery speed (Short elapsed time between order placement and product reaches the
customer)

(5) Delivery quality (condition of product after shipment)

4. Flexibility

(1) Ability to customize products

(2) Lead time to introduce new products

38

 

 

fur
de:
set
is t

ach

 

 

(3) Ability to adjust production volumes

(4) Number of engineering change orders per year

(5) Number of new products introduced each year

(6) Ability to respond to changes in delivery requirements

(7) Ability to produce a range of products

 

In achieving the two objectives of this dissertation, not only the methods and theories
from world class manufacturing literature are studied, but also theories and
methodologies fi'om related areas. For example, efficiency analysis is one method that
has been used for performance analysis and identifying “best practice” units in various
domains such as health care systems (Sherman 1984, Gerdtham, Rehnberg, and Tarnbour
1999), retail stores (Thomas et al. 1998), banks (Sherman and Gold 1985, Thomas and
Barr 1998), education systems (Bessent and Bessent 1980, McMillan and Datta 1998),
and manufacturing (Banker et al. 1984, Bymes et al. 1988 and Vargas et al. 1996).

In efficiency analysis, more specifically, data envelopment analysis (DEA), no
firnctional forms linking outputs and inputs have to be assumed. Efficiency can be
defined as the weighted sum of unit i’s outputs over weighted sum of unit i’s inputs. A
separate linear programming problem is solved for each unit. The objective for each unit
is to maximize the weighted sum of that unit’s output versus its inputs. Those units that
achieve an efficiency of 1 are those units that can produce the most amount of outputs
with the least amount of inputs compared with their peers.

Data envelopment analysis, due to its non-parametric property, requires no prior
knowledge or theory relating to the relationship between inputs and outputs, therefore,
this method is not biased toward any existing theory. It has also been used extensively as

39

a popular tool to identify the “best practices” production function. In this study, an
identification of best performing manufacturing plants can be very well firlfilled by
applying this type of analysis.

Efficient manufacturing plants will be the ones that can best utilize inputs and
produce outputs. This criterion will be considered as a pre-condition for world class
manufacturing.

The second objective of this dissertation is to systematically study the relationship
between the so-called “WCM practices” and performance. The positive relationship
between “WCM practices” and performance has been assumed in WCM literature, but
has never been empirically tested and verified. In this dissertation, not only will the
relationship between WCM practices and performance he studied, but also the
relationship will be used to verify the validity of efficiency analysis. Efficiency analysis
is a quite new technique in operations management although it has been used extensively
in other fields. One way to verify the validity of efficiency is to cross check efficiency
with the strength of the relationship between practices and performance. Efficient plants
are best performing plants that can best utilize tangible inputs to produce tangible
outputs. It can be argued that efficient plants can also utilize manufacturing practices
better than non-efficient plants. That is, the strength of the relationship between practices

and performance are stronger for efficient plant than for non-efficient plants.

40

L -9'. .“l

Figure 3.1 Efficiency as a Moderator between WCM Practices and Performance

 

 

 

Efficiency

 

+

 

 

WCM 4 WCM

 

 

 

 

 

 

 

The purpose of this dissertation is to examine the relationship between practice and
performance from a broad perspective. It is hypothesized, that adopting these practices will
lead to superior performance, similar to arguments of the "best practice" literature and
practice-based definition of WCM. Figure 3.2 depicts the framework in this dissertation.
In this figure, practice profile and performance profile are the two vital components.
However, this framework differs from prior knowledge in that the relationship between

practice and performance is moderated by efficiency.

41

Figure 3.2 Conceptual Framework for World Class Manufacturing

Practice Profile

 

 

 
  

  
   
 
 
   
  
    

Quality
Management

  

Customer
Orientation

  

Labor Force
Management

Operations

Flow

  
   
 
  

  

Operations
Capability

 
    

Supplier
Relationship

Technology

 
  

New Product
Development

 
   
   

Strategy and
Measurement

 
   

Performance Profile

 

 

 

efficiency

 

 

 

 

 

42

a

 

 

 

(‘V
d1

1'
U i

If
(I).

H 1

H;

H;

pt’;

 

 

3. 2. Hypotheses Development

To test the moderating effect of efficiency on the relationship between practices
and performance, hypothesis 1 is constructed with nine separate sub-hypothesis each for
testing for one practice dimension.
Hypothesis 1a: Efficiency is a positive moderator between Quality Management
dimension and performance.
Hypothesis 1b: Efliciency is a positive moderator between Customer Orientation
dimension and performance.
Hypothesis 1c: Efliciency is a positive moderator between Labor Force Management
dimension and performance.
Hypothesis 1d: Efi‘iciency is a positive moderator between Operations Flow dimension
and performance.
Hypothesis 1e: Efliciency is a positive moderator between Operations Capacity
dimension and performance.
Hypothesis If: Efliciency is a positive moderator between Supplier Relationship
Dimension and pejormance.
Hypothesis 1g: Efliciency is a positive moderator between Technology dimension and
performance.
Hypothesis 1h: Efficiency is a positive moderator between New Product Development
and performance.
Hypothesis 1i: Efficiency is a positive moderator between Strategy and Measurement

dimension and performance.

43

Literature from taxonomy of manufacturing strategy suggested that there are
multiple strategy types even with the best manufacturing plants. By applying the same
logic, it is reasonable to assume that there is more than one strategic group in efficient
manufacturing plants. Competitive priorities are generally viewed as the linkage between
market/consumer and manufacturing strategy (Porter 1980, New 1992, Todd 1995). Since
market is generally segmented by different groups of customers disregarding the specific
product or industry, efficient manufacturing plants have adopted different strategies that
best serve the diverse market.

This leads to hypothesis 2 which is constructed to test whether there are more than
one strategy type within efficient manufacturing plants.

Hypothesis 2: There is more than one strategic group within the efficient manufacturing
plants.

The nine practice dimensions in the conceptual framework are summarized based on
literature of practice-based definition of world class manufacturing. One of the
weaknesses in such definition is that it does not address the priority in selecting these
practices. A plant planning to benchmark the best performing plant, however, would very
much want to find out, on which practices it should place its efforts since resources are
limited. This dissertation tries to overcome this limitation by investigating how the
priorities differ across the practice dimensions among different strategic groups.

Hypothesis 3 is framed in line with prior expectation that these priorities differ
across different strategic groups. Nine sub-hypothesis are constructed one for each

practice dimension.

44

Hypothesis 3a: Different strategic groups within efficient manufacturing plants place
different priorities on Quality Management dimension.
Hypothesis 3b: Different strategic groups within efficient manufacturing plants have
difi'erent priorities on Customer Orientation dimension
Hypothesis 3c: Different strategic groups within efficient manufacturing plants have
difl'erent priorities on Labor Force Management dimension.
Hypothesis 3d: Different strategic groups within efficient manufacturing plants have
different priorities on Operations Flow dimension
Hypothesis 3e: Different strategic groups within efficient manufacturing plant have
diflerent priorities on Operations Capability dimension.
Hypothesis 3f: Diflerent strategic groups within efi‘icient manufacturing plants have
different priorities on Supplier Relationship dimension.
Hypothesis 3g: Dijferent strategic groups within efficient manufacturing plants have
difi’erent priorities on Technology dimension.
Hypothesis 3h: Diflerent strategic groups within efficient manufacturing plants have
difi'erent priorities on New Product Development dimension.
Hypothesis 3i: Diflerent strategic groups within eflicient manufacturing plants have
different priorities on Strategy and Measurement dimension.

This chapter presented a conceptual framework that integrated theory and
methodology from several different areas. This framework resolves around and facilities
a better understanding of world class manufacturing. Individual hypotheses were

development to test this framework and achieve the objectives of this dissertation.

45

Chapter 4 describes research design, sampling frame, and the methodology that will be

used to test hypotheses and the conceptual framework.

46

CHAPTER 4

RESEARCH DESIGN

4.1. Research Design

As stated in chapter 1, there are two major objectives that this dissertation intends
to achieve by following five procedures. In this chapter, five stages of statistical testing
are designed one for each procedure.

The first stage of this dissertation is to establish a set of measures for WCM
practices. Previous studies (Hayes and Wheelwright 1984, Schonberger (1990, 1996),
Giffi, Roth and Sea] (1990)) are descriptive in nature. None of them provided concrete
measurement items, not to mention validation of the measures. The second stage is to
provide efficient analysis and select efficient plants in data sample. The third stage is to
investigate the moderating effect of efficiency on the relationship between manufacturing
practices and performance. The fourth stage is to provide an empirical taxonomy on
efficient manufacturing plants in data sample. The fifth stage is to test whether strategic
groups emphasize different set of practices if there is more than one strategic group in
efficient manufacturing plants.

For the above purposes of this dissertation, a large scale empirical testing will be
useful because standardized measures, which are a necessity for making comparisons,
can be used across a broad population in order to make generalizable conclusions (Fowler
1988). A survey will be used since very few published empirical studies have undertaken
broader scale investigations to empirically test hypotheses associated with WCM.

DeVellis (1991) recommended a sequence of action when building scales:

47

First, clarify and specify what is sought to be measured, and generate an item pool.
Next, choose the format of measurement, have initial item pool reviewed by experts, and
consider inclusion of validation items. Then administer items to a development sample,
evaluate the items. Lastly, optimize scale length.

Scale development procedure in this dissertation will follow the above guidelines.

4.1.1. Scale Validation

Except for control variables, all practices and performance item measures will be
measured on a 7 point Likert scale. The respondents are asked to compare their plants
with industry average, with 1 indicating "much less" (than industry average), 4 indicating
"about the same" (as industry average) and 7 indicating "to a much greater extent" (than
industry average). The Likert scale has been preferred over other scales for several
reasons. Alternative methods do not offer specific response categories. For example, in
semantic differential scaling method, there are no interval points along the two extremes.
The respondents' position on the continuum between the two extreme is subject to mis-
interpretation by both respondents and researchers. In contrast, there are plenty of
supports in literature for the assumption of numeric, equal intervals in a Likert scale, for
the purpose of multivariate data analysis (DeVellis 1991, Nuunally 1978, Kerlinger
1973). Therefore, Likert scale has been commonly used in operations strategy literature
due to its established performance.

Initial face validation for the items was achieved through discussions with
practitioners and academics. A Q-sort of the scales was undertaken through a blind
administration of the item measures to a group of 3 practitioners and 3 academics. The

subjects were requested to examine the item measure and relate them to their respective

48

underlying constructs. Those items with a majority of incorrect linkages to underlying
constructs were dropped from the list of measures.

The next step of validation of the scales was undertaken by administrating the
instrument to the study sample and employing confirmatory factor analysis for evaluating

reliability and validity.

4.1.2. Unit of Analysis and Sample Selection

The unit of analysis in this dissertation was individual plant due to the
consideration that plant is the level of implementation for most manufacturing practices
programs, and has been used in numerous other empirical studies related to competitive
priorities (Schonberger 1983, Griffin 1988, Ebrahimpour and Withers 1992, Ahire,
Golhar, and Waller 1996). Therefore, the use of plant as the unit of analysis is strongly
supported by previous research.

It is noted that information gathered from multiple respondents at a single site is
desired in reducing the potential for bias from a single respondent (Klassen 1995).
However, the cost and time associated with obtaining multiple responses from a large
number of plants would be prohibitive in this research. Response rate will very likely be
significantly reduced with such a strategy. Miller and Roth (1994) suggested that greater
attention to informant selection can help to overcome the common method variance
problem. Therefore, only single respondents will be targeted for this dissertation. Plant
manager is considered the level of management that is most suited to complete this

questionnaire, is therefore selected as the respondent in this research.

49

4.1.3. Data Collection

The data was collected by utilizing Internet Technology: web technology and
database. The respondents were not required to mail the questionnaire back. Instead, a
copy of the questionnaire was posted on the Internet. The respondents were given a
survey ID when they received the invitation letter and they were required to go to the
specified web site to fill out the questionnaire. When the respondent first visited the web
site, he/she was asked to provide the survey ID (in order to keep track of respondents and
send follow up notice to the non-respondents) and select login name and password. With
login name and password, they were not required to fill out the questionnaire in one
session. They were able to save the answers by submitting the questionnaire. Upon
submitting the questionnaire, the respondent was notified how many questions he/she has
already filled out. When the respondent logged on using login name and password next
time, only the unanswered questions were presented to him/her. This design prevents the
respondents from modifying their answers once submitted and helps the respondents keep
track of unanswered questions. If the respondent wanted to keep a copy of what he/she
answered for the questions, he/she could select the option of "View survey" and print out
his/her own answers for each survey question.

This technique had several advantages over the traditional mail based surveys
03i11rnan 2000). The primary reason for adopting this technique was cost consideration.
By utilizing the web survey technique, the printing cost was totally avoided. The cost of
mailing was significantly reduced due to a smaller package. Time is also a big issue that
favors the decision to go on web survey. With the Internet communications, the responses

were able to reach the researcher in a few seconds rather than weeks via mail. With

50

database performing data storing and retrieving functionality, there was no need to
manually enter data into the spreadsheet from responses as needed by the traditional mail
survey. Not only was time and cost for data entry reduced, but there was increased
accuracy. The specific design of this web survey, which allowed respondent to return to
the web site and fill out in another session, added a lot of flexibility. First, fatigue with
filling out a long survey was somewhat lessened, therefore, response rate may have
increased. Second, by constantly informing the respondents of the number of questions
they have completed, the chances of unintentional missing field were reduced. Integrity
checking was added in database design to prevent the respondent entering obviously
invalid responses. For example, when asked for sales, we would expect the responses to
be in the format of numbers instead of letters. If letters were received, an error message
was sent to the respondent and the respondent was prompted to reenter the field. By
adding error checking into the database, a higher level of data integrity that is not
possible with the mail based survey was achieved.

Despite so many advantages with this new technique, it is not perfect. The biggest
concern the researchers are faced with web based survey is that the response rate may be
affected by managers' inexperience with Internet Technology. The senior managers, the
target of this research, are often in their fifties or sixties. These managers are generally
less familiar with the Internet than younger generations. They may feel uncomfortable or
reluctant to read and answer questions on the computer and using the mouse clicks. They
may prefer reading and filling out questionnaires old-fashioned way.

Although the questionnaires were to be filled out on the Internet, the managers

were contacted by mail because it was believed the mailing material would generate more

51

awareness than email. The mailing procedure employed a three-step process similar to
that recommended by Dilhnan (1979, 2000). Each letter was mailed directly to the plant
manager. The first invitation letter was sent out together with the web address where the
survey is hosted. A reminder postcard with the web-address of the questionnaire was sent
out to those plants not yet responding two week following the first mailing. Finally, four

weeks after the first mailing, another reminder postcard was sent to the non-respondents.

4. 2. Methodology
To follow the five procedures of data analysis in this dissertation, there will be
five different research methodology one for each procedure. This section details how

these five procedures will be implemented, and how to test for validity of each procedure.

4. 2.1. Establish Measurement Models

During the first stage of data analysis, the WCM practices and performance
measurement models was tested using confirmatory factor analysis. Confirmatory factor
analysis overcomes some inherent limitations associated with exploratory factor analysis
(Ahire, Golhar and Waller 1996). Exploratory factor analysis often falters when faced
with interpretability issues (Mulaik 1972). Lack of a-priori knowledge about construct
covariation dynamics may lead to the result that the interpretations of the factors may be
nothing more than tautological transformations of the names of the original variables.
Confirmatory factor analysis, on the other hand, builds factor constraints based on the
number of factors, the nature of their mutual relationships and the strength of those

relationships. Confirmatory factor analysis has gained increasing popularity over

52

exploratory factor analysis in the fields of marketing and organizational behavior due to
its conceptual strengths (Venkatraman 1989, Bollen 1989).

In this research, separate measurement models for WCM practices and
performance using CF A based on literature were constructed and tested. Data were first
analyzed for univariate and multivariate normality by examining Mardia's coefficient in
EQS. Reliability can be assessed using Cronbach's alpha or composite alpha. Convergent
validity was evaluated by the standardized factor loading for constructs, using a cut-off
value of 0.40 (Hair, Anderson, T atham, Black 1994). Discriminant validity can be
assessed by two methods. First, all inter-factor correlations for all the constructs needed
to be significant different from 1 (Challagalla and Shervani, 1996). Second, the average
variance extracted for each construct was computed and verified to ensure that it was
greater than the squared correlation between that construct and any other construct in the
model (Fomell and Larcker 1981).

After eliminating items that load on multiple constructs or have low item-to-
construct loadings, fit indices were used to assess if the measurement model is a good fit
in the data.

Absolute fit concerns the degree to which the covariance implied by the fixed and
free parameters specified in the model match the observed covariance from which the
free parameters in the model were estimated. Four of the most basic measures of absolute
fit are the likelihood Chi-square ()8), the goodness-of-fit index (GFI), the root mean
square residual (RMR) and root mean squared error of approximation (RMSEA). Chi-
square can be used to assess the discrepancy between a hypothesized model and data. x2

test become more sensitive as the number of indicators rise. Bentler and Bonnet (1980)

53

and Bentler 1990 indicated that t statistics may not be x2 distributed in small sample.
Small samples usually have a much greater tolerance towards specification errors than
large samples. Specifically, the x2 statistics will almost always be significant in large
samples (Hartwick and Barki, 1994). GFI is a goodness-of-fit measure, which is not
adjusted for model parsimony. The recommended level for an acceptable GFI is .90 or
higher. . The root mean square residual (RMR) indicates the average residual correlation,
usually deemed acceptable with a value of .080 or lower. Root mean squared error of
approximation (RMSEA) assesses model discrepancy per degree of freedom. Values of
the RMSEA less than or equal to .080 represent a reasonable fit (Hair et al. 1994).

