T"E:‘:’ 0“.

2003

This is to certify that the

dissertation entitled

INFUSION EFFECTS OF INNOVATIONS ON THE
TECHNOLOGICALLY ORIENTED NEW PRODUCT
DEVELOPMENT PROCESS

presented by

ROSANNA GARCIA

has been accepted towards fulﬁllment
of the requirements for

PHD , MARKETING
degree in

 

 

W

0 major professor

Date July 25, 2002

 

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

 

 

IJBRARY
Michigan State
University

 

 

 

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

 

AA

 

Aiii ii 8 ZUUT *
12L514

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIHCIDateDue.p65—p. 15

 

 

INFUSION EFFECTS OF INNOVATIONS ON THE TECHNOLOGICALLY
ORIENTED NEW PRODUCT DEVELOPMENT PROCESS

By

Rosanna Garcia

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Marketing and Supply Chain Management

2002

ABSTRACT

INFUSION EFFECTS OF INNOVATIONS ON THE TECHNOLOGICALLY
ORIENTED NEW PRODUCT DEVELOPMENT PROCESS

By

Rosanna Garcia

The primary objectives of this dissertation were two-fold: (1) to advance the
practitioner’s understanding of how adopted innovations can assist new product (NP)
managers in driving the success of inventions, and (2) to advance the methodological
framework of how the NP development process of technological ﬁrms can be studied as a
dynamical system. A secondary question was to determine if ‘catalytic’ innovations
signiﬁcantly increase the success of an NP program. Two methodological approaches
were used to study these objectives; a system dynamics model and a longitudinal study.
The system dynamics model was prefaced by intensive case studies of two
technologically oriented ﬁrms. The data collected in the longitudinal study were
evaluated using bivariate linear growth curves using structural equation modelling.

It was found in both studies that the successful adoption of innovations into the
ﬁrm has an indirect effect on the success of a NP program. Innovations successfully
adopted into an organization have the effect of increasing a ﬁrm’s competences in
technology knowledge utilization (exploitation) and R&D proﬁciencies. These studies
also reveal a positive relationship between technology knowledge exploitation and R&D

proﬁciencies as explanatory variables to NP program success. Thus, indirectly the

successful adoption of innovation can drive NP success by helping organizations to better
utilize their technology knowledge competences.

Technologically oriented organizations that are interested in using the adoption of
innovations as a strategy for driving greater success in their NP program should focus on
making resources readily available to the NP process, encouraging failures in NP design,
and actively involving marketing in the NP process. An important ﬁnding supported by
both the system dynamics model and the linear growth curve model was that adopting
innovations can increase technological knowledge competences. However, the quadratic
trajectories observed in the longitudinal study suggest that peaks in performance are often
reached. The continual renewal of knowledge is important in the NP development
process. Firms that structure themselves to be open to new knowledge will beneﬁt not
only in their success of adopting innovations, but also in NP program performance.

‘Catalytic’ innovations were identiﬁed in this study as innovations adopted into
the organization that act as a catalyst to develop new inventions not previously planned
prior the adoption of the innovation. Marginal support was found for the existence of
‘catalytic’ innovations in the sample population studied. Although catalytic innovations
are not an essential prerequisite to ensure new product program success, when adopted in
moderation they may substantially increase the ﬁrm’s knowledge base for building new

products. This increased knowledge can lead to increased new product program success.

DEDICATION

This dissertation is dedicated to my parents, Alice Perez Garcia and Frank G. Garcia, and
my grandparents, Rosalie Range] Perez and Marciano Duran Perez with immeasurable
thankfulness for life and its wonders. It is also dedicated to Paul W. Rummel for without

his love and support this dissertation would not have been completed.

iv

ACKNOWLEDGEMENTS

There are numerous people that have supported me in the completion of this
dissertation. First, I’d like to acknowledge my committee: Dr. Roger Calantone for
making this journey interesting and most of all fun, Dr. Abbie Grifﬁn for encouraging me
to take the journey from the ﬁrst time I threw myself at her at a Product Development
Management Association (PDMA) conference many years ago, Dr. Ralph Levine for
going beyond the call of duty and agreeing to work with me even though he had never
seen me prior to the day I knocked on his ofﬁce door seeking guidance, Dr. Cornelia
Droge for teaching me that sometimes the difference is in the details, and Dr. David
Closs for guiding me through simulation modelling. I have had the best instructors in the
process of completing this dissertation and will be forever drawing upon the knowledge
that they have kindly imparted upon me.

Several other people supported me in my efforts. Dr. John Hollenbeck ﬁelded
construct development questions, Jason Black at MIT ﬁelded SD questions, Dr. David
Lounsbury provided me his SD worksheets, and Marilyn Brookes covered on survey
support. The ofﬁce staffs at MSC are awesome, and special thanks goes to them for their
kind words of encouragement. Financial support was received from MSC, PDMA, and
Paul W. Rummel.

Besides ﬁnancial support, Paul W. Rummel provided morale support,
proofreading, typing, editing, and unending love throughout the process. Thank you for
agreeing to take this journey with me. I hope that you have learned as much from it as I

have.

My family provided endless love, encouragement, and prayers these past four
years. Without their support I would have given up before I started. Thank you for being
there at any time of the day or night.

For helping me to maintain my sanity during this process, special
acknowledgements goes to Dr. Mary Thompson, the Running F it/Dante’s Infernal Gang
(Mark, Marcia, Marcus, Mike, Amy, Ann, and Joan), and all the Turtles. Truly without
your support this process would not have been as enjoyable. Thank you for being there.

Numerous friends from California to Boston continually believed in me and
cheered me on even on my darkest days. Thank you Tom, Gwen, Jackie, Gary, Calvin,
Karen, Kathleen, Julia, Ann, Tony, Karen and many others! Thank you Dr. Rudolf
Sinkovics for making me laugh when I needed to and letting me cry when I needed to.
Thank you to future Dr. James Ramos for getting me to let my hair down on occasion.
We make great tango partners.

And to the MSC doctoral students, thank you for sharing margarita nights and

apple pie.

vi

TABLE OF CONTENTS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LIST OF TABLES x

LIST OF FIGURES xii
CHAPTER 1

INTRODUCTION 1

1.1. Background 1

1.2. The Evolving New Product Development Process 4

1.3. Model Introduction 5

1.3.1. Adaptive Capacity 5

1.3.2. Slack Resources 8

1.3.3. Marketing-R&D Integration 9

1.3.4. Successful Adoption of Innovations 9

1.3.5. Knowledge Exploitation 10

1.3.6. Marketing and R&D Proﬁciencies 11

1.3.7. The Feedback Loop 11

1.4. Methodology 12

1.5. Research Questions to be Answered 15

1.5.1. Practitioner Oriented Questions 16

1.5.2. Theoretical/Methodological Research Questions 16

1.6. Summary and Overview of the Remaining Chapters 17
CHAPTER 2

LITERATURE REVIEW AND HYPOTHESES DEVELOPMENT 18

2.1. Domain Delineated 18

2.1.1. Unit of Analysis and ‘Innovation’ Domain 18

2.1.2. Adoption vs. Diffusion of Innovation 20

2.1.3. Domain Summary 22

2.2. The Evolutionary Nonlinear Dynamic Complex System 22

2.2.1. The NPD Process as a Complex System 23

2.2.2. Modeling of Complex Systems 25

2.3. Background Literature on the Adoption of Technological Innovations ........................... 28

2.4. Hypotheses Developed 30

2.4. 1. Antecedents to the Successful Adoption of Innovations 30

2.4.1.1. Adaptive Capacity as a Construct 30

2.4.1 .1.1. Flexibility 31

2.4.1.1.2. Market & R&D Orientation 32

 

vii

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2.4.1.1.3. Absorptive Capacity 33

2.4.1.2. Adaptive Capacity as an Antecedent 34

2.4.1.3. Slack Resources 35

2.4.1.4. Marketing — R&D Integration 37

2.4.2. Outcomes of the Successful Adoption of Innovations 38

2.4.2.1. Knowledge Exploitation as an Outcome 40

2.4.2.2. Marketing & R&D Proﬁciencies as Outcomes 42

2.4.3. Antecedents to NP Program Success 43

2.4.3.1. Knowledge Exploitation as an Antecedent 43

2.4.3.2. Marketing Proﬁciencies as an Antecedent 45

2.4.3.3. R&D Proﬁciencies as an Antecedent 46

2.4.4. The Feedback Loop 46

2.4.4.1. New Product Program Success as an Antecedent 47

2.4.5. Hypotheses Summarized 50
CHAPTER 3

OVERVIEW OF METHODOLOGY ......................... 51

3.1 Chapter Summary 51

3.2 Study 1 — Case Study & System Dynamics Simulation 51

3.2.1 Case Study Data Collection 54

3.2.2 System Dynamics Model Simulation Analyses 57

3.2.2.1 Conceptualization 58

3.2.2.2 Formulation 59

3.2.2.3 Testing & Implementation 60

3.3 Study 2 — Longitudinal Study & Structural Equation Model 63

3.3.1 Longitudinal Study Data Collection 63

3.3.2 Structural Equation Modeling Analysis 64

3.4 Chapter Summary 66
CHAPTER 4

RESEARCH METHOD - STUDY 1: SYSTEM DYNAMICS MODEL 67

4.1 Introduction ......... 67

4.2 Case Studies 68

4.3 Overview of System Dynamics Model 73

4.4 Central Assumptions 80

4.5 Research Questions 82

4.6 Research Design 84

4.6.1 Simulations and Base Case 87

4.6.2 Results of Simulation 92

4.7 Results 95

4.7.1 Research Questions Evaluated 95

4.8 Hypotheses 107

 

4.9 Discussion and Conclusion

viii

111

CHAPTER 5
RESEARCH METHOD - STUDY 2: LONGITUDINAL STUDY

5.1 Introduction

 

5.2 Sampling Frame

 

 

5.3 Missing Data Imputation

 

5.4 Measurements
5.4.1 Measurement Means

 

5.4.2 Measurement Invariance, Reliability and Validity

 

 

5.5 Methodology
5.5.1 Univariate Linear Growth Curves

 

5.5.2 Bivariate Linear Growth Curves

 

5.6 Results
5.7 Conclusions

 

CHAPTER 6

CONTRIBUTIONS, LIMITATIONS AND FUTURE RESEARCH
6.1 Introduction

 

 

6.2 Research Questions Addressed & Contributions

 

6.2.1 Summary to Research Questions
6.3 Limitations and Future Research

 

6.3.1 Other Future Research

 

APPENDICES

 

 

A. Case Studies Supporting Documents
B. System Dynamics Supporting Documents

 

B.1 Differential Equations for System Dynamics Model

 

B.2 Vensim Code

 

C. Longitudinal Study Supporting Documents

BIBLIOGRAPHY

 

ix

115

115
115
117
119
125
127
128
134
139
145
156

159
159
159
172
173
175

I77

178
202
210
220
237

246

Table 2.1

Table 3.1

Table 3.2

Table 4.1

Table 4.2

Table 4.3

Table 5.1

Table 5.2

Table 5.3

Table 5.4

Table 5.5

Table 5.6

Table 5.7

Table 5.8

Table 5.9

Table 5.10

Table 5.11

Table 5.12

Table 5.13

LIST OF TABLES

 

 

 

 

Hypotheses ....... , ................................. -50
Data Gathering Methods for Case Studies ........................................................................................................ 56
Stages of Model Construction ------ 62
Simulation Legend Constant Settings 91
Means of Stock Variables From Simulations & Results of

ANOVA Analyses 93

Results of Unspeciﬁed Curvilinear Regression — Best Fitting Models ................ 108

Means of Constructs at Each Time Period ........

Results of 3-Group Structural Equation Model to Test
Measurement Invariance

 

Results of CFA for Each Time Period

126

129

130

 

 

Timel Reliabilities and Discriminant Validity Measurements

Time2 Reliabilities and Discriminant Validity Measurements

 

Time3 Reliabilities and Discriminant Validity Measurements

 

Univariate Linear Growth Curve Results

 

Means and Variances of Univariate Models ................................................................ . ................. . ...............

 

 

131

132

133

142

143

144

 

Hypotheses

Bivariate Linear Growth Curve Results M1-M6

149

 

Bivariate Linear Growth Curve Results M7-M11

 

Means and Variances of Bivariate LGC Model for Dependent
Variable Ml-M6

150

 

Means and Variances of Bivariate LGC Model for Dependent
Variable M7-M11

 

151

152

Table A.1

Table B]

Table B.2

Table B.3

Table C.1

 

Case Companies’ Description

Parameter (Constants) Used in Vensim Modeling

 

Stocks (Levels) with Initial Values at t=0

 

 

Auxiliary Variables Used in Vensim Modeling

Examples of Innovations Cited by Respondents

 

xi

Figure 1.1

Figure 2.1
Figure 2.2

Figure 2.3

Figure 2.4
Figure 2.5
Figure 3.1
Figure 3.2
Figure 3.3
Figure 4.1
Figure 4.2
Figure 4.3
Figure 4.4
Figure 4.5
Figure 4.6
Figure 4.7
Figure 4.8

Figure 4.9

LIST OF FIGURES

Overall Model of Infusion Effects of Technological Innovations
on the NPD Process 6

Fundamental Modes in Business Dynamics 27

 

Antecedents to the Successful Adoption of Technological Innovations ............ 31

Infusion Effects from the Successful Adoption of Technological

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Innovations 40
Feedback Loop , ...................................................... 47
Predicted Relationship H10 48
System Dynamics General Framework 52
Full Structural Model ....... , ..................................... 53
Latent Growth Curve Model 65
SD Model with Paths Tested 72
General Feedback Model 74
“Prosperity” Feedback Loop 74
“Revitalization” Feedback Loop » 75
“Street Smarts” Feedback Loop 75
Detailed SD Model ...................................... , .............. 79
Base Model-Overall Performance &Test Runs 86
Base Case NP Program Success & Catalytic Innovations 97

 

Sensitivity of NP Program Success to Changes in Catalytic Innovations ........ 98

Figure 4.10 Sensitivity of Catalytic Innovations to Changes in NP Program Success ....... 98

 

Figure 4.11 Base Case Slack Resources & Catalytic Innovations 102

Figure 4.12 Sensitivity of Catalytic Innovations to Changes in Slack Resources ................... 102

xii

Figure 4.13

Figure 4.14

Figure 4.15

Figure 4.16

Figure 4.17

Figure 4.18
Figure 4.19

Figure 4.20

Figure 4.21
Figure 5.1
Figure 5.2
Figure 5.3
Figure 5.4
Figure A.1

Figure A.2

Figure B.2
Figure 8.3

Figure 8.4

Base Case Fraction of Resources Allocated to Exploration Activities &

Catalytic Innovations

Sensitivity of Catalytic Innovations to Changes in Fraction of
Resources Allocated to Exploration Activities

103

103

 

Base Case Adaptive Capacity, Tech Exploit, Marketing Proﬁciencies, R&D

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Proﬁciencies 104
Sensitivity of Technology Exploitation to Changes in Adaptive
Capacity 104
Sensitivity of Marketing Proﬁciencies to Changes in Adaptive
Capacity 105
Sensitivity of R&D Proﬁciencies to Changes in Adaptive Capacity .................. 105
Base Case Marketing-R&D Integration and Catalytic Innovations ...................... 106
Sensitivity of Catalytic Innovations to Changes in Marketing-R&D

Integration 1 06
Resource Allocation Effects on NP Program Success 114
Two-Factor Polynomial Latent Growth Curve Model 136
Three-Factor Polynomial Latent Growth Curve Model 136
Bivariate Linear Growth Curve Model Example 141
Model with Hypotheses 148
Common Modes of Linear Behavior over Time 188
Common Modes of Curvilinear Behavior over Time 189
Figure B.1 f R , D as a Function of Performance Gap 218
Resources Module 230
Exploration Module 231
Exploitation Module 732
Knowledge Module 233

Figure B.5

xiii

Figure B.6
Figure B.7
Figure B.8

Figure C.1

Proﬁciencies Module

234

 

Proﬁts Module

235

 

Gap & Adaptive Capacity Module

Example of Knowledge Exploitation Framed by Adopted Innovations

Being Tracked Across Time

xiv

236

 

245

CHAPTER 1
INTRODUCTION

1.1 Background

The adoption of technological innovations has traditionally been seen as a means
of improving productivity, reducing costs, and building core competencies in the
manufacturing, distribution, and/or marketing processes of an organization (Robertson &
Gatignon, 1986). From an organizational behavioral perspective, this research has
primarily taken two different theoretical approaches: (a) studies investigating what
factors facilitate the adoption of innovations in order to speed its diffusion (Damanpour,
1991; Rogers, 1995; Zaltman, Duncan, & Holbek, 1973) and (b) studies focusing on the
effects from adopting innovations on organizational structure (Blau & Schoenherr, 1971;
Repenning, 2001). From a marketing perspective, the adoption of innovations has often
been studied to generate models of diffusion to consumers (Bass, 1969; Sultan, Farley &
Lehmann, 1990).

The innovations adopted by ﬁrms traditionally have not been viewed as a resource

for subsequently generating new innovationsl. Yet, ﬁrms regularly adopt technological
innovations to build competitive advantages, not as a means of improving internal
productivity, but as a competitive strategy to invent new products for the marketplace.
Two examples will help to demonstrate this point.

The California State Automobile Association (CSAA) is one of the largest

 

' ‘new’ innovations are those that are new to the ﬁrm or adopting entity although the innovation may
already have been introduced into the marketplace (Garcia & Calantone, 2001).

providers of automobile, home and life insurance in California. It is not a ﬁrm that one
would associate with technological advancements. However, recently it invested over
$10 million in computer and telephony technology in order to introduce new services to
their customers as a pre-emptive competitive move. By adopting technological
innovations new to an insurance company, they were able to introduce an automatic call
center and web services for processing claims. This has allowed the ﬁrm to expand their
service offerings and preempt the competition.

Similarly, Comdel, Inc., a small electronics supplier to the semiconductor industry
headquartered in Massachusetts, adopted a programmable memory chip in their user
interface. This will allow them to introduce $500,000 worth of new products to their
customers this year with that revenue predicted to grow exponentially over the next ﬁve
years. The president of Comdel, Ted Johnson, admits to continually exposing the
organization to innovations of all types: administrative, technological, and process.
Exposure serves as a catalyst for introducing uncertainty into the R&D process with the
hope of developing signiﬁcantly more innovative products than the last generation.

These two examples show the adoption of innovations as a competitive strategy
undertaken by ﬁrms for building their product portfolios, and not as a means for
improving internal distribution or manufacturing efﬁciencies. The dissertation was
motivated by this under-researched phenomenon where technological innovations are
adopted by ﬁrms for the sole purpose of facilitating the invention of new products for
marketplace distribution. Questions that arise when investigating this innovation strategy
are: What is the effect of the adoption of technological innovations on new product

success? What characterizes and differentiates the types of ﬁrms that use this new

product development strategy? How should the new product development process adapt
to accommodate the continual adoption of innovations new to the ﬁrm?

This dissertation looked at the inﬂuences of the adoption of innovations by a
technological ﬁrm on innovations invented by the ﬁrm. The primary research question it
addresses is “over time, how does the adoption of new innovations affect the new product
development process, if at all?”

A secondary issue that was addressed is how does the ﬁrm evolve and adapt its
organizational structure over time in order to effectively adopt innovations and
subsequently produce inventions. The evolutionary theory of the ﬁrm is not new to
organizational behavior researchers (cf. Nelson & Winter, 1974; Tushman & Romanelli,
1986, Miller, 1990; Aldrich, 1979) or to economic theorists (cf. Schumpeter, 1934; Perez
1983). As well, the evolutionary theory of technologies has been extensively explored
(cf. Utterback, 1996; Christensen, 1997; Ziman and colleagues, 2000). The adoption of

innovations into the ﬁrm itself has been modeled as an evolutionary process whereby

innovations are invented, and out-dated inventions undergo exnovation2 in order to free
resources for future adoption of new innovations (Kimberly, 1981). The interaction
between adopted innovations and the new product development process also can be
viewed as an evolutionary system. This dissertation looked at the evolving new
product development process as effected by the adoption of new innovations.

The remainder of this introduction is organized as follows: section 2 addresses the
new product development as an evolving process. Section 3 introduces the model that

was tested in this dissertation. Section 4 summarizes the methodology to be utilized to

 

2 ‘exnovation’ is the removal of an innovation from an organization, a term coined by Kimberly (1981).

test the hypotheses, and section 5 provides research questions that intend to be answered

from this study. Section 6 concludes this chapter with a summary.

1.2. The Evolving New Product Development Process

The new product development program process is characterized by nonlinear
relationships and complex interactions that evolve dynamically over time. Field research
supports this perception of a dynamic nonlinear evolving NPD program process. Van de
Ven, et a1 (1999) conducted a seventeen-year long longitudinal study on the ‘innovation
journey’ undertaken by organizations in their development of new products. They
concluded, “our research of a wide variety of innovations has found no support for a
stage-wise model of innovation development and no support for a linear (cyclical) model
of adaptive trial-and-error [random] learning, particularly during highly ambiguous and
uncertain periods of the innovation journey” (pg. 4). Thus, despite common practice by
NPD researchers, the innovation program process should not be considered as following
a static linear stage-gate process popular in many studies (cf. Cooper 1993).

Environmental conditions in new product development continuously change over
time. The inventing organization must continually adapt to dynamic exogenous and
endogenous inﬂuences in the marketplace. Competitive threats surface, customers alter
product requirements, and technology advances exponentially to create instabilities in the
new product development process. NPD team movement, mergers and acquisitions,
management turnover, and funding constraints (advancements) add to the dynamic
internal environment faced by NP managers. These ‘agents’ interact during the NPD

process, causing evolutionary dynamic and path dependent relationships.

Figure 1.1 presents this model of the evolutionary NPD process as inﬂuenced by
the adoption of innovations. The relationships between constructs will be further detailed

in the following section.

1.3 Model Introduction

In the model introduced in Figure 1.1, the antecedents to the adoption of
innovations are a ﬁrm’s adaptive capacity, its slack resources, and the integration
between the marketing and R&D departments. Successfully adopted innovations are
those that have been assimilated, implemented, and used effectively in the organization.
The unsuccessful adoption of innovations is characterized by a failure to implement an
innovation into the organization despite an expenditure of resources. It is proposed that
the effects of successful adoption of innovations are to increase the ﬁrm’s ability to
exploit knowledge acquired through the adoption of the innovations and to improve the
ﬁrm’s marketing and R&D proﬁciencies. These gains in knowledge and proﬁciencies
lead to greater new product program success. This new product program success

ultimately affects the ﬁrm’s slack resources and its adaptive capacity.

1.3.1 Adaptive Capacity
Evolutionary theory is concerned with how ﬁrms adapt to endogenous and
exogenous environmental factors. The adaptation process has been described as

following either ‘punctuated equilibrium’ or ‘time-paced evolution’. The punctuated

 

  

3325
8:395 sz

  
     

 

 

86:39:95
Gﬁm

new-Ewen:—
nay—assets:

     
       

  
 

homoﬁm o
couﬁaoEoEEH o

cogzaﬁmmx‘s o
@8935. 22555::—
ue $325

89533—
:35

83:32.3...—

wﬁuou—ua—Z — :—

 

v:

   

m:
:ouﬁcoco
nemasﬁaz

£335 .
bmowamu o>uEOmn< .
buzz—«U 339:;

     
    

Seesaw”.—
once—Beav—

 

 

mmoooi DAZ 05 no 22555: Rama—2.38% mo 381cm :oBEE mo .352 =Eo>O 2 oSwE

equilibrium model assumes that long periods of small incremental organizational
evolution are interrupted by infrequent periods of discontinuous, radical change
(Romanelli & Tushman, 1994; Abernathy & Utterback, 1978). The alternative
explanation of the adapting organization embraces a more gradual evolution earmarked
by continuous, although fast-paced, change over time (Eisenhardt & Tabrizzi, 1995).
Eisenhardt and Tabrizzi claim that the ability to change continuously is a critical factor in
the success of ﬁrms and that “product innovation is a primary way in which this
alternative form of adaptation can happen” (pg. 84). This dissertation takes the
perspective of time-paced evolution where ﬁrms are continually adapting to their
endogenous environments.

Many studies have been conducted on the types of organizational structures that
allow ﬁrms to quickly adapt to the environment and that facilitate the NPD process
(Burns & Stalker, 1961; Miles & Snow, 1978). Burns & Stalker emphasized that an
‘organic structure’ appears to be required for conditions of change. Organic structures
are characterized by a lack of rigid structure and deﬁnition, but with the ability to quickly
respond to unstable conditions.

For adopted innovations to act as catalysts for inventions, they ﬁrst must have
successfully been assimilated into the ﬁrm. This requires the ﬁrm to be ﬂexible and
willing to adapt to the new ideas and new knowledge disseminated with the new
innovation. Flexibility to learn quickly and to adapt to new information is essential for
technological ﬁrms that must operate in turbulent market and technological environments
typical of today’s marketplace.

To remain ﬂexible and adaptive the ﬁrm should monitor its external environment.

Adaptive organizations demonstrate this capability through a combination of their market
and R&D orientation (Jaworski & Kohli, 1993; Slater & Narver, 1995; Hurley & Hult,
1998) and through their absorptive capacity (Cohen & Levinthal, 1990). Market
orientation pertains to identifying customers' needs and wants, sometimes pre-
anticipating the consumer's wants, and using this information in new product planning.
The technology counterpart to market orientation, R&D orientation, is equally important
as ﬁrms need to effectively respond to the ever-changing technological landscape in
today's marketplace.

Fundamental to market orientation is organizational learning, which is the
assimilation of information into the ﬁrm to increase knowledge. What the ﬁrm does with
the information gained from monitoring the environment is as important as gathering the
information itself. Cohen & Levinthal address this issue by coining the phrase
'absorptive capacity'. A ﬁrm's absorptive capacity is its ability to recognize the value of
new information, assimilate it, and apply it to the development of new products. Thus,
the construct, ‘adaptive capacity’ constitutes a ﬁrm’s ability to gather and assimilate new
knowledge, and subsequently adapt its new product strategy based on this
knowledge accumulation. A ﬁrm’s adaptive capacity is precursor to its ability to

successfully adopt and implement new ideas, processes, or products.

1.3.2 Slack Resources
Slack resources are necessary for the development of inventions (Kamien &
Schwartz, 1982); likewise, the availability of resources is required for the successful

adoption of innovations by a ﬁrm. Availability of slack resources is determined by the

successful market introduction of new products to the ﬁrm’s customers. The greater the
proﬁts from new product introductions, the more resources become available to commit
to new programs or technologies. Competition for resources can ensue as more
technologies are invested in the ﬁrm. It was expected that the more slack resources
available, the greater the likelihood that new technologies adopted by the ﬁrm are

effectively assimilated into the ﬁrm.

1.3.3 Marketing and R&D Integration

Grifﬁn & Hauser (1996) postulate that developing informal cross-functional
networks between marketing and research departments enables more information to be
communicated, increases coordination between the departments in decision making, and
decreases new product uncertainty. This integration contributes to the successful
invention of new products. Using the same reasoning, it is expected that informally
integrated marketing and R&D departments should lead to an increase in the internal
communication between the two groups and increase the successful assimilation and

implementation of innovations adopted by the technological ﬁrm.

1.3.4 Successful Adoption of Innovations

“The adoption of innovation is generally intended to contribute to the
performance or effectiveness of the adopting organization” (Damanpour, 1991 , pg. 556).
This dissertation was interested in the effects on the new product development process of
the infusion of innovations. An innovation has been deﬁned as “an internally generated
or purchased device, system, policy, program, process, product or service that is new to

the adopting organization” (Damanpour, 1991, pg.556). Some innovations adopted by

organizations have the potential to signiﬁcantly impact the NP process. March (1991)
has suggested that “some new technology is so clearly superior to overcome the
disadvantages of unfamiliarity with it, it will offer a higher expected value than the old
technology”(pg. 83). This dissertation terms these types of innovations as ‘catalytic’.
Catalytic innovations are adopted into the organization and act as a catalyst to develop

new inventions not previously planned prior to the adoption of the innovation.

1.3.5 Knowledge Exploitation

Moving away from a resourced-based view, which considers knowledge as a
company asset, knowledge-based theory argues that knowledge is a process by which
organizations evolve by adapting the body of knowledge shared by its members and that
much of this process occurs at the tacit level (Nelson & Winter, 1974; Plotkin, 1994).
This viewpoint challenges the idea of knowledge as an economic asset or commodity,
and instead argues for ‘knowledge as the skilled process of leveraging resources, where
the knowledge is embedded in the organization’ (Spender, 1996, pg. 54). Knowledge
accumulation occurs at a systems level where individuals are agents for the
organization’s continually evolving knowledge base.

Extant literature suggests that a fundamental agent contributing to the
evolutionary process is the learning experiences of the ﬁrm during new product
development. During the innovation process, knowledge (both market and technological)
is accumulated and assimilated in order to facilitate the successful marketplace
introduction of new products. Exploiting the knowledge gained from the new
technologies embedded within new products increases the potential for the ﬁrm to

achieve higher levels of performance.

10

1.3.6 Marketing and R&D Proﬁciencies

A ﬁrm’s proﬁciency at technical activities and its proﬁciency at marketing
activities have been noted in numerous studies to be development process factors that
drive new product success (Cooper & Kleinschmidt 1995; Montoya-Weiss & Calantone,
1994). Technical activities include the R&D process, testing of the product, trial
production, and production startup. The marketing activities include customer

orientation, competitive monitoring, program development, as well as launch activities.

1.3.7 The Feedback Loop

‘Innovation’ as a process implies a temporal perspective. In order to study new
product development as a process, this dissertation utilizes feedback loops to model the
effects of evolution. A cyclical causal relationship exists in which the agents involved in
the NPD process inﬂuence one another and in turn, are inﬂuenced by the very agents they
have interacted with during innovation. The success or failure of new products in the
marketplace will have direct, although nonlinear, path dependent effects on the ﬁrm’s
slack resources and its adaptive capacity. Failures in the marketplace can mean either
more or less resources committed to the adoption of innovations, depending upon the
ﬁrm’s new product strategy. Successes most frequently will result in an increase in slack
resources, but with allotment constraints most likely imposed.

Likewise, the more successful a NPD project, the greater the commitment made to
that new product development program, which frequently results in decreasing the
adaptive capacity of the organization. It has been shown that ﬁrms that did not
continually renew their knowledge stock lock themselves into old technological

knowledge and lock themselves out of the ability to acquire new knowledge. This

11

phenomenon is referred to as lock-in by Cohen & Levinthal, (1990), core rigidities by
Leonard-Barton, (1995), or the Icarus Paradox by Miller, (1990). Lock-in to outdated
knowledge and lock-out of new knowledge causes the ﬁrm to lose its ﬂexibility and
openness to new ideas. A ﬁrm must have a knowledge foundation that is continually
updated in order to be able to assimilate the latest technological knowledge.

The feedback loop was evaluated using system dynamics modeling, which allows
the evaluation of the nonlinear relationships between constructs over time. This

methodology will be further discussed in Chapter 3.

1.4 Methodology

The systems model in Figure 1.1 was tested in two studies. System dynamics

computer simulations were used in study 1 to test the model. Study 2 utilized a doubly

multivariate longitudinal study3 (Bijleveld, et. a1, 1998) to further provide support for the
model introduced here. Utilizing these two methods for studying the model not only
allowed testing the external validity of the simulation model, but also provided an
opportunity for reconciling actual new product practices with the theory developed here.
System dynamics modeling has long been utilized in operations and strategic
management research (Forrester, 1987; Morecroft and Sterman, 1994; Nelson and
Winter, 1974; Repenning, 2001; Sastry, 1997). Nelson and Winter's (1974) seminal work
proposed an evolutionary theory of economic change based on the continual development
of innovations. They used computer simulations to model this dynamic system in order

to study the interactions among market structures, R & D spending, technical change, and

 

3 doubly multivariate designs survey multiple respondents in multiple groups on multiple variables over
multiple waves (Bijleveld, et a1 1998).

12

other aspects of industry performance.

Janszen (2000) recently used system dynamics modeling to study the peculiarities
of the new product process. He argues that non-linear mechanisms resulting from the
interaction of the different factors involved in NPD are responsible for the unpredictable
dynamics of the process. The new product development process can be best described as
an evolving dynamic complex system, and thus, was studied from this perspective in this
dissertation.

A system dynamics modeling software was utilized to evaluate the continuous
process as the ﬁrm adapts itself to the innovations it adopts, and subsequently, invents
new products. Ventana Systems' Vensim (2001) was used for the model test. The
internal and external validity of the model was validated using the methods suggested by
Levine, Van Sell and Rubin (1992). This involves conducting structural tests, behavioral
tests and policy tests, all of which will be elaborated upon in Chapter 3. A sensitivity
analysis was performed to learn more about the importance of the various parameters in
relation to certain structural characteristics of the model. The objective of the sensitivity
analysis was to learn more about the effects of changes in the parameters of the system.
It is the equivalent of running ‘what if scenarios (J anszen 2000).

Computer simulations allow numerous factors to be modeled into the system,
and thus, the problem becomes one of determining: what are the important exogenous
factors not explicitly deﬁned in the model under study to be set as ceteris paribus, what
are the appropriate mathematical models that explain the relationships between variables,
and what are appropriate initial values for modeling the dynamic system? In the ﬁrst

study, an in depth case study of an electronics ﬁrm involved in the semiconductor

l3

industry was undertaken to answer these questions. The results of the computer
simulation provided the insights necessary to compile a relevant and concise survey for
distribution to a large sampling frame.

The second study utilized a longitudinal survey methodology on 40+ technology
ﬁrms with 2-3 respondents per company. Data was collected at three points in time ﬁve
months apart. This time lag was partially selected for convenience4 (J ap 1999) and
partially because new product development projects typically take 9-36 months from
original concept to launch (Grifﬁn, 1997). Since the unit of analysis was the NP
development program of the ﬁrm or division, this ten-month period reﬂects the changes
within a new product development program from the infusion of technological
innovations at different stages of the adoption process.

To maximize the sample size and reduce attrition over time, a minimum of two
respondents from each company were asked to complete the survey. A modiﬁed Dillman
(1978) approach was utilized in soliciting survey response. This consisted of three waves
of mailings where the ﬁrst and third waves are the survey instruments and the second
mailing is a postcard reminder.

Six of the eight constructs in the model have been used in numerous studies,
thus, the scales were borrowed or adapted from previous empirical research. Measures
for ‘adaptive capacity’ and ‘knowledge exploitation’ have not been previously reported.
‘Adaptive capacity’, was based on the well-documented concepts of absorptive capacity,

ﬂexibility, and marketing/R&D orientation originating in the management and marketing

 

4 six months or less between sampling was suggested by Sandy lap in personal correspondence.

14

literature. Measures for these concepts have been reported in the literature, and form the
measures for the new construct, ‘adaptive capacity’. Likewise ‘knowledge exploitation’
was based on the well-cited works of Cohen and Levinthal (1990) and March (1991).
Measures for this construct have not been reported in the extant literature. The measures,
thus, were based on the theoretical models suggested by these researchers.

Convergent validity and discriminant validity were tested using conﬁrmatory
factor analysis (F omell & Larker 1981; Anderson & Gerbing 1988). Measurement
reliabilities evaluated using Cronbach’s alpha coefﬁcient (Nunnally 1978). Content
validity was ensured in the development of the survey by pretesting the instrument with
managers and other researchers knowledgeable about the new product development
process.