Incremental fit measures assess the incremental fit of the model compared to a
null model, usually the one that specifies no covariances among variables. The
recommended value for both an acceptable IFI (incremental fit index) and an acceptable
CFI (comparative fit index) is .90 or higher. CFI is relatively unaffected by sample size
(Bentler 1990) and was recommended by Bentler (1990) as the index of choice. Bentler-
Bonnet Non Normed Fit Index (NNF I) is also an incremental fit index has no substantial
association with sample size (Marsh et al. 1988).

Parsimonious fit measures provide a basis for comparison between models of
differing complexity and objectives. One applicable measure for evaluating a single

model is the normed x2 measure, which is x2 divided by the degree of freedom. The

normed x2 with a value below 2 is considered acceptable (Bentler 1990).
Individual estimates of free parameters are evaluated according to their difference

fi'om zero. The ratio of each estimate to its standard error is distributed as a 2 statistic. Z

54

value must exceed 1.96 to achieve a significance level of p< .05. For significance level of
p< .10, 2 value must be greater than or equal to 1.645.

Reliability is a measure of the internal consistency of the construct indicators,
depicting the degree to which they “indicate” the common latent (unobserved) construct.
Cronbach’s alpha with a value equal to or higher than .70 indicates a good reliability for a

latent construct.

4. 2. 2. Identifiz Efficient Manufacturing Plants
During the second stage of data analysis, data envelopment analysis (DEA) was
performed to identify efficient manufacturing plants. Data Envelopment Analysis (DEA)
was developed in 1978 by Chames, Cooper and Rhodes. Since then, it has become a
popular tool for performance evaluation. Applications of DEA have been used to evaluate
the efficiency of vendors (Weber and Desai 1996, Weber, Current, and Desai. 1998),
manufacturing firms (Banker, Chames, and Cooper 1984, Bymes, Fare, Grosskopf, and
Lovell 1988, Vargas, Whybark and Xiao 1996), health care systems (Sherman 1984;
Gerdtham, Rehnberg, and Tambour 1999), retail stores (Thomas, Barr, Cron, and Slocum
1998), banks (Sherman and Gold 1985, Thomas and Barr 1998), education systems
(Bessent and Bessent 1980, McMillan and Datta 1998) to name a few.
DEA is a linear programming based technique for estimating the relative efficiency
of organizational units where the presence of multiple inputs and outputs makes the
comparison difficult. In DEA, relative efficiency for each decision making unit (DMU)

is the ratio between weighted sum of outputs and weighted sum of inputs.

55

d /—r

in

Each DMU is allowed to adopt a set of weights which maximize this unit's
efficiency in comparison to the other units.

DEA was chosen as the tool to analyze the efficiency of manufacturing plants and
derive the best plants for several reasons. First, DEA is a non-parametric technique. That
is, it does not require any assumptions regarding the function form for the relationship
between the inputs and outputs. In this study, the production fiinction of the plants may
vary from one plant to another. WCM plants are expected to have a different production
function from non-WCM plants. Second, DEA focuses on determining "best-practice"
production function by identifying the Decision Making Units (DMUs) on the efficient
frontier rather than on central tendency properties of frontiers. In contrast to parametric
approaches whose object is to optimize a single regression plane through the data, DEA
optimizes on each individual unit. That is, instead of focusing on "average" production
function as in regression, DEA focuses on "best practices" production function. The
comparison between DEA and regression production functions is shown in Figure 4.1.
WCM plants, in this study, are defined as those plants that utilize the "best-practices"
production function so that they produce the most amounts of the outputs with the least
amounts of the inputs. These characteristics of DEA make it an ideal tool to determine
which plants in the sample are WCM plants.

There are four basic DEA models: CCR ratio model (Chames, et al. 1978), BCC
model (Banker, Chames, and Cooper 1984), Multiplicative model (Chames, et al. 1982;
Chames, et al. 1983), and Additive model (Chames, et al. 1985, 1990). The choice for
these basic models is the envelopment surface. The CCR model yields a piecewise linear,

constant retums-to-scale envelopment surface, the BBC and Additive models result in a

56

piecewise linear, variable retum-to-scale envelopment surface, while the Multiplicative
model yields piecewise log-linear envelopment surface. Figure 4.2. depicts a CCR model
with constant return-to-scale envelopment surface. Figure 4.3. illustrates a BCC model
with variable retum-to-scale envelopment surface.

Figure 4.1 Comparison between DEA and Regression

OUTPUT

14~

12-

10—

 

 

 

 

 

57

Figure 4.2 Envelopment Surface for BCC Model

 

 

 

Figure 4.3 Envelopment Surface for CCR Model

 

 

58

 

12

 

Depending on the projection path to the envelopment surface for the inefficient
DMUs, the BCC and the CCR models could have either input orientation or output
orientation (therefore BCC and CCR are called orienting models whereas ADD and
Multiplicative are called nonorienting models). In an input orientation, the projection
path focuses on maximal movement toward the frontier via proportional reduction of
inputs, whereas in an output orientation the projection path focuses on maximal
movement through proportional increase of outputs.

In this dissertation, the basic assumption of constant returns to scale is chosen.

Inputs to manufacturing plants are identified as number of factory worker, number
of non-factory worker, total raw material purchase, investment in equipment, and
investment in Information Technology. Outputs are annual change in profitability, annual
change in market share and total sales.

Efficiency of a unit iis defined as:

no
2 Oij
Weighted sum of wit i's outputs _ j =1
Weighted sum of unit i's inputs ”"
21W}
1 =1

 

Efficiency of unit 1':

Where 01'} is the value of unit i on output j , Iij is the value of unit 1' on input j ,
Wj is a nonnegative weight assigned to output j , Vj is a nonnegative weight assigned to
input j , no is the number of output variables, mis the number of input variables. The
problem in DEA is to determine optimal values for weights Wj and Vj .

A separate linear programming problem is solved for each unit in a DEA problem.
The objective for each unit is the same: to maximize the weighted sum of that unit's
outputs. For an arbitrary unit 1', the objective is:

59

"0
Max: 201'jo
J' =1

Since it is impossible for any unit to be more than 100% efficient, for each unit, it
is required that the weighted sum of the unit's outputs is less than or equal to the weighted

sum of its inputs.

:1 0ijo< g: IijVj, for k— - 1 to the number of units
j: —1

m
2 [UV] = 1
J' =1

To prevent unbounded solutions, it is required that the sum of the weighted inputs

for the unit under investigation (unit i) is equal to one.
The weights for inputs and outputs must be constrained to be nonnegative.
w120,forj=lto no, W20,forj=1to m
Afier this linear programming problem is solved for each unit, all the units

(manufacturing plants) will have an efficiency score which is less than or equal to one.

The efficient units are those that have an efficiency score of one.

4. 2. 3. Investigate Moderating Effect ofEfficiency

Moderating effect can often be modeled as an interaction term between
independent variable and moderator in multivariate analysis. Since correlation between
the practice dimensions is expected, it would be wise to construct nine models to test
each sub-hypothesis of hypothesis 1 separately. In these nine models, the four

performance variables (cost, quality, delivery, and flexibility) are the dependent

60

variables. We could either construct nine multivariate models having all four dependent
variables in the same model or thirty-six univariate models having one dependent
variable in each model. Multivariate analysis is chosen since it can reduce family type I
error rate relative to multiple univariate analysis (Huitema, 1980). There are two
multivariate techniques that could be used to test a model with multiple independent
variables and multiple dependent variables: multivariate analysis of covariance
(MANCOVA) and cononical correlation analysis. Analysis of covariance is a statistical
technique that combines regression and analysis of variance (Wildt and Ahtola, 1978).
Cononical correlation, however, is similar to factor analysis in a way that both are
variable reduction techniques that construct uncorrelated linear combinations. Cononical
correlation is a means of parsimoniously describing the number and nature of mutually
independent relationships between independent variables and dependent variables
(Stevens 1996). Since the objective is not to find out a parsimonious representation of
independent and dependent variables, rather, the purpose is to investigate interaction
effect as can be done in regression analysis but with multiple dependent variables,
MANCOVA was chosen to be the technique to test alternative hypothesis 1.

There are generally two situations where analysis of covariance is recommended.
In observational studies, the dependent variable, in addition to being influenced by the
treatment level, may be subject to the influence of a quantitative independent variable.
Analysis of covariance under this condition provides a method to adjust for the difference
between treatment levels caused by the quantitative independent variable (Cochran, 195 7,
Elashoff, 1969). In other situations, the researcher is interested in the relationship

between quantitative independent and dependent variables, but is also concerned about

61

the possible effects of some qualitative variable upon this relationship and the interaction
between the categorical and the continuous independent variable (i.e. covariate) (Wildt
and Ahtola, 1978).

This study falls into the second category. Since there are always two types of
independent variables involved in MANCOVA: categorical and continuous variables, the
efficiency is transformed into a categorical variable eff_mem by following coding:

eff_mem = 1 if efficiency<1 .00,

eff_mem = 2 if efficiency=1.00.

The model to test Hla can be constructed in analysis of covariance as follows:

Cost + Quality + Delivery + Flexibility = be + bflPl + b2*eff_mem +

b3*Pl *eff_mem
where b0, b1, b2, b; are nonscalar.

Testing hypothesis 1a is equivalent to testing that b; is positive and significant.

Since factor scores are generally considered non-transferable (Hair et a1. 1994),
summated scales of the performance variables (cost, quality, delivery and flexibility) are
established from their measurement items to represent the corresponding constructs.

To assess whether there is positive and significant interaction between efficiency
and practices, two indices are of interest. Wilk's A is the multivariate index that measures
the relationship between the interaction and the set of dependent variables (Stevens,
1996). The sign and significance of regression coefficients B indicates how interaction
and each individual dependent variable are related. Similar to steps of assessing overall F
test and parameter estimate in regression, F value of Wilk's A is always examined first to

ensure that there is at least one dependent variable significantly associated with

62

interaction. If Wilk's A is not significant, then none of the regression coefficients for

interaction need to be examined.

4. 2. 4. Taxonomy of Efficient Manufacturing Plants

The fourth stage of this study is to test whether there is more than one strategic
group contained in efficient manufacturing plants. Cluster analysis is the most commonly
adopted approach for this purpose.

Depending on how the clustering procedures are canied out, clustering algorithms
can be classified into two general categories: hierarchical and nonhierarchical.
Hierarchical procedures can be further divide into agglomerative methods and divisive
methods. In the agglomerative methods, each observation begins as a cluster. The two
closest clusters are combined into a new cluster in the subsequent steps. Divisive method
works in reverse direction. All the observations start out as one large cluster. The
observations that are most dissimilar are split off in succeeding steps. Nonhierarchical
techniques are often referred to as K-means clustering. The algorithm of K-means
clustering can be described as following:

1. Select k samples (pl, 112, pk) fiom n original samples as the initial cluster

centers for k clusters.

2. Classify 11 samples according to nearest pi,

3. Recompute m.

4. Repeat 2 and 3 until there is no change in 1.1,.

Hierarchical procedures are fast comparing to nonhierarchical procedures.

However, they can be misleading because undesirable early combinations of the

63

observations persist throughout the analysis. Nonhierarchical procedures have gained
increased popularity since they are less influenced by the outliers in the data, choice of
cluster variates and the distance measures applied. The nonhierarchical clustering results,
however, are very sensitive to the selection of initial cluster centers. There is no standard
procedure to determine the proper number of clusters in the data by nonhierarchical
procedures.

A combination of hierarchical and nonhierarchical procedures was applied in this
dissertation to overcome the weakness of either procedure as suggested by Hair et al.
(1994). First, the data was analyzed by hierarchical technique to establish the number of
clusters, profile cluster centers and identify outliers. The cluster centers generated by
hierarchical technique were then chosen as the initial cluster centers for K-means
clustering. The number of clusters generated by the hierarchical clustering procedure was
taken as the K for K-means clustering. K-means clustering algorithm therefore performs

a fine-tuning of the results by allowing the switching of cluster membership.

4. 2. 5. Compare Strategic Groups within Eflicient Manufacturing Plants
If hypothesis 2 is supported, that is, there are multiple strategic groups in efficient
manufacturing plants, it would be interesting to see whether strategic groups perform
differently on various practice dimensions. The fifth stage of data analysis involved
testing hypotheses 3 which exams how strategic groups perform differently on practice
dimensions.
Sample data was divided into groups as suggested by cluster analysis. For

hypotheses 3, nine groups of ANOVA tests each with the one practice dimension as its

64

dependent variable and clusters as independent variables were performed. A significant
test result indicated that clusters perform differently in terms of that practice dimension.
Upon recognizing the significant test result, ANOVA does not reveal specifically the
difference between which two clusters on that practice dimension leads to the significant
test result of ANOVA. Multiple range test was applied for that purpose.

This chapter presented the research design of the dissertation. Sampling fi'ame,
scale selection, and research methodologies are discussed and justified. Chapter 5 will

present the results of data analysis by applying procedures discussed in this chapter.

65

CHAPTER 5

DATA ANALYSIS

This chapter begins with a brief discussion of the pros and cons of web based
survey. Response profile is presented next to examine if there is response bias. Details of
the five date analysis procedures carrying out research design in chapter 4 are presented.
The hypothesized relationships are examined based on the results of the data analysis

procedures.

5.1. The Pros and Cons of Web Based Survey

A three-wave mailing, with one invitation letter and two reminder postcards and
some follow up phone calls employed many techniques proposed by Dillman (1978,
2000), and resulted in the return of 224 surveys. During data collection period, the
researcher did realize the advantages and disadvantages of web based survey. Comparing
to traditional paper based survey, the cost saving of web survey is obvious. The huge
savings are on the printing cost of survey and mailing cost. As discussed in the previous
chapter, the turnaround time of survey is reduced. The web survey also reduced the labor
involved and potential errors in data entering and helped collecting some extra
information very easily. For example, by adding a simple log field for each participant in
the database, the researcher is able to get the information on the how many times a
participant logged on to the web site, how many questions a participant answered when
he/she leaves the web site by running a simple query. This information helps identify

respondents from each wave, will be useful in testing for non-response bias. The database

66

also provides information on completion status and contact information of participants
who logged onto the web site but did not complete the survey. With this information, the
researcher was able to easily identify the problems if the participants did not finish the
survey and make follow up contacts with participants to encourage them finish the survey
and even relax some of the requirements based on the participants' inputs.

At the same time, the researcher did realize that there were some disadvantages
with web based survey that have been overlooked. The biggest problem the researcher
faced with was the inexperience and impatience with the web technology of the senior
managers. The researcher received a lot of phone calls and emails for logging problems.
Some participants requested to review the survey before they decided to participate.
Under these circumstances, the researcher ofien chose to email the htrnl copy of survey to
the participants and offered them the option to fax the responses back. The front end of
the database was implemented by cold fusion. The researcher believed that some of the
logging problem was associated with the instability of cold fusion server. The business
school that hosts the web survey sometimes shut the server down for backup and
maintenance. During this time period, the web pages were not available. This problem,
however, can be relatively easily mended by hosting the survey on a proprietary and
more stable web server. As stated in previous section, the respondent was asked to
establish an account and password so that they can come back to work on the surveys.
The researcher found that this design had both its benefits and drawbacks. The benefits
were obvious. The respondents were usually not able to answer all the questions in one
session. For example, they had to ask the financial people on financial indices and sales

people on sales information to get that part of questions answered. Therefore, it was very

67

common that a participant would take between two weeks and one month to finish the
survey. The drawbacks included that sometimes the managers would forget the password
and felt frustrated with logging problem and finally gave up. The web survey also tends
to be less reminding than paper based survey in the cases that the participants did not
complete the survey at once and had to come back to finish.

With all these pros and cons of web based survey, the researcher believes that
with more experience in designing such a survey, this technique will become more and
more favored by researchers with similar requirements and will eventually become the
norm in the field.

5. 2. Respondent Profile

105 respondents first logged onto the web site before the second mailing was out
and 58 of them completed 90% or more of the survey at that time. 82 respondents logged
onto web site and 79 surveys were completed between second mailing and third mailing.
After the third mailing and follow up phone calls, 68 respondents came onto the web site,
87 more surveys were completed. It should be noticed that due to the time it took the
managers to complete the survey, it was often the case that they logged onto the web site
in an earlier time frame than they completed the survey. Therefore, it was not surprising
to see that more surveys were completed in the second or third wave but more
respondents actually logged onto the web site in the first or second wave. In total there
were 255 plants that logged onto the web site. 224 of the 255 responses were more than
90% complete and deemed as usefirl responses. The 31 responses that are not considered
valid responses were either completely empty, (in which case the respondent decided not

to participate after reviewing the survey) or half-completed. After making email or phone

68

contacts with those respondents, most of them cited reason as either they were private
firms so that they could not answer some of the questions or they did not have the time to

complete such a long survey.

Some descriptive characteristics of the participants from the three waves were

listed in table 5.1, table 5.2 and table 5.3.

Table 5.1 Group 1 respondents (n=58)

 

 

 

 

 

 

 

 

 

 

 

mean Standard Deviation
Number of factory workers 581 213
Number of non-factory 331 201
workers
Plant international sales 56.7 36.9
(in Million $)

Plant domestic sales (in 149.3 101
Million $)

Number of products 70 160

Percentage of continuos 31.6 22.5

flow

Percentage of assembly line 25.7 28.3

Percentage of job 8.0 10.2
production

Percentage of batch 34.7 32.3
production

 

 

 

69

 

Table 5.2 Group 2 respondents (n=79)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mean Standard Deviation
Number of factory workers 492 310
Number of non-factory 389 314
workers
Plant international sales 66.3 103
(in Million $)

Plant domestic sales (in 128 92.7
Million 3)

Number of products 54 90

Percentage of continuos 16.9 20.4

flow

Percentage of assembly line 29.1 28.0

Percentage of job 13.7 22.9
production

Percentage of batch 40.3 30.0
production

Table 5.3 Group 3 respondents (n=87)
mean Standard Deviation
Number of factory workers 680 435
Number of non- factory 527 449
workers
Plant international sales 78.8 103
(in Million $)

Plant domestic sales (in 190 374

Million $)
Number of products 44 80
Percentage of continuos 15.1 25.9
flow

Percentage of assembly line 45.8 38.8

Percentage of job 30.0 37.0
production

Percentage of batch 42.3 35.8
production

 

 

 

70

 

 

5. 3. Confirmatory Factor Analysis

Confirmatory Factor Analysis (CFA) was used to evaluate the measurement
models for World Class Manufacturing Practices and Performances. Based on theory and
past research, CFA is a more rigorous method for assessing unidimensionality of
multiple-item constructs rather than coefficient alpha in exploratory factor analysis
(Gerbing and Anderson 1988; Calatone, Schmidt, and Song 1996).