Three data collection points allow the full structural model to be tested using
growth curve modeling techniques in EQS (Bijleveld, 1998). Structural equation
modeling (SEM) tests interdependencies and interrelationships among the constructs, as
well as evaluates the total effect of the antecedents on the dependent variables. All

methodology is further developed in Chapter 3.

1.5 Research Questions to be Answered

The primary research motivations of this study are to fully develop and test the
proposed model of the adoption of innovations as well as to explore the evolutionary
nature of the ﬁrm as it adapts to the changing internal environment instigated by the
adoption of new innovations.

The objectives of this dissertation are two-fold: (1) to advance the practitioner’s

understanding of how adopted innovations can assist new product managers in ensuring

15

the success of invented innovations, and (2) to advance the methodological framework of
how to study the NP process of technological ﬁrms as a dynamical system. Thus, two

sets of research questions ensue:

1.5.1 Practitioner Oriented Questions
PQl Can the adoption of innovations into the ﬁrm contribute to the success of new
products?

(i) Can the adoption of innovations act as a catalyst to the successful
invention of new products?

(ii) Does the successful adoption of technological innovations lead to an
increase in the number of new product development projects undertaken
by the ﬁrm?

(iii) What resources and types of organizational cultures (adaptive/rigid,
integrated/decentralized marketing-R&D, limited/moderation/excess
resources) promote the successful adoption of innovations into the ﬁrm?

PQ2 How should ﬁrms structure themselves to increase the successful adoption of
new innovations?
PQ3 What effects do the successful adoption of innovations have on lock-in (core

rigidities) and lock-out of knowledge?

1.5.2 Theoretical/Methodological Research Questions
TQl Does a relationship exist between the adoption of innovations and the
successful invention of new products in the technological ﬁrm? Is this

relationship nonlinear?

16

TQ2 Is adaptive capacity an antecedent to the successful adoption of innovations?
How should it be operationalized?

TQ4 From a methodological perspective, does system dynamics modeling better
represent the evolving new product development process compared to more
traditional linear representations?

By undertaking both sets of questions the intent of this dissertation was to advance
academic and practitioner knowledge about the innovation process. The results should be
interesting to new product managers who must understand the factors that lead to success
in the marketplace through the continual development and introduction of new products.
Academics should beneﬁt from the advancement in theory about the evolving innovation

process and the role of adopted innovations in the NPD process.

1.6 Summary and Overview of the Remaining Chapters

Chapter 1 provided an overview of the focus of this dissertation. A conceptual
model was presented which proposes relationships among the construct ‘success of
innovations adopted’, its antecedents, and its resulting outcomes. The research objectives
and questions addressed in studying this model were also introduced. Chapter 2 will
further develop the theoretical foundation of this model by presenting a literature review
on the adoption of innovations and building the hypotheses, which were tested in this
dissertation.

Chapter 3 will provide an overview of the methodology, which were used to test
the hypotheses detailed in Chapter 2. Chapter 4 details the results of the system
dynamics model, and Chapter 5 presents the results of the longitudinal survey. Chapter 6

summarizes the dissertation with conclusions and contributions.

17

CHAPTER 2

LITERATURE REVIEW AND HYPOTHESES DEVELOPMENT

2.1 Domain Delineated

The adoption of innovations has been studied from numerous perspectives:
economics (Dosi, 1990; Freeman, 1988; Perez, 1983), management science (Dewar &
Dutton, 1986; Damanpour, 1991), organizational behavior (Baldridge & Burnham, 1975;
Kimberly, 1981), and sociologically (Rogers, 1995). Because of these diverse
approaches to studying the adoption of innovations, it is important to set the domain of

this dissertation.

2.1.1 Unit of Analysis and 'Innovation' Domain

The unit of analysis for this study was the NPD program at the division level. For
ﬁrms without multiple divisions, ‘ﬁrm’ is used interchangeably with ‘division’.
Reference to a ‘division’ also pertains to the strategic business unit (SBU). In this study,
‘organization’ is used interchangeable with ‘ﬁrm’, with both terms referring to proﬁt-
seeking corporations.

This dissertation studied only technological innovations as opposed to the
managerial or administrative innovations commonly studied from an organizational
perspective. ‘Innovation’ has been deﬁned as an iterative process initiated by the
perception of a new market and/or new service opportunity for a technology-based
invention, which in turn leads to development, production, and marketing tasks striving
for the commercial success of the invention (Freeman, 1994). Technology-based

innovations are those innovations that embody inventions from the industrial arts,

18

engineering, applied sciences and/or pure sciences. Examples include innovations from
the electronics, aerospace, pharmaceuticals, and information systems industries.

Additionally, although this theory applies to product and process innovations, the
focus of this dissertation was on technological product innovations. Product innovations
are new products or services introduced to meet an external user or market need (Knight,
1967; Utterback & Abernathy, 1975). Damanpour and Evan (1984) even distinguish
between a technical innovation and a technological innovation. “Technical
innovations. . .are not merely innovations resulting from the use of technology. They are
deﬁned as innovations that occur in the technical system of an organization and are
directly related to the primary work activity of an organization. A technical innovation
can be the implementation of an idea for a new product or new service or the introduction
of new elements in an organization’s product process or service operation” (pg. 394).
Technological innovations are broader and encompass technical innovations as well as
any innovation embodying a technology.

It is important to elucidate that an invention does not become a product
innovation until it has processed through production and marketing tasks and is diffused
into the marketplace (Freeman, 1994; Layton, 1977; Smith & Barsﬁeld, 1996). This
dissertation looks at the inﬂuences of innovations adopted by the ﬁrm on the inventions
or new products created by the ﬁrm for future marketplace distribution. Innovations will
most likely originate from outside the company whereas inventions must be invented
within the ﬁrm for the focus of this dissertation. In order to distinguish between the
innovations adopted by the ﬁrm and the innovations created by the ﬁrm, the terminology,

‘innovation’ refers to adopted innovations and ‘invention’ or ‘new product’ refers to

19

created-in-house innovations, even those that the adopting ﬁrm has introduced to the

market.

2.1.2 Adoption vs. Diffusion of Innovation

Another important distinction is the focus on the adoption of technological
innovations and not on the diffusion of technological innovations into the organization.
This is important because many researchers have emphasized that the process of
innovation continues as the innovation diffuses into the marketplace. This implication is
made in Freeman’s deﬁnition whereby ‘innovation’ is not just an entity but instead a
process. Innovations evolve due to end-user customization. These innovations are
frequently referred to as ‘incremental’ innovations. This dissertation does not address the
diffusion process or how innovations evolve during diffusion. It assumes that a ﬁrm has
adopted a technological innovation and that the innovation may or may not be
customized by the adopting ﬁrm. Kimberly (1981) distinguishes between diffusion and
adoption. “Diffusion is the process whereby an innovation spreads in a population,
adoption is a process which results in a decision by potential adopters to invest resources
in an innovation” (pg. 86).

The next question to arise was: “when is an innovation considered adopted?”
Answers to this question have varied in the literature. An innovation may be considered
as having been adopted once the decision about adoption has been made (Walker, 1969;
Daft and Becker, 1978), once the implementation process has been initiated (Evan &
Black, 1967) or after it has been successfully implemented (Mohr, 1969; Rowe and
Boise, 1974). For the purposes of this study, a technological innovation is considered

once the implementation process has been initiated. The reason for this focus was to be

20

able to model how the new innovation alters the new product development process over
time from the ﬁrst decision to adopt the innovation until its full assimilation into the ﬁrm.
It is also important to deﬁne what constitutes the successful adoption of an
innovation and what comprises the unsuccessful adoption of an innovation. Most studies
on the adoption of innovations have used a positive decision to adopt as a measure of the
dependent variable, success. However, Tomatzky and Klein (1982) and Robertson and
Gatignon (1986) both argue that a measure of success should be the adoption decision
Ed. implementation of the innovation. Robertson and Gatignon use measures of the
‘depth of use’ of the technology and the ‘degree of use’. This dissertation considered the
successful adoption of an innovation as one which has a high degree of assimilation
(depth of use) and a high degree of implementation (degree of use) along with a high
degree of efﬁcacy (degree of usefulness). The unsuccessful adoption of an innovation is
characterized by a low degree of assimilation, implementation and efficacy. An
innovation should not be considered a failure if it falls low on just one of these scales.
Often an innovation is pertinent only to a single department. For example, the Linux
operating system is more useful to an engineering department and may not even be
implemented in a ﬁrm’s marketing department. This would not make it unsuccessful in
its adoption by the ﬁrm. For this reason 'dissemination', which describes the percentage
penetration through an industry or organization, was not considered in the measure of the
success of innovation adoption in this dissertation. Dissemination does not consider

assimilation or efﬁcacy.

21

2.1.3 Domain Summary

In summary, the focus of this dissertation was on the effect of the adoption of a
technological innovation on the success of new inventions created by the ﬁrm and the
evolutionary changes a ﬁrm undergoes to continually invent and market new products. A
case in point would be the CSAA example cited in the introduction. This insurance
company adopted voice-processing equipment (technological innovation) in order to
provide new services (invention) to their customers. The voice processing equipment
was most certainly adapted to the unique requirements of the CSAA, but more
importantly, for the focus of this dissertation, the ﬁrm used this technological innovation
to develop new services for their customers thereby becoming more competitive. Even
though voice processing can be considered a well-diffused technology (introduced into
the market in the late 705), it was a new innovation to CSAA that resulted in new
inventions for their customers. It was a widely implemented innovation with high
efﬁcacy for the ﬁrm and a high degree of assimilation since all employees of the ﬁrms

were ultimately affected by the new technology.

2.2 The Evolutionary Nonlinear Dynamic Complex System

The model presented in chapter one was described as an evolving nonlinear
dynamic complex system. Complex systems take on many meanings in the
organizational and management literature. Traditionally it has been used to describe an
organization with many interconnected departments, strategic business units or a large
number of employees (Blau & McKinley, 1979; Blau & Schoenherr, 1971; Mileti,
Gillespie & Haas, 1977). The greater the number of interacting members in the

organization, the greater is its complexity. It has been argued that complexly structured

22

organizations are more likely to be innovative compared to less complex ones (Aiken &
Hage, 1971; Baldridge & Burnham, 1975; Moch & Morse, 1977).

Complex systems are mainly characterized by the phenomenon that small
disturbances in the system can multiply over time because of nonlinear relationships and
the dynamic repetitive nature of the system. These small perturbations in the system can
have large impacts, although the impact may not be observed until a later point in time.

It is because of the feedback inherent in complex systems that these small seemingly
irrelevant noises can be ampliﬁed and molded into larger more inﬂuential phenomena.

Complex systems exhibit self-reinforcing positive and negative feedbacks that
invoke path dependencies. These path dependencies in turn create nonlinear relationships
among the interacting parts (agents) of the system. Disruptions originating in one part of
the system, in effect, propagate forward and backwards along the chain. For example, the
success of an innovation can create a positive self-reinforcing phenomenon called
increasing returns (Arthur, 1990). Exponentially positive returns are granted to the
product with the highest market share (e.g., VHS). Likewise, failed products rarely can
turn around their diminishing returns and face negative self-reinforcing nonlinear proﬁt

levels (e.g., Betamax).

2.2.1 The NPD Process as a Complex System

The new product development (N PD) program process can be described as an
evolutionary nonlinear complex adaptive system. The NPD program includes the full
portfolio of new product projects under development by the ﬁrm. The NPD process
rarely follows a linear path of product invention as set forth in the stage-gate process

popular in practitioner oriented reference books (Cooper, 1993). It is an evolving system

23

as innovations are continually invented by the ﬁrm to address the dynamic marketplace.
Complexity ensues as the agents continually interact with each other and the environment
causing the NPD process to adapt to the changing environment. The strategic goals of
the ﬁrm may not necessarily be changing, but the agents within the NPD process are
dynamic. Exogenous factors such as competitors, customer demands, regulatory
agencies, and/or technological developments may cause ﬁrms to alter their NPD process.
Likewise endogenous factors such as resource availability, skills sets of the NPD team,
managerial turnover, and/or organization-wide strategic changes can cause alterations in
the NPD process. Nonlinear relationships results as feedbacks from both endogenous and
exogenous factors result in further adaptations to the NPD process. The question is: how
does a technological ﬁrm develop a new product development program strategy for this
type of dynamic process? It is argued in this dissertation that one way for technological
ﬁrms to remain adaptive in evolving environments is to maintain a steady supply of
catalytic innovations in its knowledge arsenal that facilitate the invention of new
products.

In complex systems it is difﬁcult to plan how the agents in the system will interact
with each other. Two ideas or innovations, which may appear to be completely
unrelated, can join together to organize into a larger unexpected idea. The intemet
evolved in this fashion as an organization’s need to keep informed on projects (CERN)
melded with the US Defense systems need for wartime communication to form a very
complex but highly integrated vehicle for the public to share information with each other.
Many other examples of this type of organization of system agents exist in the historical

archives of new products.

24

Based on this foundation of the NPD process as a complex adaptive system, the
role of a ‘catalytic innovation’ should become more evident. A catalytic innovation is an
innovation that when introduced into the complex system will react with the other agents5
within the system. This reaction may result in a new product idea or a new process, and
may evolve into a new invention for the ﬁrm. The resulting reaction may not be
immediately evident in the system as complex systems are time and path dependent. It is
also feasible that no reaction between agents will occur at all. It is the focus of this
dissertation to look at catalytic innovations to determine if ﬁrms can increase the success
rate of their new product development programs through the adoption of these types of

innovations.

2.2.2 Modeling of Complex Systems

There are many methods of modeling complex systems (Daneke, 1999). This
dissertation used a system dynamics model to evaluate the nonlinear relationships in the
new product development process. A brief summary here will provide an explanation to
the typical types of relationships that emerge in studies of system dynamics. Sterman
(2000) identify four common modes of behavior in dynamic systems: ‘exponential
growth’, ‘S-shaped growth’, ‘growth with overshoot’, and ‘overshoot and collapse’.
Exponential growth arises from positive (self-reinforcing) feedback. The advancements
in semiconductor integrated circuits can be described as exponentially growing. Most
systems cannot maintain such a rapid growth rate over an extended period of time.

Limitations imposed on the system will impede growth. Sterman calls this constraint on

g

5 In this dissertation, agents are the factors (constructs) that will be tested. Thus, ‘knowledge exploitation’
is an agent. From this point forward, factor will be used instead of ‘agent’.

25

an otherwise positive linear or exponential relationship as the system’s ‘carrying
capacity’. Exponential growth with a carrying capacity constraint describes the S-shaped
growth curve, which has commonly been used to model the diffusion of innovations
(Rogers, 1995) and product performance over time (Utterback, 1996). S-shaped growth
models negative feedbacks that constrain growth. “Often however, there are signiﬁcant
time delays in these negative loops. Time delays in the negative loops lead to the
possibilities that the state of the system will overshoot and oscillate around the carrying
capacity” (Sterman, 2000, pg. 121). This results in S-shaped growth with overshoot.
Carrying capacity however is not ﬁxed. Erosion of carrying capacity creates a second
negative feedback also limiting growth. If carrying capacity is itself dynamic, at the peak
of the curve, carrying capacity is at its maximum. Collapse occurs in the system as the
carrying capacity is eroded. This results in overshoot and collapse. This has been used to
describe the rate of product or process innovation (Utterback, 1996). These are common
relationships predicted to be observed in this dissertation. Figure 2.1 graphically
demonstrates these fundamental modes in business dynamics.

With the domain of this dissertation addressed, the following sections will provide
further elaboration on the model of infusion effects on the NPD process introduced in

Chapter 1.

26

Figure 2.1 Fundamental Modes in Business Dynamics

Exponential Growth S-Shaped Growth

 

 

 

S-Shaped Growth with Collapse Growth with Overshoot

 

 

Linear

 

 

27

2.3 Background Literature on the Adoption of Technological Innovation

Extant literature has attempted to understand the motivations for the adoption of
innovations into organizations (Damanpour, 1991), the diffusion of innovations among
organizations (Czepiel, 1974), or the rate of organizational innovativeness by adOpting
innovations (Daﬁ & Becker, 1978; Zaltman, Duncan & Holbek, 1973). In nearly all
studies of the adoption of innovations, there is an attempt to relate innovative behavior to
organizational structure (e.g., centralization/formalization), departmental factors, or
personal characteristics of the organizational personnel (e.g., risk averse/authoritative
managers).

Empirical analyses have focused on the antecedents that facilitate the adoption of
technological innovations. To account for the variations in patterns of organizational
adoption of innovations, researchers have empirically tested four general categories: (a)
attributes of organizational structure, (b) individual characteristics of organizational
members, (c) environmental inﬂuences, and (d) the attributes of the innovation being
adopted . Damanpour (1991) and Tomatzky and Klein (1982) conducted thorough meta-
analyses on the adoption of innovations although from two different perspectives, which
are now outdated. Damanpour focused on the organizational structure of the ﬁrm
whereas Tomatzky and Klein focused on innovation attributes. Damanpour used
organizational innovativeness or rate of adoption of innovation as the dependent variable
(DV) whereas Tomatzky and Klein used adoption (yes/no) and, when available,
implementation as the DV.

Damanpour found that specialization (diversity of specialist), functional

differentiation (number of unique departments), and external communication had a

28

positive inﬂuence on the adoption of technical innovations by ﬁrms while centralization
had a negative effect. Formalization and vertical differentiation (degree of hierarchical
levels) were found to have non-signiﬁcant effects. Tomatzky and Klein found that
compatibility and relative advantage of the innovation was positively related to its
adoption. Complexity of the innovation was negatively related to adoption, whereas cost
of an innovation and trialability were found to be insigniﬁcant. Although they did not
test their propositions, Robertson & Gatignon (1986) suggested that competitiveness,
technology standardization, industry heterogeneity, professionalism, and demand
uncertainty positively relate to the diffusion and adoption of innovations. From an
environmental perspective, Baldridge and Burnham (1975) reported that high
environmental uncertainty and diversity encourages the adoption of innovations. In
hospitals, Meyers and Goes (1988) determined that the environmental factor, community
urbanization, was signiﬁcantly correlated with innovation assimilation. The antecedents
tested as predictors of innovation adoption have been numerous and the results varied.

Post-adoption outcomes are relatively less developed in the extant literature. A
few empirical studies have reported the positive effects of the adoption of innovations on
performance. Armour & Teece (1978) showed that the adoption of an administrative
innovation increased the rate of return on owner’s equity. Damanpour & Evan (1984)
reported that a balanced implementation of administrative and technical innovations lead
to higher performance in public libraries.

Most studies have ignored how the organization utilizes the innovation. Those
studies that have addressed ‘what happens after adoption’ take an organizational behavior

approach and focus on the resistance to implementation of new innovations (Repenning,

29

2001; Leonard-Barton, 1995; Klein & Sorra, 1996). It is interesting to note that in the
organizational literature, a ﬁrm’s innovativeness has been modeled as its willingness to
adopt innovations (Damanpour, 1991; Han, et al., 1998) although how this
innovativeness leads to new products is never expounded upon. The infusion effects of
technical innovations on the new product process appear to be an under-researched area.
This dissertation looks to address the infusion effects of the adoption of innovation on the

dynamic new product process.

2.4 Hypotheses Developed

In this section, hypotheses, which were used to test the model as introduced in
Figure 1.1, are formulated.
2.4.1 Antecedents to the Successful Adoption of Innovations

This section formulates the hypotheses of the model delineated in Figure 2.2.
Adaptive capacity (H1), slack resources (H2), marketing-R&D integration (H3) are
antecedents to the successful adoption of innovations into the organization. The
successful adoption of innovations has previously been identiﬁed as those innovations
that have a high degree of assimilation (depth of use), implementation (degree of use) and

efﬁcacy (degree of usefulness) within an organization.

2.4.1.1 Adaptive Capacity as a Construct

A ﬁrm’s adaptive capacity describes how well a ﬁrm can adapt to dynamic
environments. Weick (1998) found that ﬁrms with lower adaptive capabilities are more
bureaucratic in structure and inﬂexible in strategy. Lewin, Long & Carroll (1999) found

the more bureaucratic an organizational structure, the lower its absorptive capacity for

30

new knowledge and the lower its abilities to recognize new opportunities, particularly in
regards to the development of new products. New product development is viewed as one
important way that organizations can adapt to changes in markets, technology, and
competition. Adaptive capacity encompasses a ﬁrm’s ﬂexibility to adapt to changes, to
absorb new knowledge, and to have a marketing and technology orientation. These

measures are elaborated upon below.

Figure 2.2 Antecedents to the Successful Adoption of Technological Innovations

 
 
    
 
 
    
 
   
 

Adaptive Capacity

  

. Absorptive Capacity
. Flexibility

Marketing/R&D
Orientation

     
   
    
   
 
   
   

Success of
Adopted Innovations

 
  
 

Slack
Resources

Assimilation
Implementation
Efﬁcacy

 

Marketing-R&D
Integration

2.4.1.1.1 Flexibility
Flexible organizations quickly adapt to their environment and rapidly respond
with strategies and inventions to address the dynamic marketplace. These types of

organizations have been labeled as ‘organic’ (Burns & Stalker, 1961), or as ‘prospectors’

31

(Miles & Snow, 1978). These ﬁrms are characterized as lacking in bureaucratic rigidities
and high administrative intensity and rapidly build intuition and ﬂexible options in order
to learn quickly about and shift with uncertain environments (Tyre & Orlinkowski, 1994).
“In general, ﬂexibility sustains any complex system. To respond to complex and
changing markets, ﬁrms must maintain a variety of resources, routines, and robust
intellectual processes. Monolithic responses are displaced by diverse strategies and
activities. Organizational diversity, product diversity, life cycle diversity, customer
diversity, and strategic business unit diversity help ﬁrms pursue business opportunities”
(Phillips & Tuladhar, 2000, pg. 28). Flexible ﬁrms demonstrate high market and R&D

orientations.

2.4.1.1.2 Market & R&D Orientation

Market orientation refers to the generation of market intelligence pertaining to
current and future customer needs, dissemination of the intelligence across departments,
and responsiveness to the intelligence collected (Kohli & Jaworski, 1990; Narver &
Slater, 1990). The ﬁrst step in building a market orientation is the generation of market
information regarding competitors, technology evolution, and the current and future
needs of customers, latent as well as expressed (N arver, et al., 2000). Information
disseminated throughout the organization together with strategies developed and
implemented in response to the information gathered constitute the full realm of market
orientation.

Complementary to market orientation is a ﬁrm’s R&D orientation. R&D
orientation includes the generation and dissemination of technology intelligence as ways

of monitoring technological advances that may be important to help identify catalytic

32

innovations. It also refers to the ﬁrm’s responsiveness to this technology intelligence.
Berthon, et al (1999) introduced a similar concept called ‘innovation orientation’. An
innovation orientation however focuses on “openness to innovation” (Zaltrnan et al,
1973) and “a capacity to innovate” (Burns & Stalker, 1961). An innovation orientation is
equivalent to the ‘responsiveness’ component of market orientation but does not include
the generation of technology intelligence or dissemination of the information collected.
Market orientation has been closely linked with organizational learning (cf. Day
1994; Hurley & Hult, 1998; Kohli & Jaworski, 1990; Narver & Slater, 1990; Narver et al,
2000) as well as should R&D orientation. A ﬁrm may monitor the marketplace and
utilize the information gained from the intelligence collection and still not learn from the
data collection. Reasons for this lack of knowledge accumulation can be explained by a

ﬁrm’s absorptive capacity.

2.4.1 .1.3 Absorptive Capacity

Absorptive capacity is deﬁned as a ﬁrm’s “ability to recognize the value of new
information, assimilate it, and apply it to commercial ends” (Cohen & Levinthal, 1990,
pg. 128, emphasis added). Whereas marketing and R&D orientations describe a cultural
position regarding the collection of intelligence, absorptive capacity’s focus is on the
ﬁrm’s ability to use the information to accumulate knowledge and apply that knowledge
in the development of new inventions. Absorptive capacity focuses primarily on
technological information but does not preclude market intelligence.

Absorptive capacity is path dependent. Cohen & Levinthal argue that the ﬁrm’s
ability to evaluate and utilize outside knowledge is highly dependent upon the level of

prior related knowledge. They expound on this idea:

33

“By having already developed some absorptive capacity in a particular
area, a ﬁrm may more readily accumulate what additional knowledge it
needs in the subsequent periods in order to exploit any critical external
knowledge that may become available. Second the possession of related
expertise will permit the ﬁrm to better understand and therefore evaluate
the import of intermediate technological advances that provide signals as
to the eventual merit of a new technological development. These revised
expectations, in turn, condition the incentive to invest in absorptive
capacity subsequently” (pg. 136).

A ﬁrm’s absorptive capacity is highly dependent on the individuals within the
ﬁrm who act as gatekeepers of information in boundary spanning roles. Knowledgeable
gatekeepers monitor the environment and translate technical information into a form
understandable to the internal staff. Thus, while R&D orientation is a cultural aspect of
the ﬁrm, absorptive capacity addresses the structure of the ﬁrm to position capable
personnel in positions that will allow the successful assimilation of market and
technology intelligence.

To summarize, adaptive capacity explains a ﬁrm’s propensity to be structured
ﬂexibly (organically), its readiness to build marketing and R&D orientations, and its
ability to utilize the intelligence collected from monitoring the market and technology
environments for developing new products. A ﬁrm’s adaptive capacity was measured

with these three different types of scales.

2.4.1.2 Adaptive Capacity as an Antecedent

As described above, a ﬁrm’s adaptive capacity is deﬁned by how well it can adapt
to dynamic environments. Thus, ﬁrms with high adaptive capacities should have greater
success at assimilating catalytic innovations into the organization. It has been shown that

many innovations are either not adopted or not successfully implemented within the

34

organization because of internal resistance (Klein & Sorra, 1996; Repenning, 2000) or
because of the lack of ﬂexibility within the organizational structure (Kimberly, 1981).

All ﬁrms have capacity constraints. When a capacity constraint is reached the
ﬁrm becomes inefﬁcient, or reaches a point of chaos. At this point, too many options are
available for a ﬁrm to consider the appropriate action to take. If a ﬁrm has too much
ﬂexibility, it lacks the structure to set organization-wide strategic plans. If it collects too
much intelligence, the knowledge base fails to expand because of ‘information overload’.
Similarly, ﬁrms must be concerned with 'lock-in' or what Leonard-Barton (1995) refers
to as 'core rigidities.‘ This phenomenon occurs when a ﬁrm is so ingrained in a particular
technology and knowledge base that it is unable to adapt to changing market conditions
due to the rigidities in the ﬁrm. Too much knowledge is accumulated in one area,
locking the organization from learning from new intelligence accumulated. March
(1991) terms this as an imbalance in knowledge exploration compared to knowledge
exploitation. An organization may focus too much on gathering information instead of
utilizing the information to adopt innovations. Because of the imbalance in knowledge
exploration, a nonlinear relationship exists between a ﬁrm’s adaptive capacity and the
successful adoption of innovations. Thus,

H1: The relationship between a ﬁrm’s adaptive capacity and its successful adoption
of innovations follows over time a nonlinear path of S-shaped growth.

2.4.1.3 Slack Resources

Organizational slack resources determine whether the organization can afford to
adopt innovations. Slack resources have been deﬁned as those resources that have not
been committed to other organizational departments or programs. "The existence of

slack means that the organization can afford (1) to purchase costly innovations, (2) to

35

absorb failures, (3) to bear the costs of instituting the innovation, and (4) to explore new
ideas in advance of actual need” (Rosner, 1968, pg. 615). Slack resources are not just
monetary assets. They also include human, physical, and temporal resources. Is there
knowledgeable staff with available time to make decisions about new innovations? Does
the ﬁrm have the proper computer systems, equipment, and/or building sites necessary to
ensure the successful implementation of a new innovation? Are slack manufacturing
capacity, R&D capabilities, etc. available to assimilate the innovation?

In their study on the adoption of innovations, Dewar and Dutton (1986) found that
the greater the number of specialists in an organization, the more easily new technical
ideas can be understood and procedures developed for implementing them. Likewise,
Robertson & Gatignon (1986) found that R&D allocation lead to enhanced technologies.
Although Damanpour (1991) expected to ﬁnd, in his meta-analysis on the adoption of
innovations, a positive relationship between slack resources and the successﬁrl adoption
of innovations, he found a non-signiﬁcant relationship. He did however, ﬁnd a
signiﬁcant positive relationship between technical knowledge resources and adoption
success. Technical knowledge resources “reﬂect an organization's technical resources
and technical potential" (pg. 589).

F enell (1984) warns that researchers should consider whether parallel competing
innovations factor into the adoption of innovations. "Parallel innovations may pose
competing alternatives for the use of slack resources, either because two or more
innovations may have developed simultaneously in the same substantive area or because
different innovations may represent alternative ways of expending resources (Kimberly

1981)" (pg. 114). Organizations will not have unlimited funds available for adopting all

36

the innovations they believe beneﬁcial to the production of inventions. Even when slack
resources are readily available, management will very likely impose a cap on the
percentage available for new innovations. Additionally, by deﬁnition, slack resources
allow the ﬁrm to absorb failures. With an increase in slack resources, managers will be
comfortable making risky decisions that can lead to failure. It is hypothesized that as
resources increase, the rate of change in success will decrease. Thus,

H2: The relationship between a ﬁrm’s slack resources and its successful adoption of
innovations follows over time a nonlinear path of S-shaped growth.

2.4.1.4 Marketing-R&D Integration

There is strong and consistent empirical evidence, applicable across both services
and products and in both customer and industrial markets, that communication and
cooperation between marketing and R&D enhances new product development success
(Grifﬁn & Hauser, 1996; Moenart & Souder, 1990). NPD projects that balance
marketing and R&D inputs have higher rates of success (Cooper, 1984a, 1984b). Cross-
functional integration reduces language, thought, and physical barriers. This enables
more information to be communicated and utilized, increasing coordination and decision-
making (Ruekert & Walker, 1987).

Similar predictions can be made regarding the integration of marketing and R&D
and the successful adoption of innovations. Proliferation of an exchange of information
about potential market demands allows R&D to better anticipate how an innovation
adopted into the organization will facilitate the invention of new products. Likewise
R&D can provide marketing personnel with information regarding internal technology

progress that may provide solutions for customer demands. The inter-exchange of ideas

37

between departments increases idea generation and knowledge assimilation. This creates
a better understanding by both departments on how externally created innovations may be
successﬁrlly utilized within the ﬁrm for new product invention.

“It has become generally accepted that complementary functions within the
organization ought to be tightly intermeshed, recognizing that some amount of
redundancy in expertise may be desirable to create what can be called cross-function
absorptive capacities. Cross-function interfaces that affect organizational absorptive
capacity and innovative performance include, for example, the relationships between
corporate and divisional R&D labs or, more generally, the relationships among the R&D,
design, manufacturing, and marketing functions (e.g., Mansﬁeld, 1968: 86-88)” (Cohen
& Levinthal, 1990, pg. 134).

Grifﬁn & Hauser (1996) argued that marketing and R&D responsibilities in new
product development are neither independent nor static; they cannot be analyzed
separately. Responsibilities in both departments interrningle as new technological
emerges, as customer needs change, as competitors offer new products, and as
environments evolve. Thus,

H3: The greater the marketing-R&D integration, the more successful the adoption
of catalytic innovations into the ﬁrm.

2.4.2 Outcomes of the Successful Adoption of Innovations

Extant literature has focused on how an innovation moves through stages in
becoming a standard part of an organization (cf. Meyer & Goes, 1988; Pelz, 1983).
Management and organizational behavior researchers have looked at how an innovation
evolves as it assimilates into the organization forintemal use (Leonard-Barton, 1988;

Van de Ven & Rogers 1988; Van de Ven & Poole 1990). Reinvention (Rice & Rogers,

38

1980) is a common occurrence during implementation (Rogers, 1983, 1995; Tomatzky, et
al., 1980). Van de Ven and Poole examined the innovation process and proposed a
methodology that emphasizes the evolution of relationships among internal
organizational ideas, people, transactions, and outcomes over time. These studies have
all focused on the intra-organizational effects of post-adoption.

Rothwell & Gardiner (1988) introduced ‘reinnovation’ to explain the invention of
incremental innovations. These researchers noted at least twelve patterns of redesign and
reinnovation that occur throughout the new product development process from
conceptualization to diffusion. Of these twelve patterns, four involve taking existing
technology and improving upon it to invent new innovations: 'retroﬁtting’, 'new
generation', 'hybrid technologies', and 'radical technologies'. ‘Retroﬁtting’ involves the
overhaul of existing product designs with new features and improvements whereas ‘new
generation’ is a total redesign of an existing product. ‘Hybrid technologies’ upgrade
existing products with new technologies whereas ‘radical technologies’ use new
technology to introduce completely new products. The essence of these reinnovation
processes is the utilization of one technology to invent a new product. Thus, technologies
adopted by a ﬁrm are used in some manner in order to create a new product. It is not
uncommon for one technology to act as a catalyst in the invention of a new technology
(e. g., the Watt engine facilitated steamboat transit, smaller and faster semiconductors
spurred laptop inventions, and lasers are now being used as ‘scalpels’ in heart surgery).

Yet, new product researchers know very little about how the adoption of
innovations affects the new product development process over time. As previously

noted, the infusion effects of innovations on the ﬁrm’s new product development process

39

are an under-developed area of research. This dissertation looked at the effect of the
successful adoption of innovations on the ﬁrm's ability to exploit knowledge (H4), to
build on marketing proﬁciencies (H5), and to build on R&D proﬁciencies (H6).
Predictions regarding these hypotheses are described in this next section. Figure 2.3

delineates this sub-section of the model.

Figure 2.3 Infusion Effects from the Successful Adoption of Technological

Innovations

 
  
 
  

  
  

Knowledge
Exploitation

   
  
 

Success of
Adopted Innovations
. Assimilation
. Implementation
Efﬁcacy

   
     
  
  

  

  
  

NPD
Program
Success

Marketing
Proﬁciencies

  
 

 
   
  

        
       
  
   
 

  

R&D
Proﬁciencies

2.4.2.1 Knowledge Exploitation as an Outcome

The ﬁrm's ability to evaluate, implement, and assimilate knowledge for
commercial success is of strategic importance to the innovating organization (Grant,
1996; Cohen & Levinthal, 1990). This ability is frequently modeled as competitive
advantage (Burgelman, 1990; Senge, 1990). Quinn (1992) proposed that knowledge

based tangibles such as know-how, design expertise, and marketing insights largely

40

determine the competitive advantage of most products. In these studies, knowledge is
considered an asset.

However, knowledge can also be considered as a process (Nonaka & Takeuchi, 1995;
Li and Calantone, 1998; Leonard-Barton, 1995). Nonaka and Takeuchi present a
dynamic model for the creation of new knowledge starting with tacit knowledge that
becomes embedded within an innovation, ending with the reabsorption of new knowledge
into the organization from the development process. They describe this as a 'spiral of
knowledge creation'. Li and Calantone distinguish between market knowledge and
market knowledge competence. Market knowledge competence encompasses the
processes that generate and integrate market knowledge (an asset) into a new product.
This concept can apply to all types of knowledge competence.

Knowledge exploitation is a related but distinct concept from knowledge
competence. Knowledge exploitation is the utilization of knowledge competence.
Knowledge exploitation has been deﬁned as “the use and development of things already
known” (Levinthal & March, 1993, pg.105). The “essence of exploitation is the
reﬁnement and extension of existing competences, technologies, and paradigms” (March,
1991, pg. 85). Catalytic innovations that are successfully adopted by the ﬁrm bring with
them new technologies and new ideas that can lead to the future development of new
inventions. These new technologies and ideas increase the knowledge competence of the
ﬁrm. If this knowledge is exploited, new applications from existing knowledge
components will emerge (Kogut & Zander, 1992). How well a ﬁrm exploits the

knowledge gained from the adoption of new innovations is a prediction of its success in

41

the new product development process. The greater the ﬁrrn's success at assimilating new
innovations into the ﬁrm, the more knowledge exists that can be exploited by the ﬁrm.