There are several software programs that are widely adopted by operations
researchers to perform structural equation modeling. LISREL (Linear Structural
RELations) is a flexible model for cross-sectional, experimental, and longitudinal studies.
EQS places less stringent assumptions on the multivariate normality of the data (Bentler
1989, 1992). In this dissertation, EQS was used to execute structural equation modeling
including CFA and path models.

The estimation of parameters in the model was determined by maximum
likelihood (ML) estimation (Bollen 1989; Bentler 1992a; Joreskog and Sorbom 1993).
ML function is the most widely used fitting function for structural equation modeling to
date. ML estimation reproduces the covariance matrix of observed variables by using a
fitting function.

While evaluating the measurement model, the following criteria are considered
important: unidimensionality, univariate and multivariate normality, reliability,
convergent validity, and discriminant validity (Peter1981, Venkatraman 1989, Klassen
1995)

The evaluation of the measurement models was conducted as follows:

71

1) Initial runs based on the categorization of items into constructs as the survey
suggested was carried out.

2) LM suggestions as to adding parameters was examined. CF A model was justified
based on LM suggestions and theoretical backgrounds. The model was rerun and
model fit was re-evaluated.

3) Step 2 was repeated until satisfactory global fit indication was found.

The data were examined for univariate and multivariate normality. Reliability of
the constructs, convergent validity, and discriminant validity were evaluated.
Common method bias was examined by employing the error covariance matrix

and LM test results.

72

5. 3.1. CFA for WCM Practices

Table 5.4 lists all the factors and variables in CF A model for WCM practices .

Table 5.4 Factors and Variables in WCM Practices CFA Model

 

F 1: Quality Management Programs

V1: Everyone has responsibility to improve quality

 

V4: Processes on the shop floor are under statistical quality control.
V11: Total Quality Management philosophy is promoted
F2: _C_l_istomer Orientm
V13: We maintain close contacts with our customers.
V14: Results of customer satisfaction surveys are communicated throughout
organization.
V19: We have a formal "customer satisfaction" program in place.
F3: Labor Force Management
V23: Team members are encouraged to exchange opinions/ideas.
V25: Our employees have strong problem solving abilities.
V35: Labor scheduling / job assignments is handled by empowered team
F4: Operations Flow
V38: The shop floor is laid out to optimize processing sequence and flow.
V42: We use lean manufacturing production methods.
V43: We offer JIT delivery to customers.
F5: Operations Capability
V45: We dedicate to continual improvement in quality.
V46: We dedicate to continual improvement in response time.
V47 : We dedicate to continual improvement in flexibility.
V48: We dedicate to continual improvement in cost.
F6: Supplier Relationship
V56: We establish long-term contracts with suppliers.
V57: We have reduced the number of suppliers for each part family.
V58: We frequently source multiple part families from a single supplier.

73

Table 5.4. Factors and Variables in WCM Practices CFA Model

 

F7: Technology Management
V78: We plan technological upgrades to be consistent with infrastructural

upgrades
V79: Computer-aided-design (CAD) technology practice is applied
V81: Flexible manufacturing systems (FMS) technology practice is applied
F8: New Prodpct Developmen_t
V91: Customer requirements are clearly specified early in conceptual design.
V92: We use design-for-manufacture and assembly (DFMA) methods.
V97: Product designers and manufacturing staff have equal status in NPD
projects.
F9: SLategy Fornlulatiound Measurement
V101: We have clearly defined strategic manufacturing goals and objectives.
V102: Our firm's strategy leverages existing capabilities.
V103: Corporate strategy at our firm drives manufacturing decisions.
V105: Manufacturing strategy is frequently reviewed and revised.
V106: Manufacturing strategy is well aligned with corporate strategy.

 

5.3.1.1. Univariate and Multivariate Normality
Multivariate normality is one of the critical assmnptions of structural equation
modeling. Violation of this assumption could lead to downwardly biased standard errors
that would result in an inflated number of statistically significant parameters (Muthen and
Kaplan 1985; Byrne 1994). Since normal theory test statistic may not reflect an adequate
evaluation of the model if the assumption of multivariate normality is violated, the
statistical hypothesis testing can be seriously invalidated under such situation (Browne

1982, 1984; Hu, Bentler, and Kano 1992).

74

The univariate statistics for the items that represent WCM practices constructs are
displayed in table 5.5. The multivariate kurtosis for WCM practices constructs are
represented in table 5 .6. Values for skewness ranged from -1.2778 to 0.2579. Values for
kurtosis ranged from -l.0764 to 1.5572. Since all of the univariate skewness < 2.0, and
all of the univariate kurtosis < 7.0, no indications of departures from normality existed
(Cho and Bentler 1995).

The normalized estimate of Mardia's coefficient is very small (0.8176) which

indicates no violation of multivariate normality existed (Mardia 1970; Byrne 1994).

Table 5.5 Univariate Statistics: WCM Practices CFA

 

VARIABLE V1 V4 V11 V13 V14
MEAN 5.4107 4.1607 5.1964 5.6607 4.8036
SKEWNESS (G1) ~0.2960 -0.4455 -0.4544 -0.7401 -0.8269
KURTOSIS (G2) -0.8652 0.1065 -0.4955 0.1269 0.3895

STANDARD DEV 1.1641 1.6596 1.3315 1.1870 1.5781

VARIABLE v19 V23 v25 v35 V38

MEAN 4.5179 5.1607 4.3929 3.4821 4.9643
SKEWNESS ((31) -0.2466 -0.3072 .0.0379 -0.1601 -0.3802
KURTOSIS ((32) -1.0746 -0.6237 0.0122 .0.7753 -0.5826

STANDARD DEV 1.7358 1.2388 1.2374 1.9034 1.4295

 

75

Table 5.5. Univariate Statistics: WCM practices CFA

 

VARIABLE
MEAN
SKEWNESS (G1)
KURTOSIS (G2)
STANDARD DEV

VARIABLE
MEAN
SKEWNESS (G1)
KURTOSIS (G2)
STANDARD DEV

VARIABLE
MEAN
SKEWNESS (G1)
KURTOSIS (G2)
STANDARD DEV

VARIABLE
MEAN
SKEWNESS (G1)
KURTOSIS (G2)
STANDARD DEV

V42
5.0179
-1.2778
1.0706
1.7203

V48

5.5714
-0.5930
-0.2180

1.2107

V79
5.1964
-0.7206
0.8015
1.2905

V101
5.0893

-0.6712
0.9517
1.2460

V43

4.7500
—0.6358
-0.03 82

1.6321

V56

4.8929
-0.5548
-0.0493

1.4006

V81

3.7500
-0.5111
-0.2650

1.8678

V102
4.8929

-0.5618

-0. 1489
1.2228

V45

5.3571
-0.5326
-0.4991

1.2482

V57
5.0536
-0.5433
0.2494
1.3843

V91
4.5893
-O.5145
0.0758
1.6047

V103
4.6071

-0.2660

-0.8069
1.3747

V46
5.3571
-1.1443
1.5572
1.2043

V58
4.8214
-0.4929
0.1543
1.5631

V92

4.0893
-O.5623
-0.5014

1.9890

V105
4.5714

-0.3269

-O.3088
1.3507

V47
5.2500
-1 .0788
1.0666
1.2881

V78
4.1429
-0.9611
1.3489
1.5321

V97

3.3929
-0.2579
-0.9855

2.0283

V106
4.7679

-0.3329

-0.6995
1.5502

 

76

Table 5.6 Multivariate Kurtosis for WCM Practices

 

MARDIA'S COEFFICIENT (G2,P) = 4.7873
NORMALIZED ESTIMATE = 0.8176

ELLIPTICAL THEORY KURTOSIS ESTIMATES

MARDIA-BASED KAPPA = 0.0050
MEAN SCALED UNIVARIATE KURTOSIS = -0.0114
MARDIA-BASED KAPPA IS USED IN COMPUTATION. KAPPA= 0.0050

 

5.3.1.2. Goodness of F it Indices

A summary of selected fit indices for the EQS analysis is presented in Table 5. 7.

Table 5.7 Goodness-of-Fit Indices for WCM Practices Measurement Model

 

n 224

x2 989.500

Degree of Freedom (df) 368

p value .001

xz/df 2.69

BENTLER-BONETT NORMED FIT INDEX 0.923
BENTLER-BONETT NONNORMED FIT INDEX 0.912
COMPARATIVE FIT INDEX (CFI) 0.941
BOLLEN (IFI) FIT INDEX 0.935
STANDARDIZED RMR 0.057
ROOT MEAN SQ. ERROR OF APP.(RMSEA) 0.044

90% CONFIDENCE INTERVAL or RMSEA( 0.025, 0.058)

 

77

The fit indices in Table 5.7 (x2 , normed x2, CFI, NFI, NNFI, IFI, RMR,
RMSEA) indicate a strong global fit obtained in the CFA model. The normed x2 is
below 3, CF I, NFI, NNFI and IFI are greater than cutoff value .90, RMR and RMSEA are

below cutoff value .080.

5.3.1.3. Reliability and Convergent Validity

Table 5.8 listed the measurement equations with standard errors, standardized
solution and cronbach’s or for all the practices constructs.

Cronbach's or is a widely used measure for scale reliability. Typically, these
coeffiecients should be 0.70 or higher for narrow constructs, and 0.55 or higher for
moderately broad constructs such as those defined here (Van de Ven and Ferry 1979).
Cronbach's or ranges between .684 to .913 for the nine practice constructs, which
represents a reasonable scale reliability.

Convergent validity was indicated by the strong and significant item loadings. In
the standard solutions for the measurement equations, all the loadings were higher than

.50 except two items at .486 and .434. All the factor loadings were significant at .05 .

78

Table 5.8 Measurement Equations with Standard Errors, Standardized Solution

and Cronbach's alpha for WCM Practice CFA Model

 

 

 

 

 

 

 

 

 

 

 

Cronbach's on Measurement Equation Standard Error t value
.731 Quality Management Programs (F1)
V1= 1.000 F1 + 1.000 E1 Path fixed at 1.0
V4 =.838*F1 + 1.000 E2 .115 7.309
V 11 =1.224*F1 + 1.000 E3 .080 15.247
Standardized Solution
Vl = .827 F1 + .563 E1
V4 = .486*F1 + .874 E2
V11 = .884*F1 + .467 E3
.773 sttomer Orientation (E2)
V13 = 1.000 F2 + 1.000 E4
V14 = 1.615*F2 + 1.000 E5 .182 8.850
V19 =1.729*F2 + 1.000 E6 .197 8.773
Standardized Solution
V13 = .653 F2 + .758 E4
V14 = .793*F2 + .609 E5
V19 = .772*F2 + .636 E6
.774 Labor Force Management (F3)
V23 = 1.000 F3 + 1.000 E7
V25 = .972*F3 + 1.000 E8 .056 17.398
V35 = .933*F3 + 1.000 E9 .108 8.629
Standardized Solution
V23 = .889 F3 + .458 E7
V25 = .864*F3 + .503 E8
V35 = .540*F 3 + .842 E9
.685 Operations Flow (F4)
V38 = 1.000 F4 + 1.000 E10
V42 = 1.126*F4 + 1.000 E11 .103 10.937
V43 = .750*F4 + 1.000 E12 .101 7.406
Standardized Solution
V38 = .774 F4 + .634 E10
V42 = .724*F4 + .690 E11
V43 = .508*F4 + .861 E12
.902 Operations Capability (F5)
V45 = 1.000*F 5 + 1.000 E13
V46= 1.122 F5 + 1.000 E14 .065 17.210
V47 = 1.109*F5 + 1.000 E15 .072 15.408
V48 = .914*F5 + 1.000 E16 .071 12.803
Standardized Solution

 

V45 = .807*F5 + .590 E13
V46 = .939 F5 + .345 E14
V47 = .867*F 5 + .497 E15
V48 = .761*F5 + .649 E16

 

 

 

79

 

 

.884

Supplier Relationship (E6)
V56= 1.000 F6 + 1.000 E17

V57= .951*F6 + 1.000 E18
V58: 1.106*F6 + 1.000 E19
Standardized Solution

V56 = .863 F6 + .505 E17

V57 = .830*F6 + .557 E18

V58 = .855*F6 + .518 E19

.064
.072

14.778
15.323

 

.684

Technolo Mana ement F
V78= 1.163*F7 + 1.000 E21
V79 = 1.176*F7 + 1.000 E22
V81 = 1.000 F7 + 1.000 E20
Standardized Solution
V78 = .616*F7 + .787 E21
V79 = .740*F7 + .672 E22
V81 = .434*F7 + .901 E20

.216
.205

5.393
5.734

 

.753

New Prrfiuct Developmept (F8)
V91 = 1.000 F8 + 1.000 E23
V92 = 1.062*F8 + 1.000 E24
V97 = .896*F8 + 1.000 E25

Standardized Solution

V91= .841 F8 + .541 E23

V92 = .720*F 8 + .693 E24

V97= .596*F8 +.803 E25

.095
.100

11.230
8.971

 

 

.913

 

Strategy Formulation and
Measpremept (F9)

V101= 1.000 F9 + 1.000 E26
V102= .895*F9 + 1.000 E27
V103= .950*F9 + 1.000 E28
V105= 1.027*F9 + 1.000 E29
V106= 1.240*F9 + 1.000 E30
Standardized Solution
V101= .876 F9 + .482 E26
V102= .799*F9 + .602 E27
V103= .754*F9 + .657 E28
V105= .830*F9 + .557 E29
V106= .873*F9 +.487 E30

 

.059
.069
.063
.070

 

15.135
13.771
16.201
17.788

 

5. 3. 1.4. Discriminant Validity and Nomological validity

Table 5.9. listed the variances for each factor, all inter-factor covariance, all inter-

factor squared correlation and their squared error and t value.

80

 

The inter-factor correlations are significantly different from 1.00, indicating discriminant
validity. The average variance extracted for each factor is greater than the squared
correlation between that factor and any other factor in the CFA, providing a more
rigorous confirmation of discriminant validity.

All the inter-factor correlations are statistically significant and positive, which

confirmed nomological validity.

Table 5.9 Variances for Each Factor and Inter-Factor Covariances for WCM

 

 

 

 

 

Practices CFA
Factor Avg. Variance Inter-factor squared Inter-factor Standard tvalue
Extracted correlation covariance error
F1 .926
F1-F2: .291 F1-F2: .402 .076 5.299
F1-F3: .685 F1-F3: .877 .109 8.014
Fl-F4: .579 F1-F4: .810 .155 7.065
F1-F5: .498 F1-F5: .714 .098 7.291
F1-F6: .254 F1-F6: .586 .103 5.707
F1-F7: .336 F1-F7: .453 .103 4.406
F1-F8: .161 F1-F8: .521 .114 4.580
F1-F9: .238 F1-F9: .512 .090 5.666
F2 .600
Fl-F2: .291 Fl-F2: .402
F2-F3: .282 F2-F3: .453 .084 5.364
F2—F4: .188 F2-F4: .371 .086 4.329
F2-F5: .233 F2-F5: .377 .075 5.034
F2-F6: .086 F2-F6: .275 .081 3.402
F2-F7: .124 F2-F7: .221 .070 3.144
F2-F8: .128 F2-F8:.374 .097 3.855
F2-F9: .081 F2-F9:.240 .071 3.369
F3 1.212
F1-F3: .685 F1-F3: .877
F2-F3: .282 F2-F3: .453
F3-F4: .710 F3-F4: 1.026 .132
F3-F5: .618 F3-F5: .872 .111
F3-F6: .193 F3-F6: .586 .111
F3-F7: .054 F3-F7: .208 .085
F3-F8: .016 F3-F8: .188 .120
F3-F9: .147 F3-F9:.460 .097

 

 

 

 

 

 

81

 

 

 

 

 

 

 

F4 1.223

Fl-F4: .579 Fl-F4: .810

F2-F4: .188 F2-F4: 371

F3-F4: .710 F3—F4: 1.026

F4-F5: .667 F4-F5: .910 .123 7.418

F4-F6: .281 F4-F6: .709 .126 5.625

F4-F7: .093 F4-F7: .274 .097 2.832

F4-F8: .224 F4-F8: .707 .142 4.958

F4-F9: .399 F4-F9: .762 .117 6.494
F5 1.015

F1-F5:.498 F1-F5:.714

F2-F5: .233 F2-F5: .377

F3-F5: .618 F3-F5: .872

F4-F5: .667 F4-F5: .910

F5-F6: .270 F5-F6: .633 .106 5.965

F 5-F7: .162 F5-F7: .329 .089 3.698

F5-F8: .076 F5-F8: .375 .110 3.400

F5-F9: .325 F5-F9: .627 .097 6.467
F6 1.461

F1-F6: .254 F1-F6: .586

F2-F6: .086 F2-F6: .275

F3-F6: .193 F3-F6: .586

F4-F6: .281 F4-F6: .709

F5-F6: .270 F5-F6: .633

F6-F7: 318 F6-F7: .553 .126 4.401

F6-F 8: .184 F6-F 8: .699 .142 4.932

F6-F9: .283 F6-F9: .702 .114 6.171
F7 .659

Fl-F7: .336 F1-F7: .453

F2-F7: .124 F2-F7: .221

F3-F 7: .054 F3-F7: .208

F4-F7: .093 F4-F7: .274

F5-F7: .162 F5-F7: .329

F6-F7: .318 F6-F7: .553

F7-F8: .520 F7-F8: .790 .165 4.785

F7-F9: .169 F7-F9: 364 .097 3.762
F8 1.821

F1-F8: .161 F1-F8: .521

F2-F8: .128 F2-F8:.374

F3—F8: .016 F3-F8: .188

F4-F8: .224 F4-F 8: .707

F5-F8: .076 F5-F8: .375

F6-F 8: .184 F6-F8: .699

F7-F8: .520 F7-F8: .790

F8-F9: .482 F8-F9: 1.022 .142 7.218

 

 

 

 

 

 

82

 

 

 

F9 1.191
F1-F9: .238 F1-F9: .512
F2-F9: .081 F2-F9:.240
F3-F9: .147 F3-F9:.460
F4-F9: .399 F4-F9: .762
F5-F9: .325 F5-F9: .627
F6-F9: .283 F6-F9: .702
F7-F9: .169 F7—F9: .364
F8-F9: .482 F8-F9: 1.022

 

 

 

 

 

 

5.3. 1.5. Common Method Bias

Only single respondent from each plant completed the survey due to cost and time
constraints. The potential drawback with single informant is that it may yield common
method bias (Hughes, Price, and Marrs 1986). In order to test for common method bias,
the model was rerun to allow the errors to covary, that is, the zero-correlation constraints
for the relevant off-diagonal elements in the (~95 matrix were released. The LM test
provided by EQS indicated that there was only a small drop in x2 to the already well-
fitting model with @5 matrix released. The LM test result is displayed in table 5.10. The
absence of significant model improvement was an indication of the absence of common

method bias.

Table 5.10 Multivariate LM Test with theta lambda Matrix Released

 

Cumulative Multivariate Statistics

 

Step Parameter chi-square D.F . Probability
1 E13, E1 32.115 1 0.000
2 E4, E6 47.229 2 0.000
3 E17, E18 58.844 3 0.001
4 E17, E19 67.513 4 0.003

 

83

5. 3.2. CFA for WCM Performance

Table 5.11 lists all the factors and variables in CFA model for WCM performance.

Table 5.11 Factors and Variables in WCM Performances CFA Model

 

F1: Cost Performance

 

V113: manufacturing overhead cost
V114: maintenance cost
F2: Qu_ality Perform
V1 15: Product overall quality performance
V1 16: Product feature
V1 17: Product reliability
V119: product durability
F3: Delivery Performppcp
V124: Delivery availability (the probability that items will be in stock at the time
of order)
V125: Delivery dependability (delivered on the agreed upon date)
V126: Delivery speed (short elapsed time between order placement and product
reaches the customer)
F 4: Flexibility Performam
V130: Ability to adjust production volumes
V133: Ability to respond to changes in delivery requirements

 

5. 3. 2. 1. Univariate and Multivariate Normality
The univariate statistics for the items that represent WCM performance constructs are
displayed in table 5.12. The multivariate kurtosis for WCM performance constructs are
represented in table 5.13. Values for skewness ranged from -1.7272 to 0.0209. Values

for kurtosis ranged from -0.5275 to 4.2619. Since all of the univariate skewness < 2.0,

84

and all of the univariate kurtosis < 7.0, no indications of departures from normality

existed (Cho and Bentler 1995).

The normalized estimate of Mardia's coefficient is small (26.8875), which indicates no

Violation of multivariate normality existed (Mardia 1970; Byrne 1994).

Table 5.12 Univariate Statistics: WCM Performance CFA

 

VARIABLE
MEAN
SKEWNESS
(G1)
KURTOSIS
(GZ)
STANDARD
DEV

VARIABLE
MEAN
SKEWNESS
(GI)
KURTOSIS
(GZ)
STANDARD
DEV

SlOCV113
4.1250
-O.5887

1.2662

1.4308

SlOQV119
5.4286
-1.7272

4.2619

1.4654

SlOCV114

3.3571

-0.9690

0.0992

1.4724

SlODV124

4.5714

-1.0841

0.7751

1.9395

SlOQVl 15
5.4821
-0.3210

-0.5236

1.0544

SlODV125
5.0536
-0.7515

-0.9527

1.4600

85

SlOQV116
5.0714
-1.1870

4.2200

1.1807

SlODV126
4.7143
-0.8188

0.8879

1.7121

SlOQV117
5.4643
-1.1459

2.7362

1.3522

SlOFV13O
4.9107
-0.2886

-O.2638

1.3825

VARIABLE SlOFV133

MEAN 5.0357

SKEWNESS 0.0209
(GI)

KURTOSIS —0.5275
(62)

STANDARD 1.0540
DEV

 

Table 5.13 Multivariate Kurtosis for WCM Performance

 

MARDIA'S COEFFICIENT (G2,P) = 60.7630

NORMALIZED ESTIMATE = 26.8875

ELLIPTICAL THEORY KURTOSIS ESTIMATES

MARDIA-BASED KAPPA = 0.4249
MEAN SCALED UNIVARIATE KURTOSIS = 0.4207
MARDIA-BASED KAPPA IS USED IN COMPUTATION. KAPPA= 0.4249

 

5. 3. 2. 2. Goodness of F it Indices

A summary of selected fit indices for the EQS analysis is presented in Table 5. 14

86

Table 5.14 Goodness-of-Fit Indices for WCM Performance Measurement Model

 

n 224

x2 75.620

Degree of Freedom (df) 38

p value .002

xZ/df 1.99

BENTLER-BONETT NORMED FIT INDEX 0.941
BENTLER-BONETT NONNORMED FIT INDEX 0.950
COMPARATIVE FIT INDEX (CFI) 0.944
BOLLEN (IF I) FIT INDEX 0.922
STANDARDIZED RMR 0.070
ROOT MEAN SQ. ERROR OF APP.(RMSEA) 0.068

90% CONFIDENCE INTERVAL OF RMSEA( 0.062 0.074 )

 

The fit indices in Table 5.12 (x2 , normed x2, CFI, NFI, NNFI, IFI, RMR,
RMSEA) indicate a strong global fit obtained in the CFA model. The normed x2 is
below 2, CFI, NFI, NNFI and IFI are greater than cutoff value .90, RMR and RMSEA are
below cutoff value .080.

5. 3.2.3. Reliability and Convergent Validity

Table 5.15 listed the measurement equations with standard errors, standard
solutions and cronbach's a for the performance constructs. All the Cronbach's or ranges
between .723 to .843 for the nine practice constructs, which represents a reasonable scale

reliability.

87

Convergent validity was indicated by the strong and significant item loadings. In
the standard solutions for the measurement equations, all the loadings were higher than

.566 and significant at .05.

5. 3. 2. 4. Discriminant Validity and Nomological validity

Table 5.16 listed the variances for each factor, all inter-factor covariance, all inter-
factor squared correlation and their squared error and t value.
The inter-factor correlations are significant different from 1.00, indicating discriminant
validity. The average variance extracted for each factor is greater than the squared
correlation between that factor and any other factor in the CF A, providing a confirmation
of discriminant validity.

All the inter-factor correlations are statistically significant and positive, which

confirmed nomological validity.

88

Table 5.15 Measurement Equations with Standard Errors, Standardized Solution
and Cronbach's alpha for WCM Performance CFA Model

 

Cronbach's on

Measurement Equation

Standard Error

t value

 

.723

Cost Performapce(F1)
V113=1.431*F1 +
V114 =.834*F1 + 1.000 E2
Standardized Solution
V113 = 1.000*F1 + .000E1
V114 = .566*F1 + .824 E2

1.000 El

.068
.090

21.119
9.228

 

.843

Quality Performance§F21
V115= .623*F2 +

V116=.955*F2 +

V117 =1.322*F2 +
V119 =1.038*F2 +

Standardized Solution
V115= .591*F2 +
V116=.809*F2 +
V117=.978*F2 + .211E5
V119 = .708*F2 + .706E6

 

1.000 E3
1.000 E4
1 .000E5
1 .000E6

.807 E3
.588 E4

.066
.067
.069
.087

9.499

14.261
19.138
11.904

 

.823

Delivery Performppce(F3)
V124 = 1.539*F3 + 1.000E7
V125 =1.016*F3 + 1.000E8
V126 =1.518*F3 +1.000E9

Standardized Solution
V124 = .794*F3 + .608E7
V125 = .696*F3 + .718E8
V126 = .886*F3 + .463E9

.113
.090
.095

13.603
11.352
15.982

 

 

.817

 

Flexibility Performa_nce(F4)
V130 = .990*F4 + 1.000 E10
V133 =1.054*F4 +1.000 E11

Standardized Solution
V130 = .716*F4 + .698 E10
V133 = 1.000*F4 + .000 E11

 

.080
.050

 

12.403
21.119

 

89

 

 

Table 5.16 Variances for Each Factor, Inter-Factor Covariances and Squared

Correlation for WCM Performance CFA

 

 

 

 

 

Factor Avg. Variance Inter-factor squared Inter-factor Standard tvalue
Extracted correlation covariance error

F1 1.000

F1-F2: .140 F1-F2: .374 .059 6.319

F1-F3: .012 F1-F3: .109 .017 1.529

F1-F4: .083 F1-F4: .288 .061 4.698
F2 1.000

F1-F2: .140 F1-F2: .374

F2-F3: .191 F2-F3: .437 .061 7.153

F2-F4: .000 F2-F4: .000 .068 .004
F3 1.000

F1-F3: .012 F1-F3: .109

F2-F3: .191 F2-F3: .437

F3-F4: .437 F3-F4: .661 .043 15.248
F4 1.000

F1-F4: F1-F4:

F2-F4: F2-F4:

F3-F4: F3-F4:

 

 

 

 

 

 

 

 

5.3.2.5. Common Method Bias
The LM test provided by EQS indicated that there was only a small drop in x2 to the
already well-fitting model with (95 matrix released. The LM test result is displayed in
table 5.17. The absence of significant model improvement was an indication of the

absence of common method bias.

90

Table 5.17 Multivariate LM Test with theta lambda Matrix Released for WCM

Performance CFA Model

 

Cumulative Multivariate Statistics

Step Parameter chi-square D.F. Probability
1 E10, E1 1 5.432 1 0.000
2 E9, E4 7.021 2 0.000
3 E1, E3 8.311 3 0.001

 

91

5. 4. Data Envelopment Analysis

Table 5.18 listed the input and output dimensions for the DEA model

Table 5.18 Input and Output Dimensions for DEA Model

 

Inputs
1) Number of factory employees at your plant

2) Number of non-factory employees at your plant
3) Total raw material purchase in your plant last year
4) Investment in equipment in your plant over the last 3 years
5) Investment in Information Technology over the last 3 years.
Outputs
1) Average annual change in profitability over the last two years (in %)
2) Average annual change in market share over the last two years (in %)

3) Total sales (international + domestic) for your plant last year (in $)

 

Labor, expenses and equipment investment are the most commonly used input
variables in DEA analysis (Sherman and Gold 1985, Parkan 1987, Epstein and
Henderson, 1989, Oral and Yolalan 1990, Vassiloglou and Giokas 1990, Giokas 1991,
Sherman and Ladino 1995, Vargas et. a1. 1996, Athanssopoulos 1997, Schefczyk and
Gerpott, 1998, Kantor and Maital 1999, Golany and Storbeck 1999). With the emergence
of e-commence and the widespread applications of Information Technology in all
functional areas, manufacturing sections are also learning that competitive advantages
can be gained by Information Technology. E-business has had a great impact on
inventory management, forecasting and aggregate planning etc (Krajewski and Ritzman
1999). Various authors have showed that operations systems can be improved by

Information Technology. For example, F rohlich and Dixon (1999) showed that successful

92

AMT implementation must be accompanied by information systems adaptation, while
Kathuria et al. ( 1999) developed an intelligent decision support system that can allow a
firm to align its competitive priorities with the appropriate process structure and
information technology applications. Investment in Information Technology is therefore
added to the input list since operations managers have recognized the importance of IT
and are investing in IT to improve manufacturing performance.

Selection of output variables varies by the industry of the sample. For example,
studies in banking industry performance often adopt number of transactions as output
measurement in DEA (Sherman and Gold 1985, Parkan 1987, Oral and Yolalan 1990,
Vassiloglou and Giokas 1990, Giokas 1991, Sherman and Ladino 1995). In
manufacturing industry, Sales is most often employed as the output variable (Vargas et
a1. 1996, Schefcayk and Gerpott, 1998). Profitability and market share are gaining
increasing attention from operations manager as performance measures (Garvin 1989).
The way these variables are structured (as average annual change in the last two years
rather than absolute values) in this dissertation is due to both measurement considerations
and theoretical foundation. The data collection results from this dissertation indicate that
there are more missing values for the actual measures than relative improvement
measures. It seems to suggest that managers are more reluctant to give out actual
numbers of profitability and market share rather than the relative improvement in these
areas. The annual change measures, however, may be more pertinent as performance
measures in this dissertation since absolute value of market share and profitability are

determined by a lot more factors such as company history and strategy that are beyond

93

.. .r-u‘n "in."

 

the control of a manufacturing unit. However, the annual change measures better reflect
the recent performance of the manufacturing unit.
In this dissertation, Warwick Windows DEA software was used to carry out DEA.

The models are solved in two phases. In the first phase, the following model is solved:

Max: q
s.t. Zjljxy+sf=(l—wiq)x,jo i=1...m
21.2]- yrj—s:=(l+w,q)yrjo r=1 s
2.20; j=1...n, q>0

where x y- and y,j are the ith input and rth output level at DMU,
w,- and w, are user-specified priorities.
jo is the DMU being assessed.

In phase-2 the constraints of the above model are used after setting q to its optimal phase

1 value. Then the following objective function is optimized over the constraints:
, — — + +
Max. Zr, 3]. +2); s,

Where F," and F: are user-specified priorities.

Since DEA requires that any of the input and output to be non-null, those plants that did
not answer all of the questions are not included in this DEA model. Questions about
investment and sales were the questions that were most often left unanswered by the
managers. After deleting plants with incomplete responses, sample size dropped to 178.
These 178 data sets were entered into the DEA model as 178 DMUS, with input and

output fields specified. The efficiency score for each unit was given by DEA.

94

The efficiency ranged between 23.11% and 100%. The distribution of efficiency
scores of the 178 units is summarized in Table 5.19. Figure 5.1 plotted the distribution

against individual units.

Table 5.19 Distribution of Efficiency Scores

 

 

 

 

 

 

 

 

 

 

 

Range count Percentage
20-29.99 7 3.93%
30-39.99 29 16.29%
40-49.99 17 9.55%
50-5999 14 7.87%
60-69.99 1 1 6.18%
70-7999 0 0.00%
80-8999 11 6.18%
90-99.99 14 7.97%
100-100 75 42.13%

Total 178 100.00%

 

 

 

 

 

95

 

Figure 5.1 Distribution of Efficiency Score

 

 

80

 

7O

 

60

 

  

 

40

20-
0'1 I r t —I if " “

20-2999 30-3999 4049.99 50-5999 60-69.99 70-7999 80-8999 9099.99 100-100
Eflhfloncy

Count

 

 

 

 

 

 

 

 

 

 

 

From both Table 5.19 and Figure 5.1, we can see that the efficiency score is
heavily skewed to the right with 42.13% of sample data achieving a efficiency score of
100. The 75 plants in the sample with efficiency score of 100 were identified as efficient
plants.

In DEA, relative efficiency for each decision making unit (DMU) is the ratio
between weighted sum of outputs and weighted sum of inputs. The efficient DMUs
identified by DEA are those that can produce the most amount of outputs with the least
amount of inputs relative to the rest of DMUs in data sample. Therefore, it is crucial that
the participating plants in this research, if not superior to industry average, are at least

representative of industry average.

96

To validate the representativeness of data sample, a comparison on general
performance metrics between respondents of the questionnaire and a random sample
within the same SIC range from public database was made. Since unit of analysis is plant
in this research, while the general performance metrics publicly available are in firm’s
level, it is important that the metrics compared are not affected by the size of unit of
analysis. Two metrics that can be obtained from both sources and are relatively irrelevant
to the size of unit of analysis are ROA and percentage change in sales fi'om last year to
this year.

A random sample of 30 firms within the same SIC range as respondents of the
questionnaire was drawn from a public database. ROA and sales of this year and last year
of these 30 firms were available in this public database. The percentage change in sales
from last year to this year was calculated for these 30 firms. A random sample of 30
plants was selected from respondents of the questionnaire. ROA and percentage change
in sales between respondents and random sample were compared by applying two t tests.
The mean of percentage change for the 30 random selected firms was 5.26% while the
mean for the 30 respondents was 12.37%. With a p-value of .176, it was concluded that
percentage change in sales are not significantly different between respondents and
random sample. The mean of ROA for the 30 firms in public database was 8.48% while
the mean of ROA for the 30 respondents was 32.67%. The p-Value for this t test was less
than 0.01, which means that the respondents have a significantly higher ROA than the
random sample.

The above analysis shows that the respondents of the questionnaire in this

dissertation are performing equally well or better than a random sample in the same

97

industry. Therefore, it would be safe to conclude that the efficient DMUs identified from

these respondents are not biased in an undesirable way.

5.5. MANCOVA tests

There are two choices as to how to represent the practice factors: using factor
score or summated scale. Hair et al. (1994) argued that factor scores are unique to a
particular data set and should be replicated or transferred to another study. Summated
scales are preferred to factor scores when replicability or transferability is desired.
Therefore, for each practice construct, a summated scale is constructed fi'om its
measurement items.
A close examination of the content of efficiency and practice constructs reveals that
efficiency and practice could be focusing on different aspects of plant performance.
While efficiency is measuring plant's performance in pure economic sense (i.e. technical
efficiency), practices focus more on what plant does on shop floor and strategy

formulation (i.e. allocative efficiency) etc.

5.5.1 . T est of Hypotheses 1

Wilk's A for the interaction terms in the nine MANCOVA models are listed in
Table 5.20.