However, limitations exist on the ﬁrm's ability to exploit knowledge. A ﬁrm’s
carrying capacity will be reached, imposing constraints on future exploitation efforts.
Human, capital and R&D resources are limited, as are knowledge compentences and
managerial skills. Knowledge acquisition and exploitation are limited by these
constraints. Additionally, it is possible for the ﬁrm to adopt too many innovations. Too
many competing projects can result in all projects receiving little attention. Thus,

H4: The relationship between a ﬁrm’s successful adoption of innovations and its
knowledge exploitation follows over time a nonlinear path of S-shaped growth.

2.4.2.2 Marketing & R&D Proﬁciencies as Outcomes

It is expected that the successful adoption of an innovation will increase the
overall marketing and R&D proﬁciencies of the ﬁrm. It is hypothesized that information
collected during the decision making process of adopting innovations can lead to an
increase in market and R&D knowledge. "The greater the technical knowledge
resources, the more easily can new technical ideas be understood and procedures for their
development and implementation be attained" (Dewar & Dutton, 1986 pg. 1431). The
ﬁrm’s carrying capacity (limitations on employee skill sets, education, managerial
demands, etc.) will set limits on the growth in proﬁciencies. Thus,
H5: The relationship between a ﬁrm’s successful adoption of innovations and

marketing proﬁciencies in the new product development process follows over
time a nonlinear path of S-shaped growth.

42

H6: The relationship between a ﬁrm’s successful adoption of innovations and R&D
proﬁciencies in the new product development process follows over time a
nonlinear path of S-shaped growth.

2.4.3 Antecedents to NP Program Success

Numerous studies have been conducted regarding the antecedents to new product
program success (Montoya-Weiss & Calantone, 1994; Brown & Eisenhardt, 1995;
Cooper & Kleinschmidt, 1995). In this model of new product development as a complex
system, three potential antecedents were tested, knowledge exploitation (H7), marketing
proﬁciencies (H8) and R&D proﬁciencies (H9), Figure 2.3.

New product success has been accounted for in many ways (Grifﬁn & Page,
1993; 1996). They report that ﬁrms and academics use over 75 distinct measures of
product development success. Since the unit of analysis in this dissertation is the
division, success was determined on a program level. Because different divisions have
different product development strategies (Miles & Snow 1978), it is important to measure
a division’s success in growth, efﬁciency and effectiveness of the new product
development program, and ﬁnancial performance (Grifﬁn & Page, 1996). The actual

measures to be used will be delineated in Chapter 3 of this dissertation.

2.4.3.1 Knowledge Exploitation as an Antecedent

Drucker (1985) traces a ﬁrm's competence in new product development to its
processes of generating knowledge about its customers and competitors and integrating
such knowledge with technology. The greater its ability to utilize the knowledge
generated, the greater the success in the NPD process. Leonard-Barton (1992) termed

this ability as a ﬁrm's core capability. "[Core] capabilities continually spawn new

43

products and processes because so much creative power is focused on identifying new
opportunities to apply the accumulated knowledge base" (pg. 118).

Knowledge exploitation can be viewed as a core capability of the ﬁrm. Core
capabilities differentiate a company strategically. It is hypothesized that a ﬁrm that
exploits knowledge innovates new products with greater success. However, all core
capabilities “simultaneously enhance and inhibit development” (Leonard-Barton, 1992,
pg. 112). Core capabilities can result in core rigidities. Core rigidities evolve as
knowledge exploitation causes ﬁrms to build capabilities in one speciﬁc area. Technical
and managerial systems, skills and knowledge sets (assets), and values become ingrained
to the ﬁrm. Firms become 'locked-in' to the old technologies and patterns of development
activities. Thus, the cumulativeness and path dependencies of new product development
can lead to technological, organizational, and institutional boundary limitations (Cohen &
Levinthal, 1990; Alange, et al., 1998).

To break the path dependency of core rigidities in new product development,
companies need to unleam, or abandon earlier practices and behaviors that were found
necessary, or even crucial, in other development projects. “An organization that engages
exclusively in exploitation will ordinarily suffer from its obsolescence. The basic
problem confronting an organization is to engage in sufﬁcient exploration to ensure its
current viability, and at the same time, to devote enough energy to exploitation to ensure
its [future] viability” (Levinthal & March, 1993, pg. 105). Some organizations are able to
use new projects as agents of renewal and organization-wide learning (Leonard-Barton,
1992), although most ﬁrms ﬁnd this very difﬁcult. It usually takes an environmental jolt

(Meyer, 1982) or catastrophe (Bak, 1996) to move a ﬁrm off its self-reinforcing path of

44

reliance upon tried and true technology onto a path of exploring and exploiting riskier

technology. "One of the central ﬁndings of complexity theory is that robust (dynamic)

systems evolve toward the balance between order (the pull of exploitation) and disorder

(the pull of exploration) ..." (Lewin, et al., 1999, pg. 540).

Thus, it is hypothesized that over time ﬁrms that utilize their knowledge
competence (exploitation of knowledge) will increase the success of their new product
development projects. However, knowledge exploitation may begin to cause
complacency in the ﬁrm’s new product development process, negatively affecting new
product success until an environmental jolt or catastrophe causes the ﬁrm to re-evaluate
its knowledge exploitation behavior. This will result in an adjustment in knowledge
exploitation over time. Thus,

H7: The relationship between a ﬁrm’s knowledge exploitation and new product
development program success follows over time a nonlinear path of S-shaped
growth.

2.4.3.2 Market Proﬁciencies as an Antecedent
Marketing and R&D proﬁciencies have been empirically linked to the success of

new products (Montoya-Weiss & Calantone, 1994). Cooper (1979) identiﬁed the notion

of maximizing proﬁciencies in NPD activities rather than just undertaking the activities
noted as key drivers of NPD success. Using a meta-analysis of forty-seven ﬁrms,

Montoya-Weiss & Calantone (1994) found that proﬁciency in market related and

technological activities are key determinants for driving new product success.

As early as 1956, Carter & Williams noted that marketing proﬁciencies are associated
with technically progressive ﬁrms. Likewise, Kuczmarski (1988) found that skill at

marketing activities is potentially critical for ensuring the success of new products. Thus,

45

building marketing proﬁciencies are essential for ensuring NP success. Marketing
activities requiring proﬁciency include customer orientation, competitive monitoring,
program development, as well as launch activities. Thus,
H8: The greater the marketing proﬁciencies, the greater the success of the ﬁrm’s new

product development program.
2.4.3.3 R&D Proﬁciencies as an Antecedent

Quality of execution of technological activities - the actual physical development,
product testing, trial production, and production start-up - has been linked to new product
development program performance (Cooper, 1979; Cooper & Kleinschmidt, 1995). R&D
strength has been noted as a major determinant of new product advantage (Li &
Calantone, 1998). Day (1994) considered R&D strength a major internal capability and
suggested that strong R&D provides a technological base critical to new product
development. Technical activities requiring competence can include the R&D process,
testing of the product, trial production, and production startup. Thus,
H9: The greater the R&D proﬁciencies, the greater the success of the ﬁrm’s new

product development program.
2.4.4 The Feedback Loop
The feedback loop as detailed in Figure 2.4 models the relationship between the

success of the new product development program as an antecedent to the ﬁrm’s adaptive
capacity (H10) and slack resources (H11). Since this dissertation models the new product
development process over time, it is necessary to determine the effects of program
success on future development projects. The feedback loop modeled here looks

speciﬁcally at slack resources and adaptive capacity over time.

46

Figure 2.4 Feedback Loop

 

 

     

      
 

Adaptive

Capacity NPD Program

Success

   

H11

  
   

Slack
Resources

 

2.4.4.1 New Product Program Success as an Antecedent

A ﬁrm’s success can be its demise. This phenomenon has been called core
rigidities (Leonard-Barton, 1995), stale in the saddle (Miller, 1991), or the ‘Icarus
paradox’ Miller (1990). Successful ﬁrms can experience downward trajectories because
their “victories and ...strengths. . .seduce them into the excesses that cause their downfall.
Success leads to specialization and exaggeration to conﬁdence and complacency, to
dogma and ritual” (Miller 1990, pg.3). Technologically based companies can move from
being pioneering entrepreneurs to ‘escapists’ where the focus on inventiveness becomes
an unproductive pursuit of technology for its own sake (Miller, 1990).

Leonard-Barton (1995) reasoned that core rigidities form because attacking the
rigidities “means undermining the current economic foundations of the ﬁrm —
cannibalizing current product lines, making obsolete current knowledge bases and skills,
lessening the values of current assets” (pg. 34). The politics of power and organization
behavioral routines that become ingrained also contribute to the strength of core

rigidities. The inertia of current practice can overwhelm concerted efforts to change a

47

ﬁrm’ strategy (Dougherty & Hardy, 1996; Harman & Freeman, 1984; Johnson, 1988).
These dependences on the status quo result in the decrease in a ﬁrm’s absorptive
capacity, its ﬂexibility, and its marketing and R&D orientations, and hence its adaptive
capacity.

Sustained success in a ﬁrm’s established new product development program can
lead to the Icarus paradox (Miller, 1990). However, an environmental jolt, internal or
external (Meyer, 1982), or catastrophe, or even intelligent management can cause the
ﬁrm to recognize the destructiveness of core rigidities and put a concerted effort back on
building core capabilities. Over time, this leads to a negative relationship between new
product program success and adaptive capacity (Figure 2.4). Thus,

H10: The relationship between a ﬁrm’s new product development program success and

its adaptive capacity follows over time a nonlinear path of inverted S-shaped
growth.

Figure 2.5 Predicted Relationship H10

Adaptive Capacity

 

 

New Product Program Success

48

As previously noted, a ﬁrm’s goals will determine how it accounts for success in
the new product development program. It can rate success based on proﬁts, market
share, number of new products launched, etc. Regardless of the ﬁrm’s method of
determining success, greater slack resources result from greater success of new products.
Grifﬁn and Page (1996) found that Prospectors6 determine new product program success
through the growth derived from the program. One measure of success Prospectors use is
‘degree today’s products lead to future opportunities’ (Grifﬁn & Page, 1996, pg. 490). If
a ﬁrm believes that today’s products will lead to new products, it will allocate the
resources necessary for evaluating future technologies that may spur new technologies.
Even if ﬁnancial slack resources are not readily available, growing ﬁrms will frequently
obtain venture capital or other entrepreneurial ﬁrnding for future new product
development. Thus, ﬁnancial success alone is not necessary to increase slack resources.
As long as the ﬁrm continues to have the successes it predicts from the new product
program, slack resources will usually be made available.

However, all resources are limited no matter how much wealth or potential
success a ﬁrm has. Most ﬁrms will put a cap on the resources available for any single
function within the ﬁrm. Additionally, Dougherty & Hardy (1996) found that in mature
organizations, resources do not ﬂow smoothly to new product projects, particularly where
prevailing practices supported established activities. Thus,

H11: The relationship between a ﬁrm’s new products program success and slack
resources follows over time a nonlinear path of S-shaped growth.

 

6 ‘Prospector ﬁrms ‘value being ﬁrst with new products, markets and technologies even though not all
efforts prove to be proﬁtable’ (Grifﬁn & Page, 1996, pg. 482).

49

2.4.5 Hypotheses Summarized

Table 2.1 summarizes the hypotheses that were tested in this dissertation. The

methodology used to test these hypotheses is presented in Chapter 3.

Table 2.1 Hypotheses

 

H1:

H2:

H3:

H4:

H5:

H6:

H7:

H8:

H9:

 

H10:

H11:

The relationship between a ﬁrm’s adaptive capacity and its successful
adoption of innovations follows over time a nonlinear path of S-shaped
growth.

The relationship between a ﬁrm’s slack resources and its successful adoption
of innovations follows over time a nonlinear path of S-shaped growth.

The greater the marketing-R&D integration, the more successful the adoption
of innovations into the ﬁrm.

The relationship between a ﬁrm’s successful adoption of innovations and its
knowledge exploitation follows over time a nonlinear path of S-shaped
growth.

The relationship between a ﬁrm’s successful adoption of innovations and
marketing proﬁciencies in the new product development process follows over
time a nonlinear path of S-shaped growth.

The relationship between a ﬁrm’s successful adoption of innovations and
R&D proﬁciencies in the new product development process follows over time
a nonlinear path of S-shaped growth.

The relationship between a ﬁrm’s knowledge exploitation and new product

development program success follows over time a nonlinear path of S-shaped
growth.

The greater the marketing proﬁciencies, the greater the success of the ﬁrm’s
new product development program.

The greater the R&D proﬁciencies, the greater the success of the ﬁrm’s new
product development program.

The relationship between a ﬁrm’s new product development program success

and its adaptive capacity follows over time a nonlinear path of inverted S-
shaped growth.

The relationship between a ﬁrm’s new products program success and slack
resources follows over time a nonlinear path of S-shaped growth.

 

50

 

CHAPTER 3

OVERVIEW OF METHODOLGY

3.1 Chapter Summary

The hypotheses delineated in Chapter 2 were tested in two studies. Study 1
utilized a case study for data collection for a system dynamics model and study 2 utilized
a longitudinal survey for data input to a structural equation model (SEM). In study 1, two
case studies were used to identify the analytical relationships between constructs and to
determine starting values for system dynamics modeling, which uses computer
simulations to evaluate the interaction effects of nonlinear relationships. This method
allowed testing of the hypotheses in a controlled setting and for modeling the feedback
process. Additionally, ﬁeld support for the predicted underlying theoretical relationships
between factors described in Chapters 1 and 2 provided validity for the structural
equation model to be tested in study 2.

In study 2, the data collected from the longitudinal study was analyzed using a
growth curve model in structural equation modeling. Empirical data collection allowed
for external validation of the hypotheses. The nonlinean'ties, with respect to time, of the

relationships between constructs were the major focus of study 2.

3.2 Study 1 - Case Study & System Dynamics Simulation

The intent of the case study was twofold: (1) to provide both qualitative and
quantitative data for input into the system dynamics model, which is summarized in
Figure 3.1 and further detailed in Chapter 4, and (2) to provide validity to the structural

equation model, which is shown in Figure 3.2.

51

Figure 3.1 System Dynamics General Framework

R&D-Marketing

     

/ Integration
Adoption of
Catatlytic
Innovations Technology
Slack @ + _ Knowledge
Resources B 1
+ - n
prosperity "revitalization" +
‘ Marketing &
R&D
Proﬁciencies

Performance
Gap +
- R1

"success trap"

b9

"street smarts"

   

 

Technology

NP Success + . .
Explortation

Notes: Arrows represent causal relationships with direction noted as a (+) or (-).
Loops labeled R indicate a reinforcing (+) feedback. Loops indicated with B,
indicate balancing (-) feedback.

52

mmooosm
Emuwoi DAZ

 

a:

In:

   
       

momocomomoi
Dag

86:20on
weavers—Z

cosgoﬁxm—

03536ch

 

     

332% 333255
23830 no mmooosm

IVE

 

 

:ocﬁwﬁﬁ
meanness:

mOOuSOmOMH
82m

bmommmo

o>smew<

2:5.—

 

 

382 538% a; mm 2:3

53

The exploratory nature of the case study facilitated identiﬁcation of the functional forms
of relationships that occur between the variables over time, both under normal and
extreme conditions. The case study also allowed collection of quantitative data that was
used as starting values for the system dynamics model.

Since no previously known empirical research had been conducted on the infusion
effects of technological innovations on the success of the new product program process
prior to this dissertation, the case studies provided a better understanding of this
phenomenon. Identifying the characteristics of catalytic innovations through discussions
with industry experts prior to the longitudinal survey increased the potential for obtaining
content validity. The case study data collection techniques and system dynamics

simulation techniques are described in the section below.

3.2.1 Case Study Data Collection

Creswell (1998) suggests that ﬁve different traditions of qualitative inquiry and
research design exist - biography, phenomenology, grounded theory, enthnography, and
case studies. He advocates that each method has distinct uses and each should be
approached uniquely. "Case study is useful for studying a bounded system such as a
process, activity, event, program or multiple individuals" (pg. 112). In this dissertation,
the case study approach was used to evaluate the new product development process of a
single technology ﬁrm. This phase of the study followed the framework of Eisenhardt
(1989) for conducting inductive research. Although a priori speciﬁcation of constructs is
not standard in case studies, it can help shape the initial design of theory-building
research and has frequently been used in building theory in the new product development
research (Colarelli-O'Connor, 1998; Leonard-Barton, 1988). When constructs are found

54

to be signiﬁcant, researchers then have “a ﬁrmer empirical grounding for the emergent
theory” (Bourgeois & Eisenhardt, 1988, pg. 536).

Pettigrew (1988) suggests that in selecting cases to be studied, researchers should
choose cases that are extreme situations but are likely to extend the emergent theory. The
ﬁrst ﬁrm studied in study 1 is a small electronics company specializing in radio
frequency power supplies for the semiconductor industry (employees = 75). This
company is one of the seventy subsidiaries owned by a private corporation headquartered
in New Jersey. For the ﬁscal year 2000, this company was rated the top earning (ROI) of
all the seventy subsidiaries owned by the parent. In order to remain competitive in a very
volatile industry, 20-30% of annual revenues stemmed from new products introduced
within the last year. Due to their successful focus on new products, this company
provided an interesting case study for evaluating the effects of the adoption of
technological innovations on the new product development process. The second ﬁrm
studied involved in the welding industry. They are a smaller ﬁrm (employees = 25) and
most of their revenues stem from existing products or incremental improvements to their
current product line. This company has recently experienced several years with low
proﬁt margins.

In case studies it is important to obtain multiple sources of information in order to
build a wholistic picture of the problem being studied. Yin and Campbell (1989)
recommends six types of information gathering techniques: interviews, direct
observations, archival records, documentation, participant observations, and physical

artifacts. For this case study, four methods of data gathering were used - interviews,

55

direct observations, archival records, and documentation. Table 3.1 summarizes these

techniques, as they took place chronologically within this study.

 

Table 3.1 Data Gathering Methods for Case Studies

 

. Semi—structured interviews with key employees
. New product team meeting observations

. Group meeting with key employees

. Documentation review

. Archival data collection (as permitted)

. F ollow-up interviews with key employees as needed

 

A preliminary study of three additional companies was conducted in the summer
of 2000. In addition to CSAA previously mentioned in the introduction, engineers from
Leapfrog, an electronic toys designer, and EDC BioSystems, a biomedical ﬁrm, provided
insights into their company's unique NPD process. Each of the interviews lasted IV: to 2
hours. Based on these interviews, an initial model of the infusion effects of technological
innovations was reﬁned to greater reﬂect the reality of the new product development
processes used in industry.

In-depth studies of the two case ﬁrm’s new product operations were conducted
over a one-week time period. Interview questions for the semi-structured interviews are
included in the Appendix. Semi-structured interviews were held with key employees
including the president, vice president, marketing manager, and senior engineers from
each company. Typical of technology-based ﬁrms, both case study ﬁrms are engineering

driven companies where the marketing manager has a limited role in new product

56

development activities. Marketing primarily oversees marketing communications
activities such as advertising, trade shows, etc. The vice president, thus, addressed the
major marketing issues pertaining to new product development activities (Workman,
1993)

After interviews with key employees at the ﬁrst case study ﬁrm, a group session
was conducted in order to reach group consensus on the new product development
process and to clarify questions surfacing from the individual interviews. Observations
during NP team meetings supplemented the data collected during interviews. Archival
data and documentation was used as made available. Since both companies are privately
owned, this data was limited. The qualitative and quantitative data collected was then

used for setting the initial conditions in the simulation analysis.

3.2.2 System Dynamics Model Simulation Analyses

Simulation techniques for modeling dynamic nonlinear systems are common in
studying organizational behavior (Morecroft & Sterman, 1994; Repenning, 2001; Sastry,
1997). System dynamics modeling is used for modeling feedback loops and overall
interactions between nonlinear relationships. Sastry (1997) emphasizes that this type of
simulation differs from other formal modeling techniques in several important ways.
First, system dynamics highlights feedback processes or circular causal relationships in
which variables inﬂuence and interact with each other. Second, behavioral decision
making is explicitly represented in the model as decision-makers are bounded by
rationality with imperfect information. Lastly, because system dynamics models
approximate continuous-time processes, they can be used to explore the effects of time

lags.

57

The purpose of using simulation in this dissertation was to test the emergent
relationships between the constructs over several time frames. Additionally feedback
loops can easily be modeled using this type of simulation. The model in Figure 3.1 was
simulated using system dynamics modeling software. This type of simulation involves
building mathematical equations of the relationships and testing them under various
conditions. This method followed the guidelines set by Levine, et al., (1992). The stages

involved are shown in Table 3.2 and discussed in the section below.

3 .2.2. 1 Conceptualization

The ﬁrst step in system dynamics model construction is ‘conceptualization’. This
step includes formulating the questions to be answered, describing the behavior of the
problem to be addressed, specifying a time frame and forming a causal diagram of
variable relationships. A major intent of system dynamics modeling is to model realistic
behaviors and to answer practical problems. The primary question, of course, addressed
the effect of the infusion of technological innovations on the NPD program process.
Firms actively seek ways to increase the success in their new product development
programs. It was hypothesized that adopting innovations may be one means of driving
NPD program success. The models previously discussed in Chapter 1 and 2 describe the
behavior of the NPD program, which were evaluated to address the research questions
pr0posed.

Figure 3.1 shows the causal diagram form of the model. It is predicted in this
model that the successful adoption of technological innovations will positively inﬂuence
the success of the NP program success (direction of relationships is shown with a + or —

in the ﬁgure). This relationship is established by increasing the internal knowledge base

58

of the ﬁrm from adopting new innovations, thereby increasing the ﬁrm’s ability to exploit
this newfound knowledge and to build its marketing and R&D proﬁciencies. Knowledge
exploitation combined with greater proﬁciencies will then lead to greater NP program
success.

The internal boundary of the model is deﬁned by those activities endogenous to
the NP program process. Activities and managerial decision making occurring outside
the realm of the program process are exogenous to this model. The unit of analysis was
the program level that includes all new product development projects undertaken by a

division (or a ﬁrm) within the speciﬁed time frame (Repenning 2000).

3.2.2.2 Formulation

Every system dynamics model is characterized by state variables and rate
variables. State variables (also called ‘stocks’) can be thought of as an 'inventory' of
goods, and hence, are also called stock variables. Rate variables are continuously

changing and are directly manipulatable. system dynamics distinguishes between
state variables, ...and variables that represent rates of change [rate variables]. . .The
distinction is important because state variables, which represent the properties of the
organization that have been accumulated over the organization's history and characterize
the system, can not be changed instantaneously" (Sastry 1997, pg. 240). Exogenous
variables are the third type of variable in a system. These variables set the boundaries for
the simulation model. The state, rate and exogenous variables will be further detailed in
Chapter 4.

The next step in the ‘formulation’ stage is to develop a ﬂow diagram of the model

with its associated computer (mathematical) equations. This is followed by selecting

59

‘reasonable parameter values’ for initial starting values for the simulation. These steps
cannot be completed until after collection of the qualitative and quantitative data from the
case study. Interviews with the case study ﬁrms provide the insights necessary to
develop the analytical models that are used in the simulation. The semi-structured
interviews asked the respondents to describe the NP program process within their ﬁrm
and to provide ‘best estimates’ for the initial settings for exogenous variables that were
subsequently used in this study.

Asking respondents to ‘graphically deﬁne’ the relationships between variables
over time is also common in system dynamics modeling data collection. This may
require showing respondents nonlinear plots such as Figure 2.1 as well as a linear plot,
which can be found in Figure A.1 in Appendix A. Respondents were than asked to graph
the relationship between variables that they experienced within their ﬁrm. If respondents
did not feel comfortable answering this type of question without considerable prompting,
the question was dropped. A copy of the workbook developed for this part of the case

study is in Appendix A.

3.2.2.3 Testing & Implementation

Levine, Van Sell and Rubin (1992) established three types of tests to verify
system dynamics models: structural tests, behavioral tests, and policy tests. Structural
tests include structural veriﬁcation, parameter veriﬁcation, dimensional consistency, and
boundary adequacy. Behavioral tests include behavior reproduction, behavior prediction,
behavior anomaly, surprise behavior, extreme policy, generalizability, and boundary
sensitivity. Policy tests include changed behavior prediction, boundary commission, and

system improvement.

60

The purpose of these tests is to determine the robustness of the model to normal
and extreme situations and to ensure that no relevant paths have been omitted from the
model nor have extraneous paths been added. Internal and external validity involves
adding/deleting exogenous and endogenous variables to the model to determine their
effects on the system (J anszen 2000). Sensitivity analyses (what-if scenarios) were also
conducted on the model. This involves manipulating the rate and exogenous variables
from normal to extreme values. This testing process did not change the constructs or the
hypotheses that are to be tested in this dissertation.

The system dynamics software Vensim (2001) was utilized for running the
simulation. Sterman (2000) advocates, "Without simulation, even the best conceptual
models can only be tested and improved by relying on the learning of feedback through
the real world... This feedback is very slow and rendered ineffective by dynamic
complexity, time delays, inadequate and ambiguous feedback, poor reasoning skills,
defensive actions, and the costs of experimentation. In these circumstances simulation
becomes the only reliable way to test hypotheses and evaluate the likely effects of
policies" (pg. 37). However, he stresses that system dynamics modeling is not a panacea;
theoretical studies must be integrated with ﬁeldwork. Thus, the second phase of this
research included a longitudinal study with structural equation modeling analysis of the

data.

61

 

Table 3.2 Stages of Model Construction (Levine, et al, 1992)

 

Conceptualization

— Deﬁnition of questions to be addressed

- Description of problem behavior or reference mode

— Speciﬁcation of time horizon

— Speciﬁcation of the dynamic hypothesis in causal diagram form

Formulation

- Postulation of detailed structure-selecting state variables, rates, etc.

— Developing a ﬂow diagram and associated computer equations

— Selecting reasonable parameter values from a knowledge of the system

Testing

— Testing the dynamic hypothesis: Do the basic proposed mechanisms generate and
qualitatively reproduce the problem behavior?

- Retesting of model assumptions: Does the model include import variables, and
does it appear realistic enough for use?

Implementation

- Testing the sensitivity to perturbations: sensitivity to changes in variables and in
loop structure

— Testing responses to different policies

— Translation of study insights to an accessible form

 

62

3.3 Study 2 - Longitudinal Study & Structural Equation Model
A longitudinal study provided quantitative data for input into the structural
equation model (SEM), which is shown in Figure 3.2. The data collection techniques and

the model are described in this section.

3.3.1 Longitudinal Study Data Collection

Data collection was conducted at three different points in time for the longitudinal
study. The ﬁrst data collection took place in July of 2001 followed with a second in
December 2001, and the third in May 2002. The sample population was technological
ﬁrms that had introduced new products to the marketplace within the last twelve months.
Firms that have not introduced new products recently (i.e., start-ups) were excluded
because they were not able to provide information on their new product program success
at the time of the ﬁrst data collection. Permission was obtained from the parent company
of the two ﬁrms involved in the case studies. This company owns approximately 70
engineering driven manufacturing subsidiaries. These companies are involved in
industries ranging from heat-treating equipment to structural foam products for ﬁrneral
homes.

In anticipation of panel mortality in survey participants, three respondents from
each of the companies were requested to complete the survey at each time frame; the
President/CEO, Engineering Manager, and the Marketing Manger. Thus, even with
personnel changes, the same survey was administered at the second and third data
collection. This type of data collection where multiple groups (ﬁrms with multiple
respondents) are measured on multiple variables in multiple waves is called doubly

multivariate designs (Bijleveld, et. al 1998). In this dissertation the groups were not

63

evaluated separately (between groups designs, which are popular in the social sciences)
but instead, the focus was on the relationships between constructs over time.

The survey was pre-tested with at three ﬁrms prior to being administered in the
ﬁeld. Measures used and their sources are shown in the Appendix C. These measures
are all on a seven-point Likert scale unless otherwise indicated. Five of the eight
constructs utilized scales that had been previously reported upon in management and
marketing literature. Measures for ‘adaptive capacity’ were taken from the extant
literature on ﬂexibility, absorptive capacity and marketing orientation. ‘Knowledge
exploitation’ was based on the theoretical models of Cohen and Levinthal (1990) and
March (1991). Measures for “Successful adoption of innovations” are based on

Tomatzky & Klein (1982).

3.3.2 Structural Equation Modeling Analysis

When change is of interest in a study, longitudinal data collection is required.
There are numerous ways in which to collect longitudinal data and evaluate it. The data
collected in this dissertation was analyzed using EQS following the guidelines for
structural equation modeling for linear growth models as suggested by Bijleveld, et al.
(1998) and elaborated on by Duncan, et a1. (1999). Linear grth curves postulate a
theoretical structure for the variances and covariances of the factors as well as the means.
In growth curve modeling, if a construct is plotted over time, "we see a curve that starts at
an intercept and develops over time with either constant slope (linear growth) or with
slopes that vary over time as well (curvilinear growth)" (pg. 248). Figure 3.3 graphically

demonstrates this type of model. The latent vector Zl represents the intercept and the

latent variable Z2 represents the slope of the growth trajectory. The paths from V999 to Zl

64

and Z2 are equal to the means of the latent constructs. D1 and D2 represent the variances

of the intercept and slope, respectively. Yl-Y3 represent the measure at each data
collection; in this case there are three time periods. This type of model is explained in
detail in Chapter 5.

The two step modeling approach suggested by Anderson and Gerbing (1988)
where conﬁrmatory factor analysis precedes the full structural model was utilized.
Convergent validity, discriminant validity, and reliability were tested prior to running the

full model.

Figure 3.3 Latent Growth Curve Model (Bijleveld, et al. 1998)
V 999

 

 

 

 

 

notes: Y1 = manifest variable at time=1, Y2 =manifest variable at
time=2, Y3 = manifest variable at time=3, Z1 = latent mean intercept;
Z2 = latent mean slope, D1 = intercept variance, D2 = slope variance,
V999 = EQS code for mean, E1-E3 error of manifest variables

65

3.4 Chapter Summary

This chapter has brieﬂy summarized the two different methodologies that were
used to test the model as presented in Chapter 2. The ﬁrst study used system dynamics
modeling to evaluate the NP development process as a system. The second study used a
longitudinal panel study design to evaluate how the constructs change over time. In both
of these studies time is an important factor, and as such, how the NP process evolves over
time was an area of interest in this dissertation.

The remainder of this dissertation will cover details of conducting the two studies
and the results of the studies. Chapter 4 details the system dynamics model and Chapter

5 details the linear grth curve models. Chapter 6 provides a summary of ﬁndings.

66

CHAPTER 4

RESEARCH METHOD - STUDY 1: SYSTEM DYNAMICS MODEL

4.1 Introduction

In study 1, system dynamics (SD) modeling was used to evaluate the effects of
the adoption of innovations on the new product development process. SD modeling is
based on the work of Forrester (1968) and was more recently formalized by Coyle
(1996), Maani and Cavana (2000), and Sterman (2000). It combines information
feedback theory and behavior decision theory in order to map an organization’s operating
policies, information ﬂow, and decision-making processes (Morecroﬁ, 1985; Sterman,
1987). By simulating real world behavior that may be difﬁcult to capture in static
models, the system dynamics approach focuses on how processes evolve over time and
how policies might be changed to improve the ﬁrm’s performance. Organizational
structures and behaviors are evaluated through analysis of a combination of positive and
negative feedback structures. Because it can approximate continuous time processes
rather than discrete time periods, SD modeling can be used to explore the effects of time
lags at work in decision-making as well as the resultant effects of the decision.

Moreover, in system dynamics, decisions and policies are conceptually
differentiated. It would be fair to say that in the initial model building process, system
dynamicists emphasize capturing the policies (decision rules) used in the situation being
modeled (Forrester 1968; Forrester, 1994). Those policies can be conscious or
unconscious, rational or irrational. Policies in the model are reﬂected by equations and

table functions, both of which may represent nonlinear relationships. Decisions for

67

actions and adjustments to changing conditions are based on those policies. Decisions,
when made, are concrete applications of those policies or decision rules (Levine, 2002).
SD modeling has been used in several studies evaluating NPD programs
processes (Black and Repenning, 2001; Milling, 1996; Repenning, 2001) and in
organizational decision-making (Carroll, et a1, 1994; Levine, et. al, 2001; Morecroft and
Sterman, 1994; Sastry, 1997). For this study, Vensim, an off-the-shelf programming
package by Ventana Systems (2001), was used to build the model of approximate
differential equations that represent the dynamic changes occurring in a new product
development program over time. The unit of analysis is the multi-project NPD program
of technologically driven ﬁrms. The time unit of analysis is one quarter, which is a
standard ﬁnancial gauge as quarterly results are commonly reported, especially for public
ﬁrms. A 20-year analysis, or 80 quarters, was selected to simulate the average length of

time top managers may require to observe the long-term effects on the NPD program.

4.2 Case Studies

Because decisions and policies are represented explicitly, formulations should
reﬂect the existing understanding of behavioral decision-making involved in the
processes being modeled. To ensure that variables are meaningful and relationships
observable, real world observations should inform the modeling as much as possible
(Sastry, 1997; Sterman, 2000). Accordingly, the model in this study is based on case
studies of two technologically driven ﬁrms, and is further supplemented with information
from 55 technology-based ﬁrms from nine different countries. Both case study ﬁrms are
technologically driven manufacturing companies, one in the semiconductor industry and

the other in the induction welding industry. The semiconductor ﬁrm had experienced its

68

best performance in about 10 years during the 2000-2001 ﬁscal year. The induction
welding industry was struggling but had managed to make a proﬁt in the 2000-2001
ﬁscal year after several previous years of being 'in the red'.

Extensive discussions over several months with the presidents and engineering
managers of the two companies provided a foundation for modeling these ﬁrms’ new
product development programs and the decision-making criteria endogenous to the
programs. Additional conversations took place with the top manager of several of the
ﬁrms participating in the longitudinal study. Occasionally, a CEO/President would call
me to "just let me know" about their current environment and "their speciﬁc NP process".
These conversations were also used for insights into building the SD model. Important
modiﬁcations and reﬁnements to the model were undertaken throughout the course of
these discussions. Appendix A shows the interview protocol for the case studies and a
sample worksheet, which was used to help managers articulate the nonlinearities between
construct relationships as hypothesized in Chapter 2.

As the SD model is based on a speciﬁc type of organization, a brief introduction
to the two ﬁrms is warranted. The ﬁrms modeled in this study are engineering driven
technologically oriented organizations. They have less than 100 employees of which
70% are manufacturing oriented and the remaining are engineering staff or top
management. Little if any middle management exists, due to the small company size.
The ﬁrms are between 20-30 years old and have annual revenues under $25 million. The
NP program is not formalized with written documents but follows an ‘unwritten’ standard
procedure. A CTON P Engineering or the president of the company heads the NP

process. Competitors are typically well known. These ﬁrms would be considered more

69

organic than mechanistic. Keeping this proﬁle in mind will be important when the
boundaries of the simulation are introduced in section 4.6. A summary description of
these two companies is in Appendix A, Table A. 1.