Table 5.20 shows that there is Significant interaction between six out of nine
WCM practices and efficiency. The results indicate that three of the nine hypotheses are

not supported. The three hypotheses are H1 f, H1 g and th.

98

In order to assess the other six hypotheses, regression coefficients need to be
examined for these six practices. Table 5.21 lists the regression coefficients for

interaction terms of the six models remaining to be tested.

Table 5.20 Wilk's lambda for Nine MAN COVA Models Testing Hypothesis 1

 

 

 

 

 

 

 

 

 

 

 

Practice Wilk's A F Sig-

P1: Quality Management Programs .761 13.443 .000
P2: Customer Orientation .822 9.283 .000

P3: Labor Force Management .806 10.304 .000
P4: Operations Flow .824 9.109 .000

P5: Operations Capability .900 4.746 .001

P6: Supplier Relationship .958 1.869 .118

P7: Technology .969 1.376 .244

P8: New Product Development .983 .747 .561
P9: Strategy and Measurement .818 9.531 .000

 

 

 

 

 

99

Table 5.21 Regression Coefficients for Interaction Terms in MANCOVA

Pl: Quality Management Programs

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Dependent B Std. Error t Sig.
variable
Cost .622 .179 3.468 .001
Quality .005 .152 .033 .974
Delivery .322 .181 1.784 .076
Flexibility .648 .140 4.614 .000
P2: Customer Orientation
Dependent B Std. Error t Sig.
variable
Cost -.522 .154 -3.392 .001
Quality .279 .131 2.132 .034
Delivery .125 .159 .786 .433
Flexibility .470 .129 3.630 .000
P3 : Labor Force Management
Dependent B Std. Error t Sig.
variable
Cost .331 .150 2.205 .029
Quality .183 .128 1.433 .154
Delivery .460 .142 3.245 .001
Flexibility .552 .106 5.209 .000

 

 

 

 

 

100

 

 

 

Table 5.21 Regression Coefficients for Interaction Terms in MANCOVA

P4: Operations Flow

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Dependent B Std. Error t Sig.
variable
Cost .145 .139 1.040 .300
Quality -.01 1 .119 -.095 .925
Delivery .469 .122 3.832 .000
Flexibility .51 1 .098 5.223 .000
P5: Operations Capability
Dependent B Std. Error t Sig.
variable
Cost -.279 .160 -1.747 .082
Quality -.118 .137 -.861 .390
Delivery .224 .151 1.488 .139
Flexibility .526 .125 4.199 .000
P9: Strategy and Measurement
Dependent B Std. Error t Sig.
variable
Cost -.059 .188 -.312 .755
Quality —.303 .153 -1.983 .049
Delivery .697 .185 3.766 .000
Flexibility .712 .137 5.190 .000

 

 

 

 

 

101

 

 

 

 

5. 5. 2. Hypothesis 1 Test Result and Analysis

The results from Table 5.20 and Table 5.21 provide us with an assessment of the
nine sub-hypothesis of hypothesis 1. The result and its implication are discussed in this
section.
Hla

The interaction between Quality Management Programs (QMP) and efficiency is
significant on performance (p= .000). The interactions between QMP and three of the
four performance variables are positively significant. These three performance variables
and their regression coefficients are: cost (.622), delivery (.322), and flexibility (.648).
Efficiency as a positive and significant moderator of the relationship between QMP and
performance is supported. That is, Hla is supported.
HI b

The interaction between Customer Orientation (CO) and efficiency is significant
on performance (p= .000). The interactions between CO and two of the four performance
variables are positively significant. These two performance variables and their regression
coefficients are: quality (.279), and flexibility (.470). Efficiency as a positive and
significant moderator of the relationship between CO and performance is supported. That
is, Hlb is supported.
H] c

The interaction between Labor Force Management (LFM) and efficiency is
significant on performance (p= .000). The interactions between LFM and three of the
four performance variables are positively significant. These three performance variables

and their regression coefficients are: cost (.331), delivery (.460), and flexibility (.552).

102

Efficiency as a positive and significant moderator of the relationship between LFM and
performance is supported. That is, ch is supported.
HId

The interaction between Operations Flow (OF) and efficiency is significant on
performance (p= .000). The interactions between OF and two of the four performance
variables are positively significant. These two performance variables and their regression
coefficients are: delivery (.469), and flexibility (.511). Efficiency as a positive and
significant moderator of the relationship between OF and performance is supported. That
is, Hld is supported.
H] e

The interaction between Operations Capability (OC) and efficiency is significant
on performance (p= .001). The interaction between OC and one of the four performance
variables is positively significant. This performance variable and its regression
coefficients are: flexibility (.526). Efficiency as a positive and significant moderator of
the relationship between OC and performance is supported. That is, Hle is supported.
H1 f

The interaction between Supplier Relationship (SR) and efficiency is not
significant on performance (p= .118). Efficiency as a positive and significant moderator
of the relationship between SR and performance is not supported. That is, Hlf is not

supported.

103

H] g

The interaction between Technology and efficiency is not significant on
performance (p= .244). Efficiency as a positive and significant moderator of the
relationship between Technology and performance is not supported. That is, ng is not
supported.
HIh

The interaction between New Product Development (NPD) and efficiency is not
significant on performance (p= .561). Efficiency as a positive and significant moderator
of the relationship between NPD and performance is not supported. That is, th is not
supported.
H1i

The interaction between Strategy and Measurement (SM) and efficiency is
significant on performance (p= .000). The interaction between SM and two of the four
performance variables are positively significant. These two performance variables and
their regression coefficients are: delivery (.697), and flexibility (.712). Efficiency as a
positive and significant moderator of the relationship between SM and performance is
supported. That is, Hli is supported.

The further implications and significance of the MANCOVA testing of hypothesis

1 and the ensuing results are discussed in chapter 6.

104

5. 6. Cluster Analysis

Cluster analysis was performed on the 75 WCM plants to test hypothesis 2 that
there is more than one strategic group in WCM plants. The four performance constructs -
- cost, quality, delivery and flexibility which manifested the plant's strategic orientation
were chosen as cluster variate -- the set of variables representing the characteristics used
to compare objects in the cluster analysis.

5. 6. I . Hierarchical Procedure

Since most statistical packages use agglomerative methods versus divisive
methods and agglomerative method act as divisive method in reverse, agglomerative
method was applied for hierarchical clustering procedure. There are five popular
agglomerative procedures depending on how the distance between clusters is defined: 1)
Single linkage, 2) complete linkage, 3) average linkage, 4) Ward's method, and 5)
centroid linkage. In Single linkage, the distance between two clusters is defined as the
minimum distance between two individual observations each from one cluster while in
complete linkage, it is defined as the maximum distance between two individual
observations each from one cluster. The distance is defined as average distance from all
individuals in one cluster to all individuals in another in average linkage and distance
between their centroid in centroid method. In the Ward's method, the distance between
two clusters is the sum of squares between the two clusters summed over all variables.
Ward's method is most popular since it tends to produce clusters with approximately the
same number of observations.

The optimal number of clusters can be determined by examining agglomeration

coefficient. The agglomeration coefficient increases as the number of cluster decreases.

105

 

The optimal number of clusters would be the one that has highest percentage increase in
agglomeration coefficient to the lower number of clusters (Aldenderfer and Blashfield
1984).

Hierarchical clustering is run using SPSS by applying Ward's method as cluster
method and squared Euclidean distance as distance measure. Five outliers are identified.
Therefore the sample Size for clustering analysis dropped from 75 to 70. Agglomeration
schedule is examined to determine what is the best number of clusters. As can been seen
from Table 5.22, agglomeration coefficient has increased most significantly when
number of clusters decreased from three to two. Therefore, three is considered the best
number of clusters based on the agglomeration coefficient.

Three is also the most favored number of clusters in strategy literature. Porter
(1980) classified strategy into three groups (cost, segmentation and focus) conceptually.
Miller and Roth (1994) empirically classified American manufacturers also into three
groups. The three groups by Miller and Roth's taxonomy are caretakers, marketers, and
innovators.

Based on both empirical evidence and theoretical considerations, three was
chosen as the number of strategic groups in our study. The three cluster centers resulted
from Ward's method are listed in Table 5.23. The four performance variables in each
cluster center are ranked by their magnitude to provide us an idea how each strategic

group in WCM place their priority among the four performance variables.

106

 

Table 5.22 Agglomeration Coefficients from Ward's Method

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Number of Agglomeration Percentage Increase in
Clusters Coefficient Agglomeration Coefficient to
lower number of clusters

10 23.91 38.77
9 33.18 28.18
8 42.53 23.87
7 52.68 24.05
6 65.35 22.2

5 79.86 23.99
4 99.02 22.85
3 121.65 58.27
2 192.53 33.49
1 256.94

 

Table 5.23 The Cluster Centers from Ward's Method

 

 

 

 

 

 

Cluster 1 Cluster 2 Cluster 3
(n=24) (n=29) (n=17)
Cost
Mean (rank) 2.58 (4) 4.72 (4) 4.47 (2)
Quality
Mean (rank) 5.30 (1) 5.90 (3) 5.03 (1)
Delivery
Mean (rank) 5.05 (2) 6.36 (1) 3.22 (4)
Flexibility
Mean (rank) 4.33 (3) 5.95 (2) 4.00 (3)

 

 

 

 

 

107

 

 

5. 6.2. K-means Clustering

The three cluster centers computed from Ward's method are selected as the initial
cluster centers for K-means clustering. The distances between cluster centers are listed in
Table 5.24.

After specifying the number of clusters as three, the K-means clustering
procedure was run to fine-tune the clustering result. It turned out that none of the
observations switched membership after K—means clustering. This indicated the stability
of cluster solution. The validity of the cluster solution was further investigated by

examining analysis of variance (ANOVA) and discriminant analysis results.

Table 5.24 Distances Between Cluster Centers

 

 

 

 

Cluster 1 2 3

1 3.04 2.67
2 3.04 3.80
3 2.67 3.80

 

 

 

 

 

 

5. 6. 3. Validity and Hypothesis 2 Result

K-means clustering procedures in SPSS also print out the result of ANOVA test
using cluster membership as independent variable and cluster variates (i.e. strategic
variables) as dependent variables. The ANOVA test fiirther examines whether all of the
four strategic variables (i.e. cost, quality, delivery, and flexibility) have significant power
in differentiating the three clusters. The results of ANOVA test are shown in Table 5.25
from which we can see that all four strategic variables have significantly differentiated

the clusters.

108

Table 5.25 AN OVA Result for K-means Clustering

 

 

 

 

 

F si g
Cost 56.12 .000
Quality 12.18 .000
Delivery 1 18.84 .000
Flexibility 67.38 .000

 

 

 

 

 

Discriminant analysis using the original clustering variables as independent
variables, the cluster membership as dependent variable can serve another test for validity
of cluster solution (Dowling and Midgley 1988). Discriminant analysis assigned 98.6%
(69/70) of the plants to their correct clusters, which is another indication of strong
validity of cluster solution.

These results demonstrated that even using different procedures for clustering,
assignments of plants to clusters were relatively stable and reliable. Clusters were also
quite well differentiated by the strategic variables. The degree of reproducibility and
stability is strong enough for us to believe that despite the relative small sample size, the
cluster solution is adequately robust and empirically and theoretically coherent. The
strong results of validity test of cluster analysis leads us to believe that hypothesis 2 is
supported. That is, there is more than one strategic group in efficient manufacturing

plants.

109

 

5. 7. ANO VA and Multiple Range Tests

One way ANOVA tests serve two purposes. First, hypotheses 3 that strategic
groups within efficient manufacturing plants have different priorities in WCM practices
can be tested through ANOVA. Second, these ANOVA tests can also be a good validity
test for cluster solution achieved in section 5.6. Since the dependent variables in ANOVA
will be WCM practice constructs which are different from clustering variables (strategic

variables), the significant result of ANOVA will also validate the clustering solution fi'om

section 5.6.

5. 7.1. One Way ANOVA Tests

Nine one way ANOVA tests were carried out using WCM practices as dependent
variable, clustering membership as independent variable. All of the WCM practices
except Supplier Relationship and New Product Development proved to be highly

significant (p< 0.05). Table 5.26 shows the results of ANOVA tests.

 

 

 

 

 

 

 

 

 

 

Table 5.26 AN OVA Results

Factors F p
F 1: Quality Management Programs 52.88 0.000
F2: Customer Orientation 3.59 0.033
F 3: Labor Force Management 42.26 0.000
F 4: Operations Flow 22.97 0.000
F 5: Operations Capability 16.40 0.000
F6: Supplier Relationship 2.28 0.110
F7: Technology 8.19 0.001
F 8 : New Product Development 0.33 0.717
F9: Strategy and Measurement 9.42 0.000

 

 

 

 

110

 

5. 7.2. Multiple Range Tests

ANOVA results from Table 5.28 show that the three WCM strategic groups have
different means on seven of the nine WCM practices. In order to examine which pairs of
means are different for these seven WCM practices, post hoc tests are required. Tukey
multiple comparison tests are performed on these seven WCM practices. Table 5.27

shows the results of Tukey multiple range tests.
Table 5.27 Tukey Multiple Range Test Results on Seven WCM Practices

F 1: Quality Management Programs

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Cluster i Cluster j Absolute Mean Std. Error Sig.
Difference

1 2 1.679 .163 .000

2 3 .794 .181 .000

3 1 .885 .188 .000

F 2: Customer Orientation

C luster i Cluster j Absolute Mean Std. Error Sig.
Difference

1 2 .072 .271 .962

2 3 .698 .300 .059

3 l .771 .312 .042

 

 

 

 

 

111

 

 

F 3: Labor Force Management

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Cluster 1 Cluster j Absolute Mean Std. Error Sig.
Difference
1 2 1.730 .209 .000
2 3 1.620 .231 .000
3 1 .110 .240 .890
F4: Operations Flow
Cluster 1 Cluster j Absolute Mean Std. Error Sig.
Difference
1 2 1.523 .298 .000
2 3 2.034 .330 .000
3 1 .511 .343 .302
F5: Operations Capability
Cluster i Cluster j Absolute Mean Std. Error Sig.
Difference
1 2 .170 .283 .821
2 3 1.704 .313 .000
3 1 1.534 .325 .000
F 7: Technology Management
Cluster 1 Cluster j Absolute Mean Std. Error Sig.
Difference
1 2 1.059 .262 .000
2 3 .457 .290 .263
3 1 .603 .301 .119

 

 

 

 

 

112

 

 

 

 

F 9: Strategy Formulation and Measurement

 

 

 

 

Cluster 1 Cluster j Absolute Mean Std. Error Sig.
Difference

1 2 .628 .230 .022

2 3 1.068 .255 .000

3 1 .440 .264 .226

 

 

 

 

 

 

 

The descriptive statistics of nine WCM practices dimensions for the three clusters

and ranking of the three clusters on practices based on the results of multiple range tests

are shown in Table 5 .28.

High, Middle, and Low are used to indicate ranking based on multiple range test
results. If two clusters differ on the mean of certain practice, they received different
rankings; otherwise they received the same ranking. For example, Tukey tests reveal that
Quality Management Programs score differ between any pairs of the clusters, therefore,
the three clusters receive ranking as "high", "middle" and "low" separately. For
Technology Management, cluster 1 and 2 have different means, however, cluster 3 has
the same mean as both cluster 1 and cluster 3. Therefore, cluster 1 and 2 have different
rankings ("low" and "high") and cluster 3 receive a ranking "high-low" to differentiate
from the case that cluster 3 has different means from cluster 1 and 2. There is no
Significant difference found for Supplier Relationship and New Product Development.
Since all three clusters score generally high on Supplier Relationship (mean>4.5) and
generally low on New Product Development (mean<3.9), all three clusters received

r anking of "high" on Supplier Relationship and "low" on New Product Development.

113

Table 5.28 Descriptive Statistics of WCM Practices for the Clusters and Multiple

Range Test Results

F1: Quality Management Programs

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Cluster N Mean Rank Std. Error
1 24 3.39 Low 0.14
2 29 5.07 High 0.08
3 17 4.27 Middle 0.16
Total 70 4.3 0.1 1

F 2: Customer Orientation

Cluster N Mean Rank Std. Error
1 24 4.22 High 0.28
2 29 4.15 High 0. 12
3 1 7 3.45 Low 0. 18
Total 70 4 0. 12

F3: Labor Force Management

Cluster N Mean Rank Std. Error
1 24 3 .5 Low 0. 14
2 29 5 .23 High 0.12
3 1 7 3 .61 Low 0.24
Total 70 4.24 0.13

 

114

 

 

 

 

 

 

——-|
—

C11

1

CT

”I

'1 CT, V1

1--

[CFC/”’7”

 

F4: Operations Flow

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Cluster N Mean Rank Std. Error
1 24 4.24 Low 0.2
2 29 5.76 High 0.16
3 17 3.73 Low 0.35
Total 70 4.74 0.17
F5: Operations Capability
Cluster N Mean Rank Std. Error
1 24 5.42 High 0.11
2 29 5.59 High 0.19
3 17 3.88 Low 0.36
Total 70 5.1 l 0.15
F6: Supplier Relationship
Cluster N Mean Rank Std. Error
1 24 4.63 High 0.13
2 29 5.02 High 0.23
3 17 5.33 High 0.28
Total 70 4.96 0.13

 

 

 

 

 

 

115

Fun...