Based on information gathered during the case studies, two major reﬁnements
were made to the model as originally proposed in Chapter 1, 1.1. These modiﬁcations
include: (1) adding ‘technology knowledge acquisition’ as a mediator between successful
adoption of innovations and technology exploitation activities, and (2) modeling adaptive
capacity as an ampliﬁer/moderator. The ﬁrst reﬁnement concerns the role of knowledge
in the NPD process. Madhavan and Grover (1998) argue that “from the idea-generation
phase to the launch phase, the creation of new knowledge can be viewed as the central
theme of the NPD process” (pg. 2). Nonaka (1994) supports this by suggesting that
“innovation is a key form of organizational knowledge creation” (pg. 14). A primary
reason for adopting innovations is to acquire knowledge currently not available in-house
(Dewar and Dutton, 1986). The NPD manager’s task is to manage the transition of
knowledge embedded in adopted innovations to knowledge embodied as new products.
Thus, technology knowledge acquisition itself plays a critical role in the NP process and
is added as a mediator in the model, see Figure 4.1.

Discussions with representatives of the case study ﬁrms explicated the notion
that exploitation is path-dependent upon exploration activities; a ﬁrm will not have the
knowledge for exploitation (development) projects unless it has ﬁrst conducted
exploration (research) projects to develop the appropriate knowledge base. Knowledge
gained from research activities feeds the knowledge base required in the development of

new products. This knowledge base can be accessed by all participants in the new

70

product process. In this revised model, gains in technology knowledge lead to increases
in the ﬁrm’s ability to build new products as well as in its marketing and R&D
proﬁciencies.

The concept of adaptive capacity was discussed with the managers of the case
study. These managers did not believe that a ﬁrm’s adaptive capacity (ﬁrm’s
willingness, ﬂexibility and ability to adapt to its environment) acted as an antecedent to
the emergence of catalytic innovations during exploration activities. However, adaptive
capacity d_ic_l act as a moderator to the use of different knowledge stocks during the
exploitation process. Not all knowledge is treated equally. The current environment can
substantially inﬂuence the utilization of knowledge of different ages. Sometimes older,
familiar practices and procedures are favored over newer knowledge.

An engineering manager from one of the case study ﬁrms explains; “ If you can
get away with using old [technology] knowledge, then do it. It is proven and reliable not
only to the design engineers, but also to manufacturing and sales. If competition is

putting on pressure, then you probably need to use new technology. It's purely an

economic decision.” Adaptive capacity may act as an ampliﬁer1 between knowledge
acquisition and its utilization in the form of technology exploitation, marketing
proﬁciencies, and R&D proﬁciencies. Thus, in Study 1, adaptive capacity is removed as

an antecedent to adopted innovations and instead becomes an ampliﬁer (see Figure 4.1).

 

' ampliﬁers in SD modeling are equivalent to moderators in the marketing literature.

71

 

.vm BE .mk .NK 8 xSEmhoEEsmbEEe no
we 58396 MSESCV $8.535 wk 3:: 3353 .3 ~83: nﬁemnuasew was:

      
                 

coumuwowﬁ

Qumm
-wcuoﬁmz

momoaomomem

    

£83305:

    
  

coummmswg‘ on» Sm
momoeomomocm owuogoam All town—mopm
wcuoxhwz kwoﬁogoop. 3 mo mmooosm

 

cosﬁmoﬁxm mooSOmom

82m

   
  

awe—0:531 o>tn§o<

 

 

cause 25 SE 382 am 3 8:3

72

4.3 Overview of System Dynamics Model

Figure 4.2 shows the basic premises of the SD model. The +/- notations on the
arrows indicate the direction of the relationship between constructs in the model. This
ﬁgure also illustrates the four primary feedback loops of interest in this study.
Determining the polarity of a loop is important in order to trace the effect of a small
change in a variable as it propagates around the loop. Starting from any variable, if the
feedback effect reinforces the original change, it is a positive loop; if it opposes the

original change, it is a negative loop. The polarity of the loop is the sign of the loop gain

from the starting position, 6x2) , until it returns to itself, 6x: (Sterman, 2000):
Polarity 0/1001) = SGN(ax,0 /a x1!) = SGN[(ax,0 /a x; )(axff /a x;_, )...(ax,0 /a x,’ )1 (1.0)

It is common to name a loop to facilitate a general understanding of its role in the
larger model. The ﬁrst feedback loop, R1 “success trap”, is reinforcing or positive. In a
positive feedback process, a variable continually feeds upon itself to reinforce its own
growth or its collapse. When this occurs, it is known as a “vicious cycle”. In the loop
R1, success in exploitation (development) results in a positive performance level, thus
causing managers to focus less on exploration and more on the tried and true routes to
maintaining proﬁt levels — i.e., more exploitation activities. However, in this loop there
is no opportunity to build new knowledge for developing innovative state-of-the-art
products. Eventually, products become obsolete and the system can collapse upon itself.
This models the phenomena described in Chapter 2 as a ‘lock-in’ to existing knowledge
so that a ‘lock-out’ of old knowledge occurs.

The second reinforcing loop, R2 “prosperity”, is singled out in Figure 4.3. It is

positive controlling the amount of slack resources that become available to exploration

73

Figure 4.2: General Feedback Model

R&D-Marketing

           

Integration
+
+ Adaptive
Adoption of ........ Capacity
Catatlytic ’
Innovations Technology
Slack 1:2 + _ Knowledge
Resources B 1 E
+ H ' H i
PIOSPG‘W "revitalization" *7
Marketing &
Performance R&D
Gap Proﬁcrencies
RI -
B2
"success trap" +
' "street smarts"
NP Success + ““1"?“ng
$__———-Exp101tatron

+

Notes: Arrows represent causal relationships with direction noted as a (+) or (-).
Loops labeled R indicate a reinforcing (+) feedback. Loops indicated with B,
indicate balancing (-) feedback.

Figure 4.3: “Prosperity” Feedback Loop

+
Ad0pti0n of R&D Marketing
+ Catatlytic Integration
Innovations
Slack
Resources Technology
+ Knowledge
"prosperity" @/
NP Success
+ \JTechnolog'
\dExploitation

74

Figure 4.4: “Revitalization” Feedback Loop

R&D-Marketin g

/ Integration

Adoption of +

Catatlytic Technology
Innovations Knowledge

l @

Performance
Ga

"revi talization" +

Technologl
Exploitation

NP Success

+

Figure 4.5 “Street Smarts” Feedback Loop

V R&D-Marketing
Adoption of Integratmn
Catatlytic +
+ o
Innovat1ons Technology
Knowledge
_ \.
Performance B2 Marketing &
Gap R&D
- "Street smarts" Proﬁciencies
NP Success +
V+\_ Technology
Exploitation

75

projects. In this loop, slack resources provide the means to adopt innovations that result
in new technology knowledge. The new knowledge allows more new products to be
invented (technology exploitation) feeding NP success that ultimately feedbacks to
increase the stock of slack resources, thus slack resources feeds its own growth. In times
of abundant slack resources (prosperity) the ﬁrm is able to undertake more exploration
projects, even speculative ones, without jeopardizing overall program performance.

The ﬁrst balancing (or negative) loop BI is labeled “revitalization”. It is singled
out in Figure 4.4. Negative feedback loops are characterized by goal-directed behavior.
“Such terms as self-goveming, self-regulating, self-equilibrating, homeostatic, or
adaptive, all implying the presence of a goal, deﬁne negative feedback systems”
(Goodman, 1974, pg. 37). In the system modeled here, the NPD program’s goal is to
diminish any performance gap resulting from underachieving a target. The more target
performance exceeds actual performance, or the greater the gap, the greater the focus is
put on exploration activities through the adoption of catalytic innovations. A
concentrated focus on exploration projects “revitalizes” the product portfolio with newer
and better products, thus growing proﬁts. Greater proﬁts decrease the performance gap
experienced in the previous time period, meeting the goal. This negative loop has the
NPD program revitalizing the product portfolio when NP performance suffers.

The adoption of catalytic innovations also has the effect of increasing marketing
and R&D proﬁciencies through the new knowledge acquired (feedback loop B2, "street
smarts", which is singled out in Figure 4.5). Sufﬁcient levels of marketing and R&D
proﬁciencies are necessary to understand how to proﬁtably invent and market the type of

products that will be successful in the marketplace. These "street smarts" are an

76

important part of any NPD program. The greater the proﬁciencies, the greater the ability
exists to develop new products (technology exploitation) and the greater the NP success.
This success closes any performance gap that had original driven the emphasis on
innovation. Thus, a decrease in the focus on the adoption of innovations, or a negative
feedback, results.

Figure 4.6 shows a more detailed view of the model in the form of a stock and
ﬂow diagram consistent with system dynamics models. In such diagrams, stocks (or
states) are denoted by rectangles and represent properties or assets of the organization
that have accumulated over time. Flows are symbolized by arrows with “valves” and
represent action or activity within a system. Inﬂows to and outﬂows from the stock cause
it to increase or decrease accordingly. That is the only way a stock can ever change.
Moreover, ﬂows are where the actions occur, but those actions are the results of the
policies that have been captured by the model to represent how the actors will react to
changes in the situation (Levine, 2002).

Flows, unlike stocks, may change instantaneously. For example, Innovative
Knowledge Stock represents the technology knowledge of the NPD program accumulated
(and lost) over the course of the simulated history of the organization. A ﬁrm builds a
stock of technology knowledge through learning during research activities. Knowledge
stocks can experience an “outﬂow” through a forgetting process (Ash and Smith-Daniels,
1999). Accordingly, a stock is a “snap shot” at a single point in time of its state resulting
from inﬂows and outﬂows.

The NPD process can be examined by starting with the stock, Slack Resources.

Slack Resources, Resources to Exploration, and Resources to Exploitation are allocated

77

to Exploration Projects and Exploitation Projects. Successful exploration projects lead
to ‘successful adoption of catalytic innovations’ and to increases in Innovative
Technology Knowledge Stock, which provide the knowledge for development projects.
This stock ages over time into Mid-Knowledge and Old Knowledge. The use of
innovative or aging technology knowledge is moderated by the ﬁrm’s Adaptive Capacity.
Knowledge is used to build Marketing Proﬁciencies and R&D Proﬁciencies. Successful
exploitation projects lead to gains in Products Launched. The product portfolio ages over
time resulting in varying proﬁt levels for the aging products. Products sales result in
gains to the Quarterly Proﬁt stock. Proﬁts lead to increases in Slack Resources, which
may be used toward research and development projects, thus closing the loop in the NPD
activities.

A third type of variable is used in SD models. ‘Auxiliary’ variables are neither
stock nor ﬂows but are intermediate concepts added to the model to add clarity. They are
often broken out ﬂow equations in order to allow a better understanding of the factors
affecting the model. Three important auxiliary variables in this model should be noted:

(1) ‘fraction of resources allocated to exploration versus exploitation activities’, fR/D, (2)
‘ performance gap’, G, and (3) ‘catalytic innovations’, PC. The ﬁrst two set the next

quarter decision-making criteria, and the third is an outcome of undertaking exploration
projects. Equations for the ﬁrst two auxiliary variables are given in Appendix B. 1 .7.1
and the equation for catalytic innovations is in Appendix B.l.2. 1. The remaining
auxiliary used in the model are listed in Table 3.3 and are brieﬂy explained in Appendix
B.2 which lists the Vensim code used in creating this model. Formulas for all the

variables, stock, ﬂow, and auxiliary as well as the constants, are in Appendix 32.

78

Figure 4.6 Detailed SD Model ‘

frut '.
Ill" ‘I‘l‘rL\'(‘ 4

 
   
   
   
 
 
    
     

don’t use-lose

I
t/ixrrr bu nun (',l,)ltll't’
cxplor

 

slack to
c x lure Resources

    
   
   

or

Exploration -- - of

e xplore resources
resources

c xploration

 
   
 
    
  
 
  
 

quarter

frac slack \

resources to <frac resource 9%
explora‘ion allocation to I'L‘.\(Ill I'('(€.\

cxplorallon>

   
 
  
 

SI ac k Resources
slac k
resources
in

resources
exploitation 1

<quurlcr>

 

quarterly
slack

 

u
slut: k to
exploit

   

for
Exploitation allocation of
perfurm~ slack “‘Wm‘
r'nnvurxiun

(It'luv ('rp/urldon -[ “$64058
dixlrlbutum

<j’rur'.
(Ivorian c >

      
 
 
     
   
  
  
  
  
 

GAP exploit resort red
TA RG ET- ACTUAL) pro 'e

('75

  

delay/or target
frac resource ’
allocation to P f
exploration er orrnance

‘ Goal: target ---——A-;»— we
1011! change in perf
conversion goal

   

. perception

Adaptive ' Perceived time

Capacity -’—‘— Perfomrancc:
Chan

89 I "‘7‘; T;
adaptive capacity

    

         
 
  

perceived
performance change

  

actual

     
   
 

 
   

drain

I adaptiVe

 
 

IIVL'I'It’r

perception
pvlfu rmum'c /pruf1t ap

   
  
   
   
     
  
 
   

fruc outdated
, . .

r

           
  

frat." decrease
in AC quarterly (by)?
c.\ ) are .rro "(c
/ 7 I I j r

Exploitation
Projects

<L|Lmr1cr>

outdated
exploitaiton
ects

(Technology ”expenses

rof'
Exploration> D

It
<qunrrcr> outﬂow
profits ’) a] 0
increase . l t]

3 Quarter Profits exP ora on>

<rmal approved

  
   
 
 

 

prujects>
profit I 0+ prnfiio- [0 prnj'il I _5 {ﬁr-(£35,113
/prmlu¢-, \ /pr()dlu:r proﬂt6-10 /,,,-,)(,u(., proﬂtnew P

proﬁt 10+

    
 
    

\ﬂﬂtl-S

 

 

__roduct

inﬂow4 6-10 yr

   

 

  
 
   

 

 

Prod 10+ .
T ug trig aging
‘ prndé- [0 ob solete3 prod! -5
aging
PH", 1 0+ frucrlearll3 E f rucdcurliz fracdea tlrl 1

Legend: italics-constants; bold-auxiliary variables; stocks in boxes, flows represented byv

79

      

<quru‘tcr>\ [rat failed
at lorutinn

funded
exploration

 
 
  
    
 

 
 
  
   
    

   
  

 

  
 

 

 

 
   
   
  

“P“‘I‘m a at Technology
resourtes I Exploration failed
Xplomtmn projects
Exploration
- successful

    

 
 
  

     
  
 
 

 

     

  
 
  

    

 

  
   
  

      

   
   
  
  
   
  
  
 
 

 

 

 

"mm a ‘/ Requiring explore projeCIs
Fu din outdated 1,, . . . t
g exploration ‘ '- uy “"16 <successfu1
3 l I projects exploitation
rC‘PIOIIJIIUtI new exploration f rue fruc. decrease pfOJCClS>
a pro Cs! would“ 1 inregran‘or
<qu artcr> . tplora tum Marketing
‘ .3 ‘ ——:— RD .—
IIallOn 3M outﬂow ~ '
mg <frac resource catalytic integration Integration integration
allocation to Innovations
exploration> rnktg-r&d
It ‘I I k"” wle dge constraint Converter
expﬂfr); :23:ch catalytic (”0.19075 integration/projects
J A <successful
exploitation
<successfu1 ‘ projects>
‘tplore projects>
It}! >
knowledge .
"ﬂ” “1’?" 'n . . exploit projects
lull“, :1 explortauon aging old
rropcts> knowlwedgc i outdated
’1 knowledge failed aging mid to > ‘ knowled
ro 'ectx aging new to old I 8°
‘C i" ,... I K
:031 .,L LELLL Innov L . L;L. Kn M121: vL L . Old
/perce;mun ‘ knowledge Kn°W1°d8¢ aging innov ow ge aging mid Kn°W1°d8¢
time acquired: knowledge
failed proj
:cived “’iﬁe‘o‘mg .nforgetting mid
nce chanlt¢ I ‘ I ' edge know‘P—dse
I forgetting V y
forgetting . factor ‘ £3
‘acror new "”d
.r 1c ”(Wand kM‘V’edge liftzxrifge discount factor 4
‘ . on _
exp/0‘10”” funded exploitation \ AC Old Know
projects

discount factor AC
free. full knowledge/ ld-Know

rt) '21:! constant
p J discount factor AC ‘______<Adapt1ve

     
  
 
  

n-I Know Capacrty>
failed exploitation
o'ects
pr J lossfncror <new AC ampliﬁer
R&D eﬂects

exploitation projects

 

  
     
 
   
   

  

R&D
constraint proficiencies/
\P' 0110“ knowledge
C(PHE’L'I‘ p-I'OdllCIS/ C'Clnve'Tcr

(”ects <AC ampliﬁer
effects>

Marketi‘

Profic ie no ies ou tﬂow
mkt g

<prafic' Icncie \/ Frogs: nc y
knowledge loss fac tar ’
mktg

C ()nvertcr>

  

mktg
corstralnt

 

 

 

4.4 Central Assumptions

Four core assumptions are embodied in the model’s structure. One central
assumption of the model is that because the NPD program only has the ability to
undertake a set number, n, of projects per quarter, it must decide whether to concentrate
the NPD program efforts on exploration versus exploitation activities. This decision, the
fraction of resources allocated to exploration (research) vs. exploitation (development),

f R , D , is a function of past performance (Lant, 1992):

f R / D = f (target performance — actual performance) (2.0)

A positive performance gap indicates that the NPD program is not meeting
expectations. The more positive the performance gap, the greater the emphasis on
exploration activities (see Figure 4.2, loop Bl). A negative performance gap indicates
that expectations are being exceeded, and thus the NPD program will concentrate on
exploiting the existing knowledge base (Christensen, 1997; Miller, 1990). New
exploration projects will be undertaken in order to try to close the gap. When NPD target
performance equals actual performance, NPD managers will allocate equal proportions of
resources to new exploration projects and new exploitation projects (Levinthal and
March, 1981).

A related key assumption is that exploration projects are supplemented through
slack resources. Slack represents resources, time, and energy that can be devoted to
pursuing organizational goals which otherwise might have been channeled into other
things. Levinthal and March (1981) further exempliﬁes, “slack is the difference between
the potential performance of an organization and the performance actually received. . .It

includes such manifest inefficiencies as over-designed equipment, over-qualiﬁed

8O

personnel, undiscovered improvements in current technology, and relaxed managerial
control procedures” (p. 309). Organizational slack accumulates when performance
exceeds the target.

The third key assumption is that resources available for research and
development projects consist of ‘budgeted’ resources and ‘slack’ resources. Budgeted
resources are assumed constant over timez. These are resources minimally required to
maintain a NPD program. They are often budgeted by top management as an annual
expense of running a NPD program. Even if slack resources are not available, budgeted
resources allocated to the program will keep it functioning.

The fourth assumption concerns knowledge acquired through exploration and
exploitation activities. It is assumed that exploration activities lead directly to increases
in technology knowledge stock and indirectly to new products. As “exploration includes
things captured by terms such as search, variation, risk-taking, experimentation, play,
flexibility, discovery, innovation” (March 1991, pg. 71) exploration projects are riskier
with more volatile outcomes. The payoffs from exploration must compensate for the
uncertainty in the outcomes of these types of projects. The necessary result is
signiﬁcantly higher quantity of knowledge acquired from exploration projects compared

to exploitation projects. The model also assumes that a small percentage (10% as

modeled in this study3) of successful exploration projects result in catalytic innovations.

 

2 This assumption was based on input from several NP managers in addition to the managers involved in
the case studies. A pool of budgeted resources are made available quarter. Additional resources, or ‘slack’
resources, which are a function of revenues, are added to the pool as they become available.

3 As the concept of ‘catalytic’ innovations is new to the NP literature, this number is based on reports that
10% of NPD projects in ﬁrm portfolios are new-to-the-world (Grifﬁn, 1997). This number has been
consistent since originally being reported by the Booz Allen Hamilton study in 1982.

81

These catalytic innovations have two effects: (1) substantially increasing the knowledge

base and (2) acting as a catalyst for new exploration projects to be undertaken.

4.5 Research Questions

In Study 1, the following questions were addressed. (1) How does the successful
adoption of catalytic innovations affect NP program success? (2) How do changes in
slack resources affect the successful adoption of catalytic innovations? (3) How do the
allocation decisions between exploration and exploitation activities affect catalytic
innovations? (4) What role does adaptive capacity play in the utilization of different types
of knowledge in the innovation-centric organization? (5) How do changes in marketing-
RD integration affect the success of the NP program?

A fundamental objective of SD modeling is to gain a greater understanding of
how a ﬁrm’s policies can affect its performance in various situations, particularly with
respect to time. As product lines age and technology becomes obsolete, NP managers
must make decisions on how to adjust the NP process. The role of catalytic innovations
in creating policies to maximize NP program success is the major interest of Study 1.

The case studies conﬁrm that adopted innovations are typically associated with a
particular research project. The successful adoption of catalytic innovations is thus an
outcome of a successful exploration project. However, exploration projects require
funding. Cyert and March (1963) contend that slack resources are a necessary (but not
sufﬁcient) requirement for exploration activities. The existence of slack resources means
that the organization can afford to absorb failures and to explore new ideas in advance of
actual need (Rosner, 1968). “Slack encourages search activities that cannot be justiﬁed

in terms of their expected return for the organization. They are initiated because of their

82

attractiveness to some individuals or subunits, and tolerated because of the organization’s
current success in achieving targets” (Levinthal and March, 1981, p. 309). How the
successful adoption of catalytic innovations changes with the increased or decreased ﬂow
of slack resources is a research question addressed in Study 1.

Because resources are limited, only a set number of projects can be undertaken in
the typical NP program. Available resources must be allocated to exploration (research)
and exploitation (development) activities. As modeled in Study 1, when targets are being
met, resources are primarily allocated to exploitation activities. When targets are not
being achieved, a greater focus is put on ﬁnding better, more innovative products through
exploration activities. Firms must locate the proper balance point between new
exploration projects and new exploitation projects. How these allocation decisions affect
catalytic innovations was also explored in this study.

The acquisition of new knowledge itself does not guarantee the success of
exploitation projects. As the case studies elucidated, not all technological knowledge is
treated equal by the innovation-centric organization. At different phases of the NPD
process, in different technological environments, and in different industries, the recency
of knowledge acquisition can play a greater or lesser role. For example, highly intensive
knowledge industries (i.e., biotechnical, pharmaceutical) may value newer knowledge of
much greater importance to new product generation compared to older knowledge.
Similarly, low intensity knowledge industries (i.e., food, packaged consumer products)
can value older, well-known knowledge more than un-proven newer knowledge in new

product development. Thus, knowledge is modeled as an aging chain (c.f. Sterman,

83

2000, chapter 13) whereby the age of the knowledge can have an impact on the projects
undertaken by the ﬁrm.

Adaptive capacity is modeled as an ampliﬁer to the aging knowledge stocks, to
the marketing proﬁciencies, and R&D proﬁciencies. When adaptive capacity (AC) is
positive, technologically oriented ﬁrms will put more value on new innovative
technology. When AC is negative, a ﬁrm values older technology more. In a neutral

state (AC = 0), innovative technology stock has the same value to the organization as

aged technology knowledge stock. This is modeled as d AC , a ‘discount multiplier’, or
moderating factor. When AC is positive, old knowledge is discounted by d AC and when

AC is negative, newer knowledge is discounted by I - d AC .

A ﬁnal policy that examined in Study 1 is the effect that marketing-RD
Integration has on the successful adoption of catalytic innovations. Chapter 2 suggests
there is a positive relationship between these two variables. However, a question of

interest is: how important is this relationship to the dynamic NP process?

4.6 Research Design

Sterman (2000) stresses that the following questions must be asked of any model
concerning robustness and sensitivity: (1) Are the policy recommendations sensitive to
plausible variations in assumptions about parameters, aggregation (consistency between
constructs), and model boundary? (2) Is the model robust in the face of extreme
variations in input conditions or policies? and (3) Does the model exhibit surprise

behaviors or anomalies? These questions are addressed through the research design.

84

In determining the research design, it is important to ﬁrst deﬁne the model
boundaries. As previously discussed, this study focuses on the effects of the adoption of
innovations on the NPD process of technologically oriented organizations. The primary
relationships modeled in this study examine the antecedents to catalytic innovations and
their impact on NPD performance. Factors exogenous to the NPD program not modeled
here include: competitive inﬂuences, environmental turbulence (both technological and
marketing), rate of technology change, organizational structure, and ﬁrm ownership
(public/private), to name a few. Several factors endogenous to the NPD process, but
beyond the scope of this dissertation, are also outside of the model boundaries; e. g.,
availability of human capital, product portfolio mix, and cost of resources. The impact of
these factors on the adoption of innovations is left for other studies.

Aggregation, or consistency between constructs, was tested by splitting the
system into seven sub-modules: (1) resources, (2) exploration activities, (3) knowledge
gain/lose, (4) exploitation activities, (5) proﬁciencies, (6) proﬁts from products, and (7)
performance gap and adaptive capacity. Each sub-model was tested alone and then
coupled into the entire model. Additionally, these sub-models were also reconciled with
managers of the case ﬁrms to verify the rationality of the decision rules as modeled4.
These modules, along with the Vensim code used to model the differential equations, are
shown in the Appendix B.

In order to test the robustness of the model, 32 simulations were run to determine

the effects of changing conditions on the model. The ﬁrst simulation establishes a base

 

" Reconciliation of SD models with vested parties is one method of determining if the models accurately
represent the phenomenon of interest (Levine, et al., 1992).

85

case to which all other runs are compared. Additionally, two test runs were conducted,
the ﬁrst simulating a NP program where only exploration activities occur (fR/D = 1) and
the second run where only exploitation activities occur (fR/D = 0). March (1991) warns of
the detriments of engaging exclusively in exploration or exploitation. "Adaptive systems
that engage in exploration to the exclusion of exploitation are likely to ﬁnd that they
suffer the costs of experimentation without gaining many of its beneﬁts. Conversely,
systems that engage in exploitation to the exclusion of exploration are likely to ﬁnd
themselves trapped in sub optimal stable equilibria" (pg. 71). The base model was run
under conditions of only exploration and then only exploitation activities. Figure 4.7
shows the overall performance results of the base case and the test runs. As can be seen
in the graph, focusing solely on exploration or exploitation very quickly leads to a poor

performing NP program.
Figure 4.7 Base Model Overall Performance & Test Runs

Proﬁts Overall Performance
60

. . 1 5
44.5 or ' - ”TIT I 5T-—

 

29 I ' L ‘ t ‘1‘ - --“~ - —A '— : .LLT—
1

~‘ l
. » l a

13.5 __ ‘_| l !___
. . . l 1

l

 

8121620242832364044485256606468727680
Time(Quarter)

 

Eprit 2 3 3 3 4. 4 l 2 3 3 3 3 1 ¥ peﬂbﬁmnce
Explore 2 2 2 2 2 2 2 2 2 2 2 ..‘L 2 2 peﬂbrnnncc
Base Case3 3 3 3 3 3 3 3 3 3 3 3 3 4. perlbrmance

86

 

w

4.6.1 Simulations and Base Case

Parameters (constants) used in the models are shown in the Appendix, Table A. 1.
Initial starting values for the stock variables are identiﬁed in the Appendix, Table A2.
These values remain the same across all simulations. It must be noted that the parameters
and starting values are based on input from the case studies. The case study ﬁrms were
small engineering-driven, privately owned companies with informal NP programs.

Any unusual or unrecognized behaviors were resolved by working through each
module of the system to determine the origination point of the anomaly. All
discrepancies between the model behavior and the ﬁrms’ NP program as described during
the interviews were resolved. Many informal ‘back and forth’ discussions occurred with
the managers of the case studies to clarify processes and decision-making rules. Initial
conditions and constant values (see Table B. 1, Appendix B) were also veriﬁed with the
ﬁrm’s managers. Extreme variations in initial conditions and decision-making rules (i.e.
under what conditions to change from an exploitation to an exploration focus) were also
tested. The starting values and other constants were varied to determine the sensitivity of
the model to these changes. In any situation where the system was found to be highly
sensitive to a change in a constant or decision rule, the constant (or rule) was set at the
average value as reported by the case studies ﬁrms. For example, these ﬁrms on average
undertake one new development project per engineer quarterly. Thus, the model used six
new projects per quarter as one company had three engineers on staff and the other had
ten engineers on staff. Starting values and constants may be signiﬁcantly different for
different types of organizations, and thus, the SD model would require revisions to reﬂect

this type of ﬁrm. This limitation is further discussed in Chapter 6.

87

The thirty-two simulations were used to test the model for boundary conditions.
They were run under ‘high’ and ‘low’ setting for ﬁve variables found to have the greatest
impact on the successful adoption of catalytic innovations: slack resources, marketing-
R&D integration, adaptive capacity, NP program success, and fraction of resources
allocated to exploration activities. Thus, 32 simulations were conducted, which

constitutes a 5 variable by 2 level research design (25). To set the high and low settings,

constants were added to the functions used to model the phenomena. It is the constants
themselves that are varied, not the mathematical functional relationship.

The stocks: technology knowledge exploitation, marketing proﬁciencies, and
R&D proﬁciencies, were found to have little sensitivity on catalytic innovations.
Accordingly, they were set high for all the simulations. The insensitivity of these
constructs to the SD model will be discussed further in the results section of this chapter.
It is important to note that ‘insensitivity’ does not mean lack of relationship.

High slack resources indicate that 100% of proﬁts are allocated back into the NP
program (SLKR = 1). High marketing-RD integration indicates that the interaction
between the two departments occur on a regular basis (MRDI = 1), and high adaptive
capacity (ACAP =1) indicates that the ﬁrm’s past performance will determine its
willingness to utilize new knowledge over older knowledge; dated knowledge becomes
discounted. When NP program success is set as ‘high’ (NPPERF = 1), new products
entering the marketplace are always proﬁtable; competitive factors do not play a factor in
the proﬁtability of the ﬁrm. When the fraction of resources allocated to exploration
activities is set high (FRAC = 1), it varies as a function of the performance gap as

previously described.

88

When slack resources are set ‘low’, 50% of NP program proﬁts are allocated back
into the NP process (SLKR = 0.5 ). Low marketing-RD integration indicates that there is
little interaction between R&D and marketing departments. It is set at 50% of the high
level (MRDI = 0.5). Low adaptive capacity, or more accurately no adaptive capacity
(ACAP = 0), indicates that the ﬁrm does not consider past performance in its decision on
how to use its existing knowledge base. A low NP program success represents that the
ﬁrm is only able to sell their products in the marketplace at ‘/4 of what they would be able
to in a ‘high’ proﬁtable environment although expenses remain the same (NPPERF =
0.25). When the fraction of resources allocated to exploration is modeled low, this
indicates that 50% of slack resources are always allocated to exploration activities and
50% are allocated to exploitation activities (FRAC = 0.5). It is not a function of the
performance gap as when modeled ‘high’. These high and low settings were chosen
based on discussions with the ﬁrms in the cases, one that was performing extremely well
and the second that was struggling. In these analyses, the dependent variable was taken
as the successful adoption of catalytic innovations, and thus it was not set as either high
or low. Table 4.1 shows the settings for each run.

The base case represents the system in a 'typical' environment as deﬁned by the
case study ﬁrms. The base case was modeled with all high settings. It sets the
perspective from which the remaining simulations are compared. Thus, a brief
explanation is necessary to set this reference frame. In all the simulations a two-year
equilibrium period (8 quarters) is set where resources are equally allocated between
exploration and exploitation activities. After this point in time, the independent and

dependent variables are allowed to vary. As can be seen in Figure 4.7 for the base case,

89

from quarter 8 (Q8) to about Q36, performance is gradually increasing as the ﬁrm
overachieves its goals. The abundance of slack resources is primarily allocated to
exploitation activities as the ﬁrm is performing well with its current product line. After
about Q3 6, the firm’s ability to keep up its current rate of growth begins to erode.
Subsequently, fewer slack resources become available and fewer projects can be
undertaken. What limited slack resources are available are allocated to new exploration
(adoption of innovation) projects. At approximately Q60, the rewards of these
exploration activities are observed as new products enter the marketplace driving up

performance again.

90

Table 4.1 Simulation Legend — Constants Settings (H = high, L = low)

 

 

 

 

 

Run SLKR MRDI ACAP NPPERF FRAC

Base H H H H H

2 H H H H L

3 H H H L H

4 H H L H H

5 H L H H H

6 H L L L L

7 H L L L H

8 H L H L H

9 H L L H H

10 H H L L L

1 1 H H H L L

12 H L H L L

13 H H L H L

14 L H H H H

15 L H H H L

16 L H H H H

17 L H L H H

18 L H L H L

19 L H L L L

20 H H H H H

21 H H H H H

22 H H H H H

23 L L L L L

24 L L L L H

25 L L L H L

26 L L H L L

27 L L L H H

28 L L H H L

29 L L H L H

30 H L L H L

31 H H L L H

32 H L H H L

Legend

SLKR: Slack Resources Constant High = 1 Low = 0.50
MRDI: Marketing-R&D Integration Constant High = 1 Low = 0.50
ACAP: Adaptive Capacity Constant High = 1 Low = 0.00
NPPERF: New Product Performance Constant High = 1 Low = 0.25
FRAC: Fraction of Resources Allocated to High = 1 Low = 0.50

Exploitation Project Constant

91

 

4.6.1.1 Results of Simulation

The purpose of conducting the simulations was to observe how the model reacted
under various conditions. First, surprise behavior and anomalies from the base case
where checked. None were observed in these simulations. Secondly, ANOVAs were run
to determine how the stocks themselves would vary under these different conditions.
Mean values for the 80 time periods were calculated for each of the stocks of primary
interest and subjected to the ANOVA analysis. The ANOVA results comparing the 32
simulations (31 + base case) are shown in Table 4.2.

Evaluating the effects of the various conditions on the successful adoption of
catalytic innovations indicate three distinct groups of results. Group 1 (n = 10) has the
lowest mean for the successful adoption of catalytic innovations. Group 1 has low levels
on 3 or more out of the 5 constants that were allowed to vary. All the members of group
1 have low levels of Marketing-R&D integration. Eighty percent have low performance
levels as a constraint.

In group 2 (n = 16) 75% of the group has high marketing-R&D integration.
Sixty-two and 1/2% have low settings on slack resources and 62‘/2 % have low settings on
adaptive capacity. All the members of group 3 (n = 6) have high settings for both slack
resources and NP program success. This is an expected result as slack resources are
needed to undertake the adoption of innovations and slack resources are generally more
available when NP program success is high. Although the inﬂuence of marketing-RD
integration on the successful ad0ption of innovations may be an artifact of the
programming of the SD model, the relationship is worth more investigation in future

research.

92

N:

a...

S.

i N... 1

 

 

 

 

 

 

13.: as New N:
R: E 3.: NE Sam 85 8.” 3o :2 N:
2.: v3 5.: mom we: :5 86 3o :2 N 2
3: £5 3.: 3:. 3.3 :2- a: N; .2 N 2
$2 $6 3.: N3 :le :2- 8.” Be a: N E
on: :3 mo. : EN :3 Be m: 86 a: N :
8.: a; 8.: E. 2.5. 8.0 m: 86 :2 NS
2.: 86 mm. : 8:. Ex 2 .o 8e 85 a: N m
3.: o? 8.: Re. ”new 33 02 Ed 91 so: 385
8.: a? 2.: 43. Mesa. :6 8e :3 NE _ am
on: m; 8.: :w 2.x cod 8.» Gd :2 _ R
8.: Rd 3.: EN 8.x £5 E .5 mod N: _ em
8.: :3 E . : «3. MBA cod 0% :3 N3 _ X
2.: m; 8.2 E. 8.x cod is 86 NE _ mm
3.2 $5 3.2 :2 Nova :2- S.” 6.0 m: _ mm
am: who 8.: :5. 23 Se m: 86 m3 _ S
3.: as a: 8.4 ”EN Se 93 mod 2: 2w
2.: as 2. : of. an: Se 86 ace 9.: _ a
3.: Ed 5.: EN 2.x 86 as Be a: _ a

mom: mz Bibs: SEQ: um: :03 3022 N20 ZEN 9:53 xoﬁm .5455 Ex

 

 

mowing <>OZ< mo $38M um 2:23:85 88m $33.2; xooem me $802 me 033.