F 7: Technology Management

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Cluster N Mean Rank Std. Error
1 24 3.46 Low 0.23
2 29 4.52 High 0.18
3 17 4.06 High-low 0.13
Total 70 4.04 0.12

F8: New Product Development

Cluster N Mean Rank Std. Error
1 24 3.87 Low 0.32
2 29 3.7 Low 0.23
3 1 7 3.55 Low 0.19
Total 70 3 .72 0.15

F9: Strategy Formulation and Measurement

Cluster N Mean Rank Std. Error
1 24 4.68 Low 0.21
2 29 5.3 High 0.16
3 17 4.24 Low 0.12
Total 70 4.83 0.11

 

 

 

 

 

116

 

 

 

 

5. 7.3. Hypothesis 3 Test Result and Analysis

Hypothesis 3 states that WCM strategic groups have difference scores on nine
practice dimensions. The assessment of hypothesis 3 can be made from the result of
ANOVA tests which are shown in table 5.26.
H_3g
ANOVA result shows that the three strategic groups in WCM have different scores on
Quality Management Programs. That is, H3a is supported.
H312
ANOVA result shows that the three strategic groups in WCM have different scores on
Customer Orientation. That is, H3b is supported.
113p

ANOVA result shows that the three strategic groups in WCM have different
scores on Labor Force Management. That is, H3c is supported.
Ed

ANOVA result shows that three strategic groups in WCM have different scores
on Operations Flow. That is, H3d is supported.
fig

ANOVA result shows that three strategic groups in WCM have different scores
on Operations Capability. That is, H3e is supported.
1131'

ANOVA result shows that three strategic groups in WCM have the same scores

on Supplier Relationship. That is, H3f is not supported.

117

onl

E

on‘

'13:
I'J’

011

 

1.?

ANOVA result shows that three strategic groups in WCM have different scores
on Technology. That is, H3 g is supported.
H3b

ANOVA result shows that three strategic groups in WCM have the same scores
on New Product Development. That is, H3h is not supported.

3.3.1

 

ANOVA result shows that three strategic groups in WCM have different scores

on Strategy and Measurement. That is, H3i is supported.

 

The implications from multiple range tests are discussed in chapter 6.

118

CHAPTER 6
DISCUSSION OF RESULTS

Five statistical procedures were carried out to achieve the two major objectives of
this dissertation. In-depth examination of these test results adds more knowledge to
literature and provides managerial insights for practitioners. This chapter builds on the
findings of previous chapters by discussing the implications from MANCOVA testing
interaction between practices and efficiency, cluster analysis, ANOVA and multiple
range tests. Contributions are presented in the following section (6.3). Finally, the last
section (6.4) discusses limitations of this dissertation and future research possibilities that
arise as a result of this study.

6.1. Implications of Hypothesis 1 Test Results

Efficiency was proposed as a performance measurement tool for selecting the
“best performing” manufacturing plants due to the consideration that neither practice
based definition of WCM or performance based definition of WCM is sufficient. Practice
based definition assumes that once a manufacturing plant adopts a set of practices, it will
achieve WCM status. Performance based definition assumes that there is only one type of
strategy existing in WCM. Selecting most efficient plants relating outputs to inputs in
economic sense does not require prior knowledge of how the inputs and outputs are
related. Neither does the inclusion or exclusion of a specific practice or performance
dimension affect the result.

It was discovered from the results of hypothesis 1 that efficiency not only
measures the ratio of output to input in the economic sense (capital inputs and sales and

productivity improvement outputs), but also positively moderates the ratio of

119

manufactming performance to manufacturing practices. That is, an economically efficient
plant is also the one that has the strongest relationship between manufacturing
performance and manufacturing practices.

This finding is interesting since the manufacturing plants are indeed most
concerned with how effective manufacturing practices are rather than interested in
calculating how many practices they have adopted (as Schonberger's l6 principles do) if
in fact these practices do not necessarily improve performance.

The findings of hypothesis 1 also correspond to practice dimensions proposed by
most WCM literature. Efficiency was found to have a positive interaction with the most
commonly endorsed practice dimensions -- Quality Management Programs, Customer
Orientation, Labor Force Management, Operations Flow, Operations Capability, Strategy
and Measurement (see Table 3.1.for how literature supports each practice dimension).
The interaction between efficiency and three relatively recent practice dimensions
(Supplier Relationship, Technology, and New Product Development) was not found to be
significant. However, this does not mean that WCM plants are not adopting these
practices. For example, ANOVA and multiple range test results show that all three
clusters in WCM plants score high in Supplier Relationship dimension. Non-significant
interaction indicates that the strength of relationship between Supplier Relationship and

performance was not found significantly different between efficient manufacturing plants

and less efficient plants.

120

6. 2. Implications of Hypothesis 2 and 3 Test Results

Combining the results from Table 5.23 and Table 5.28, the differences among the
three clusters are listed below.

The importance cluster 1 places on performance metrics from high to low follows
this sequence: Quality, Delivery, Flexibility and Cost. This sequence is the same as the
sequence proposed by sand cone model (Ferdows and De Meyer 1990). Sand cone model
proposes that if you follow this sequence, the tradeoffs between various performance
metrics will be minimized. However, this cluster is not the best performing cluster in
terms of competitive performance as sand cone model may have predicted. The
relationship among the performance metrics could be more complicated than that
proposed by the sand cone model. It may also be mitigated by the strategy the
manufacturing plant adopts. Compared to the other strategic groups that are also efficient,
this group aims to provide reliable and durable products and fast and on time delivery
without a strong focus on cost reduction. Cluster 1 can be labeled as “Quality and
Delivery (QD)” based on its strategy that is manifested by its competitive priorities.

Multiple range test results as shown in table 5.28 indicated that members of this
group, aiming to provide high quality with customization, have invested their time and
resources highly on three of the nine practice dimensions. These three dimensions are:
customer orientation, operations capability and supplier relationship. It is worth noting
that this group did not focus their attention around Quality Management Programs as one
might expect. A close examination of the measurement items of Quality Management
Programs and Customer Orientation reveals that while Customer Orientation Program

intends to define quality by customer satisfaction, Quality Management Programs focuses

121

on reducing reject rates. The seemingly paradoxical result that “Quality and Delviery”
group does not focus on Quality Management Programs suggests that defining
customers’ needs is the key to achieving quality competitiveness in industry.

The importance cluster 2 places on performance metrics from high to low follows
this sequence: Delivery, Flexibility Quality, and Cost. Cluster 2, compared to the rest of
the efficient manufacturing plants, placed high priorities on all the practice dimensions
except New Product Development. As a result, cluster 2 performs best on all four
dimensions of competitive performance compared to the other two clusters. This cluster
meets the criteria of both practice-based and performance-based definitions of WCM. In
addition, this cluster meets the pre-condition defined in this dissertation: technical
efficiency.

Since this group of manufacturing plants meets multiple criteria of WCM, which
are defined either in literature or in this dissertation, it is considered to be the only group
that deserves the title of “WCM”. Up to now, this dissertation proposed a new definition
of WCM. This definition not only integrated the requirements of both practice-based and
performance-based definition, but also imposed a new criterion: technical efficiency. A
plant has to be efficient in the economic sense. The validity of efficiency is established by
its moderating effect on the relationship between practices and performance, which is
verified by test results of hypothesis 1.

The four competitive advantages, sequenced by the priority that cluster 3 places
on them are Quality, Cost, Flexibility, and Delivery. Cluster 3, among all three clusters
that are efficient, is the only one that places cost in such a high priority. This cluster,

compared to the other two clusters, focuses on a “Cost/Value (CV)” leadership.

122

Multiple range tests reveal that CV group focuses highly on three out of nine
practice dimensions to achieve its cost/value leadership. These three practice dimensions
are Quality Management Programs, Supplier Relationship and Technology. CV also
emphasized Quality Management Programs, but not to a great degree. Those practice
dimensions that CV group chose not to invest in aggressively are Customer Orientation,
Labor Force Management, Operations Flow, Operations Capability, New Product
Development, Strategy Formulation and Measurement. The list of practices chosen by
CV suggests that CV finds its niche in cost/value leadership by ignoring the long term
and expensive programs and focusing only on the quick and most effective cost reduction
programs. Technology program has a reputation of being risky, less profitable, and
paying off slowly. Therefor managers with only short-term goals may not favor it.
However, the adoption of Technology Program by CV group shows that keeping up with
current technology does help with keeping cost low and quality high.

The taxonomy analysis of efficient manufacturing plants indicates that these
plants, although all technically efficient, have adopted different approaches toward the
so-called “WCM practices” due to the differences in their strategies.

It shows that to achieve the “WCM status”, a manufacturing plant still needs to
adopt almost all the practice programs and performs well on all the performance areas.
However, Cost/value leadership (CV) and quality and delivery (QD) groups managed to
find their niche by investing in only a small set of practice programs and be efficient. Due
to the limited resources, this means a lot to a plant wanting to benchmark and become
more efficient. It does not have to benchmark a WCM plant if it does not plan to become

one. Instead, based on its own strategy, the plant can selectively choose a small set of

123

 

 

1.“ .‘gi'gt .

practices to engage in for the purpose of achieving leadership in certain performance

areas.

6.3. Contributions

This dissertation makes four major contributions to operations strategy literature.

First, this research represents one of the first studies to synthesize and validate the
dimensions of WCM practices in literature. Comprehensive measurement models, which
may provide a foundation for further empirical testing in this area are proposed and
empirically tested.

Second, efficiency as defined in this dissertation is based on the traditional ideas
in economics relating outputs to inputs. It overcomes the drawback in some empirical
studies that purport to study the relationship between practices and performance but have
questionable ways of defining WCM. These studies have problems of tautology since
they rely on either practices or performance to define WCM. One can claim that the
examination of the relationship between practice and performance in this dissertation is
more robust. It reconciles the notion of technical efficiency with the notion of world
class.

Third, results of cluster analysis and comparison among the three clusters within
efficient manufacturing plants indicate that these manufacturing plants have adopted
different strategies to become efficient. What is more important is that these groups have
different priorities on WCM practices. This further indicates that the approach of current
WCM literature, which does not realize that differences exist in priorities on

manufacturing practices within best performing plants due to strategy differences, is

124

 

inadequate. This research points to the need to further investigate how the strategy
differences affect the relationship between practices and performance.

Finally, the research findings suggested that to be termed WCM requires all three
criteria: 1) engage in WCM practices, 2) be technically efficient and 3) perform well on

all the performance areas.

6. 4. Limitations and Future Research Directions

The major limitation of this dissertation is that the sample size is relatively small
so that carrying out a research design that yields more information and tests more
relationships among the constructs in one model was not be possible.

This dissertation represents one of the first efforts that integrated the dimensions
of WCM practices in literature and empirically tested the measurement model for WCM
practices. Further efforts in offering and testing measurement models of WCM practices
on different data sets will help providing a valid and comprehensive model to facilitate
future empirical studies in the area of WCM.

This dissertation is the first study that utilized efficiency analysis, taxonomy
analysis to solve the questions arising from the context of world class manufacturing.
This study identified three strategic groups within the efficient manufacturing plants and
showed that strategic difference affects plants’ decision as to which practices they should
pay more attention to. Strategy could make a lot of differences in more than resource
employment decision. For example, strategy may account for the differences in the slope
between performance and practices that this dissertation was unable to study due to the

smallness of the sample.

125

 

This dissertation successfully identified a group of manufacturing plants that
perform well on all performance areas. However, whether the trade-off among the
performance areas exist is still unclear. With a bigger sample size, models that test the
relationship among the performance dimensions will provide us more insight into this
debate.

The recommendation that technical efficiency be used as one of the pre-requisites
of WCM is a relatively new idea in operations strategy. The choice of input and output
variables will significantly affect the quality of the results. Efficiency analysis has a great
potential in performance evaluation in manufacturing strategy due to its simplicity and
few assumptions. Yet experience from further studies using this tool will help perfect this

process.

126

 

APPENDIX A

 

MEASUREMENT INSTRUMENT

127

APPENDIX A

MEASUREMENT INSTRUMENT

 

Michigan State University
World Class Manufacturing / Benchmarking Study

 

 

Please refer any questions regarding this study to Cynthia Wang, Michigan State
University (517) 353-6381 All returned surveys will remain confidential.

 

 

First Name

 

 

Last Name

Title

 

 

 

 

_———-l
_l—I_.L—

Company
Address
City

 

 

 

 

 

 

 

State
Zip Code

 

 

 

_

 

 

 

Phone number

 

Please enter your

 

al.—T—

survey ID on the
letter you

received from us

 

 

 

 

 

 

 

 

 

Email address [ ]
Password [................... I
Confirm [““uuu I
password

 

 

 

128

 

 

Plant Characteristics

 

 

 

 

 

average annual change in labor productivity over the %
last two years (in %)

average annual change in sales over the last two years %

average annual change in profitability over the last two %
years

average annual change in market share over the last two %
years

What has been the market share for your primary %

products over the last two years (In %)?

 

What is your primary SIC code?

 

 

What is your secondary SIC code?

 

 

 

How many factory employees work at your plant?

 

 

How many non-factory employees work at your plant?

 

 

 

 

 

 

 

 

What was the total international sales for your plant last S
year (in S)?
What was the domestic sales for your plant last year (in S
S)?
What was the total raw material purchase in your plant S
last year (In S)?
How much did your plant invest in equipment over the S

 

last 3 years (in S)?

 

How much did your plant invest in information S

 

technology over the last 3 years (in S)?

 

ROI

 

 

ROA

 

 

 

Number of different major products produced at your

 

plant?

 

 

129

1““ Juan-.3-..
'i .

 

 

Plant Characteristics

 

How would you characterize your major product line?

@make to stock
Omake to order
OEngineer to order

OAssemble to order

 

 

 

 

 

Percentage of continuous flow (24 hour operation) in %
your plant
Percentage of assembly line in your plant %
Percentage of job shop (wide variety of custom products) %
in your plant
Percentage of batch (medium sized lots of production) in %

your plant

.3. .hhu - xv.”
3::

 

 

 

130

Comparing your plant to the standard or average in your industry, indicate the extent to

which the following statements apply.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Quality Management Don't Much About To a
Programs Know less the much
same greater
extent

Everyonehas O 10 2 3O 4O 50 6O 7O

responsibility to improve N/A 0

puality.

Continuous improvement 0 1 O 2 3 O 4 O 5 O 6 O 7 O

of quality is expected in N/A 0

all work processes.

Workers are financially O 1 O 2 3 O 4 O 5 O 6 O 7 O

rewarded for quality N/A 0

improvement.

Processesontheshop O 10 2 3O 4O 50 6O 7O

floor are under statistical N/A 0

quality control.

Weusestatistical O 10 2 30 4O 50 6O 7O

techniques to reduce N/A 0

variance in processes

We measure process 0 1 O 2 3 O 4 O 5 O 6 O 7 O

capability (e.g., Cplg) N/A 0

We use visual control 0 1 O 2 3 O 4 O 5 O 6 O 7 0

techniques (e.g., charting N/A 0

defects, schedule, etc.).

SPCmethodisused O 10 2 3O 4O 50 6O 70
N/A 0

Operators inspect their 0 1 O 2 3 O 4 O 5 O 6 O 7 O

own work N/A 0

QualityFunction O 10 2 3O 4O 50 6O 7O

Deployment is applied N/A 0

TotalQuality O 10 2 3O 4O 50 6O 70

Management philosophy N/A 0

is promoted

Design of Experiments is O 1 O 2 3 O 4 O 5 O 6 O 7 0

used N/A 0

 

 

131

Comparing your plant to the standard or average in your industry, indicate the extent to
which the following statements apply.

 

 

 

 

 

 

 

 

 

 

 

Customer Don't Much About To a
Orientation Know less the much
same greater
extent
We maintain close 0 1 O 2 3 O 4 O 5 O 6 O 7 0
contacts with our N/A 0
customers. r
Results of customer 0 1 O 2 3 O 4 O 5 O 6 O 7 O ‘
satisfaction surveys are N/A 0 1"
communicated throughout
organization. .‘
We actively create 0 1 O 2 3 O 4 O 5 O 6 O 7 O <
opportunities for N/A 0
customer-employee "T
interaction. to“
Manufacturing operations 0 l O 2 3 O 4 O 5 O 6 O 7 O
are organized around N/A 0
specific customers.
Mfg. operations are 0 1 O 2 3 O 4 O 5 O 6 O 7 O
organized around product N/A 0
families with similar
customer needs.
We produce at a rate close 0 1 O 2 3 O 4 O 5 O 6 O 7 O
to the customer's rate of N/A 0
use.
Wehaveaforrnal O 10 2 30 4O 50 6O 70
"customer satisfaction" N/A 0
rogram in place.

 

132

Comparing your plant to the standard or average in your industry, indicate the extent to

which the following statements apply.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Labor Force Don't Much About To a

Management Know less the much
same greater

extent

We actively engage in O 10 2 3 O 4 O 5 O 6 O 7 O

skill development N/A 0

Direct labor technical O 10 2 3 O 4 O 5 O 6 O 7 O

competency is high N/A 0

Employees are cross 0 1 O 2 3 O 4 O 5 O 6 O 7 O

trained to perform a N/A 0

variety of activities

Teammembersare O 10 2 3O 4O 50 6O 7O

encouraged to exchange N/A 0

opinions/ideas.

Supervisors and teams 0 1 O 2 3 O 4 O 5 O 6 O 7 O

communicate frequently. N/A 0

Ouremployeeshave O 10 2 3O 4O 50 6O 70

strong problem solving N/A 0

abilities.

Shop floor employees are 0 1 O 2 3 O 4 O 5 O 6 O 7 O

rotated among jobs. N/A 0

Employees are rewarded O 1 O 2 3 O 4 O 5 O 6 O 7 O

for learning new skills. N/A 0

Weuseproblemsolving O 10 2 3O 4O 50 6O 7O

ability as a criterion in N/A 0

employee selection

Weuseknowledgeand O 10 2 3O 4O 50 6O 7O

skill level as a criterion in N/A 0

employee selection

Weuseabilitytoworkin O 10 2 3O 4O 50 6O 70

a team as a criterion in N/A 0

employee selection

Weuseworkvaluesand O 10 2 3O 4O 50 6O 7O

ethics as a criterion in N/A 0

employee selection

Production scheduling is O 1 O 2 3 O 4 O 5 O 6 O 7 O

handled by empowered N/A 0

teams

Quality assurance is O 1 O 2 3 O 4 O 5 O 6 O 7 O

handled by empowered N/A 0

teams

 

133

 

 

 

 

 

 

 

Labor Force Don't Much About To a
Management Know less the much
same greater

extent

Skills certification and O 1 O 2 3 O 4 O 5 O 6 O 7 0

training is handled by N/A 0

empowered team

Laborscheduling/job O 10 2 3O 4O 50 6O 7O

assignments is handled by N/A 0

empowered team

Hiring/firingofteam O 10 2 3O 4O 50 6O 70

members is handled by N/A 0

empowered team

Performance reviews are 0 1 O 2 3 O 4 O 5 O 6 O 7 O

handled by empowered N/A 0

teams

 

 

Comparing your plant to the standard or average in your industry, indicate the extent to

which the following statements apply.

 

 

 

 

 

 

 

 

 

 

Operations Flow Don't Much About To a
Know less the much
same greater
extent
Theshopfloorislaidout O 10 2 30 4O 50 6O 70
to optimize processing N/A 0
sequence and flow.
Weuseapullsystemin O 10 2 3O 4O 50 6O 70
production. N/ A 0
We produce in small lot 0 10 2 3 O 4 O 5 O 6 O 7 0
sizes in this plant. N/A 0
Wehavea"focused O 10 2 3O 4O 50 6O 7O
factory" (e. g., product or N/A 0
process focused)
production system.
Weuselean O 10 2 3O 4O 50 6O 70
manufacturing production N/A 0
methods.
WeofferJITdeliveryto O 10 2 3O 4O 50 6O 70
customers. N/A 0
Ourproductionflow O 10 2 3O 4O 50 6O 7O
utilizes manufacturing N/A 0

cells.

 

 

134

Comparing your plant to the standard or average in your industry, indicate the extent to

which the following statements apply.

 

 

 

 

 

 

 

 

 

Operations Don't Much About To a

Capability Know less the much
same greater

extent

We dedicate to continual O 1 O 2 3 O 4 O 5 O 6 O 7 O

improvement in quality. N/A 0

We dedicate to continual O 1 O 2 3 O 4 O 5 O 6 O 7 O

improvement in response N/A 0

time.

We dedicate to continual O 1 O 2 3 O 4 O 5 O 6 O 7 O

improvement in N/A 0

flexibilig.

We dedicate to continual O 1 O 2 3 O 4 O 5 O 6 O 7 O

improvement in cost. N/A 0

We develop measurement 0 1 O 2 3 O 4 O 5 O 6 O 7 0

systems that encourage N/A 0

continual learning.

We develop operations 0 1 O 2 3 O 4 O 5 O 6 O 7 O

that are able to respond N/A 0

rapidly to changes in

product and market.

We continually reduce O 1 O 2 3 O 4 O 5 O 6 O 7 O

variations and mishaps. N/A 0

 

135

 

 

Comparing your plant to the standard or average in your industry, indicate the extent to
which the following statements apply

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Supplier Don't Much About To a
Relationship Know less the much

same greater

extent

Our suppliers deliver on a O 1 O 2 3 O 4 O 5 O 6 O 7 O __
just-in-time basis. N/A 0 E
Werequiredaily O 10 2 3O 4O 5 0 6O 70 E
shipments from suppliers. N/A 0 L
Weuse'totalcost O 10 2 3O 4O 50 6O 7O 1
purchasing' in N/A 0 '
procurement decisions . *-
Our suppliers are certified 0 1 O 2 3 O 4 O 5 O 6 O 7 O J;
or pre-qualified. N/A 0 -"
We establish long-term O 1 O 2 3 O 4 O 5 O 6 O 7 O
contracts with suppliers. N/A 0
We have reduced the O 1 O 2 3 O 4 O 5 O 6 O 7 0
number of suppliers for N/A 0
each part family.
We frequently source 0 1 O 2 3 O 4 O 5 O 6 O 7 0
multiple part families N/A 0
from a single supplier.
We procure complete 0 1 O 2 3 O 4 O 5 O 6 O 7 0
systems / modules from N/A 0
major suppliers.
We require major O 1 O 2 3 O 4 O 5 O 6 O 7 O
suppliers to contribute to N/A 0
cost and quality
improvement efforts.
We no longer require 0 1 O 2 3 O 4 O 5 O 6 O 7 O
incoming inspection. N/A 0
Wehaveahighdegree of O 10 2 3 0 4O 50 6O 7O
mutual trust with our N/A 0
suppliers.
Wepursuejoint O 10 2 3O 4O 50 6O 7O
investments with N/A 0
suppliers.
We provide financial O 1 O 2 3 O 4 O 5 O 6 O 7 O
assistance to suppliers. N/A 0
We provide technological O 1 O 2 3 O 4 O 5 O 6 O 7 O
assistance to suppliers. N/A 0

 

136

 

 

 

 

 

 

 

 

 

Supplier Don't Much About To a

Relationship Know less the much
same greater

extent

We provide training in O 1 O 2 3 O 4 O 5 O 6 O 7 0

quality issues to supplier N/A 0

personnel.

We use buyer-supplier O l O 2 3 O 4 O 5 O 6 O 7 O

councils. N/A 0

Wesharerealtime O 10 2 3O 4O 50 6O 70

production schedule N/A 0

information with

suppliers.

Weshareourcost O 10 2 3O 4O 50 6O 70

information with our N/A 0

major suppliers.

Werequirecost O 10 2 3O 4O 50 6O 70

information sharing by our N/A 0

suppliers .

We select suppliers based 0 1 O 2 3 O 4 O 5 O 6 O 7 O

on design/innovation N/A 0

capabilities.

We select suppliers based 0 1 O 2 3 O 4 O 5 O 6 O 7 O

on abilities to N/A 0

accommodate volume /

mix changes.

 

137

 

Comparing your plant to the standard or average in your industry, indicate the extent to

which the following statements apply.

 

 

 

 

 

 

 

 

 

 

 

 

 

Technology Don't Much About To a
Management Know less the much
same greater
extent

Wedesignanddevelop O 10 2 3O 4O 50 6O 7O

proprietary equipment for N/ A 0

our own use.

We develop dedicated O 1 O 2 3 O 4 O 5 O 6 O 7 O

technologies for specific N/A 0

product families/customer

segments.

We improve present 0 1 O 2 3 O 4 O 5 O 6 O 7 0

equipment before N/ A O

considering new

equipment

We present human work 0 1 O 2 3 O 4 O 5 O 6 O 7 0

methods before N/ A O

considering new

automation

We seek simple, flexible, O 1 O 2 3 O 4 O 5 O 6 O 7 O

low-cost, readily available N/A 0

ecpripment

We plan technological O 1 O 2 3 O 4 O 5 O 6 O 7 O

upgrades to be consistent N/A 0

with infrastructural

upgrades

Computer-aided-design O 1 O 2 3 O 4 O 5 O 6 O 7 0

(CAD) technology N/A 0

practice is applied

Computer-aided- O 1 O 2 3 O 4 O 5 O 6 O 7 0

manufacturing (CAM) N/A 0

technology practice is

applied

Flexible manufacturing 0 l O 2 3 O 4 O 5 O 6 O 7 0

systems (FMS) technology N/A 0

practice is applied

Robotics is used. 0 IO 2 3 O 4 O 5 O 6 O 7 0
N/A 0

Automated storage 0 1 O 2 3 O 4 O 5 O 6 O 7 O

/retrieval systems N/ A O

tehcnology practice

(AS/RS) is applied

 

138

 

 

 

 

 

 

 

 

 

 

Technology Don't Much About To a

Management Know less the much
same greater

extent

Computer aided process 0 1 O 2 3 O 4 O 5 O 6 O 7 O

planning (CAPP) N/A 0

technology practices is

applied

Rapid prototyping (e.g., O l O 2 3 O 4 O 5 O 6 O 7 O

stereolithography) N/A 0

technology practice is

applied

AdvancedMRPHsystems O 10 2 3O 4O 50 6O 70

is used N/A 0

Enterprise Resource O 1 O 2 3 O 4 O 5 O 6 O 7 0

Planning system (ERP) is N/A 0

used

Product data management 0 l O 2 3 O 4 O 5 O 6 O 7 O

G’DM) systems is used N/A 0

EDI links to customers 0 1 O 2 3 O 4 O 5 O 6 O 7 O

and suppliers is used N/A 0

 

 

139

‘ I
I '.-. wish!"

 

 

Comparing your plant to the standard or average in your industry, indicate the extent to

which the following statements apply.

 

 

 

 

 

 

 

 

 

 

New Product Don't Much About To a
Development Know less the much
same greater
extent
Weemphasize early 0 10 2 3O 4O 50 6O 7O
supplier involvement in N/A 0
product development.
Customer requirements 0 l O 2 3 O 4 O 5 O 6 O 7 O
are clearly specified early N/A 0
in conceptual design.
Weusedesign-for- O 10 2 3O 4O 50 6O 7O
manufacture and assembly N/A 0
(DF MA) methods.
Manufacturing 0 10 2 3O 4O 50 6O 7O
involvement and sign-off N/A 0
are required for new
roducts.
We practice job rotation O 1 O 2 3 O 4 O 5 O 6 O 7 0
between design and N/A 0
manufacturing
flgineering.
Wehavecreatednew O 10 2 3O 4O 50 6O 70
ways to coordinate design N/A 0
and manufacturing issues.
Our product designers O 1 O 2 3 O 4 O 5 O 6 O 7 0
make use of N/A 0
manufacturability
_gu_idelines / checklists.
Productdesignersand O 10 2 30 4O 50 6O 70
manufacturing staff have N/A 0
equal status in NPD
rojects.

 

140

 

 

Comparing your plant to the standard or average in your industry, indicate the extent to

which the following statements apply.

 

 

 

 

 

 

 

 

 

 

 

 

 

Strategy Formulation Don't Much About To a

and Measurement KM“ "’35 the mm"
same greater

extent

Performance 0 10 2 3O 4O 50 6O 7O

improvements are gauged N/A 0

against "best practice"

companies.

Performance measures are 0 1 O 2 3 O 4 O 5 O 6 O 7 0

directly related to N/A 0

customer requirements.

Weusebenchmarking O 10 2 3O 4O 50 6O 70

studies as a basis for N/A 0

performance

improvement.

We have clearly defined 0 1 O 2 3 O 4 O 5 O 6 O 7 O

strategic manufacturing N/A 0

goals and objectives.

Ourfirrn'sstrategy O 10 2 3O 4O 50 6O 7O

leverages existing N/A 0

capabilities.

Corporatestrategyatour O 10 2 3O 4O 50 6O 7O

firm drives manufacturing N/A 0

decisions.

Manufacturing strategies 0 1 O 2 3 O 4 O 5 O 6 O 7 O

and goals are clearly N/A 0

communicated to all

employees.

Manufacturing strategy is O 1 O 2 3 O 4 O 5 O 6 O 7 O

frequently reviewed and N/A 0

revised.

Manufacturing strategy is O l O 2 3 O 4 O 5 O 6 O 7 0

well aligned with N/A 0

corporate strategy.

Ourmanufacturing O 10 2 3O 4O 50 6O 7O

strategy emphasizes N/A 0

shortest delivery lead

times

Ourmanufacturing O 10 2 3O 4O 50 6O 7O

strategy emphasizes N/A 0

highly customized product

 

141

 

 

 

 

 

 

Strategy Formulation Don't Much About To a
and Measurement KM“ "’55 the “WC"
same greater

extent

Ourmanufacturing O 10 2 3O 4O 50 6O 7O

strategy emphasizes a N/A 0

product with better

features than those offered

by the competitors

Ourmanufacturing O 10 2 3O 4O 50 6O 7O

strategy emphasizes 100% N/A 0

conformance quality

Ourmanufacturing O 10 2 3O 4O 50 6O 7O

strategy emphasizes being N/A 0

the lowest cost producer

 

142

 

 

Please rate your performance relative to your principal competition

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Manufacturing Don't Much About Much
Performance Know worse the better
same

initial purchase cost 0 1 O 2 3 O 4 O 5 O 6 O 7 0
N/A 0

manufacturing overhead 0 l O 2 3 O 4 O 5 O 6 O 7 0

cost N/A 0

maintenance cost 0 l O 2 3 O 4 O 5 O 6 O 7 0
N/A 0

Product overall quality 0 1 O 2 3 O 4 O 5 O 6 O 7 0

performance N/A 0

Productfeature O 10 2 3O 4O 50 6O 70
N/A 0

Product reliability 0 1 O 2 3 O 4 O 5 O 6 O 7 0
N/A 0

Product conformance O l O 2 3 O 4 O 5 O 6 O 7 0
N/A 0

product durability O 1 O 2 3 O 4 O 5 O 6 O 7 0
N/A 0

Product serviceability O 1 O 2 3 O 4 O 5 O 6 O 7 0

(speed, courtesy and N/A 0

competence of repair)

Customer satisfaction 0 l O 2 3 O 4 O 5 O 6 O 7 0
N/A 0

Impactofbrandname O 10 2 3O 4O 50 6O 70
N/A 0

Delivery accuracy (correct 0 1 O 2 3 O 4 O 5 O 6 O 7 0

items were delivered) N/A 0

Delivery availability (the O 1 O 2 3 O 4 O 5 O 6 O 7 O

probability that items will N/A 0

be in stock at the time of

order)

Delivery dependability O 1 O 2 3 O 4 O 5 O 6 O 7 O

(delivered on the agreed N/A 0

upon date)

Delivery speed (short 0 1 O 2 3 O 4 O 5 O 6 O 7 O

elapsed time between N/A 0

order placement and

product reaches the

customer)

Delivery quality 0 1 O 2 3 O 4 O 5 O 6 O 7 0

(condition of product after N/A 0

shipment)

Manufacturing Don't Much About Much

 

143

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Performance Know worse the better
same

Ability to customize 0 l O 2 3 O 4 O 5 O 6 O 7 0

products N/A 0

Lead time to introduce O 1 O 2 3 O 4 O 5 O 6 O 7 0

new products N/A 0

Ability to adjust O l O 2 3 O 4 O 5 O 6 O 7 0

production volumes N/A 0

Number ofengineering O 1 O 2 3 O 4 O 5 O 6 O 7 0

change orders per year N/A 0

Number of new products 0 1 O 2 3 O 4 O 5 O 6 O 7 0

introduced each year N/A 0

Ability to respond to O 1 O 2 3 O 4 O 5 O 6 O 7 0

changes in delivery N/A 0

requirements

Ability to produce a range 0 1 O 2 3 O 4 O 5 O 6 O 7 O

of products N/A 0

Productivity 0 1 O 2 3 O 4 O 5 O 6 O 7 0
N/A 0

Profitability O 1 O 2 3 O 4 O 5 O 6 O 7 0
N/A 0

Marketshareofmajor O 10 2 3O 4O 50 6O 7O

product/product line N/A 0

growthrateinunitsales O 10 2 3O 4O 50 6O 70
N/A 0

 

 

 

Thank you for participating in this World Class Manufacturing / Benchmarking study!

144

BIBLIOGRAPHY

145

 

BIBLIOGRAPHY

Aczel, A. D. Complete Business Statistics, 1999. Irwin Mcgraw-Hill.

Ahire, S. L., Golhar, D. Y., and Waller, M. A., 1996. Development and Validation of
TQM Implementation Constructs. Decision Sciences, 27(1), 23-55.

Aldenderfer, M. S., and Blashfield, R. K., 1984. Cluster Analysis. Beverly Hills: Sage
Publications.

Anderson, J. C., Cleveland, G., Schroeder, R. G., 1989. Operations strategy: a literature
review. Journal of Operations Management, Vol. 8, No. 2, April, 133-158.

Athanassopoulos, A. D., 1997. Service Quality and Operations Efficiency Synergies for
Management Control in the Provision of Financial Services: E Branches. European
Journal of Operational Research, Vol. 98, 300-313.

Bacharach, S. B., 1989. Organizational Theories: Some Criteria for Evaluation. Academy
of Management Review. Vol. 14, No. 4, 496515.

Banker, R. D., Chames, A., and Cooper, W. W., 1984. Some Models for Estimating
Technical and Scale lnefficiencies in Data Envelopment Analysis. Management
Science, Vol. 30, Iss. 9, 1078-1092.

Bentler, P. M., 1989. EQS: Structural Equations Program Manual. Los Angeles: BMDP
Statistical Software.

Bentler, P. M., 1990. Comparative Fit Indexes in Structural Models. Psycological
Bulletin, 107, 238-246.

Bentler, P. M., 19923. EQS: Structural Equations Program Manual. Los Angeles: BMDP
Statistical Software.

Bentler, P. M., 1992b. On the fit of models to covariances and methodology to the
bulletin. Psychological Bulletin, 112, 400-404.

Bentler, P. M., and Bonett, D. G., 1980. Significant tests and Goodness of Fit in the
Analysis of Covariance Structures. Psychological Bulletin, 88, 588-606.

Bessent, A. M., Bessent, E. W., 1980. Determining the comparative Efficiency of
Schools through Data Envelopment Analysis. Educational Administration Quarterly,
16, 57-75.

146

 

Beaumont, N. B. and Schroder, R. M., 1997. Technology, Manufacturing Performance and
Business Performance amongst Australian Manufacturers. Technovation, 17(6), 297-
307.

Bollen, K. A., 1989. Structural Equations with Latent Variables. John Wiley, New York,
NY.

Boyer, K. K., 1998. Longitudinal linkages between intended and realized operations
strategies. International Journal of Production and Management, Vol. 18, No. 4, 356-
373.

Bozarth, C., McDermott, C., 1998. Configurations in manufacturing strategy: a review
and directions for future research. Journal of Operations Management, 16, 427-439.

Browne, M. W., 1982. Covariance Structures. In D. W. Hawkins, eds., Topics in
Multivariate Analysis. Cambridge: Cambridge University Press, 72-141.

Browne, M. W., 1984. Asymptotic Distribution Free Methods in Analysis of Covariance
Structures. British Journal of Mathematical and Statistical Psychology, 37, 62-83.