93

x008 _ v 00:00:. 302 I mama—m2 9.00% 86:06:05 wEBu—BE

I hoax—Ex: 3007. 3856.008 Odd -mOmeM 2003 5:36—93 30.05007 xm :Umh x008 0300.305. 038525 I 3022 ”xoﬁm £00000 03530
I m<U Home. 2008 eceﬁwﬁﬁ Day—meeetae I #75:)— !008 80.50000 x00? I ¥U<I~m x0309; bﬂzxeg 0:20.805: £5300é>4<b<u 6=0w0q
G u 5 M 90000 I m x2 I 5 N 90000 I m E: u 5 ~ 9.000 I _ Sodv a :0 .<>OZ< :0 e003 £28382: anew 00:00:05 “0:00.003

 

 

 

 

 

 

 

 

8.: 8.8 2:: 30. 2.8 8.8- 02 £8 a: as: 8520
3.2 8.2 8.: 9% 8.8 :88- 8.8 3.: an: :82 885
8.2 83: 8.: 88 8.2 88 a: 8.: on: N 2
8.2 8.2 8.: 2% 2.8 8.8 8.8 8.: 8: N a
8.2 8.2 8.: com 3.8 8.8- 8.8 8: 3.: N. m
8.: 02: 8.: $8 8.8 8.8 83 8.: of N a.
8.2 VS: 8.: 88 3.8 8.8- a: 8.: on: N N
3.: 8.: :.: 88 2.8 an- 8.8 :3 8.: Noam
8.: 8.8 2.: «S. 8.8 £8- as ”8.8 a: :82 8:80
8.: :2: 8.: x: 8.2 8.8- :8 8.: 8.: N 8
2.: 8.8 :.: 8+ 8.8 8.8 8.0 8.8 ”I N a
8.: 8.2 8.: 88 2.2 88 a: 8.: 8.: N 2
8.: 8.8 :.: 3:. 2.8 :2- N: 8.8 2.: N 8
8.: 8.8 8.: 3:. 83m 88 8.8 :8 8: N mm
3.: a; 8.: E. 8.8 28 is 8.8 8.: N a
.2.._....m8.N.N.N..:.. 8.8 2.: 8... 8.8 88 8.0 8.8 :2 N om
2.: 3.8 8.: :.:. 8.8 88 z s 8.8 a: N 2
Nam: 82 L085: 389. um: mom: 3020: N20 29.. 9752 06.3w 505.5 9;:

 

 

m0mbee< <>OZ< m0 $150M 0% 3052086 805 33093 xooem $0 00002 3.58 N6 2an

94

4.7 Results

This section reports on the ﬁve research questions posed in section 4.5. These
questions were evaluated in two ways: (1) by evaluating the change trajectories of the
constructs with respect to time and (2) by conducting sensitivity analyses to determine
the impact on the dependent variable of interest by the change in an explanatory variable.
Sensitivity testing is the process of changing the assumptions about the value of a
constant set in the model and examining the resulting output for the impacts to this
change. It is similar to the ANOVA analysis previously described except here the focus
is on one explanatory variable. Hundreds of simulations with the constant varying over a
range of values can automatically be performed using Monte Carlo simulations in
Vensim. Conﬁdence intervals show the spread of values that the stock takes from
varying the constant. These sensitivity analyses are shown in Figures 4.9-10, 4.12, 4.14,
4.16-18, and 4.20. The different shadings in the graphs indicate 50-100% conﬁdence
intervals. The lightest center shading indicates the 50%CI and the outside bands
100%CI. The centerline running through the 50%CI represents the mean value at each
time point of the Monte Carlo simulations. Each sensitivity analysis was run using 200

Monte Carlo simulations.

4.7.1 Research Questions Evaluated

To address the research questions, the base case was evaluated for its evolution
with respect to time and the variables of interest. The following discussion uses the base
case in each evaluation. The ﬁrst research question looked to address what effect (if any)
do the adoption of catalytic innovations have on NP program success. Figure 4.8

presents a viewpoint of how the auxiliary variable, adoption of catalytic innovations, and

95

the stock variable, NP program success, change over time.5 Curve 1 shows the change
trajectory of catalytic innovations, and curve 2 shows the trajectory of the NP program
performance with respect to time. By plotting both of these curves together, it can be
seen how both variables change together. Figure 4.8 demonstrates that adopting catalytic
innovations in moderation may increase NP program success. A decrease in NP program
success is seen when catalytic innovations are high. This is not necessarily an effect of
the catalytic innovation but more of an impact of the focus on exploration projects from
which catalytic innovations emerge. A propensity toward exploration projects results in
fewer exploitation projects, and hence, fewer new products are introduced to the market.
More resources are also used in exploration projects. A research focus, as opposed to a
development focus, and the resulting drain on slack resources leads to a decrease in NP
program success. Too much of a research focus can hurt achieving optimal NP program
success as shown in Figure 4.8. This type of oscillation in NP program success has been
observed in numerous innovative companies such as Hewlett-Packard and 3M especially
in the mid-19805.

Figure 4.9 shows the sensitivity of catalytic innovations on NP program success.
The peak indicates 100% of all adopted innovations are catalytic innovation representing
maximum knowledge yield from adopting an innovation. The lowest curve indicates that
no adopted innovations are catalytic; all innovations result in standard knowledge yield.
The middle curve represents the mean from varying the catalytic innovations from all
(100%) to none (0%). The 100% scenario assumes that every exploration project results

in a catalytic innovation so that large knowledge gains are achievable through every

 

5 Note the two different scales along the y-axis, one for each variable. These numbers are representative of
the relationship and should not be taken at face value.

96

exploration project. This new knowledge is then easily and cost effectively turned into
new products. If this could be achieved, NP performance could be maximized.

Figure 4.10 shows the effect of changes in NP program success on the number of
resulting catalytic innovations. These evaluations lend support for the theory proposed in
Chapters 1 and 2 that the successful adoption of innovations can have a positive effect on
NP program success. High NP program success can signiﬁcantly increase catalytic
innovations, whereas low NP program success leads to very few catalytic innovations.
This is expected because high NP program success can result in excess slack resources,

which can then be allocated to exploration projects.

Figure 4.8 Base Case NP Program Success & Catalytic Innovations

 

0.4 projects/Quarter
4O proﬁts

 

 

 

 

 

 

 

 

*/
i
r

0.3 projects/Quarter
20 proﬁts

 

 

 

 

 

 

__\:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.2 projects/Quarter I
0 proﬁts i

 

8 l 6 24 32 40 48 56 64 72 80
Time (Quarter)

1!.

Catalytic Innovations % 1 IL 1 t 4 3 3 4 ‘0 4 projects/Quarter
Npsuccesszzezzz zzezzzzzzpmﬁts

 

97

Figure 4.9 Sensitivity of NP Program Success to Changes in Catalytic Innovations

NP Program Success

 

200

150~ --s

 

 

100

 

 

 

50 _, "

 

 

 

 

 

 

 

 

 

 

Figure 4.10 Sensitivity of Catalytic Innovations to Changes in NP Program

20

Success

catalytic innovations

40 60
Time (Quarter)

80

 

0.3908

0.3436

0.2963 -- w k

0.2490 - __

 

 

 

 

 

 

 

 

0.2017

20 40 60

98

80

Question 2 of the research questions presented in this chapter asked how do
changes in slack resources affect the successful adoption of catalytic innovations. As
slack resources grow, more exploration and more exploitation activities can be
undertaken. Slack resources fall to zero when actual performance falls short of the
targeted performance. Slack resources are necessary for the emergence of catalytic
innovations because they provide the funding for exploration activities, the vehicle by
which catalytic innovations arrive. Yet, slack resources alone are not sufﬁcient for
ensuring the emergence of catalytic innovations. This is seen in how the adoption of
catalytic innovations changes over time.

In Figure 4.11 from time=8 to approximately time=40, there is a slight increase in
catalytic innovations when slack resources are high; however, from time 40 to
approximately time=64, there is a signiﬁcant increase when there are no slack resources.
The failure to reach the targeted goal results in an increased focus on exploration
projects. Starting at about time period 40, there is a shiﬁ in focus from exploitation
activities to exploration activities because performance levels are not being met. Recall
loop B1 (Figure 4.4) where the performance gap is positively related to an increase in the
adoption of innovations — the greater the gap, the greater the focus on looking for new
technologies. This results in a substantial increase in catalytic innovations. Thus,
although slack resources have fallen to zero, a ﬁrm may still put a focus on adopting
innovations in order to try to improve overall NP program success.

Figures 4.12 shows the sensitivity of catalytic innovations to changes in slack
resources. Superimposing Figure 4.11 on 4.12 shows that catalytic innovations are not

driven by the availability of slack resources but by the ﬁrm’s need to close a performance

99

gap through innovative new products. Catalytic innovations are found even when there
are no slack resources available. This dual inﬂuence is shown in Figure 4.2 where both
slack resources and the performance gap has positive effects on the adoption of catalytic
innovations.

Question 3 focused on the relationship between allocation of resources and
catalytic innovations. Figure 4.13 shows how percentage of resources allocated to
exploration activities affects catalytic innovations. This ﬁgure demonstrates very well
the dual inﬂuence on catalytic innovations; from 8Q to approximately 38Q less than 50%
of resources are allocated to exploration projects yet a slight rise is occurring in catalytic
innovations. This gain is from the abundance of slack resources during this time period
as observed in Figure 4.11. After 38Q the number of catalytic innovations grows even
higher because of the greater than 50% resource allocation to exploration projects.

Figure 4.14 shows the sensitivity of catalytic innovations to changes in resource
allocation. By comparing Figures 4.12 and 4.14, it appears that catalytic innovations may
be more sensitive to the lack of slack resources than to insufﬁcient focus on research
activities. Catalytic innovations are more sensitive to lack of slack resources (lowest
valley at 0.2) compared to a reduction in resources allocated to exploration (lowest valley
at approximately 0.265). However, catalytic innovations are more sensitive to resource
allocation decisions when modeled high than when slack resources are modeled high.

Question four looked at the role of adaptive capacity as an ampliﬁer to knowledge
utilization. Adaptive capacity has been deﬁned as a ﬁrm’s willingness and ﬂexibility to
be open to new ideas. When adaptive capacity is low, the ﬁrm cannot (or will not) easily

adapt to environmental changes. When it is high, it is open to making changes as its

100

environment evolves. Figure 4.15 shows how adaptive capacity changes over time (curve
1) along with the number of technology exploitation projects undertaken (curve 2). As
adaptive capacity dips, the number of exploitation projects undertaken increases. As
adaptive capacity increases, the number of projects decreases.

Adaptive capacity was also modeled as an ampliﬁer to marketing proﬁciencies

(curve 3 in Figure 4.15) and R& D proﬁciencies (curve 4). Similar dampened effects are
also seen with these two variables. However, Figures 4.16-18 show these variables are
highly insensitive to adaptive capacity. Adaptive capacity does not appear to be a
moderator to these variables.

Research question 5 asked how the changes in marketing-R&D integration
affected the success of the NP program. The effect of marketing-RD integration on
catalytic innovations appears to have no effect on the success of catalytic innovations
(see Figure 4.19). This success is most likely an artifact of the model itself. As modeled,
marketing-RD integration increases with each successful exploitation project but erodes
slowly. Thus, it does not act as a limitation to the successful adoption of innovations;
only the upside of the two departments interacting is seen. The sensitivity analysis in
Figure 4.20 shows that catalytic innovations are only slightly affected with changes in

marketing-RD integration. This model ﬁnds little conclusive support for the relationship

between these two variables.

101

Figure 4.11 Base Case Slack Resources & Catalytic Innovations

 

 

 

 

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 

6 resources 5 l
0.4 ro'ects/ uarter
/ I I \
‘i‘ -4: .27 \\
3 resources _ _____ ___4 _ :5 X: _r g__ _4_\“\-2-/*/2‘
0.3 projects/Quarter / ‘ I
ITS . .2:
0 resources 3 ' /T/
0.2 projects/Quarter j l l 3 3 / ,

 

 

 

 

 

 

 

 

 

 

 

 

 

8 16 24 '56 64 72 80

 

Slack Resources 3 2 2 3 3 3 3 2 4 1 3 1 3 resources

Catalytic Innovatiom 2 2 2 2 2 2 2 2 2 2 2 projects/Quarter

Figure 4.12 Sensitivity of Catalytic Innovations to Changes in Slack Resources

catalytic innovations
0.4

 

I

0.332

 

 

 

0.264 ——;—— '

 

 

 

 

 

0.196 t _ ’ ‘q’ —‘—”

 

 

 

 

0.128

 

0 20 40 60 80
Time (Quarter)

102

Figure 4.13 Base Case Fraction of Resources Allocated to Exploration Activities
& Catalytic Innovations

 

 

 

 

 

 

 

 

 

(1.32an 1111‘31 1
0.4 projects/Quarter i , i i ’ ’ ' A
_L 1-12 . -.-2 '3.”
1 i i i Y //4—~\
l 1 I * * ; l N
. 5 ' - \\1

. %e2q)lorepq 1 1 5 J
8.3 projeCts/Quarter 7 .1- ‘iﬁm" T Lﬁﬁp _H gﬁhzk

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.2%e>cplmepmj |

0.2 projects/Quarter

 

8 16 24 32 40 48 56 64 72 80

 

Frac. Resource Allocation to Exploration 3 3 3 3 3 3 3 3 3% Dmnl
Catalytic Imovations 2 2 2 2 projects/Quarter

u
m
m

p
m
m
M

a

Figure 4.14 Sensitivity of Catalytic Innovations to Changes in Fraction of
Resource Allocated to Exploration Activities

catalytic innovations

 

 

 

 

 

 

 

 

 

 

 

0.3952

0.3344

0.2736 —
If" ‘

0.2128 7’ 22 4 -44- __1. ---_ _ hm-“ _ _3
f 1

0.152 ‘
0 20 4O 60 80

Time (Quarter)

103

Figure 4.15 Base Case Adaptive Capacity, Technology Exploit, Marketing
Proﬁciencies, and R&D Proﬁciencies

 

4 adaptcap 3 ; I T f f
10 projects ‘ ‘
20 proﬁciencies
20 proﬁciencies

0 adaptcap
5 projects
10 proﬁciencies
l4 proﬁciencies

 

 

 

 

 

 

 

 

-4 adaptcap
0 projects
0 proﬁciencies ' : l : . ~ 1 ‘ I . . I ‘ |
8 proﬁciencies l . l l l l l 1 | l l l
8 12 16 20 24 28 32 36 4O 44 48 52 56 60 64 68 72 76 80
Tune (Quarter)
Adaptive Capacity 3 3 3 3 3 3 3 3 3 3 adaptcap
Technology Exploitation 4 2 2 2 2 2 2 2 2 projects
Marketing Proﬁciencies 2 2 2 2 2 2 2 2 2 proﬁciencies

R&D Proﬁciencies —4-—-*+—-—+--+——+—~4—~+~"—4*-~4——— proﬁciencies

Figure 4.16 Sensitivity of Technology Exploitation to Changes in Adaptive
Capacity

Technology Exploitation
8.946

 

7.552 ~- ; 4
6.157 l 4 .-im-.-__._. ___

4.762 3““ *—ﬁ 4 ——~- 3 ~ —_~ “______

 

 

 

 

3.368
0 20 40 60 80

Time (Quarter)

104

Figure 4.17 Sensitivity of Marketing Proﬁciencies to Changes in Adaptive
Capacity

Marketing Proﬁciencies
1 8.05

 

14.79 #

 

 

11.52 ’—

 

 

8.263 ' —‘

 

 

 

 

Time (Quarter)

Figure 4.18 Sensitivity of R&D Proﬁciencies to Changes in Adaptive Capacity

"R&D Proﬁciencies"
16.95

 

14.86

 

12.77

 

 

10.67

 

 

 

 

 

8.583
0

Time (Quarter)

105

Figure 4.19 Base Case Marketing-RD Integration & Catalytic Innovations

 

20 integration 1 1 1 1
0.4 projects/Quarter 1 1 '

 

 

1

1 .
10 integration 7 _1 _ - 1 1 ,1 . / 'i- 1
0.3 projects/Quarter 1 1 ' r 1

 

 

 

 

 

 

1
“1*
1
P————-‘—1-3¥4—.—#——d———

0 integration . 1 1 1
0.2 projects/erter 1 1 1
8 16 24 32 40 48 56 64 72 80

 

 

 

 

 

 

 

 

 

 

 

Marketing-RD Integration 3 3 3 3 4 3 3 3 3 3 3 3 integration
Catalytic Innovatiom 2 2 2 2 2 2 2 2 2 2 2 projects/Quarter

 

Figure 4.20 Sensitivity of Catalytic Innovations to Changes in Marketing-R&D
Integration

catalytic innovations

 

 

 

 

 

 

 

 

 

 

0.3968 1
0.3376
0.2784
1 1
0.2192 1 ~ —*~ 1 —-1
1 i , ‘
; 1
1 : 1 1
0.16 ‘ 3 '* '
0 20 40 6O 80

Time (Quarter)

106

4.8 Hypotheses

This section reports on the results from testing the hypotheses presented in
Chapter 2. As the programmer sets the SD model, the relationships as described in the
hypotheses are either explicitly or implicitly modeled. Thus, submitting this set of
simulated data to statistical analysis may not be relevant to Study 1. For the purpose of
this dissertation they are analyzed here for completeness. The values are relative, thus,
coefﬁcients from the resulting regressions are not reported.

Hypotheses Hl-Hll were tested using curvilinear regression estimation
techniques in SPSS 10.0.7 (SPSS, Inc., 1999). The data generated in the 32 simulations
as reported in Table 4.2 were used for inputs to the curvilinear regression. Each
hypothesis was ﬁrst tested for the nonlinear relationship proposed in the hypotheses. The
data were then submitted to an unspeciﬁed curvilinear regression whereby SPSS will ﬁt

numerous trajectories to the data and report ﬁt statistics to the curvilinear estimation.
The best ﬁt to the data is determined by examining R2’ the measure of the proportion of

variance explained, and the F-statistic for signiﬁcance. The best ﬁtting unspeciﬁed
regression is shown in Table 4.3.

Hypotheses H1-H3 consider the antecedents adaptive capacity, slack resources
and R&D integration. In this simulated data, an s-shaped relationship (ln Y = b0 + b1 /
x) exists between a ﬁrm’s adaptive capacity and the successful adoption of catalytic
innovations (H1: F = 4.95, R2 = 0.009, p S 0.05) and between a ﬁrm’s slack resources
and the successful adoption of catalytic innovations (H2: F = 15.7, p S 0.001 , R2 =

0.063). Although both H1 and H2 found s-shaped relationships, the R2 values were small

enough to consider the relationships inconsequential.

107

Table 4.3 Results of Unspeciﬁed Curvilinear Regression — Best Fitting Models

 

 

 

Hyp. Type Equation F p R2
H1 cubic y = b0 + b1 :1: x .1. b2 :1: x2 .1. b3 :1: x3 13.33 .000 0.069
H2 linear y = b0 + bl * x 76.39 .000 0.253
H3 cubic y = b0 + b1 II! x .1. b2 :1: x2 + b3 :1: x3 5.91 .001 0.032
H4 inverse y = b0 + (b1 / x) 116.15 .000 0.176
H5 quadratic y = b0 + b1 :1: x + b2 :1: x2 15.93 .000 0.056
H6 inverse y = b0 + (b1 / x) 23.10 .000 0.041
H7 linear y = b0 + b1 * x 3598.37 .000 0.440
H8 cubic y = b0 + b1 .1. x + b2 :1: x2 342.22 .000 0.130
H9 quadratic y = b0 .1. bl :1: x .1. b2 :1: x2 .1. b3 :1: x3 30.00 .000 0.019
H10 growth In y = b0 + b1 * x 2771.09 .000 0.377
H11 linear y = b0 + b1 * x 1963.38 .000 0.506
P1 inverse y = b0 + (b1 /x) 58.01 .000 0.177
P2 quadratic y = b0 .1. b1 :1: X .1. b2 :1: x2 6492.05 .000 0.587
P3 inverse y = b0 + (b1 / x) 235.39 .000 0.093
P4 quadratic y = b0 + b1 :1: X .1. b2 :1: x2 986.68 .000 0.177

 

 

The unspeciﬁed curvilinear regression for H1 found support for a linear

relationship (y = b0 + blx) between slack resources and the adoption of innovations (F =

76.39, p = 0.001, R 2 = 0.253). Hypothesis H3 (marketing/R&D integration — catalytic

innovation) was not supported, as a linear relationship does not exist (F = 1.78, p =

0.183). Although the relationship can best be ﬁt with a cubic model, the R2 is low.

Hypotheses H4 — H6 look at the adoption of catalytic innovations as an

antecedent to technological knowledge exploitation, marketing proﬁciencies, and R&D

proﬁciencies. The s-shaped trajectory of H4 is supported (H4: F = 111.17, p 5 0.001,

R 2 = 0.170), although an inverse regression better ﬁts the data. Even though H5 and H6

108

 

can be described by s-shaped curves, the relationships can be considered inconsequential

(H5: F = 13.38, p s .001, R2 = 0.024; H6: F = 14.62, p s .001, R2 = 0.026).
Hypotheses H7- H9 test the s-shaped relationship between the antecedents
technology knowledge exploitation, marketing proﬁciencies, R&D proﬁciencies and the

dependent variable, NP program success. Only H7 is supported (H7: s-shaped
relationship - F = 1372.18, p = .000, R2 = .231). H8 and H9 proposed linear

relationships that were found but the st were low (H8: F = 10.21, p S .001, R 2 = 0.002;

H9: F = 24.79, p = .000, R 2 = 0.005). Even the unspeciﬁed nonlinear relationships with

higher R 2 values, as reported in Table 4.4, are inconsequential.
Hypotheses H10 and H11 test the feedback relationships from NP program

success to adaptive capacity and slack resources. The s-shaped relationships are
supported including the inverted s-shaped for H10 (H10: F = 1930.04, p .000, R2 =

0.297; H1]: F = 1782.76, p = .000, R2 = 0.482). H10 is better described by an inverse

relationship (y = b0 + b1 / x) and a linear relationship was found to best describe H11.
As previously mentioned, ﬁve new paths were introduced to the model as shown

in Figure 4.1. These paths were also tested using curvilinear regression. Paths P1 and P4

were supported as an s-shaped relationship was found to be signiﬁcant (Pl: F = 111.17, p
= .000, R2 = 0.170; P2: F = 6076.52, p = 0.000, R2 = 0.571) but P3 and P4 resulted in
low R25 (P3: F = 195.92, p = .000, R2 = 0.041; P4: F = 413.18, p = 0.000, R2 = 0.083).

Again, all of these relationships can be explained with greater precision with unspeciﬁed
nonlinear equations (Table 4.4). In comparing the model in Chapter 2 with the model in

Figure 4.1, Pl replaces H1, P2 replaces H4, P3 replaces H5 and P4 replaces H6. For each

109

of the four new paths modeled, the R2 is considerably higher than the R2 for the path it
replaced (see Table 4.4 for these comparisons). These results lend support for the
improved model.

To evaluate the moderating relationship of adaptive capacity, (P5), an interaction
term was generated. Multiple linear regression was conducted on each of the dependent
variables, technology exploitation, marketing proﬁciencies, and R&D proﬁciencies. The
interaction term (technology knowledge acquisition * adaptive capacity) was signiﬁcant
for all three regressions (p = 0.000) lending support for P5 that adaptive capacity will act
as a moderator to knowledge utilization in the NP process. However, this is not
conclusive, as it may actually be an artifact of how the SD model was programmed.

It must be noted that even though many of the hypotheses are not supported this
does not indicate an incorrect SD model. First it must be understood that the simulated
data was obtained by deliberately modeling in ‘low’ or no relationships between
constructs. Thus, marginal effects in the model will lead to inconsequential effects in the
simulated data regression analysis. Most importantly, as previously noted, in retrospect
building hypotheses for simulated data is inappropriate. The question to ask is whether
the model can accurately represent the phenomena of interest. Sterman (2000) contends
that “no model can be veriﬁed or validated because all models are wrong. All models,
mental or formal, are limited, simpliﬁed representations of the real world. They differ
from reality in ways large and small, inﬁnite in number” 03g. 846). Levine et al., (1992)
and Sterman suggests ways of testing models to uncover ﬂaws and improve models so
that they closely as possible reﬂect the reality of the system being studied. They suggest:

testing boundary assumptions, reconciling the model structure with vested parties, testing

110

extreme conditions, checking for behavior anomalies and surprise behaviors, and
conducting sensitivity analyses. All of these steps were taken to ensure that this model as
accurately as possible reﬂects the evolving NP processes of small engineering driven

ﬁrms.

4.9 Discussion and Conclusions

A fundamental question introduced in Chapter 1 is whether the adoption of
innovations can be tied to the success of a ﬁrm's NP program. The case studies and
resulting system dynamics simulations shows support for this link. However, it is not
necessarily the innovation itself that helps to promote a healthy NP program, but more
accurately, it is the knowledge gained from adopting the innovation that leads to a more
successful program. Catalytic innovations also may contribute to the success of new
products. Since they bring unanticipated knowledge gains, more NP projects can be
undertaken by the organization.

This study supports the March (1991) proposition that a ﬁrm may hinder its
performance by exclusively engaging in either exploration (research) or exploration
(development). Since the adoption of innovations is a part of most technologically
oriented ﬁrms’ research process, too much of a focus on catalytic innovations may
possibly be detrimental to NP performance. Adopting innovations requires resources in
forms of man-hours and ﬁnancing. If adequate resources are not available to support
these activities, the ﬁnancial bottom line may be affected.

The results of this study also suggest a decision-rule criterion for determining the
proper balance between exploration and exploitation activities. In this model a 50/50

allocations was not seen to maximize NP program success (see Figure 4.21). In this

111

study, NP program success is maximized with about a 25-75 split on exploration/
exploitation activities. When too many resources are allocated into expense research
activities, the ﬁrm cannot take on exploitation activities that allow the creation of new
products. Obviously NP proﬁts cannot be generated until the products reach the
marketplace, thus, a major focus on exploitation activities appears to be more important
than emphasizing exploration activities especially when resources are limited.

This study also supports Cyert and March’s (1963) contention that slack resources
are necessary (but not sufﬁcient) for the success of the adoption of innovations. This is
particularly true for catalytic innovations whose outcome may not always be successful.
Exploration activities have highly variant results on the NP process; sometimes they may
be wildly successful, occasionally ruinously unsuccessful, but usually moderately
successful (March, 1991). Firms must be able to have sufﬁcient resources so that
unsuccessful exploration does not become a disadvantage to the ﬁrm. They must also
have a propensity towards exploration.

System dynamics models provide a fundamental opportunity for studying a ﬁrm's
strategic policy. One of the policies examined in this study was the lockout of new
knowledge due to the lock-in to core competencies (Leonard-Barton, 1992). As proposed
in Chapter 1 and 2, a ﬁrm may become ensnared in a "success trap" when performance
targets are being exceeded. It may take on a 'if its not broken, don't ﬁx it' attitude and
will focus on exploiting an existing knowledge base through a focus on development.
This can lead to lower performance levels as products become outdated. This study
shows that a ﬁrm must balance exploration and exploitation activities so as not to end up

in the success trap. Exploration activities are required for a sustained ﬂow of new

112

knowledge, and exploitation activities are required for continual improvement of the
existing product line.

The case studies suggest that a ﬁrm may utilize its knowledge stock based on its
past performance. Newer knowledge is emphasized more when competition increases or
NP performance is in possible jeopardy. Older knowledge is emphasized when there is
no threat to performance. This may also increase the potential for a ﬁrm to fall into the
success trap. A ﬁrm may increase the success of the adoption of innovations by
arranging for adequate slack resources and encouraging R&D marketing integration. The
most important outcome of the adoption of innovations is that the acquisition of new
knowledge that can be used to develop more competitive products for the marketplace.
Adaptive capacity, or a ﬁrm's willingness to be open to new ideas, does not appear to be a
requirement for the successful adoption of innovations, but instead may play a more
consequential role in how knowledge is utilized once it has been obtained. Future
research evaluating how a ﬁrm's changing adaptive capacity affects decision-making
policies in the NP process may be worth investigating.

Although catalytic innovations are not an essential prerequisite to ensure NP
success, they may substantially increase the knowledge base of the ﬁrm for building new
products. A ﬁrm can foster an environment for the emergence of catalytic innovations by
encouraging research projects, even though they may fail, and by supporting exploration

activities with the necessary resources.

113

Figure 4.21 Resource Allocation Effects on NP Program Success

NP Program Success

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 1 1 1 1 1 n 1
6° 11111111111111
1 1 1 '1 1 ‘= 1 1 1 1
__ .LL ,1“, i 1 i D a] 7 21- _ 2-1.2- - - “T”. i W. _. L... n 4.; L1__ L ,__1 AL. __
451.111.. 1111111
1 1 1 1 1 1 1 1 1 1 1
1 . . ; | 1 ‘ '
3° 7 *— 1 1' “ 1 111 ' ““1 T1“: F: . .
/ 1 1 1 1 1 1 1 . 2 2 2 a - *2/41:
. 1 1 1 1 .
15 f “‘1 1 "“17" 1’ ‘1F1m1—"1’F1‘ 1'—"'—'"—'* “‘31—'33”-
/ 1 1 1 1 1 1 a 1 ' 1
1 1 ‘ 1 , 1
o 1 1 1 1 l 1 1 1
8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 80
T”rme(Quarter)
base3 1 1 i i t 1 4 4 1 1 1 1. + i i proﬁts
ﬁac. resource allocation 50/50 2 2 2 2 2 2 2 2 2 2 2 proﬁts

frac. resource allocation 25/75 2—-—2—-2—2—-2——2-—-2—2——2—2—2—2- proﬁts

114

CHAPTER 5

RESEARCH METHOD — STUDY 2: LONGITUDINAL STUDY

5.1 Introduction

Study 2 involved conducting longitudinal panel data collection in order to test the
hypotheses as outlined in Chapters 2 and 3. Measures at time 1 and their corresponding
change scores at times 2 and 3 were collected in order to observe the change over time
that occurs in the NP development process. Chapter 5 reports on the data collection, the
structural equation modeling linear growth curve methodology, and the results of the data

analyses.

5.2 Sampling Frame

The hypotheses were tested using longitudinal data collection over three time
periods based on the sampling frame of 101 companies/divisions owned by a
manufacturing conglomerate based in New Jersey, USA. This sample is a unique family
of engineering and technology-based companies that manufactures a diverse line of
products with a primary focus on the metals industry. The ﬁrms are also involved in
other manufacturing industries such as optics, semiconductors and lighting displays.
Divisions are located in Australia, Belgium, Brazil, England, France, Germany, India,
Japan, Korea, Mexico, Taiwan and Turkey. In all of these companies, excellence in
engineering is the key shared characteristic.1

The VP of Marketing of the parent company notiﬁed each of the one hundred

 

1 from company web page. Consistent with non-disclosure, identity of company is not available

115

ﬁrms/divisions of the intended study in a letter. Each was asked to participate in the
year- long study. In July 2001, survey 1 was sent to the President/General Manager, VP
Engineering/CTO and the VP Marketing/Marketing Manager of the 101 ﬁrms/divisions,
as these individuals are most familiar with their new product development processes.
Eighty-seven surveys were returned representing seventy-three ﬁrms/divisions. Many of
the surveys were completed jointly by several people in key positions in the new product
process. The other 28 companies either declined to be involved in the study or responded
that their primary function was sales/support, thus they were not involved in new product
development and could not answer the survey. Of the 73 ﬁrms responding, 55 ﬁrms
provided information regarding innovations they had adopted into their organization
within the last thirty-six months. As the study centered on the adoption of innovations,
only those ﬁrms providing this information were administered survey 2.

Forty-three ﬁrms responded to survey 2, which was sent in early December, 2001.
One company had ceased operation, one had merged with another sister company, one
company had a change in President and the new President did not feel qualified to
respond, and one respondent had suddenly died since the ﬁrst survey. The other eight
ﬁrms were considered nonrespondents. In April 2002, survey 3 was delivered to the 53
ﬁrms that had originally responded to survey 1 with information regarding adopted
innovations. Only the two ﬁrms that were no longer in existence were excluded from the
original 55 respondents. Forty-two ﬁrms responded to survey 3. The ﬁnal count was 38
ﬁrms responding to surveys l, 2,and 3, six ﬁrms responding to surveys 1 and 3, and four

ﬁrms responding to surveys l and 2. This resulted in a total sample size of 49 ﬁrms or a

116

48.5% response rate for the entire study. Missing data was imputed for ten ﬁrms not

responding at times 2 or 3. This method will be explained in the next section.

5.3 Missing Data Imputation

Missing data is a classic problem in longitudinal studies. Missing data may occur
due to several different reasons; non-response by subjects unable or unwilling to respond
to the study, missing occasions when a subject participates in the study during a number
of occasions but not all times, item non-response, and/or failure to obtain measurements
at equal time periods. In addition to the reasons why data may be missing, the
mechanism by which missing data are generated is also important. Little and Rubin
(1987) describe three different mechanisms. Data that are missing completely at random
(MCAR) does not depend on its value or the value of any other variables. Data that is
missing at random (MAR) may depend on the value of any of the manifest variables but
is unrelated to the scores on any of the latent variables. Data that are non-ignorable
missing is related not only to the values of the manifest variables but also to the latent
variables (Bijleveld, et. a1, 1998). When data is assumed to be MCAR, a common
approach is to run computations using listwise (LD) or pairwise (PD) deletion and to
generalize the results to the population from which the sample was drawn. This leads to
consistent although not necessarily efﬁcient estimates. If the data is MAR, LD and PD
estimates may also yield biased results. Another method of addressing the missing data
problem is to replace each missing value with the observed mean of the variable. Mean

imputation (MI) can lead to biased variance and covariance estimates.

117

An alternative solution to missing data is based on full-information maximum
likelihood (Arbuckle, 1996). This imputation method uses all the information of the
observed data including estimates of the population means, variances and covariances.
Matrices of the estimates of the means and covariances are obtained from conﬁrmatory
factor analysis. These matrices are used to generate the missing data for an individual

case using the following formula:

E1x. xm>=ﬁ.+i.,.i;1.<x.—ﬁ.> (5.1)

 

where

- xu is the vector of unknown measurements which are being calculated,

- xm is a vector containing the measurements that are observed for the case,

- [1m and [1,, are estimates of the elements of population mean, ,u’ ,

corresponding to the measured and unmeasured values, respectively.

A

- me contains the row and columns of the population covariance matrix, 2‘ ,

corresponding to the measured covariance values, and

A

- 2,0,, is a submatrix obtained from Z. by deleting the rows corresponding to the

measured values and deleting the columns corresponding to the unmeasured
values.
An example will help to illustrate this technique. Marketing proﬁciencies were
measured by four variables at time 1 and the same four variables at time 2. Thus, the
following matrices were estimated for the mean and covariances based on the available

data from all respondents at time 1 and time 2.