Byrne, B. M., 1994. Structural Equation Modeling with EQS and EQS/Windows. Sage
Publications, Thousand Oaks, California.

Bymes, P., Fare, R., Grosskopf, S., Knox Lovel, C. A., 1988. The effect of Unions on
Productivity: U. S. Surface Mining of Coal. Management Science, Vol. 34, Iss. 9,
1037-1053.

Calantone, R. J., Schmidt, J. B., and Song, M. X., 1996. Controllable Factors of New
Product Success: A Cross-National Comparison. Marketing Science, 15 (4), 341-358.

Challagalla, G. N., Shervani, T. A., 1996. Dimensions and Types of Supervisory Control:
Effects on Salesperson Performance and Satisfaction. Journal of Marketing, J an., 90-
105.

Chames, A., Cooper, W. W., 1978. Measuring the efficiency of decision making units.
European Journal of Operations Research, 2(6), 429-444.

Chames, A., Cooper, W. W., Golany, B.; Seiford, L., 1985. Foundations of Data
Envelopment Analysis for Pareto-Koopmans Efficient Empirical Production
Functions. Journal of Econometrics, Vol. 30, Iss. 1/2, 91-107.

Chames, A., Cooper, W. W., Seiford, L., Stutz, J., 1983. Invariant Multiplicative
Efficiency and Piecewise Cobb-Douglas Envelopments. Operations Research Letters,
Vol.2, No. 3, 101-103.

147

 

Chames, A., Cooper, W. W., Seiford, L., Stutz, J., 1982. A multiplicative Model for
Efficiency Analysis. Socio-Economic Planning Sciences, Vol. 16, No. 5, 223-224.

Chames, A., Neralic, L., 1990. Sensitivity Analysis of the Additive Model in Data
Envelopment Analysis. European Journal of Operational Research, Vol. 48, Iss. 3,
332-341.

Cochran, W. G., 1957. Analysis of Covariance: Its Nature and Uses. Biometrics, 13, 261-
281.

Collins, R., Cordon, C., Julien, D., 1996. Lessons fiom the 'Made in Switzerland'
Study:What Makes a World-Class Manufacturer? European Management Journal, 14(6),
576-589 '

Collins, RS, and Schmenner, R., 1993. Achieving rigid flexibility: Factory focus for the
19905. European Management Journal, 11(4), 443-447.

Davies, D. L., and Bouldin, D. W., 1979. A cluster separation measure. IEEE
Transactions On Pattern Analysis And Machine Intelligence, Vol. PAMl-l, No.2.
April.

De Meyer, A., Nakane, J., Miller, J. G., Ferdows, K., 1989. Flexibility: The next
competitive battle -- The manufacturing futures survey. Strategic Management
Journal, 10(2), 135-144.

DeVellis, R. F., 1991. Scale Development: Theory and Applications. Newbury Park:
Sage Publications.

Dilhnan, D., 1988. Mail and Telephone Surveys: The Total Design Method. New York:
John Wiley and Sons.

Dilhnan, D., 2000. Mail and Internet Surveys: The Tailored Design Method. New York:
John Wiley and Sons.

Dowling, D. R., and Midgley, D. F., 1988. Identifying the coarse and fine structure of
market segment. Decision Sciences, Iss. 19, 830-847.

Ebrahimpour, M., and Withers, B. E., 1990. Employee Involvement in Quality
Improvement: A Comparison of American and Japanese Manufacturing Firms
Operating in the US. IEEE Transactions on Engineering Management, 39(2), 142-
148.

Elashoff, J. D., 1969. Analysis of Covariance: A Delicate Instrument. American
Educational Research Journal, 6, 383-401.

148

._ . D. I,
, 'n sic."

,’
- 2‘.

 

 

Epstein, M. K., Henderson, J. C., 1989. Data Envelopment Analysis for Managerial
Control and Diagnosis. Decision Sciences, Vol. 20, 90-119.

Everett, B., 1975. Cluster Analysis. New York: Wiley.

Ferdows, K., DeMeyer, A., 1990. Lasting improvements in manufacturing performance
in search of a new theory. Journal of Operations Management 9, 168-184.

Flynn, B. B., Sakakibara, S., Schroeder, R. G. (l995,a), "Relationship between JIT and
TQM: Practices and Performance," Academy of Management Journal, 38(5), 1325-
1360.

Flynn, B. B., Schroeder, R. G., Sakakibara, S. (l995,b), "The Impact of Quality
Management Practices on Performance and Competitive Advantage," Decision
Sciences, 26, 5, Sept./Oct., 659-691

Friedman, H. P., and Rubin, J., 1967. On some invariant criteria for grouping data.
Journal of American Statistics Association. Vol., 62, 1159-1178.

Frohlich, M. T., Dixon, J. R., 1999. Information Systems Adaptation and the Successfirl
Implementation of Advanced Manufacturing Technologies. Decision Sciences, 30(4),
921-957.

F omell, C. and Larcker, D. F. Evaluating Structural Equation Models in Unobservable
Variables and Measurement Error. Journal of Marketing Research, Vol. XVIII, 39-50.

Fowler, F. J ., Jr., 1988. Survey Research Methods. Newbury Park: Sage Publications.

Garvin, D. A., 1993. Manufacturing strategic planning. California Management Review,
Summer, 85-106.

Gerbing, D. W., Anderson, JG, 1988. An Updated Paradigm for Scale Development
Incorporating Unidimensionality and Its Assessment. Journal of Marketing Research,
25, 186-192.

Gerdtham, U-G, Rehnberg, C., Tambour, M., 1999. The Impact of Internal Markets on
Health Care Efficiency: Evidence from Health Care Reforms in Sweden. Applied
Economics, Vol. 31 (8), 935-945.

Giffi, C., Roth, A., Seal, G. M., 1990. Competing in World Class Manufacturing:
America's 21St Century Challenge. Business One Irwin, Homewood, IL.

Giokas, D., 1991. Bank Branch Operating Efficiency: A Comparative Application of
DEA and Loglinear Model. Omega, Vol. 19, No. 6, 549-557.

149

 

Golany, B, Storbech, J. E., 1999. A Data Envelopment Analysis of the Operational
Efficiency of Bank Branches. Interfaces, Vol. 29, Iss. 3, 14-26.

Griffin, R., 1988. Consequences of Quality Circles in an Industrial Setting: A
Longitudinal Assessment. Academy of Management Journal, 31(2), 338-358.

Hair, J. F., Anderson, R. E., Tatham, R. L., Black, W. C., 1994. Multivariate Data
Analysis with Readings. Prentice Hall, New Jersey.

Hall, R. W., 1987. Attaining manufacturing excellence. Homewood, IL: Dow Jones-
Irwin.

Hall, R. W., and Nakane, J., 1990. Flexibility: Manufacturing battlefield of the 90's:
Attaining manufacturing flexibility in Japan and the United States. Association for
Manufacturing Excellence.

Hartwick, J. and Barki H., 1994. Explaining the role of user participation in information
system use. Management Science, 40(4), 440-465.

Hayes, R. H., Pisano, G. P., 1994. Beyond world class: the new manufacturing strategy.
Harvard Business Review, January- February, 77-85.

Hayes, R. H., Schmenner, R. W., 1978. How should you organize manufacturing?
Harvard Business Review, January-February, 105-119.

Hayes, R. H., Wheelwright, S. C., 1979. Link manufacturing process and product life
cycles. Harvard Business Review, January-February, 133-140.

Hayes, R. H., Wheelwright, S. C., 1984. Restoring Our Competitive Edge: Competing
Through Manufacturing. Wiley, New York.

Hill, T., 1989. Manufacturing Strategy: Text and Cases. R. D. Irwin, Homewood, IL.

Hu, L. T. and Bentler, P. M., 1995. Evaluating model fit. In R. H. Hoyle (Ed.), Structural
Equation Modeling: Concepts, Issues and Applications. Sage Publications, Thousand
Oaks: CA., 76-99.

Hu, L. T., Bentler, P. M., and Kano, Y., 1992. Can Test Statistics in Covariance Structure
Analysis Be Trusted? Psychological Bulletin, 112, 351-362.

Hughes, M. A., Price, E. L., and Marrs, D. W., 1986. Linking Theory Construction and
Theory Testing: Models with Multiple Indicators of Latent Variables. Academy of
Management Review, 11 (1), 128-144.

Huitema, B. E., 1980. The Analysis of Covariance and Alternatives. New York: Wiley.

150

 

Joreskog, K. G. and Sorbom, D., 1993. New Features in LISREL 8. Chicago: National
Educational Resources.

Kantor, J. and Maital, S., 1999. Measuring Efficiency by Product Group: Integrating
DEA with Activity-Based Accouting in a Larege Mideast Bank. Interfaces, Vol. 29,
Iss. 3, 27-36.

Kathuria, R., Anandarajan, M., Igbaria, M., 1999. Linking IT Applications with
Manufacturing Strategy: An Intelligent Decision Support System Approach. Decision
Sciences, 30(4), 959-991.

Kerlinger, F. N., 1973. Foundations of Behavioral Research. Holt, Rnehar and Winston,
Inc., New York, NY., 439-441.

Klassen, R. D., 1995. The Implications of Environmental Management Strategy for
Manufacturing Performance. Doctoral Dissertation, University of North Carolina at
Chapel Hill.

Kotha, S., Ome, D., 1989. Generic manufacturing strategies: a conceptual synthesis.
Strategic Management Journal 10(3), 211-231.

Krajewski, L. J., and Ritzman, L. P., 1999. Operations Management -- Strategy and
Analysis. Reading, MA: Addison-Wesley.

Marsh, H. W., Balla, J. R., and McDonald, R. P., 1988. Goodness-of-Fit Indexes in
Confirmatory Factor Analysis: The Effect of Sample Size. Psychological Bulletin,
vol. 103, 391-411.

McMillan, M. L., Datta, D., 1998. The relative efficiencies of Canadian universities: A
DEA perspective. Canadian Public Policy, Vol. 24 (4), 485-511.

Miller, J ., Roth, A., 1994. A taxonomy of manufacturing strategies. Management
Science, 40, 285-304.

Muthen, B., and Kaplan, D., 1985. A Comparison of Some Methodologies for the Factor
Analysis of Non-Nonnal Likert Variables. British Journal of Mathematical and
Statistical Psychology, 38, 171-189.

Nakane, J., 1986. Manufacturing futures survey in Japan, a comparative survey 1983-
1986. Tokyo: Waseda University, System Science Institute.

New, C. C., 1979. Manufacturing Strategy in the 1980's. Cranfield Inaugural Lecture,
Cranfield School of Management, June 1979.

New, CC, 1987. Competitive Edge manufacturing Workshop Documentation,
Cranfield/DTI.

151

Th

 

I'll... ".thin...‘ M ’Otfl" m- .
l

 

New, CC, 1992. World class manufacturing versus strategic trade-offs. International
Journal of Operations and Production Management 12 (6), 19-31.

New, C. C., Sweene, M. T., 1984. Delivery Performance and Throughput Efficiency in
UK Manufacturing Industry. SSRC (ESRC) Research Report, Contract HR5212/2,
Cranfield/SSRC.

Nobel M. A., 1995. Manufacturing strategy: testing the cumulative model in a multiple
country context. Decision Sciences, Vol. 26, No. 5, September-October 693-721.

Nuunally, J ., 1978. Psychometric Theory. McGraw Hill, New York.

Oral, M., Yolalan, R., 1990. An Empirical Study on Measuring Operating Efficiency and
Profitability of Bank Branches. European Journal of Operational Research, Vol. 46,
No.3, 282-294.

Parkan, C., 1987. Measuring the Efficiency of Service Operations: An Application to
Bank Branches. Engineering Costs and Production Economics, Vol.12, 237-242.

Peter, P. J ., 1981. Construct Validity: A Review of Basic Issues and Marketing Practices.
Journal of Marketing Research, 18, 133-145.

Porter, M., 1980. Competitive Strategy. Free Press, New York.

Porter, M. E., 1981. Competitive Strategy: Techniques for Analyzing Industries and
Competitors. Free Press, New York.

Porter, M. E., 1985. Competitive Advantage: Creating and Sustaining Superior
Performance. Free Press, New York.

Roth, A.V., and Miller, J. G., 1990. Manufacturing strategy, manufacturing strength,
managerial success and economic outcomes. In J. E. Ettlie, M.C. Burstein &A.
F iegenbaum (Eds), Manufacturing Strategy: The Research Agenda for the Next
Decade, Boston, MA: Kluwer Academic Publisher.

Roth, A. V., and Miller, J. G., 1992. Success factors in manufacturing. Business
Horizons, 35(4), 73-81.

Sakakibara, S., Flynn, B. B., Schroder, R., and Morris, W. T., 1997. The Impact of Just-
in-Time Manufacturing and Its Infrastructure on Manufacturing Performance.
Management Science, Vol. 43, No. 9, 1246-1257.

Schefczyk, M., Gerpott, T. J ., 1998. Determinants of Corporate Efficiency in a Declining

Industry - An Empirical Analysis of German Foundries. Management International
Review, Vol. 38, Iss. 4, 321-344.

152

 

Schmenner, R. W., 1993. Production/Operations Management (5th ed.). New York:
MacMillan.

Schonberger, R. J ., 1983. Work Improvement Programs: Quality Control Circles
Compared with Traditional Western Approaches. International Journal of Operations
and Production Management, 3(2), 18-32.

Schonberger, R. J., 1986. World Class Manufacturing: The Next Decade. Free Press,
New York.

Schonberger, R. J ., 1990. Building a Chain of Customers. Hutchinson, London.

Schonberger, R. J., 1996. World Class Manufacturing: The Next Decade. Free Press,
New York.

Schroeder, R. G., Anderson, J. C., Cleveland, G., 1986. The content of manufacturing
strategy: an empirical study. Journal of Operations Management, Vol. 6, No. 4,
August.

Sherman, H. D., 1984. Hospital Efficiency Measurement and Evaluation: Empirical Test
of a New Technique. Medical Care, Vol. 22, 922-938.

Sherman, H. D., Gold, F ., 1985. Bank Branch Operating Efficiency. Journal of Banking
and Finance, Vol, 9, 297-315.

Sherman, H. D., Ladino, G., 1995. Managing Bank Productivity Using Data
Envelopment Analysis (DEA). Interfaces, Vol. 25, No. 2, 6-73.

Skinner, W., 1969. Manufactruing: Missing Link in Corporate Strategy. Harvard
Business Review, May-June, 136-145.

Skinner, W., 1974. The focused factory. Harvard Business Review, 52(3), 113-121.

Stevens, J., 1996. Applied Multivariate Statistics for the Social Sciences. Lawrence
Erlbaum Associates, Inc., Publishers, Mahwah, New Jersey.

Swink, M., Hegarty, W. H., 1998. Core manufacturing capabilities and their links to
product differentiation. International Journal of Operations & Production
Management, Vol. 18, No. 4, 374-396.

Taninecz, G., 1997a. Best practices & performances: manufacturers tackling leading-
edge initiatives generally reap the best results. IW, December 1, 28-43.

Taninecz, G., 1997b. World-class manufacturers: those plants believed to be "world
class" exhibit best practices and best results. IW, December 1, 44-47.

153

 

Thomas F. S., Barr, R. S., 1998. Benchmarking the productive efficiency of US. banks
Financial Industry Studies, Dec. 1998, 11-24.

Thomas, R. R., Barr, R. S., Cron, W. L., Slocum, J. W., 1998. A process for evaluating
retail store efficiency: A restricted DEA approach. International Journal of Research
in Marketing, Vol 15 (5), 487-503.

Thorndike, R. L., 1953. What belongs in a family? Psychometrika, vol. 18, 267-276.
Todd, J ., 1995. World-Class Manufacturing. McGraw-Hill, London.

Tou, J. T., and Gonzales, R. C., 1974. Pattern Recognition Principles. Readings, MA:
Addison-Wesley.

Van de ven, A., and Ferry, D., 1979. Measuring and Assessing Organizations. Wiley,
New York.

Vargas, V. A., Whybark, D. C., and Xiao C. Z., 1996. Comparing Manufacturing Output
and Practices in China, Hungary, the USSR and USA. International Journal of
Production Research. Vol. 34(5), 1429-1445.

Vassiloglou, M., Giokas, D., 1990. A Study of the Relative Efficiency of Bank Branches:
An Application of Data Envelopment Analysis. Journal Operational Research
Society, Vol.41, No.7, 14-26.

Venkatraman, N., 1989. Strategic Orientation of Business Enterprises: The Construct,
Dimensionality, and Measurement. Management Science, 35(8), 942-962.

Vickery, S. K., 1991. A theory of production competence revisited. Decision Sciences,
Vol. 22, 635-643.

Vickery, S., Droge, C., Markland, R, 1993. Production competence and business strategy:
Do they affect business performance? Decision Sciences, 24(2), 435-455.

Voss, C. A., 1995. Alternative paradigms for manufacturing strategy. International
Journal of Operations and Production Management 15 (4), 5-16.

Voss, C., Blackrnon, K., Hanson, P., and Oak, B.,1995. The Competitiveness of European
Manufacturing—A Four Country Study. Business Strategy Review, Spring.

Ward, R, Bickford, D., Leong, K., 1996. Configurations of manufacturing strategy,
business strategy, environment, and structure. Journal of Management, 22 (4), 597-
626.

154

 

Weber, C. A., Current, J. R., and Desai, A., 1998. Non-Cooperative Negotiation
Strategies for Vendor Selection. European Journal of Operational Research, Vol. 108,
208-223.

Weber, C. A., Desai. A., 1996. Determination of Paths to Vendor Market Efficiency
Using Parallel Coordinates Representation: A Negotiation Tool for Buyers. European

Journal of Operational Research, Vol. 90, 142-155.

Wildt, AR, and Ahtola, O., 1978. Analysis of Covariance. Beverly Hills, CA.

155