118

 

 

 

 

"3.23‘ "1.71 1.02 0.93 1.081.93 1.301.191.24"
4.33 1.02 2.27 1.74 13311.20 2.51 1.75 1.40
3.86 0.93 1.74 2.60 1.431084 1.83 2.30 1.57
A. _ 3.70 A. _ 1.081.331.43 2.121105 1.421.57 2.42
1’ — 3.70 and Z " 1.93 1.200.841051326206 1.49167
4.74 1.30 2.51 1.83 1.421206 3.53 2.26 1.82
5.19 1,191.75 2.3015711492263112.23
_4.30_ _1.24 1.40 1.57 24211.67 1.82 2.23 3.65 _

In this study, case #1 had missing data for survey 2, however, data for all

variables were available for survey 1. Thus, for this individual,

 

 

 

 

 

 

 

13.23“ ”3.70T 11.71 1.02 0.93 1.08 ‘
A 4.33 , 4.74 . 102227174133
‘1’": 3.86 1”“: 5.19 1‘2”": 0.931.742.601.43 ;
_3-70- _4.30_ _1.081.33 1.43 2.12 1
"1.93 1.20 0.84 1.05‘ "2' "2.261“
. 1.30 2.51 1.83 1.42 A 6 —— 6.358
2”" : 1.191.752.301.57 iand x“ = 6 ”that EU" x’") = 6.095
_1.24 1.40 1.57 2.42_ -5- _5.447_

 

 

 

 

 

 

In this manner missing data was imputed for case #l's marketing proﬁciencies
data at time 2 in order to have a complete data set. As previously noted, six cases had
missing data for survey 2 and four had missing data for survey 3. This imputation
method was used to calculate estimates of the missing variables for these cases. Overall,

this results in 10.20% of imputed data in this study.

5 .4 Measurements

Two rounds of questionnaire pre-testing were conducted using measures either

adapted or developed from the literature. In the ﬁrst round, ﬁve managers each with

119

more than ten years experience in NPD commented on the survey and the cover letter.
Their comments and suggestions resulted in revisions before conducting pretests in two
companies.

In survey 1, seven point Likert scales (1 = strongly disagree to 7 = strongly agree)
were used throughout, except for the performance scales (which were —3 to +3, as
speciﬁed below). In surveys 2 and 3 in order to collect change scores data, seven point
Likert scales (-3 = signiﬁcantly decreased to +3 = signiﬁcantly increased) were used
when repeated measures data were collected.

In the ﬁnal analyses some questionnaire items were not used in surveys 2 and 3
because: (1) exploratory factor analysis (EFA) revealed more than one factor and only the
ﬁrst was chosen; (2) tests for measurement invariance over time revealed inconsistencies
in a measure, or (3) subsequent conﬁrmatory factor analysis (CF A) suggested that the
measure be dropped. The ﬁnal measures used in the model analyses are presented in
Appendix C. The following list of constructs and associated items were used in the ﬁnal
analysis of the longitudinal study:

Adaptive Capacity (t=1, Cronbach's alpha (or ) = 0.809; t=2, 01 = 0.745; t=3, or =
0.817). One of the goals of this study was to operationalize the concept 'adaptive
capacity', which this study had previously deﬁned as a ﬁrm's willingness, ﬂexibility and
ability to adapt to its environment. Adaptive capacity was based on concepts of
absorptive capacity (Cohen and Levinthal, 1990), ﬂexibility and marketing/R&D
orientation from management and marketing literature. In order to separate the speciﬁcs
of a marketing absorptive capacity from an R&D/technical absorptive capacity, ﬁve

constructs were included in survey 1: market orientation (5 items based on J aworski and

120

Kohli, 1993), R&D/technology orientation (5 items based on Jaworski and Kohli),
strategic ﬂexibility (5 items — 3 based on J aworski and Kohli and 2 new), technical
absorptive capacity, and marketing absorptive capacity (4 items each, all new based on
Cohen and Levinthal, 1990). After tests of construct validity, described in the next
section, only three of the items measuring strategic ﬂexibility were used in surveys 2 and
3. The two items based on Jaworski & Kohli's 1993 study on market orientation. and
one new measure, which emphasized the ﬁrm's ability to adapt to changes in its
environment, were used in the ﬁnal analyses.

Slack Resources (t=1, a = 0.809; t=2, a = 0.741; t=3, 01 = 0.852). Slack
resources are those resources not previously committed to other organizational
departments or programs. It was measured by four items based on Calantone, eta1.,
(forthcoming). These were used to measure the abundance of capital equipment, material
supplies, R&D resources and advertising resources available for new product
development.

Marketing-R&D Integration (t=1, or = 0.899; t=2, a = 0.894; t=3,01 = 0.913).
Four items based on Li and Calantone (1998) were used to measure the degree to which
marketing concepts were considered in engineering/R&D decisions. During the survey
pre-test it was found that most of the ﬁrms in this study only had an informal marketing
departments, and therefore, the marketing department was not involved in the
engineering process. However, when further pressed, respondents conﬁrmed that it was a
matter of standard policy to consider customers' needs, pricing issues, competitive
positions and market conditions when building new products. Thus, the four questions

used in this study reﬂect this implicit use of marketing issues in the NP process. This is

121

an important distinction that should be recognized in future uses of measuring this
construct.

Successful Adoption of Innovations (t=1, a = 0.713; t=2, or = 0.823; t=3, a =
0.875). Both Tomatzky and Klein (1982) and Robertson and Gatignon (1986) argue that
a measure of success for the adoption of innovations should be the implementation of the
innovation within the ﬁrm. Eight measures were developed for this study that focused on
the use of innovations within an organization aﬁer adoption. As this study's unit of
analysis was the NPD program, each respondent was ﬁrst asked to list three innovations
that the ﬁrm had adopted in the last thirty—six months. These three innovations were
tracked throughout the remainder of the study.

Secondly, the eight measures developed for this study were then asked with
regards to the innovations identiﬁed in the ﬁrst section. The average of the responses
across the three innovations was used as the ﬁnal score for each measurement. Surveys 2
and 3 followed a similar format, but listed with each question the innovations identiﬁed
by the respondents in Survey 1 for ease of recall.

For a list of innovations cited by these ﬁrms see Appendix C. After exploratory
factor analysis and evaluations of construct validity, only ﬁve of the original eight
measures were used in surveys 2 and 3. The ﬁnal analysis used three of these items. The
ﬁnal list of measures used in also in Appendix C.

Knowledge Exploitation (t=1, 01 = 0.843; t=2, a = 0.827; t=3, a = 0.805).
Knowledge exploitation was measured by ﬁve items based on March's 1991 concept of
technology exploration/exploitation. These items did not hold up against tests of

construct validity and reliability. Thus, these measurements were eliminated from

122

Surveys 2 and 3 and four new measures were created for the subsequent studies. After
the CF A, one item was dropped resulting in three items in the ﬁnal analyses.

The four new questions were asked twice in survey 2; once to get an initial
measure and the second time to determine their relationship to the innovations identiﬁed
in survey 1. The ﬁrst set of questions was in part 1 of survey 2 and the second set of
questions was in part 2 of survey 2, which focused on the outcomes to the adoption of
innovations. Although not preferred, measuring both the initial condition and the effects
of the adoption of innovations on knowledge exploitation at time 2 was a way of
correcting the unusable items in survey 1. The wording of the questions and their
location within survey 2 helped to guide the respondent to answer these questions from
the two different time frames.

Surveys 2 and 3 framed the questions with respect to the three innovations
identiﬁed in survey 1 so that each question was asked up to three times, once for each
innovations reported in survey 1. The responses were averaged across the three
innovations and were used as measures of program-level effects. See Figure CI in
Appendix C for an example of the question presentation.

Marketing Proficiencies (t=1, a = 0.803; t=2, or = 0.808; t=3, a = 0.781). Eight
items were originally used to measure marketing proﬁciencies in survey 1. Five of the
items were borrowed from Song and Parry (1997). Three additional items were added
based upon the interests of the parent company sponsoring the longitudinal study. As
these last three items were directly related to the marketing scope of the company, it was
not surprising that these three items had the best construct reliability and were retained

for the remainder of the study. The marketing proﬁciency measures in survey 2 and 3

123

were framed with respect to the innovations being tracked. These measures were
averaged across the three responses pertaining to the three innovations in the same
manner as previously described for the 'knowledge exploitation' scale.

R&D Proficieneies (t=1, a = 0.889; t=2, a = 0.805; t=3, a = 0.839). Four
measures for R&D proﬁciencies were borrowed from Song and Parry. A ﬁfth was added
for this study but was deleted after survey 1. The four remaining items were used to
measure a ﬁrm's ability to conduct engineering/R&D activities after adopting an
innovation. These four items were asked for each of the three adopted innovations and
averaged for a program level measure as previously described for the 'knowledge
exploitation' scale.

New Product Program Success (t=1, 01 = 0.912; t=2, 01 = 0.912; t=3, a = 0.938).
In survey 1, ﬁfteen measures were used to evaluate the success of the new product
program. Six measures were borrowed from Calantone, et a]. (forthcoming), which were
originally based on Grifﬁn and Page (1996). Two additional items were based on Grifﬁn
and Page (1996). The ﬁrst eight measures were indicators of NPD program success for
proﬁts and sales relative to competitors and ﬁrm objectives. These eight measures used
semantic scales anchored —3 (a great failure) to +3 (a great success). Three of these eight
measures were eliminated because an EFA indicated two orthogonal constructs. The
remaining ﬁve items were repeated in surveys 2 and 3 although only four were used in
the ﬁnal analysis based on the construct validity results.

The remaining seven questions pertaining to new product performance were only
administered in survey 1. These questions were primarily used in developing the system

dynamics model in Study 1 and were thus not repeated in surveys 2 and 3.

124

5.4.1 Measurement Means
Table 5.1 shows the means, standard deviations, and conﬁdence intervals of the

constructs at each time point. The means were calculated by averaging across the items
used to measure each construct. The standard errors and conﬁdence intervals are also
reported.

The Durbin-Watson test for ﬁrst-order serial correlation was also conducted. The

d statistic, deﬁned as

 

(:

3

A2
“I

l=|
is the ratio of the sum of squared differences in successive residuals to the calculated sum

of square residuals (Gujarati, 1995). The residuals, 13, , are calculated from the ﬁrst-order

regression analysis in SPSS. The DW test sets boundary limits for the range of d to

determine the existence of positive or negative autocorrelation. The lower bound, dL, and
the upper bound, dU, are determined from tables established by Durbin and Watson. For
this study dL is set at 1.503 and dU is set at 1.585. If 61 falls within the range of dU < d
<4-dU , no autocorrelations exists. If d falls within the range of 0 < d< 1.503, positive

autocorrelation exists and if (1 falls within the range of 4- dL < d < 4, negative

autocorrelation exists. From Table 5.1 it can be seen that only the construct, ‘successful

adoption of innovations’, was positively autocorrelated on time periods 1 and 2.

125

 

 

 

 

 

 

modv: Hm EmoEFEm 8806:? .m :5 N 8:5 8:89am: mm: .N :5 _ 88: 3:08.52 N3 iota—oboe 3:8 .830 “we :8 3.6.8 0535?: gawk/-5950 .v

Ed; 83? 0368: .«o ems“: .3083 age: 82 5083 72 .3 mos—g 0338: we age: mu: nm+ 1 m1 888 &55 mu: 3 .m

$6; 83.; 0368: mo $53 .3826 EEon AZ .388 7: .ml. 82g :88 037.8: mo owﬁc ml wm+ 1 ml 833 owﬁﬁo Nu“ «m .N

_m .m1_ mo 2QO .3326 got 82 :8 £088 5; 282.1.“ E .—
36
$0.: mva wood onmd mood gm: at; Sam: 6:359:
26
TB; vmm; mmm.m m3: *mmv; Vmod moo; oz; 4.05389:
REES
Go .888 G3 .28 63 .E. 8 63 £8 8.: .88 a; .38 $8.483 33.8.8 888:8
688 am. 8 80. c a: 8 c :.8 5. 8 so. 8 A3. 8 m 9% Be
oﬁm 36 mod 36 3.x. mmd 3N 26 ml :88
REES
$5. .8. 3 £2 .88 s: an. s a; .38 8.2 .83 :3 .38 88 .42: an.» .88 8:638:
$9: 8m: 8m. 8 am. 8 am. 8 $8 and am. 8 N 9% 28
mm; 3am mmé wwm No.0 mud Nm.m $6 m": :88
R285
6 z .m .3.. Z a; .38 :88 .E. 8 $3.28 :3 .38 a: .38 $3 .38 $2.98 85380
5.: 688 :88 5.: 8:8 628 an: :8: .98 93
3.0 may 3.: mod mw.m mmd mm.m end 7.“ 52:

22335: :ouﬁwouﬁ
mmooozm 36:22.35 moﬁﬁxowem 5:88—8me .«o 53803.4 Dag mooSOmoM bmommmu
:52on 82 Doug wEBxSE owuorsog ﬂammooosm $58852 xoﬂm o>ummu<

 

 

 

 

 

 

 

 

 

 

@288 23H comm E 325200 80 882 :.m 2an

126

5.4.2 Measurement Invariance, Reliability and Validity

Measurement invariance is the ﬁrst step in construct validity for longitudinal
studies. It is important for future analyses that the items measure the same concepts
across time. In order to test measurement invariance, 3-group SEM models were
conducted for each of the eight constructs. In the multi-group SEM the factor loadings,

A, and the measurement errors, ('99 are constrained to be equal across all three time

periods. Measures that were not consistent over time were eliminated from the model. In
all models, the factor loadings were found to be invariant. Only the measurement error
for one item of 'successful adoption of innovation' was variant for timel and time3. This
constraint was released. As CFI is sensitive to sample size (Hu and Bentler, 1995), one
of the CPI was found to be 1.0. A larger sample size is required to obtain a better
estimate of this ﬁt index. Although bootstrapping is an occasionally used to 'correct' for
small samples, it is not an option in multi-group analysis. The results of the multi-group
analyses are presented in Table 5.2.

Conﬁrmatory factor analysis (CF A) was used to verify the validity and
unidimensionality of the measures (Anderson and Gerbing, 1988; Bollen, 1989). CFAs
were conducted for each timeframe. However, due to the small sample size running all
the constructs together did not allow sufﬁcient power to make conclusive analyses on the
measurement model. Thus, for each time frame two CFAs were run; the ﬁrst with the
construct 'successful adoption of innovations' and its three antecedents, and the second
with the construct 'NP program performance' and its three antecedents. Since this study
will use bivariate linear growth curves to evaluate the hypotheses, this method is

sufﬁcient to determine reliability and validity of the constructs that will be modeled

127

together. Table 5.3 show the measurement models ﬁt well with the lowest comparative
ﬁt index (CFI) of 0.905, NNFI = 0.877 and the highest RMSEA = 0.103. All
standardized coefﬁcients not demonstrating convergent validity with loading less than 0.5
were eliminated. This resulted in three measures for 'adaptive capacity' and three
measures for 'ad0ption of innovations' compared to the four invariant measures found for
each in the multi-group analysis. All other constructs did not change from the multi-
group analysis. The next step was to assess construct reliability and validity following
Anderson and Gerbing (1988). Cronbach’s alphas, reported above and in Tables 5.4, 5.5
and 5.6, ranged from 0.713 to 0.938, thus, overall the alphas were satisfactory (Nunnally
1978)

To test discriminant validity, the procedure in Anderson and Gerbing (1988) was
adopted to determine whether the correlation estimate’s conﬁdence interval included 1.0.
Each construct passed this test: Tables 5.4, 5.5 and 5.6 report these correlations and

standard errors. The ﬁnal measures used are shown in Appendix C.

5.5 Methodology

The primary focus of Study 2 was to evaluate how a dynamic variable that is
changing with respect to time may affect another dynamic variable. Linear growth
curves have been recently introduced into the social sciences as a means of investigating
the analysis of change over time and a variable's relation to select predictors, which
themselves may be changing over time (Curran, 2000; Meredith and Tisak, 1990;Willet
and Sayer, 1994). Linear growth curve models (LGC or LGM) are often equated to

hierarchical level models as they allow the modeling of intraindividual and

128

 

 

 

v v m m w v v v 853.3 8
895m
mm H «N 9 H OWN 3 H omN Wm H on». om H wN NV n ems“ mm H om». om H om». - EU
88.8 388 an; at; an; $88 so; :88
58.8 .83: 48.8 58.8 .858 58.8 ._ 8.8 58.8 8 £8
Vmod oood mwod coed owed awed wood 3o. <m2mm
93.0 o H ad 3&6 co; m ﬂ ad wvmd ovod amod EZZ
mwad wood wmod 84 Sad Nvmd ovod «36 E0
3826 £83.86qu 5:835
Eacwem momocﬂocoi “#:8228on 20:863me mo “8333‘ ﬂag mooSOmom guano
AZ Gag mauve—32 035?qu Emmooosm 853852 88% gamma/w

 

 

 

 

 

 

 

 

 

 

85:82: 30883302 “mob 8 6on cosmscm _mgwosbm 95.5% 80 $38M “N6 033.

129

 

 

 

 

 

 

 

 

 

03:3

0.3 u we“ 38 u mg. 8.3 n _MN ~23 n _Nc. 28 u we“ 38 u _NQ - £0

a: S 58.8 828 .038 am; .088 888588 :28 .888 and .888 5 £8

E; 83 Sod N2; wood wood <m2mm

$3 £3 83 ES $3 $3 £22

33 :3 33 25.0 :3 8? Eu
$5 2.6 $80 26 9:6 :80
$88 388 888 88:. 38: 3:5

 

 

85¢ 25 :23 é <5 .8 33QO ”mm 038

130

xEmE .532 so Guano 3%ng acouﬂobou .Raowmmu 2: mac? 3:32 SEQ @5385

 

 

 

82.8 83.8 22.8 3.80%
NR... 33 $2 $2- Samoa m2 8
62.8 82.8
a»... 33 mg- 8383on as? E
62.8 83568on
new... $3 85332 8
comuﬁmoaxm
new... 83328 5
$2335:
3; .8 AS 8 .8 .6 8:83.
was- 82 _ammooozm A8
62.8 88.8 88825
32. :6... 388-85%“: 6
:88
8a.: 888 82888 8% E
25... £088 Ease/s. 3
m8 8 8 6 6 8 sunbeam

 

 

 

38805302 821.5 “58853 a. 88:33.08 :58 ”in 23

131

x32: .832 co $85 @3285 8032280 .38me 2: mac—m Eton“: «an? M5385

 

 

 

$2.8 32.8 32.8 308%
2.2 22 $3 $38- gem m2 :8
3:8 $2.8 meoaaoaoa
3»... $3- 85o- B2 6
22.8 36532;
8a... Sod 3:332 A8
coaﬁmoaxm
5:. owBEQQ 5
32:32::
32.8 82.8 We 8:33.
83. £3 _emsooam A8
82.8 88.8 8:325
83 82 388-3332 5
6:8
:2. 33. ...838£ 82% av
£0350
m3... gang 8
98 S 68 6 Q 8 8388

 

 

 

38505302 825 $5855 a. aﬁzﬁé 85$ ”2 2.3

132

xEmE 532 :o $8.5 “58:33 macaw—2.80 .38me 06 mac—m 3:82 man—m $28985

 

 

 

 

:88 $38 323 388m
an? 83 come ES- 8&5 .2 A8
$2.8 32.8 $653on
a: mad. 39¢ 93 S
:28 8.820888
E... 33 wags: Q8
couﬁmoaxm
m3... 3.3399 6
32:39::
E _ .8 $28 88.8 .8 882%
m5... :3 Rod 83 _emmooosm :8
$28 88.8 8:235
2.2 8% wood mamaéeﬁz 5
83.8
«mm... .83 8.238% ﬁsm S
.3330
Sn... 33%.... E
as 6 68 6 A8 6 S 3 8385

 

 

35805302 823/ “55858.9 a. 83333 Eng ”3 2.3

133

interindividual differences over time. LGCS permit the evaluation of the general shape of
individual growth trajectories and of the estimates of individual growth parameters with
respect to time (Level 1 within-person analysis). All members of a population are
assumed to have trajectories of the same functional form but different members may have
different values of the individual growth parameters. LGCs also provides estimates for
the means, variances and covariances of the individual growth parameters across all
members of a population (Level 2 between-person analysis). It is at Level 2 that
predictors of change are modeled to estimate the effect on interindividual growth
parameters.

Linear growth curves were used to evaluate the hypotheses as proposed in
Chapter 2. LGCs require repeated measures in balanced data panels. In order to obtain
repeated measure data, the change scores collected from survey 2 were added to the
initial scores obtained from the data collection at time 1 for reporting time 2 measures.
Similarly, the change scores collected in survey 3 were added to the time 2 scores to
obtain time 3 measures. Evaluation of the linear growth curves occurred in three stages.
The ﬁrst stage, measurement invariance, has been previously described. After validating
measurement invariance across time, the second step is to conduct univariate linear
growth models, and the ﬁnal step is to evaluate the bivariate linear grth curves. This

section will describe these last two steps.

5.5.1 Univariate Linear Growth Curves
Linear growth curves use information about the covariance and mean structure of
repeated observed measures to deﬁne one or more underlying latent growth factors

(Meredith and Tisak, 1990, Stoolmiller, 1995). Conceptually, it considers the observed

134

measures taken over time to be indicators of an unobserved true growth trajectory. The
ﬁxed and random components of the grth trajectory are estimated via the means and
variances of the latent grth factors. These types of models are shown in Figures 5.1
and 5.2.

Linear growth can be modeled with two latent factors as shown in Figure 5.1.
The ﬁrst factor, F1, represents the intercept of the grth curve and the second factor,
F2, represents the slope/shape of the curve. The manifest variables, v1, v2, v3, reﬂect the
measurements taken at each point in time. Fixing the factor loadings from F1 to v1, v2,
and v3 to 1.0 and to zero from F2 to v1, sets the starting point ("initial status" or
intercept) of the growth trajectory curve. The factor loadings relating the manifest
variables (v1, v2, v3) to the shape factor (F2) can be free or ﬁxed, whichever best
captures the functional form of the grth trajectory over the time points. The shape
corresponds to the "rate of change" in the variable or the rate of increase or decrease over
the time period modeled. For example, ﬁxing the paths at from F2 to v1, v2, and v3 at O,
1, and 2 respectively, represents linear growth. Freely estimating the path from F2 to v3
allows the shape of the trajectory to be determined by the data. Such a model is referred
to as an unspeciﬁed two-factor model (Chan, 1998). In this type of model, the loadings
plotted against the observed time interval gives a visual representation of the nature or
shape of the change trajectory. The time points do not need to be equally spaced apart for
LGC models but do require balanced data in that all individual data must have been
collected at the same time. In this study, the time periods are spaced ﬁve months apart
for reasons discussed in Chapter 3. The balanced data panel was achieved through the

missing data imputation previously described.

135

Figure 5.1 Two-factor Polynomial Latent Growth Curve Model

 

 

 

 

 

 

's *
* . *
m1 ms

... . Intercept Slope ‘_ *
d1 F1 F2 ds

1 at:
O 1

V1 V2 V3

E1 E2 E3

Notes: v1, v2, v3 represent the same manifest variable at t=1, 2, 3, F1 is the
latent variable intercept, F2 is latent variable slope, mi is the mean
intercept, ms is mean slope, di is the intercept variance, ds is the slope
variance. Ris is intercept-slope covariance.

Figure 5.2 Three-factor Polynomial Latent Growth Curve Model

, * Ric * *
RlS Rsc

mc

ds*

. dc *
§ .

1 1N1 .

 

 

 

 

 

V1 V3
E1 E2 E3

136

Three or more time points allow an opportunity to test for nonlinear trajectories
(Duncan, et al., 1999). The LGC in Figure 5.2 demonstrates a quadratic function of the
observed time metric. In this type of model, a third latent factor is added to represent the
curvilinear trajectory. The paths from F3 to the manifest variables are set at 1, 2 and 4 to
represent quadratic growth. The error variances can either be freely estimated or ﬁxed at
certain values and are often constrained to be equal across repeated measures. When the
model is under-identiﬁed, the error variances are usually set to zero (Duncan, et a1. 1999).

In LGCs, the means and variances of the latent variables are freely estimated.
This is accomplished by evaluating the augmented moment matrix or using the V999
code within EQS (Bentler, 1995). The latent factors are allowed to covary as indicated

by the double-headed curved arrow between factors. The mean of the intercept factor,

m, is equal to the mean of the latent scores of the individual growth curves at time 1.
Variables, mS and rnC also model the group means for the latent factors, F2 and F3. The

variances of the disturbances of the latent variables, d] and d2, are equal to the variances

of the latent factors themselves.

Standard assumptions of LGCs are: (a) the variance of the latent means is zero;
var(m.) = var(m2), (b) the variances of all latent variable have zero means, E(d1) = E(d2)
= 0, (c) the means and variances of latent variables do not covary; cov(m1,m2) = cov(ml,
d1) = cov(m 1, d2) = cov(mz, d1) = cov(mz, d2) = 0, and (d) the error variances do not
covary with each other or any variables except the measured variables they directly
affect; cov(El, d1) = cov(E2, d1) = cov(E3,d|) = cov(El , d2) = cov(E2,d2) = cov(E3, d2)

=0.

137

Each of the constructs in the model deﬁned in Chapter 2 was ﬁrst tested under the
univariate LGC to determine the shape of the change trajectory. In each case, a quadratic
LGC was conducted ﬁrst to determine if change followed an s-shaped growthz. In order
to just identify the model in Figure 5.2 the error variances must be set to zero (Duncan, et
al., 1999). The factor loadings are ﬁxed at l, 2, and 4 to model quadratic growth. The
saturated nature of this model does not allow for testing of overall model ﬁt using the
standard model ﬁt indices in structural equation modeling (CF I, NFI, RMSEA, etc.).
With three repeated measures, there are 3 variances, 3 covariances, and 3 observed means
and the model estimates 3 variances, 3 covariances, and 3 means for the intercept, slope,
and quadratic latent factors, thus, representing 0 degrees of freedom. However, the test
statistics for the three latent growth factors can be evaluated.

Table 5.7 shows the results of the EQS models run under maximum likelihood
estimation. The only construct that was found to have a signiﬁcant coefﬁcient on the
quadratic latent variable, F3, was the 'successful adoption of innovations'. Subsequently,
the remaining constructs were tested under a linear growth model where F3 is eliminated
as depicted in Figure 5.1. Unspeciﬁed two-factor models were run with the 'shape' path
freely estimated.

Small sample size bias has been shown to be a problem in using ﬁt indexes for
evaluating goodness of ﬁt statistics in SEM (Hu and Bentler, 1995). In evaluating the
univariate growth models in several cases, CFI were evaluated at 1.0 with an RMSEA =
0.000, suggesting a perfect ﬁt. In these instances, to better estimate the goodness of ﬁt,

simulations were conducted using bootstrapping with 150 cases and 5 replications in

 

2 It is not possible to conﬁrm an s-shaped grth with three data points. Only the quadratic trajectory can
be tested.

138

EQS. The resulting ﬁt parameters reported in Table 5.7 indicate if this simulation
technique was used to estimate the ﬁt indexes and parameter estimates.

In this table, the path from F3 to v3 is labeled 'time'. As can be seen from the
factor loadings, all of the coefﬁcients for this path have 90% conﬁdence intervals that
include 2.0. This indicates a linear grth trajectory over time for these constructs.
Signiﬁcant effects were found on the intercept and slope effects, F1 and F2, for all the
constructs. Table 5.7 also reports the quadratic function factor loading for each construct
although ultimately it was not modeled except for 'successful adoption of innovations'.
Table 5.8 shows the variances and covariances of the latent variables. The estimate of
the covariance between time 1 and 2 for the construct, 'NP program success' was the only
one shown to be signiﬁcant. This can be interpreted as the initial starting point
determines the slope factor at time 1 for measuring NP program success. The positive
correlation (0.559, p s .05) indicates that the higher the initial reported value for NP

success the greater the rate of change over time.

5.5.2 Bivariate Linear Growth Curves

Multivariate latent growth models can be obtained by combining two or more
univariate models. Bivariate models were used to test the hypotheses presented in
Chapter 2 (replicated in Table 5.9 for convenience). As the name implies, bivariate
modeling combines two LGCs in order to evaluate the effect of the change in one
variable upon another. lBy regressing the intercept, slope and shape factor of one LGC
upon another, the relationship between each other can be tested. The intercept, slope and
shape factors have the same meaning as in univariate LGC modeling. An example of one

of the bivariate models used in Study 2 is shown in Figure 5.3

139

The hypotheses presented in Chapter 2 propose speciﬁc growth trajectory
relationships over time. If the hypothesis suggests nonlinear s-shaped growth, the
dependent variable was modeled as a quadratic trajectory. If the hypothesis suggests a
linear growth relationship between the constructs, the DV was modeled with a linear
trajectory. The growth trajectory of the explanatory variable was set based on the
univariate LGC and the grth trajectory for the dependent variable was set based on the
hypotheses. For example, hypothesis 1 proposes a nonlinear relationship with the DV,
successful adoption of innovations. The univariate analysis indicated that adaptive
capacity is linearly growing over time. Thus, a linear growth model was used for the
adaptive capacity's time trajectory and a quadratic model was used for the successful
adoption of innovations' time trajectory to test the relationship between these constructs.

Figure 5.3 demonstrates the model used in M1, M2, M3, M7, M8, M10 and M1 1,
which were created to test hypotheses H1, H2, H3, H7, H8, H10 and H1 1, respectively.
Quadratic growth trajectories for both the antecedent and DV were used to model M4,
M5, and M6. A quadratic LGC modeling the antecedent and a linear LCG for the DV
were used for M3. Linear LGCs for both the antecedent and DV were used to model M9.

In a bivariate model, the paths between the endogenous factors, [3“ , and the means
of the latent variables for the DV are of primary interest, and thus, are freely estimated.

The results of the bivariate modeling are detailed in the next section.

140

Figure 5.3 Bivariate Linear Growth Curve Model Example

E1 E2 E3

V1 V2 V3
1 . .
1 V"‘v 2
* . 2k
ml] H132
at: at:
dil ‘—dsz
’63] * * 1652

, *
m13\‘_ mcs*
* di3 Slope (105*
‘Q 4
1 1‘

F4
.,

E4 E5 E6

 

 

 

 

 

 

 

 

Notes: v1, v2, v3 represent the some measure at t=1, 2, 3, respectively, for variable A.
v4, v5, v6 represent the same measure at t=1, 2, 3, respectively, for variable B.

141

330E330: 53385.0. 33: c03=3>0 .—
wod w a 3 “305%; 1.

 

 

 

 

 

 

 

 

30:5 30:5 30:5 30:55 336330 30:5 30:5 . 30:5 : bouuomaah
98.8 88.8 £88 $8.8 $8.8 $8.8 €38 28.8 0:28....
Sod- $3. :80 Sod- $33.0- 38.0 Ego- god. .2..
:28 3:8 $8.8 8:8 :38 68.8 888 9.3.8
1.32 1.43.0 3.820 :23 .3502 :83 1.08.0- 133.0 0%.... .2..
G28 2:8 3:8 $2.8 3:8 m $2.8 $8.8 $2.8 2.8.5.5
**mNo.o **mvo.m “InvoHﬁ 3,004» tamaw **mNN.m 1.336 **Nmm.m «is
$3.8 82.8 82.8 :28 82.8 28.8 €28 .
13o: 13R: :23 tom: .2: m 1.48.: :83 ...... 2: 08: ea
_
NEAR ammuﬂcu 2.33 STE £3”? 3.7M». STE R: J3 2.33-30
3:5 Gone :03 2:2 H Ammo ammo Gas
.088 . 35.8 . 58.8 . .898 . 50925.8 58.8 58.8 . .898 5 .39..
Rod 83 . $3 owe. Sod 02.0 88o am; «maﬁa:
83d ~83 . :3 ”So m wood ﬁg Sad Sad 22
33 23¢ Sod 88o co: £3 mood god Eu
30026 , W W m:o_3>o:5 5333005 .
8.2me 86:22.58: w 38:08.88: m :osﬁaaxm 80 85:23 :9 “ 803835 3:330
_mZ _Qowm 3300335 _H0wv0355: 5330005 _ -w::0v_3: 103—m M 020355

 

£30m 0>.S.U £380 30:5 03:32:: ..5 m 033%

142

 

3.2 w a 3 030%:me I.

 

 

 

8:8 228

1.030 0: :.0-

5:8 2.88

1.03.0 . 103.0

_

A0008 _ 20:8

03102.5 L imam;

000026 _ 0050:2505
same: 02 : :3:

30:8
0:0

9358
:wde

33.8

18:».—

0: :8
0.2.0

388
10:0
:88
0.1.5005

36:06:05 3:86—95

wEEv—3E

030039205

 

228

**N@m.ou

€008 . ,
30.0- .

5.08 w .
*LNNd W .
2:8 N 888
3000.0 02.0
25.8 @008
1.92.0 180.0
628 W 23.8

simood **mmm.~

22355: M 55330::
mo :o::o:<

08m
_Ewm0oosw M, 3:50:33: ,

€28
000.0-
228

1.0on

3:8
3:00.:

MOOSOmOM
0:20

 

8:8
0:0-
3:8

“Inmmmd

$38
3w EV.—

35330 W,
030350ﬂ ,

0mm
Anna—003.50

02
5:43.350
0:

:5» 03am
02
238:3:
we

.53 0.50—m

=0 ..3>
E00000:—

 

 

0500035 03:33:05 mo 000:0:0> 03 2302 ”Wm 033:.

143

.5508 :0m0:m-m
.80 800 8008800 0 08: :05 850:8 0008008 :00? :00 800000 80808 “00:08 500 0.8.8 0 00080: 8:80:08 0: H

850% :0Q0:m-m 00802: :0 80: 000880: 0
08: 00>0 850:8 800000 9:80:00 0: :00 0000000 80808 80800850: 808008 500 0.8.8 0 00830: 8:80:28 0:,:.

80808 808820050: 8608 500 0,88 0:: :0 0000000 08 00:00am 0:: .m0_000:0:08 088 08 00:00am 0:5
80808 8080230: «00:08 508 0.8.8 08 .80 800000 05 ..808w 05 $8808.88 88:00:08 08 08008 0:,H

.8308 :000:m-m (:0 :80: 808800 0 08:
0050 050:8 800000 80808 8088205: 80:08 500 :00 80:80:98 0300—505: 0.88 0 :0080: 8:80:28 05.

858w :080:m-m (:0 80: 0008800 0 08: 00>0 850:8 000008 80880—0050:
8:008 50: 0:: 8 00:80:88: Q88 :80 80:058.: 0:30:00 .80 00:80:00 8800000 0,88 0 505:0: 8:80:28 0: H

.5308 38:08 (:0 :80: 800880: 0 08: 00>0 850:8 0000000 8088—05: “00:08
50: 08 8 00:80:88: 8:00:08 :00 80:088: 0:30:00 .80 80:83 880000000 0.88 0 000580: 8:80:38 0: H

850% :080:m-m .80 :80: 8008800 0 08: :86
850:8 00:80:98 03007505: 0: :00 80:88:: 0:30:00 .80 00:00:00 :880003 0.88 0 800250: 8:80:28 05.

8.8 08 08: 80:88:: 0:50:00 .80 00:00:00 0:: 80000000 808 08 80:088.: Q88-w0:0:008 08 08008 0: H

.8508 :0a0:m-m :0 80: 308800
0 08: 80>0 850:8 80:88:: 0:30:00 :0 00:00:00 8800000 0: :00 0008008 0.00:0 0.88.: 0 000550: 8:80:08 0: h

850% :000:m-m .80 808 000880: 0
08: 00>0 850:8 80:088: 0:30:00 80 00:00:00 880000000 0: :00 3:00:00 9:80:00 0,88 0 800550: 8:80:88 0: H

”:3

5::
”2,:
”mm

”E

”:3

”mm

”:3
”mm

”N:

”:2

 

0000885: 06 030k

144

5.6 Results

The results of the bivariate analyses are shown in Tables 5.10-5.13. Except for
M4, all of the models tested showed CFIs and NFIs greater than or equal to 0.900, which
indicated a good ﬁt (Bentler, 1995). The RMSEAs ranged from 0.043 to 0.208
(excluding M4). Except for M4 and M6, the RMSEA 90% conﬁdence intervals included
0.05, the maximum cutoff suggested for this goodness-of—ﬁt criteria (Browne and
Cudeck, 1993). Overall, the models show very good ﬁt to the bivariate LGC model. In
order to evaluate Akaike’s Information Criteria (AIC). Each model was also compared to
a null model (H0) where no relationships between constructs existed. The lower the AIC,
the better the ﬁt of the model. In all cases, the AIC was consistent with the ﬁt indices.
All indices are reported in Tables 5.l0-5.13.

The intent of this study was to evaluate how the dynamic nature of a variable may
affect a dependent variable over time. In order to test this type of relationship, paths were
established from the intercepts, slopes and shapes (curvatures) of one LGC to another
LGC. For a hypothesis to be supported, at least one of the [3 paths linking the LGCs to
the other must be signiﬁcant, an_d_ secondly, the means of the intercept, slope and/or shape
factor of the dependent variable must be signiﬁcant to show that the model follows the
trajectory modeled. If both criteria are not achieved, there is no support for the
relationship.

Time effects must be taken into consideration when interpreting signiﬁcant
relationships in bivariate LGCs. Intercepts indicate timel status, slopes indicate the rate

of growth (or decay) over the time frame between data collections, and shapes indicate

145

any change in growth direction over time. A few of the [3 paths established in Figure 5.3

are interpreted as follows:

(a) intercept —) intercept (BM; ): the 'initial status' (starting point) of a dependent
variable is affected by the initial status of the explanatory variable. A positive [3
indicates the higher (lower) the values at t=1, the higher (lower) the value of the DV
at t=1. A negative B indicates the higher (lower) the start values at t=1 , the lower
(higher) the initial value of the DV at t=1.

(b) slope —) slope (651,52 ): the rate of change of a dependent variable is affected by the
rate of change of the explanatory variable. A positive relationship indicates that the
faster a variable is changing over time (change may be + or -) the faster the change
in the DV( change can be + or -). Thus, the rate of change of the DV becomes
ampliﬁed or dampened due to the effect of the explanatory variable.

(c) intercept —> slope ((3,132 ): the initial status of an explanatory variable affects the
rate of change of the DV. A positive relationship indicates a greater grth rate
and a negative relationship indicates a dampen growth rate. For example, a
negative relationship indicates that higher initial levels of a variable result in less
steep growth trajectories in a positively growing DV. This negative relation does
not indicate a decrease in the DV over time but instead implies smaller rates of
positive change.

Figure 5.4 shows the model with the hypotheses and their outcome. Five of the
eleven hypotheses were supported. H1, H4, H5, and H7 were not supported because
either the paths, the means of the latent variables, or both were not signiﬁcant. In M1, no

paths between the LGCs were signiﬁcant. For M4 and M7, at least one path between the

146

LGCs was signiﬁcant except the mean on the shape (curvature) latent variable was
insigniﬁcant suggesting a linear relationship opposed to the s-shaped relationship
proposed in H4 and H7. In M5, although two paths were signiﬁcant, none of the means
of the DV were signiﬁcant suggesting no relationship between the adoption of
innovations and marketing proﬁciencies over time. Hypotheses H10 and H11 were also
not supported and will be discussed later.

Support was found for H2, which proposes an s-shaped relationship between slack
resources and the adoption of innovations. The means for the intercept, slope and shape

for the DV, successful adoption of innovations, were all found to be signiﬁcant (see
Table 5.12). The mean of the shape factor has a negative coefﬁcient (mc = -0.146, p S
0.05), suggesting growth where a peak is achieved and a downturn occurs, possibly 3-

shaped3 . In M2, signiﬁcant relationships were found on two paths between LGCs (see

Table 5.10). A negative relationship exists between the intercept of slack resources and
the intercept of the successful adoption of innovations (BiIJZ = -0.182, p S 0.01). The

more slack resources available to the NP process at t=1 , the lower the reported success in
the adoption of innovations. However, the rate of change in the availability of resources

for NP projects was positively related to the intercept of the successful adoption of
innovations (Bsmz = 0.235, p S 0.05). These two effects drive the intercept in opposite

directions with the slope effect possibly being greater.
In M3, support was found for the linear relationship between marketing-R&D

integration and the adoption of innovations as proposed in H3. First the means for the

intercept and slope factors for the DV were positive and signiﬁcant (m = 5.227, p S001;

147

               
     

gnaw

$825
83on m2

    

 

mﬂocﬂowoﬁ

mﬂocomomoﬁ

hmtemdsv. H Aka. ...emtomesm B: H w: .6802

ﬂag

wEExS—Z

:23onme
owumV—Boqv.

acumewoﬁ:
Odom

Q95 -wcuovtﬁz

mm

 

      
       

    

@383.
323235
mo mmooosm

moocsomom

.aa . seam

mm:
82::

m:
:17.

 
    

bﬁmmmo
o>smmv<

82:2

 

 

momoﬁomxz 5:5 6on Sm oSwE

148

3:62: 332252 c: 5: ”>0 .8 oamsmumo 45.9835 :0 Augmizov again—u ”>9 no oaoﬁnmm 458825
:0 32m"? ”>Q co EQBBEHQ 83835 no 50235": ”ﬁntomnsm Q. 396 w a 8 38556 1.... ”mod w a “a aﬁomﬁma 3 6.. W a 8 Ecoﬁcwﬁ.

 

 

 

 

 

antenna tor—0&3 votoamsm
3:233 ”cm 8: ”mm 8: Hum votoaqsm “mm 88.83% ”NI 8: :2 8852;:
$88
.88 n N. _S
8.8 88.8 88.8 a: 2.8
.83. u N. ..S .88 u N. _S :82 u N. _S .1.de u SS
:88 5.: 8.8 88.8 28.8
:88” N. S :85" N. an 18.2 n N. _S N... _ :8 u ES :83- u SS .2 a
5.8 8.88 8.8- $8- 8.8 m _ .m- 85 0:.
92: mod Rd. 3.2. 2.x- mud- D:
8.8 n ck 8.8 u .N_~ 8.8. n NN§ 8.: u _Ncm 8.: u _N8 8.2 n _N_.« 22.....- Eu
8.8.8 .428 $8 .88 :88 .828 :28 58.8 53 .888 9:2 .888 _u 88
womd ~36 036 $56 owed Rﬁo «£22
Sod ooad #36 Sad omad wmod EZ
wmad mvmd wowd wood Sod ~36 EU
86:32.85 momoaomomoi cosﬁmoaxm meosgoaﬁ 32855:
Gaga 3:852 03535ch 80 5303.. 82855: .«o cosmovxx
- 228.3% - 82835: - 98:92:: - 28:83:: mo qouqoc< - - £3qu
mo sonnet/N. .8 :osaog. 80 5303‘ QﬁmmamHB—BE 88:8me Mom—m o>smmc<
82 22 32 m2 m2 :2

 

 

 

 

 

 

 

 

82- :2 .3388 0:5 5305 32.: @355 no; 2%...

149

3338 538238 0: ”oz M>Q so 233nm”. 503025 :0 3:39.53 0523"? ”>0 so 0523"?- £53085
:0 25.3": SD so 583::nﬂ 503083 no 52:85”: ”35333 Q Jed w 5 an «525in 2.... ”mod w 5 “a Ewe-«Em; .2.. 5: w 5 “a EnocEwF.

 

 

 

 

 

 

3:253 3:253 W H _
Ho: ”5 :5 8: 6:5 3:253 ”53 3:253 ”mm W 3:253 8: ”S5 8352::
$8.8 5 :.8
3.38.8 u N. :.NN. W 1:28 u N. :8.
$1.8 $8.8 62.8 W
$.38 u N. ..5 32.8- n N. :5 .338 u N. .5
:28 _._ 82.8 W 8 _ .8 a: :.8 a: .8
£5.88- ” N. :5 _ :88 u N. :8. :82 u N. :5 .388 u N. :5 .388- n N. :5 8
2: 8.2 8.x- 88- 88- 3:57.
am W. 5N- M £8. H 2.2. EM. 0:.
mm. a u ON...“ 8. a u NNQ W 8.: n NN_~ 8.: n NN§ 8.2 n N8 2.5...- so
:88 .888 888.888 32.8 .888 82.8 .888 68.8 .888 5 88
23 W 58 88.8 88.8 2: .o 3:35
~88 W ”8.8 . 88.8 W. 83 82 Ez
38.8 W ~88 M88 82 88.8 :8
335305 W £2525 0352?. W C35: C35: 3825 3325 3585
32m - 3325 W - mmooozm . mmooosm 52 - 52 - 36352585 52 - 33:25:25
88585 52 :8:on 52 3:036:85 085 5:832 53—3055
:2 V 22 a 82 w: w E

 

 

:2- 52 .233: 2:5 £305 58:: 0853 N: 2 2.5

150

E @20on 8: £80.85 ﬂaunt—ma 8: n .0:

:3 w a a 388:? ..i

 

 

 

 

 

85 w a a 288:»; :
o; w a a “308%... ..
$5.8 98.8 :28 82.8 88.8
.2 .88.? m :LS? 2 3- I... :2- 152.? 3. 582.525
38.8 €88 88.8 28.8 68.8
___:mmod- Sod- .2 :53? 38.8 M 3%.? OE 5885.59
$2.8 88.8 838 82.8 6:8
.3 238 H ...._....3~.o ........m$._ .. :82 $8888 .5. .3..=8.§8
m $5.8 $88 338
woo. moo. .. £838 $3 .3388 So. a. .228 :3
$8.8 E _8 £38 :88 $38
moo. ***vam.o .Iiowmd owed ...IbNoN «..zwwomé mt Ava—23 .2;
82.8 :3. 8 a: _8 :28
..._.....8o._ 89o woo. woo. 1.538 :Lvod 6 6328.8 a:
93.8 88.8 38.8 I $8.8 $8.8
***OMN.O- cad owed- .Ibvﬁo- ..iQEd- aging—m
$3.8 38.8 22.8 98.8 85.8 83.8
***0Nm._ voﬁd “.2."?de ..chvQ— .125va Iiovﬁé ma JAG—m
528 83.: E: 83.8 68.8 8&8
awed $2 3....me :mem :3”? 2.88% _5 £832:
22 m2 «2 $88 22 . N2 _ #2 8
- - , 3.9 .0225, aggmammz-,,-mﬂammawmaﬁmaﬂawxwi .3

 

151

E “Va—owes. 8: 28205 53$:me 8: n .3.
:3 w a a “508%; 2:
35 w a a 288:»; t.
o; w a s 888%; ..

 

 

 

83.8 I I :38
ES- .2 wood. 5. 638.39
€88 l i 23.8
_.. _ m _ .o .2 :35- am $88,359
98.8 $2.8 $28 28.8
.33.? .3 I853 :.38 ...IoSd 5. Angina;
38.8 38.8 .I I $5.8
Sod *1356 moo. on. A0933 .5»
an _ .8 88.8 5 _.8 A _ _ :8 €28
***omh.o .1325de 3.....mmvd ***wvv.o *mmmd mm. Ava—33 a?»
23.8 63.8 88.8 6&8 $8.8
***wom.~ flow—A fimwﬁ; ***moﬁ.~ ***hmo.~ G Ana—023:5 55>
$8.8 88.8 I I 88.8
:12? 1&3. MS? 2.. .255
82.8 § _.8 22.8 as .8 8&8
***m©v.o- *fLLVd **m~m.o iibamd *MImNoA m8 inc—m
$3.8 $2.8 $38 38.8 8&8
1.1:: .v ***va.v tibiae hvod- .5th _E . 259.3:—
_ E 82 82 w: E

 

 

 

 

 

 

 

22-22 63mg; Eggnoa .58 .252 004 0533mm 8o moocmcﬁr 98 B522 Hmﬁm 2an

152

mS = 1.048, p S 0.01). Secondly, a positive and signiﬁcant relationship exists 0.01; mS =
1.048, p S 0.01). Secondly, a positive and signiﬁcant relationship exists between the
slope factors of these two constructs (B5152 = 0.411, p S 0.05). This indicates the greater

the change in integration activities, the greater the rate of change in the successful
adoption of innovations.
Hypothesis 6, proposing that the successful adoption of innovations result in a
non-linear s-shaped grth trajectory for R&D proﬁciencies, was marginally supported.
Although CFI and NF] passed goodness of ﬁt criteria (CFI = 0.93 8, NFI = 0.912),

RMSEA did not, RMSEA = 0.208, 90%CI = (0.124, 0.289). The high RMSEA appears
to be the result of the marginally signiﬁcant [3 coefﬁcients4. Both tests for signiﬁcance

were passed. The means of the intercept and slope of the DV were signiﬁcant (m = 2.08,
p S 0.10, ms: 1.529, p S 0.01, mc= -0.210, p S 0.01) and three B paths were found to

be signiﬁcant. In M6, the intercept means were positively related (BilJz = 0.480, p S
0.05), the curvature factor mean from the successful adoption of innovations to the slope

of R&D proﬁciencies was positive (Belg = 4.966, p S 0.10), and the slope factor mean

of the antecedent positively affected the curvature mean of the DV (le,c2 = 0.413, p S

0.10). Thus, the intercept, slope and curvature factor means of the DV were affected by
the growing explanatory variable.
An interesting effect found in M6 is that in the univariate analysis, the mean

quadratic factor was insigniﬁcant for R&D proﬁciencies (-0.059, n.s.), but when

 

4 High RMSEAs appear to be characteristic of many LGC models. Researchers have questioned the
appropriateness of ﬁt indices in the context of linear growth curve models, (c.f. Kaplan 1998, SEMNET
discussion, 12/21/98). High RMSEAs may not be indicative of poor ﬁt. Ale may be better indicators of
best ﬁtting models.

153

regressed upon the successful adoption of innovations the mean becomes highly
signiﬁcant (-0.210, p S 0.01). This suggests that the R&D proﬁciencies follow a
nonlinear trajectory when affected by the successful adoption of innovations.

Both H8 and H9 were supported. For M8, the resulting mean intercept for the
DV, NP program success was not signiﬁcantly different from zero (scale —3 to +3) and
the slope mean was 0.396, p S 0.01 (see Table 5.13). For M9, mean intercept for the DV
was 0.746, p S 0.01, and the slope mean was 0.313 p S 0.05.

In testing the B paths, the initial starting point for marketing proﬁciencies was
related to the initial starting point reported for NP program performance (M8: Bil,i2 =

0.235, p S 0.10). The greater the proﬁciencies at t = 1, the greater the NP program

performance reported at time 1. In both M8 and M9, the slopes between the LGCs were
positive and signiﬁcant (M8: B5152 = 0.419, p S 0.05; M9: B31952 = 0.356, p S 0.05).

This indicates a positive relationship whereby the faster that proﬁciencies grow (or
decay), the faster the growth (or decay) in NP program success. It is interesting to note
that in the univariate analysis, NP program success’s slope mean was 0.537 (p S 0.05).
However, when regressed upon marketing proﬁciencies the slope decreased from 0.537
(p S 0.05) to 0.396 (p S 0.01) and when regressed upon R&D proﬁciencies the slope
decreased to 0.313 (p S 0.05). This suggests slower grth when the effects of
proﬁciencies are considered

Although M10 found evidence to support a nonlinear relationship between NP
program success and adaptive capacity, H10 was not supported. H10 proposed an
inverted relationship where as NP success increases, adaptive capacity decreases. This

growth trajectory was not found. Hypothesis 11 was also not supported as it followed an

154

inverted relationship that was not proposed! Even though the hypotheses were not
supported, the M10 and M11 demonstrate interesting characteristics. For both adaptive
capacity and slack resources, univariate LGCs indicate linear grth trajectories for these
variables. However, when regressed upon NP program success, they both take on a
quadratic grth trajectory where a peak is reached and a shift in trajectory direction

occurs.

In M10 the intercept, slope, and shape means were all signiﬁcant (mi = 4.834, p S

0.01, ms: 0.417, p S 0.01, mc = -0.082, p S 0.05), see Table 5.13. M10 showed a
positive relationship between the intercepts of NP program performance and adaptive
capacity. The higher the initial status of NP program success, the higher the reported
intercept for adaptive capacity (Bmz = 0.458, p S 0.05), see Table 5.11. A negative
relationship exists between the initial status of NP program success and the slope of
adaptive capacity (Bug = -0.154, p S 0.10). This indicates that higher initial levels of

NP program success reported are associated with less steep (but still positively
increasing) growth trajectories in adaptive capacity over time. This negative relation
does not indicate a decrease in adaptive capacity over time but instead implies that higher
initial levels of NP program success are associated with smaller rates of positive change
in adaptive capacity.

As a DV, slack resources showed an inverted s-shaped relationship (M11). It has

a positive intercept (m = 4.147, p S 0.01), a negative slope (mS = -0.469, p S 0.01), and

a negative mean curvature (mc = -0.151, p S 0.01). A negative relationship exists

between the intercept on NP program success and its initial status on slack resources

155

(anz = -0.654, p S 0.01). The higher the initial status reported for NP success, the lower

the reported availability of slack resources. Yet, a positive relationship exists between

the rates of change between these two variables. The greater the rates of change in NP

program success, the greater the positive rate of change in slack resources (B5152 = 0.526,

p S 0.01). A positive relationship was also found on the B coefﬁcient linking the
intercept factor of NP program success and the shape factor of slack resources (Bus; =
0.085, p S 0.01). M6 was the only other model that showed signiﬁcant paths to all three

latent factors of the dependent variable.

5.7 Conclusions

Despite only ﬁnding support for ﬁve of the eleven hypotheses many of the
concepts introduced in Chapters 1 and 2 were supported. First, although adaptive
capacity was not found to be an explanatory variable to the successful adoption of
innovations, it was found to be affected by NP program performance. However, a
positive relationship was revealed as opposed to the negative relationship proposed.
Extant literature has suggested that as ﬁrms become more successful, they can become
more rigid in structure (Miller, 1990; Leonard-Barton, 1992; Christensen, 1997). This
was not found in the time frame studied here. Adaptive capacity followed a positively
growing quadratic trajectory and NP program success a positive linearly growing
trajectory when regressed upon each other. The bivariate analysis showed that as the rate
of change of NP program success increased, the rate of change in adaptive capacity also
grew. NP managers may become more ﬂexible and adaptive to changes in the

environment when NP success rates are increasing. Firms may see their success due to

156

addressing the changing needs of the marketplace and continue to open themselves to
meeting market requirements.

Secondly, support was found for slack resources and a marketing-R&D
integration as explanatory variables for the successful adoption of innovations into the
NP process. Greater rates of change in marketing-R&D integration are associated with
greater positive rates of change in the adoption of innovations. This suggests that the
adoption of innovations within a ﬁrm may possibly be more successfully integrated into
the ﬁrm when both the marketing and R&D/engineering departments are involved.
When marketing issues become part of the evaluation process for the adoption of
innovations the potential for its successful implementation within the ﬁrm increases.

In regards to operationalizing the construct, ‘marketing-R&D integration’, it is
important for researchers to recognize that many technologically oriented ﬁrms do not
have formal marketing departments or if they do, marketing personnel may not interact
with engineering personnel. This does not mean that the engineering department does not
consider marketing issues during the NPD process. Measurements for this construct
should reﬂect this situation.

Thirdly, only at t=1 was a relationship found between slack resources and
innovation adoption success. The negative relationship observed is consistent with the
notion of 'slack ' resources allowing greater failures in the innovation process (Levinthal
and March, 1981). Slack resources are necessary to support failures in product
development that are inherent in engineering driven ﬁrms. However, this study also

found that an unlimited supply of resources will not necessarily lead to wildly successful

157

adopted innovations as success appears to reach a saturation point as the observed
quadratic trajectory suggests.

Lastly, this study provides empirical evidence that the adoption of innovations can
be used indirectly as a strategic opportunity for a ﬁrm to build competences in the NP
process. As a predictor variable, the successful adoption of innovations was found to
have a positive effect on exploitation of technological knowledge (M4) and on R&D
proﬁciencies (M6). Additionally, both knowledge exploitation and R&D proﬁciencies
were found to be positively related to NP program performance success (M7 and M9).
The adoption of innovations may be an overlooked strategy by NP managers (or at least
NP researchers) for driving the success of the NP program. It is important to make the
distinction that it is the knowledge embedded in the adopted innovation that can become
embodied as successful new products. The adoption of innovations into a ﬁrm does not
itself drive NP success. Firms may beneﬁt from actively evaluating the innovations they
adopt to ensure that it can effectively integrate the innovation into the NP process.
Knowledgeable personnel as well as an 'innovation champion’ may help to attain the
highest knowledge yield from adopted innovations.

The major goals of study 2 were to gain a better understanding of the dynamic NP
process and to determine if major components of the NP process were nonlinearly
related. The result of this study show that the adoption of innovations may be one way to
build NP process proﬁciencies, and indeed when time is taken into consideration, the

relationships are nonlinear.

158

CHAPTER 6

CONTRIBUTIONS, LIMITATIONS AND FUTURE RESEARCH

6.1 Introduction

The primary objectives of this dissertation were two-fold: (l) to advance the
practitioner’s understanding of how adopted innovations can assist NP managers 'in
driving the success of inventions, and (2) to advance the methodological framework of
how the NP development process of technological ﬁrms can be studied as a dynamical
system. A secondary question was to determine if ‘catalytic’ innovations signiﬁcantly
increase the success of an NP program. These questions and several others were set forth
in Chapter 1. The answer to these questions a summary of the ﬁndings from studies 1
and 2 are presented in this chapter. Study 1, detailed in Chapter 4, used case studies and
system dynamics modeling to test the relationships between the constructs as deﬁned in
Chapter 2. Study 2, detailed in Chapter 5, used a longitudinal repeated measures design
to test the relationships between constructs using linear grth curve methodology in

EQS.

6.2 Research Questions Addressed and Contributions

The ﬁrst set of research questions presented in Chapter 1 addressed practitioner
issues.
PQl. Can the adoption of innovations into the firm contribute to the success of new
products?
(i) Can the adoption of innovations act as a catalyst to the successful

invention of new products?

159

(ii) Does the successful adoption of technological innovations lead to an
increase in the number of new product development projects undertaken by
the firm? 1
(iii) What resources and types of organizational cultures (adaptive/rigid,
integrated/decentralized marketing-R&D departments, scarce/unlimited
resources) promote the successful adoption of innovations into the firm?
The intent of this dissertation was to investigate the infusion effects of the successful
adoption of innovations on the NP process. Studies 1 and 2 both addressed the overall
question presented in PQl. It was found in both studies that the successful adoption of
innovations into the ﬁrm has an indirect effect on the success of a NP program.
Innovations successfully adopted into an organization have the effect of increasing a
ﬁrm’s competences in technology knowledge utilization (exploitation) and R&D
proﬁciencies. These studies also reveal a positive relationship between technology
knowledge exploitation and R&D proﬁciencies as explanatory variables to NP program
success. Thus, indirectly the successful adoption of innovation can drive NP success by
helping organizations to better utilize their technology knowledge competences.

An important relationship not originally proposed for study neither in the research
questions nor in the research model, was how does the adoption of innovations affect the
ﬁrm’s knowledge base. As no studies have previously reported on the eﬂects of adopting
innovations for the purpose of improving the NPD process, this relationship did not
become evident until after the case studies had been conducted. The case studies
revealed that ﬁrms accumulate knowledge banks that are utilized to develop new

products. Without the proper knowledge stock in place, the ﬁrm cannot develop

160

competitive products for the marketplace. Just one method of continually renewing the
knowledge stock is through the adoption of innovations. Although March (1991) and
Levinthal and March (1993) allude to the need to continually renew a knowledge base,
these concepts have not previously been applied to the NP process or to the role that the
adoption of innovations can play in this knowledge maintenance process.

The notion that technological knowledge requires ‘management’ by innovation-
centric organizations was also revealed. As knowledge becomes outdated it may need to
be ‘shelved’ as its usefulness in creating competitive products declines. However,
utilizing ‘new’ knowledge may not always be optimal, as a ﬁrm may not have the
opportunity to build engineering and manufacturing competences if it is continually
evaluating and integrating new knowledge into the organization. The management of
‘new’ and ‘old’ knowledge and exploitation of aging knowledge may be important
strategies for NP managers. It is possible that radical inventions require newer
knowledge banks compared to the development of incremental inventions that can be
developed using older knowledge. Knowledge management in the NP process is an area
for future research.

Sub-issue (i) presented in PQl concerns the role of the ‘catalytic’ innovations in
the NP process. The ﬁrst question that must be asked is whether ‘catalytic’ innovations
actually exist. ‘Catalytic’ innovations were previously deﬁned in Chapter 1 as
innovations adopted into the organization that act as a catalyst to develop new inventions
not previously planned prior the adoption of the innovation. Marginal support was found
for the existence of ‘catalytic’ innovation in the sample population studied. Respondents

of the longitudinal study were asked to identify three innovations that had been adopted

161

into their organizations within the past 36 months. Respondents were then asked to
respond to the following question: “This innovation has allowed us to build new products
we had not previously considered prior to its adoption”. At time 1, the mean response to
this question was 5.47 (0' = 1.24) based on a Likert scale of l to 7 with 7 being ‘strongly
agree’. Twelve of the 49 (24.5%) respondents rated this item at the highest scale. In the
system dynamics model, sensitivity analyses found little evidence of ‘catalytic’
innovations effects NP program success. Together these ﬁndings show some support for
the existence of catalytic innovations. Yet they should not be taken as conclusive
evidence that ‘catalytic’ innovations do or do not exist. The ﬁrms in the study may not
have reported on innovations that have signiﬁcantly altered their NP strategy or may not
be able to recognize these types of innovations. More likely, ‘catalytic’ innovations, like
radical innovations are rare (Garcia and Calantone, 2002), and are not easily observable.
Historical studies on the development processes of radical innovations may be a better
method for studying catalytic innovations. This is an area for future research.

Sub-issue (ii) of PQl questions whether the successful adoption of innovations
can lead to an increase in the number of new product projects undertaken by the ﬁrm. As
previously discussed, the renewal of a ﬁrm’s technology stock is an outcome of the
successful adoption of innovations. Consequently, the greater the ﬁrm’s technology
stock, the greater its ability to undertake more new NP projects. The availability of
knowledge stock does not guarantee more NP projects; it just enables the ﬁrm to develop
more new products if other resources, such as personnel and capital, are available.

Sub-issue (iii) addresses which factors drive the successful adoption of

innovations. The suggested predictors were adaptive capacity, marketing-R&D

162

integration, and slack resources. Both studies 1 and 2 found no support that a ﬁrm’s
adaptive capacity affects it ability to successfully integrate an adopted innovation into the
ﬁrm. The case studies revealed that adaptive capacity might in fact be a moderator to
knowledge utilization. Firms that can more easily adapt to dynamic environments are
more likely to make use of ‘new ‘technological knowledge compared to ﬁrms that have
lower adaptive capacities. Subsequently, adaptive capacity was modeled as an ampliﬁer
to the utilization of ‘old and ‘new’ knowledge stocks in the SD model. Sensitivity
analyses showed that adaptive capacity had relatively small effects on both success of
adopting innovations and NP program success.

Study 2 was consistent with Study 1 in revealing that adaptive capacity is
probably not an explanatory variable to the adoption of innovations. Study 2 did show a
positive relationship between a ﬁrm’s NP program succss and it adaptive capacity. The
greater the NP success reported by a ﬁrm, the greater its rate of growth over time in
adaptive capacity. This was contradictory to the proposed relationship suggesting that
the more successful a ﬁrm becomes, the more rigid it becomes to new ideas (Leonard-
Barton, 1988). As adaptive capacity is a new construct to marketing/N P literature,
further theoretical and empirical analyses are needed to better deﬁne its role in the NP
process.

The second explanatory variable studied was marketing-R&D integration. The
empirical results of the longitudinal study showed that the greater the growth in inter-
departmental integration, the greater the growth in the successful adoption of innovations.
The more consideration given to marketing issues in the NP process, the better the

assimilation of an innovation in to the ﬁrm. As previously noted in Chapter 5, marketing

163

and R&D integration does not necessarily mean a structured interaction between
departments. Many of the engineering companies in this study do not have formal
marketing departments, yet they consider marketing issues when developing new
products.

The third explanatory variable to the successful adoption of innovations, slack
resources, has been noted to be a necessary but not sufﬁcient criteria for allocating
resources to innovation activities (Cyert and March 1963, Rosner, 1968). The results of
the system dynamics model lent support for this theoretical concept. The number of
innovations adopted into the ﬁrm was highly sensitive to the amount of slack resources
available for these types of activities. A lack of resources resulted in fewer adopted
innovations, which resulted in fewer new products.

A different effect was found in study 2. The initial status of slack resources
reported at t=1 had a negative effect on the initial status of success in adopting
innovations at t=1. So, the greater the resources available, the less the success at
adopting innovations the was ﬁrm reported. These results add to the contradictory results
obtained from previous empirical studies on the role of slack resources in the adoption of
innovations (cf. Dewar and Dutton (1986) showed a positive relationship, and
Damanpour (1991) showed no relationship). The previous studies had looked at
individual innovations. This current study is unique from these previous studies in that it
considered the NPD process for more than one innovation and at the program level. This
wholistic approach may lead to the negative effect observed. On the program level,
greater slack resources allow the ﬁrm to be more innovative and adopt more innovations,

which ultimately lead to greater failures. Overall, even though greater slack resources

164

lead to more individual innovation successes, they lead to more failures when considered
at a program level. This negative relationship is supported by Levinthal and March
(1993) who contend that slack resources allow the ﬁrm more opportunities to fail in NP

development.

PQ2. How should firms structure themselves to increase the successful adoption of
innovations?

A summary of the ﬁndings from PQl will also answer PQ2. Technologically
oriented organizations that are interested in using the adoption of innovations as a
strategy for driving greater success in their NP program should focus on making
resources readily available to the NP process, encourage failures in NP design and
actively involve marketing in the NP process. An important ﬁnding from both studies
was that adopting innovations can increase technological knowledge competences.
However, the quadratic trajectories observed in study 2 suggest that peaks in performance
are often reached. The continual renewal of knowledge is important in the NP
development process. Firms that structure themselves to be open to new knowledge will
beneﬁt not only in their success of adopting innovations but also in NP program

performance.

PQ3. What effect does the successful adoption of innovations have on the lock-in
(core rigidities) and lock-out of knowledge?
This question was addressed in Study 1 using the system dynamics model. Lock-
in has been described as the phenomenon whereby a ﬁrm becomes so ingrained in a

particular technology and knowledge base that it is unable to adapt to changing market

165

conditions (Leonard-Barton, 1995). Leonard-Barton describes this as core competences
evolving into core rigidities. A similar phenomenon is lock-out of knowledge, which is
also closely related to ‘absorptive capacity’ (Cohen and Levinthal, 1990). Cohen and
Levinthal argue that the ability to evaluate and utilize new knowledge is a function of the
level of prior related knowledge. A ﬁrm must have an adequate foundation upon which
to build new knowledge. A ﬁrm may lock itself out of the ability to acquire new
knowledge for various reasons, which may include lack of appropriate personnel,
complacency in the current status (lock-in), or simply poor managerial insights.

The effects of lock-in and lock-out from a knowledge perspective were examined
in study 1. It was shown that ﬁrms that did not continually renew their knowledge stock
lock themselves into old technological knowledge and lock themselves out of the ability
to acquire new knowledge. A ﬁrm must have a knowledge foundation that is continually
updated in order to be able to assimilate the latest technological knowledge. March
(1991) referred to this as the balance between exploration and exploitation activities.
Firms that do not continually take on exploration activities lock themselves out of new
technology, whereby decreasing the ﬁrm’s competitive advantage in developing new
products (exploiting that knowledge base). The results of study 1 showed that ﬁrms that
do not continually rebuild their knowledge stock eventually lose all competitive
advantage and soon expire. As the adoption of innovations is one way of renewing the
knowledge stock, lock-in and lock-out can be avoided by the continual adoption of

innovations into the organization.

166

TQl. Does a relationship exist between the adoption of innovations and the
successful invention of new products in the technological ﬁrm? Are these
relationships nonlinear?

The ﬁrst part of TQl has been answered in responses to the previous questions. Strong

evidence from both studies 1 and 2 showed that a positive relationship, although indirect,

exists between the adoption of innovations and the successful invention of new products
in the technologically oriented ﬁrm. What has not previously been addressed is the
second part of TQl. System dynamics modeling (study 1) creates differential equations
to model the effects between constructs over time. Any nonlinear relationships are
explicitly modeled in. Thus, it is not an appropriate methodology with which to answer
this question.

Study 2 showed the existence of nonlinearities between many of the constructs
not just between the adoption of innovations and NP program succss. As the focus of the
study was on the change in variables over time, it is important to distinguish that time is
an important factor in this study. It was found that a variable’s growth trajectory over
time may take on a signiﬁcantly different picture when evaluated with respect to another
dynamic variable.

For example at time 1 of the study 2, slack resources and NP program success are
negatively correlated and highly signiﬁcant (r = -O.421, p S 0.005). At time 2 there is no
relationship (r = -0.057, p = 0.70) and at time 3 there is a marginally signiﬁcant positive
relationship (0.219, p S 0.10). Substantially different results about the relationship
between slack resources and NP program success can be obtained based on when the data

was collected. The linear/nonlinear growth curve analysis allows a more complete story

167

 

to be presented. Slack resources follows an inverted quadratic trajectory over time
(P‘intercept = 3.473, uslope = -0.573, premature = -0.089, all signiﬁcant at p S 0.05) when

regressed upon NP program success. Yet, a positive relationship exists between the rates
of change between these two variables. The faster NP program success changes over
time, the faster the change in slack! resources over time. During the data collection
period, slack resources happened to be decreasing over time; thus, the faster NP program
success grows, the faster slack resources become unavailable for NP activities. If slack
resources had been growing over time, the positive relationship would indicate more
resources become available as NP program success grows.

During the time of this study (July 2001-May 2002), the business environment for
the United States was extremely turbulent. Many ﬁrms were re-strategizing due to the
recession caused in part by the terrorist attack on New York City in September 2001.
Taking this environment into account, it does not appear unreasonable that even though
ﬁrms may have had increasing NP success, less slack resources were available for NP
projects. Slack resources were being dedicated to other areas of the organization in order
to weather the recession. The greater the success in new products meant that the NP
program could sustain itself without additional dedicated resources, and therefore slack
resources could be allocated to other programs in greater need.

Thus, the empirical analyses have found signiﬁcant support for the existence of

nonlinearities in the relationships studied.

168

TQ2. From a methodological perspective, does system dynamics modeling better
represent the evolving new product development process compared to more
traditional linear representations?

The intent of conducting two studies was to obtain two uniquely different perspectives of

the dynamical NP process that could not be obtained through cross-sectional research

designs. In hindsight, TQ2 is an inappropriate question, as no one methodology is

‘better’ than another. Each has its own advantages and disadvantages. As the limitations

of each methodology will be explained in the next section, the advantages of the two

methodologies in this study will be brieﬂy examined here.

System dynamics (SD) modeling’s primary advantage is its ability to simulate
how a process evolves over time and how policies and decision rule changes may affect
the goals inherent in the process. It allows the long-term impact of policy changes to be
studied in a laboratory environment. Insights can be obtained that might not have been
readily evident from methodological analyses that only allow short-term time periods.

SD modeling allows sensitivity analyses to determine important effects of
changing, eliminating or adding a construct. Cross-sectional data collection and even
longitudinal data collection does not allow for the collection of new data should a
interesting concept appear during the course of the study. This was notable in this
dissertation as knowledge was recognized as playing a more important role than
originally modeled. As the longitudinal data collection had already been started it was
not practical to try to collect the additional data required to evaluate this construct.
Modeling a knowledge stock in the SD model was a much easier method of gaining a

better understanding of the role of knowledge in the NP process.

169

SD modeling also provides an excellent opportunity for researchers to ‘test out’
the rationality and validity of their models prior to collecting data in the ﬁeld.
Simulations can test the internal consistency of a theory by ensuring that the behavior of
the system theorized can in fact be generated by the underlying assumptions outlined (cf.
Sastry, 1997). SD modeling also provides a 'laboratory' environment in which to
discover implications of theoretical assumptions that may not be intuitively obvious
(Sterman, 2000).

Additionally, current events have reinforced the increasing difﬁculties that
researchers have in the collection of data through mail surveys. ‘Testing’ a model prior
to collecting expensive primary data is extremely beneﬁcial. Simulation techniques
provide an excellent method for researchers to expand the knowledge base without
gathering large quantities of data. Validation of the model still needs to occur with real
world concurrence, thus simulation modeling can not eliminate the need to conduct
ﬁeldwork.

Longitudinal data also has its beneﬁts despite the difﬁculty in obtaining balanced
data sets. As previously demonstrated it can show a much more complete picture of a
evolving process compared to cross-sectional data. Even though it usually is conducted
under a much shorter time span than simulation models, it provides empirical support for
what is actually occurring within a ﬁrm’s NP process over a speciﬁc time frame. Chapter
5 provided an example on how to use linear/nonlinear growth models to analyze the
dynamic NP process.

A different set of research questions can be asked when using longitudinal designs

that cannot be addressed with cross-sectional data or even with system dynamics

170

modeling. As Chapter 4 showed, the hypotheses introduced in Chapter 2 were not
relevant to a system dynamics modeling approach. The hypotheses, which focused on
the nonlinearities in the relationships between the constructs, also could not have been
answered using cross-sectional data. When change itself is the object of study, the only
way to investigate it is by using longitudinal data. Longitudinal studies allow
nonlinearities between relationships to be tested.

A motivating factor for conducting this dissertation was to gain a better
understanding of how ﬁrms evolve and adapt their organizational structure over time in
order to effectively adopt innovations and produce inventions. To study the evolutionary
nature of this process, longitudinal data must be used. Although simulation techniques
have the advantage of being able to model extended time frames, actual data collection
provides a picture of the changes actually taking place in the NP process of innovation-
centric ﬁrms.

TQ3. Is adaptive capacity an antecedent to the successful adoption of innovations
and how should it be operationalized?

Both studies showed little support for any relationship between adaptive capacity and the

successful adoption of innovations. In study 1, adaptive capacity was modeled as an

ampliﬁer to marketing proﬁciencies, R&D proﬁciencies and technology knowledge

utilization (exploitation). In conducting sensitivity analyses, it was found to have a

limited effect on these constructs.

In study 2, adaptive capacity was found to be related to a ﬁrm’s NP program
success. The higher the initial status of NP program success, the higher was the reported

adaptive capacity. A negative relationship exists between the initial status of NP program

171

success and the slope of adaptive capacity (B5152 = -0.154). This indicates that higher
initial levels of NP program success reported are associated with less steep (but still
positively increasing) grth trajectories in adaptive capacity over time. This negative
relation does not indicate a decrease in adaptive capacity over time but instead implies
that higher initial levels of NP program success are associated with smaller rates of
positive change in adaptive capacity. This ﬁnding supports the idea that the greater a
ﬁrm's NP program success, the less ﬂexible it becomes in being able to adapt to its
environment. NP success can dampen the grth rate of adaptive capacity.

The relationship observed may have been a result of how the construct was
operationalized. It was measured with four items (only three passed CFA analysis). Two
were based on Jaworski and Kohli’s (1993) market orientation measures and two were
created for this study. The Cronbach alphas for the three measures were strong (t=1 a =
0.827; t=2 a = 0.761; t=3 a = 0.822) and an EF A revealed one factor. The three
measures retained focused on a ﬁrm’s ﬂexibility in responding to customer’s needs and
competitors infringement on customers. Future research should retest the validity of the
measures used to study this concept of a ﬁrm’s ability, willingness and ﬂexibility to adapt
to its environment. Adaptive capacity does appear to be related to a ﬁrm’s marketing

orientation strategy and its NP program success.

6.2.1 Summary to Research Questions

These six research questions summarize the results of the two different
methodologies used to study the infusion effects of the adoption of innovations on the
evolving new product development process. System dynamics modeling emphasized the

policy and decision making rules of the process whereas the longitudinal study examined

172

the nonlinearities in the relationships of NP process attributes with respect to time. Each
study provides unique contributions to the understanding of how the NP process changes
over time. However, both studies were conducted within a limited scope, which will be

discussed in the next section.

6.3 Limitations and Future Research

As previously noted, the two methodologies used in this dissertation have their
distinct advantages, but they also have distinct disadvantages. System dynamics
modeling simulates a process as established by the modeler. As the SD model developed
was based on case studies of two small engineering ﬁrms (employees < 100), the process
simulated reﬂects but a very small subset of the population of technologically oriented
ﬁrms. Both of these ﬁrms did not have a marketing department that was actively
involved in the NP development process. The role of marketing-R&D integration may
have been more pronounced in the model if organizations that fully integrate these two
departments had been studied. Although SD models should be veriﬁed with outside
sources, they still reﬂect the perspective of the modelers.

Additionally, in building the SD model, boundaries were set that reﬂected only
endogenous issues to the NP program. Exogenous issues such as environmental
turbulence, competition and rate of technology change were not modeled into the system.
These important factors can change the results of the decision making process within the
NP process. One study, looking at the effects of competition, showed that the policies
regarding resource allocation to research and development activities can be signiﬁcantly
altered based on competitive pressures (Garcia, et al., 2002). Effects of these types of

exogenous factors should be modeled in future SD model based research.

173

 

Limitations to the longitudinal data analysis are similar to those for the system
dynamics model. The sample was drawn from a unique set of small engineering driven
companies (employees < 1000). These companies are all privately owned by a United
States based parent company. Although allowed to maintain autonomy in many
decisions, these companies must still report to corporate headquarters. Many of the
companies also do not have a documented NP process in place. Fundamentally different
results may have been obtained if the sample population had been large companies with
more formalized NP programs.

Several of the companies participating in the longitudinal study were
internationally located. The sample size was too small to conduct a multigroup analysis
on the differences between the internationally based ﬁrms and the domestic located
companies. The sample size (n = 49) also posed problems in obtaining goodness-of-ﬁt
indices for the conﬁrmatory factor analysis and the univariate linear growth curves. A
larger sample size would have allowed better evaluation of the goodness-of-ﬁt of the
model.

Linear growth curves (LGCs) require balanced data sets across the time frame
studied. As longitudinal data is very difﬁcult to obtain, this requirement of LGC
necessitates missing data imputation. There is no ideal method for imputing data and
future research should investigate the effects of using different imputation methods on the
ﬁnal results. Additionally, with data taken at only three time points, limitations exist on
the types of growth trajectories that can be tested in LGC modeling. More data points

across time would allow a more robust testing of the growth trajectories.

174

Utilizing SEM to evaluate linear growth curves is a relatively new methodological
approach. It has primarily been used in psychology studies on individual growth rates.
Curran (2000) writes that "it is important to note that the more complicated these grth
models become, the greater the variety of ways there are to approach the model building
strategy. There is not necessarily a right or wrong way to build and probe these models,
but whatever steps are taken should be clearly articulated, well justiﬁed, and closely
guided by the theoretical questions of interest" (pg. 27). This dissertation has used this
guideline in building the models investigated. Future analyses of the use and
interpretation of linear grth curves in modeling marketing/N P development issues will

lead to a better understanding of this methodology.

6.3.1 Other Future Research

In addition to the suggestions for future research already noted, the results of this
dissertation have facilitated thoughts for other research directions. The role of
knowledge management in the new product development process, especially the use of
‘old’ and ‘new’ knowledge for building competitive products, should be further explored.
The president of one of the case studies recently conﬁned in me that he had recently
realized the important role that older technology can play in the development of new
products. He has vowed to investigate how he can use technology that had previously
been shelved to revitalize his new product portfolio. Exploration and experimentation
with newer technology would continue within the ﬁrm but would now take on different
perspectives.

A potentially interesting study that developed from the system dynamic model

would be to focus on resource allocation between exploration (research) and exploitation

175

(development) activities. The outcome of the SD modeling provided some insights on
how to balance these two distinctively different objectives. Future research in this area
would help NP managers better understand when and how to allocate resources to R&D
projects.

This dissertation has only alluded to the role of radical innovations in the NP
process. However, this dissertation was motivated by an interest in learning how radical
innovations ‘emerge’. One thought was that catalytic innovations can cause a ﬁrm to
take on a new perspective in designing new products that may result in radical
innovations. This dissertation found little support for this proposition. This in itself is
new knowledge gained, which can be used in future studies on the emergence of radical
innovations.

It would be remiss not to add that future research should continue to look at the
ﬁrm as an evolving system. System dynamics modeling and linear growth curves are just
two methodologies for studying how processes change over time. This is an area rich

with future research opportunities.

176

APPENDICES

APPENDIX A

APPENDIX B

APPENDIX C

177

APPENDIX A

Case Studies Supporting Documents

178

Table A] Case Companies’ Descriptions

 

 

 

 

 

 

 

 

 

 

 

 

 

Co. A Co. B
Primary Business Semiconductor Industry Welding Machines
Suppliers
Number of Employees: 75 15
Revenues: less than 20 million less than 5 million
Age: 30 10
Location: Massachusetts, USA New Jersey, USA
Sales Regions: US, Korea, Europe, US, Canada
Japan
Engineering Staff: 10 2 + outsourcing
Sales & Marketing 5 2
Staff:
Formal NPD Process: No No

 

179

 

Guidelines for Semi-Structured Interview for Study on the Infusion Effects of

Technological Innovations on the New Product Development Process

Goal of research: The intent the of the interviews was to determine how technology-
based ﬁrms use new-to-the-ﬁrm technologies to invent new products. Why innovations
are adopted and how they are utilized within a ﬁrm were key points to understand. The
qualitative data gathered was used as input to the written survey and a systems dynamics

model.

Methodology: Interviews were arranged with the President, Vice President, Marketing
Manager, the Chief Engineer and another Engineering Manager at the Case Firml. At
this company, both the President and Vice President have engineering responsibilities
with new products. Interviews were also arranged with the President and Marketing
Manager of Case Firm2. Each interview took 2-3 hours. In the semi-structured interview
not all the questions were asked of each individual. An overlap in questions was used in
order to look for any discrepancies or unusual practices within the NPD program that

might need further clariﬁcation.

For systems dynamics modeling, it is important to focus on how the process changes over
time. In order to look at the dynamic relationships, ‘graphic modeling’ exercises1 were
conducted to focus on the important relationships within the model proposed in the
dissertation. Approximately 1-2 hours were spent on the interview questions and 05-10

hours were spent on the modeling exercises.

 

' David Lounsbury, a student of Dr. Levine, was kind enough to share his ‘graphical modeling’
exercises.

180

Interview Procedures:

Each interview started by explaining the Human Subjects in Research policy and
obtaining the signature of interviewee indicating that the interview process is being
recorded and may be stopped at any point upon request of the interviewee. A copy of

the signed form was given to the interviewee in the event they wished to contact

UCIRHS at a later date.

The discussion was started by brieﬂy describing the primary interests of this study is
to look at the new product development process of the ﬁrm. The goal of the ﬁrst part
of the interview was to look at the overall new product process. The second half of
the interview put more emphasis on the adoption of new technologies as it pertains to
systems dynamic modeling, which looks at how the relationship between factors

important to the model change over time.

An effort was made to keep the interview conversational in order to listen to the
“story” the interviewee had regarding their involvement in new product development.
It was important to capture both successful adoption events as well as failed adoption
events and how the NPD process was affected by the introduction of these new

technologies.

If time did not allow for both the semi-structured interview and the systems dynamics
exercises, the worksheets were left with the respondent and returned at a later date or
another interview time can be scheduled. Again, not all questions of either Part 1 or

Part 2 were asked of all respondents.

181

Part 1: Semi-structured Interview Questions

. After the lead in and the 15‘ question, the interview did not follow any particular
schedule. An attempt was made to focus on as many of the questions within this list
in the allotted time, but in no particular order. The ﬂow of the conversation was
dictate in whatever sequence the questions were asked. Additional questions not

listed here may have been added if necessary. The list of questions asked were:

1. Where do ideas come from regarding new products? Can you give me an
example?
2. Describe a new technology brought in-house within the last year to help build

new products which you consider a success. How did it help to build a new

product? Can you describe the process?
3. Can you describe the development process this product took? For example, tell
me what happened aﬁer this idea was presented to the company? What happened

next?

4. Was this a completely new product idea for the company or a revision of an

existing one?
5. Who was involved in bringing this technology in house? What role did they play?
6. How did it change the way you did ﬁrings? Changes in the engineering
department? Changes in marketing? Changes in manufacturing? How did its

importance change over time?

7. What did it allow you to do differently that you may not have been about to do

previously?

182

 

10.

ll.

12.

l3.

14.

15.

16.

17.

What is your deﬁnition of the successful adoption of an a new technology into the

ﬁrm? (Assimilation, Implementation, Efﬁcacy)?

What problems did you encounter in using this new technology?

What are the biggest challenges you face when building new products? How has

this changed over the last year? Last ﬁve years? Is this cyclical?

What different types of challenges do you experience from a totally new product

idea compared to an incrementally new product idea?

You described a new technology which resulted in a successful (failed) new
product. Could you please describe a new technology brought in house that did

not (did) lead to a successful new product?

You described two technologies that were brought in house, one lead to a
successful new product and one did not.. What criteria requirements do you have
for new technologies? Deﬁnitive project use? Budget constraint? Personnel

constraint?

How do you deﬁne success (failure) of a new product?

Could you draw me a ﬂow chart or time line of the typical NPD process in your

organization?
In this ﬂow chart, I see you have (not) included reference to marketing
involvement. When does marketing get involved in the new product development

process? [H3]

When does marketing get involved with product development? How often do

team members interact? Does this change over time?

183

 

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

When would marketing get involved in determining which new technologies are
adopted by the ﬁrm?

How does involving marketing impact how a new technology is used in house?
Does this change based on the success or failure of a new product Deleted? Can
you provide an example? (un-modeled path NPD program success —>
R&D/MKTG Integration).

How does this change with failed new products? Can you give me an example?

How do engineers monitor technological changes in your industry? trade shows?

publications? customers? suppliers [Adaptive Capacity]

How do the engineers keep current with new technology? attend seminars?

obtain degrees? etc. [Adaptive Capacity]

How do these activities (trade shows, etc.) impact new product development

activities?
What % of your existing product line is protected by patents? How many new
products have you introduced in the last year that are protected by patents?

[Adaptive Capacity]

How does marketing monitor the market place? trade shows? marketing research

activities, etc. [Adaptive Capacity]

How does marketing keep current with new marketing ideas? seminars? other

education? [Adaptive Capacity]

184

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

How do these marketing activities (trade shows, etc.) impact new product

development activities?

Has there been a recent year where the ﬁrm was very successful in introducing

new products? (not all products — just new products)

Why was this a successful year? How did the success impact the ﬁrm? Did you
in/decrease R&D activities & spending? in/de marketing activities & spending?
[hl 1]

Did this success lead to a more structured or more ﬂexible organization?

Can you recall a recent year where you did not introduce many new products?

How did this less than successful year impact the ﬁrm?

How did it affect R&D activities and spending? Marketing activities & spending?
Was anyone let go? Were any experts or other specialists hired? Were any perks

taken away? Did conference attendance in/decrease?

What effects did the failure have on the NPD process? More rigid requirements?
More ﬂexibility to explore new technologies? [H10]

Has there been a time you missed a market opportunity for a new product? What

happened? Has anything been done to avoid this situation in the future?
How are resources assigned to new technology decisions. Are there constraints
on how much is allocated? Who makes the decisions on what new technologies

get funding or resources?

How does this change over the life of a product - from when the idea is ﬁrst

conceived until it is taken out of the market?

185

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

How much time does the ﬁrm spend in R&D on improving existing technologies?

How much time is spent on evaluating new technologies for future use?

What determines how many new products are undertaken yearly?

How does your ﬁrm respond to new competitive offerings, price changes, or new

technology? How long does it take the ﬁrm to react?

What determines when you bring a new technology in house? Typically how

long is a technology used within the company?

What happens when your company has too many projects? What caused this

situation? [H4]

Did this effect the R&D activities in any way? [H5]

How about the marketing activities? [H6]

What role do specialists or product champions play in ensuring the success of a

new technology introduced into your company?

What are the most important internal organization issues that contribute to the

success of new products in your organization?

How would you describe your ﬁrms NP program success compared to your

competitors.
Are there other companies that you might share R&D information or new

technology ideas with but do not necessarily result in new products? Would you

please list those companies. Are these old relationship or a new relationships?

186

Part 2 : Systems Dynamics Modeling

Part two of the interviews introduced the idea of nonlinear relationship in
different behaviors of the NP process. The following workbook was shared with the
respondent. They were asked to draw in the relationships as they experienced them in the
workplace. The drawings were usually accompanied by discussion as the respondent
frequently had questions about how to complete this workbook.

When a relationship did not make sense or was not relevant to the respondent, it

was skipped.

187

1. Relationships in the NP Process

Depicting the reference mode of a variable for a given time horizon is considered
essential to the development of all system dynamics model-building (Sterman, 2000).
The reference mode for most dynamic variables can be depicted using a relatively small
number of distinct patterns of behavior. The most common modes of dynamic behavior

are shown below.

Figure A.l Common modes of Linear Behavior over Time

 

 

 

4-----------_----_---__-__ L‘
‘
~
‘
‘
s
‘
~
\
~
~
~
\
~
~
‘
~
‘
§
§
‘
‘I
~.
I

 

 

 

 

 

 

 

 

 

188

 

£266 uoamcm-m

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mmam=ou ucm 602390 605555? 5390
x. m. ...e
. ...x ....~
ea... s.\ . \I.. .s....
/ SN :.:. . .x ..
...z... . .s. tittoitztiit.‘ .1.
963$ .30 co=m___om0 £265 _mzcocoaxm
at... ..
... i \x .. a . ..\
t. / x. s ..x.
lento-cola oooooo . \ a.» ...s I N.
..... t. .. .t . S. , ... ..S
..... . a .. 2 .. ..
w .\ .. x ...f x \
. «r. as“ a ...K r/ .... .
..... ......\.\ ...z\ r .. S...
S
a
.2
n
.2.
S
08E. B>o STE—Em Qua—3.50 mo 8822 208800 N.< "mama

 

 

189

 

SLACK RESOURCES

SLACK RESOURCES

VISUALIZE Slack Resources

Consider Slack Resources made available for the adoption of new technologies. Sketch a
reference mode that you believe best describes its shape over the speciﬁed time horizon
in the real world. Refer to the Figures Al & A2 (common modes of behavior) for
possible ways of sketching the reference mode. Keep in mind that these modes of
behavior may be combined with each other. Why do you believe the reference mode for

Slack Resources is shaped this way?

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Successful Adoption
Initial concept Discontinuance
TIME
Failed Adoption
Initial concept Discontinuance
TIME

190

 

Now, using the diagram template below, write in the names of variables that you believe

impact Slack Resources over time and that — in turn — Slack Resources impacts over time.

Indicate how you believe these variables are associated with a postive change in Slack
Resources. Circle (+) if a positive change in a variable you identiﬁed would be expected
to cause a positive change in Slack Resources. Circle (-) if a negative change in a variable

you identiﬁed would be expected to cause a positive change in Slack Resources.

Note that phrases such as ‘an increase in,’ ‘growth of,’ ‘reinforcement of,’ ‘higher,’ or
‘intensiﬁcation of’ can be substituted for ‘a positive change in’ and that phrases such as
‘a decrease in,’ ‘depletion of, ’dampening of,’ ‘lower,’ or ‘weakening of can be

substituted for ‘a negative change in.’

 

Slack CD/"\+
6 Resources ..

 

 

 

191

 

VISUALIZE R&D-Marketing Integration
Consider the dynamics of R&D-Marketing Integration for the adoption of new

technologies. Sketch a reference mode that you believe best describes its shape over the
speciﬁed time horizon in the real world. Refer to the Figures A.l & A.2 (common modes
of behavior) for possible ways of sketching the reference mode. Keep in mind that these
modes of behavior may be combined with each other. Why do you believe the reference

mode for R&D-Marketing Integration is shaped this way?

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Successful Adoption
Z
SE
S
(3
LL]
[_.
E
O
E
E
E
Q
08
04
Initial concept Discontinuance
TIME
Failed Ado tion
Z
2
E
O
LLI
I—
E
(D
E
i
<1
2
Q
08
ad
Initial concept Discontinuance
TIME

192

 

Now, using the diagram template below, write in the names of variables that you believe
impact R&D-Marketing Integration over time and that — in turn — R&D-Marketing

Integration impacts over time.

Indicate how you believe these variables are associated with a postive change in R&D—
Marketing Integration. Circle (+) if a positive change in a variable you identiﬁed would
be expected to cause a positive change in R&D-Marketing Integration. Circle (-) if a
negative change in a variable you identiﬁed would be expected to cause a positive change

in R&D-Marketing Integration.

Note that phrases such as ‘an increase in,’ ‘growth of,’ ‘reinforcement of,’ ‘higher,’ or
‘intensiﬁcation of’ can be substituted for ‘a positive change in’ and that phrases such as
‘a decrease in,’ ‘depletion of, ’dampening of,’ ‘lower,’ or ‘weakening of’ can be

substituted for ‘a negative change in.’

 

R&D- G)/’\C
6) Marketing -

/ Integration 6)

 

 

 

193

VISUALIZE Organizational Flexibility (Adaptive Capacity)

Consider the dynamics of Organizational Flexibility for the adoption of new
technologies. Sketch a reference mode that you believe best describes its shape over the
speciﬁed time horizon in the real world. Refer to the Figures A.1 & A.2 (common modes
of behavior) for possible ways of sketching the reference mode. Keep in mind that these
modes of behavior may be combined with each other. Why do you believe the reference

mode for Organizational Flexibility is shaped this way?

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Successful Adoption
>4
:1
:i
m
E
LL]
.4
LL.
Z
9
I—
<.'
E
Z
<
U
ad
0
Initial concept Discontinuance
TIME
Failed Adoption
>~
:
d
E.
X
LI-I
h—l
LL.
Z
2
[—.
<‘.
E
Z
<
0
ad
0
Initial concept Discontinuance
TIME

194

Now, using the diagram template below, write in the names of variables that you believe
impact Organizational Flexibility over time and that — in turn — Organizational Flexibility

impacts over time.

Indicate how you believe these variables are associated with a postive change in
Organizational Flexibility. Circle (+) if a positive change in a variable you identiﬁed
would be expected to cause a positive change in Organizational Flexibility. Circle (-) if a
negative change in a variable you identiﬁed would be expected to cause a positive change

in Organizational Flexibility.

Note that phrases such as ‘an increase in,’ ‘growth of,’ ‘reinforcement of,’ ‘higher,’ or
‘intensiﬁcation of’ can be substituted for ‘a positive change in’ and that phrases such as
‘a decrease in,’ ‘depletion of, ’dampening of,’ ‘lower,’ or ‘weakening of can be

substituted for ‘a negative change in.’

 

Organiz. 63/—\
(+) Flexibility —

 

 

 

195

 

VISUALIZE Market Proficiencies

Consider the dynamics of Market Research Activities for the development of new
innovations. Sketch a reference mode that you believe best describes its shape over the
speciﬁed time horizon in the real world. Refer to the Figures A.l & A.2 (common modes
of behavior) for possible ways of sketching the reference mode. Keep in mind that these
modes of behavior may be combined with each other. Why do you believe the reference

mode for Market Research Activities is shaped this way?

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Successful Ad0ption
:1:
U
a:
<
LLI
E
g
<
2
Initial concept Discontinuance
TIME
Failed Adoption
I
U
o:
<
LL]
”é
[—
DJ
:2.
a:
<
2
Initial concept Discontinuance
TIME

196

 

Now, using the diagram template below, write in the names of variables that you believe
impact Market Research Activities over time and that — in turn — Market Research

Activities impacts over time.

Indicate how you believe these variables are associated with a postive change in Market
Research Activities. Circle (+) if a positive change in a variable you identiﬁed would be
expected to cause a positive change in Market Research Activities. Circle (-) if a negative
change in a variable you identiﬁed would be expected to cause a positive change in

Market Research Activities.

Note that phrases such as ‘an increase in,’ ‘growth of,’ ‘reinforcement of,’ ‘higher,’ or
‘intensiﬁcation of’ can be substituted for ‘a positive change in’ and that phrases such as
‘a decrease in,’ ‘depletion of, ’dampening of,’ ‘lower,’ or ‘weakening of can be

substituted for ‘a negative change in.’

 

Market G>/—\+

(9 Research _
Activities

 

 

 

197

VISUALIZE Number of New Technologies Adopted

Consider the dynamics of Number of New Technologies Adopted with the development
of new innovations. Sketch a reference mode that you believe best describes its shape
over the speciﬁed time horizon in the real world. Refer to the Figures A.1 & A.2
(common modes of behavior) for possible ways of sketching the reference mode. Keep
in mind that these modes of behavior may be combined with each other. Why do you

believe the reference mode for Number of New Technologies Adopted is shaped this

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

way?
Successful Innovation
m
B
0
O
v—1
8
0
Lu
I-
3
LL]
2
LL:
0
:4:
Initial concept Discontinuance
TIME
Failed Innovation
CI)
.93.
O
O
...I
2
0
LL]
E-'
3
LL]
Z
LL:
0
an:
Initial concept Discontinuance
TIME

198

 

Now, using the diagram template below, write in the names of variables that you believe
impact Number of New Technologies Adopted over time and that — in turn — Number of

New Technologies Adopted impacts over time.

Indicate how you believe these variables are associated with a postive change in Number
of New Technologies Adopted. Circle (+) if a positive change in a variable you identiﬁed
would be expected to cause a positive change in Number of New Technologies Adopted.
Circle (-) if a negative change in a variable you identiﬁed would be expected to cause a

positive change in Number of New Technologies Adopted.

Note that phrases such as ‘an increase in,’ ‘growth of,’ ‘reinforcement of,’ ‘higher,’ or
‘intensiﬁcation of’ can be substituted for ‘a positive change in’ and that phrases such as
‘a decrease in,’ ‘depletion of, ’dampening of,’ ‘lower,’ or ‘weakening of’ can be

substituted for ‘a negative change in.’

 

Number of G)/’—\+

New _
G Technologies

f Adopted ®

 

 

 

199

NEW TECH UTILIZED

VISUALIZE New Technology Utilization

Consider the dynamics of New Technology Utilization with the development of new
innovations. Sketch a reference mode that you believe best describes its shape over
the speciﬁed time horizon in the real world. Refer to the Figures A.1 & A.2
(common modes of behavior) for possible ways of sketching the reference mode.
Keep in mind that these modes of behavior may be combined with each other. Why

do you believe the reference mode for New Technology Utilization is shaped this

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

way?
Successful Adoption
Q
LLI
bl
d
I—
D
I
0
LL]
{—1
3
LE
Z
Initial concept Discontinuance
TIME
Failed Adoption
Initial concept Discontinuance
TIME

200

Now, using the diagram template below, write in the names of variables that you believe
impact New Technology Utilization over time and that — in turn — New Technology

Utilization impacts over time.

Indicate how you believe these variables are associated with a postive change in New
Technology Utilization. Circle (+) if a positive change in a variable you identiﬁed would
be expected to cause a positive change in New Technology Utilization. Circle (-) if a
negative change in a variable you identiﬁed would be expected to cause a positive change

in New Technology Utilization.

Note that phrases such as ‘an increase in,’ ‘growth of,’ ‘reinforcement of,’ ‘higher,’ or
‘intensiﬁcation of’ can be substituted for ‘a positive change in’ and that phrases such as
‘a decrease in,’ ‘depletion of, ’dampening of,’ ‘lower,’ or ‘weakening of’ can be

substituted for ‘a negative change in.’

 

New %+

® Technology -

Utilization
’/—I <9

 

 

 

201

APPENDIX B

System Dynamics Modeling Supporting Documents

202

Table 8.1
Table B.2
Table B.3
Section 8.1

Section 8.2

Figure 8.1
Figure B.2
Figure B.3
Figure 8.4
Figure 8.5
Figure 8.6
Figure B.7
Figure B.8

Appendix B Contents

Parameters (Constants) used in Vensim Modeling
Stock (Levels) with Initial Values at t=0
Auxiliary Variables used in Vensim Modeling
Differential Equations for System Dynamics Model
B. l .1 Resource Equations

B. l .2 Exploration & Exploitation Equations

B. l .3 Technology Knowledge Equations

B. 1 .4 Proﬁciencies Equations

B. l .5 Marketing-R&D Integration Equations

B. 1 .6 Product/Proﬁt Equations

B. 1 .7 Performance Gap & Adaptive Capacity Equations
Vensim Code

B2] Resource Module

B.2.2 Exploration & Exploitation Modules

823 Knowledge Module

B.2.4 Proﬁciencies Module

B.2.5 Product/Proﬁt Module

B.2.6 Performance Gap &Adaptive Capacity
fR/D Graphical Function

Resource Module

Exploration Module

Exploitation Module

Knowledge Module

Proﬁciencies Module

Proﬁts Module

Performance Gap and Adaptive Capacity Module

203

 

_v
mn—

 

 

 

 

Table B.1 Parameters (Constants) used in Vensim Modeling

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Symbol Label Deﬁnition Constants
CAC unit conversion 1 unit in performance gap .10
becomes 0.10 units of performance/
adaptive capacity adaptive cap
C D exploit resources / resources required for 1 1
projects development project resource/projec
t
C E quarterly cost/explore quarterly cost of maintaining 0.125 expense
pr0j ect a exploration pr0ject units
C1 converter integration/ units of Marketing-RD 1 integration
products integration from 1 successful unit
exploitation project
C K knowledge/ project knowledge units required for 5 knowledge
constant 1 development project units
CP converter units of performance from 1 1 performance
performance/ products unit of proﬁts
C P_ K proﬁciencies/ units of Marketing or R&D .20 proﬁciency
knowledge converter Proﬁciency gained from 1 units
unit of knowledge
C P_ P converter products/ # of products resulting from a 2
projects development project
CSR perform/slack # performance gap allocated 1
conversion to slack resources
C R explore resources/ resources required for l 2.5 resources/
project research project project
f, fracdeathl % of new products at end of 10%
PLC quarterly
f2 fracdeath2 % of products 1-5 years old at 10%
end of PLC quarterly
f3 fracdeath3 % of products 6-10 years old 25%
at end of PLC quarterly
fAC frac decrease in AC % decrease in Adaptive 1%
Capacity per quarter
feat frac catalytic % of exploration projects 10%

 

 

resulting in catalytic innov

 

 

204

 

Table B.1 Parameters (con’t)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.sz frac fail % of development projects 11%
failing to reach completion
flaunc/zedz frac success launch % of products launched 95%
rate] successfully
flose frac decrease % of resources lost quarterly 5%
because they are unused
fmid forgetting factor mid % of mid knowledge 1%
forgotten every quarter
fM loss factor marketing % of Marketing Proﬁciencies 1%
lost every quarter
fMRD frac decrease % erosion in marketing-R&D 10%
integration integration/quarter
fnew forgetting factor new % of new knowledge 1%
knowledge forgotten every quarter
four-D frac outdated % of ‘shelved’ development 1%
exploitation projects that become outdated
foam frac outdated % of ‘shelved’ development 1%
exploration projects that become outdated
fR frac failed exploration % of research projects failing 12%
to reaching completion
f RD-prof loss factor R&D % of R&D Proﬁciencies lost 10%
every quarter
kc knowledge — catalytic knowledge increase from a 7 knowledge
projects catalytic innovation units
k D knowledge - exploit knowledge increase from 1 knowledge
projects development project units
kf knowledge — failed knowledge increase from 0.25
projects failed development projects knowledge
units
k R knowledge — explore knowledge increase from 2.5 knowledge
projects research project units
"T total approved projects # of projects/quarter that have 6 +f(slack
received approval and funding resources)

 

 

2 Historically, the fraction of failed development projects has been noted anywhere from 20-80% (Cooper,
1993). These companies record a much lower % failed as new products brought into development have
designated customers; if a product fail they fail after launch not during the development phase.

205

 

 

 

 

 

i .'

 

Table B.1 Parameters (con’t)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

product 1-5 years old

 

R resources man-hours, capital, ﬁmding 15 resource
for new projects units/quarter

r1 aging new prod product aging time new 4 quarters
products

12 aging prodl-S product aging time products 20 quarters
1-5 years old

r3 aging prod6-10 product aging time products 20 quarters
6-10 years old

14 aging prod10+ product aging time until 32 quarters
obsolescence

To delay on exploit time to complete a 2 quarters
exploitation project

70 delay for target delay time to relay 2 quarters

adjustment information regarding

performance gap

mum,” aging new to mid time for new knowledge to 8 quarters
become mid knowledge

mm“, aging mid to old time for mid knowledge to 12 quarters
become old knowledge

mom aging old time for old knowledge to 20 quarters
become outdated and obsolete

rQ perception time delay time in receiving info 2 quarters
regarding performance

27; delay time time to complete a research 4 quarters
project

15,”), z-SR_R delay exploit (explore) time delay for distributing 2 quarters

distribution slack resources to exploitation

(exploration) projects

17, proﬁtnew/product relative proﬁt units ﬁom l 1.2
new product

17 2 proﬁtl -5/product relative proﬁt units from 1 1.0
product 1-5 years old

173 proﬁt16-10/product relative proﬁt units from 1 0.80
product 1-5 years old

174 proﬁt10+/product relative proﬁt units from 1 0.75

 

206

 

 

 

 

 

 

 

 

Table B.2 Stock (Levels) with Initial Values at t=0

 

Deﬁnition

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Stock Initial Values -
all simulations
Adaptive Capacity, NP program ability to adapt to 0
AC(t) environment (moderator)
Exploitation Projects approved development projects awaiting 0
Requiring Funding, resources
N R( 1)
Exploration Projects approved research projects awaiting 0
Requiring Funding, resources
N D( 0
Innovative Stock of technology knowledge less than 5
Knowledge, K,(t) 5 years old
Marketing Level of Marketing Proﬁciencies 0
Proﬁciencies, MP(t) maintained
Marketing-RD Level of Marketing-RD Integration 0
Integration, MRD(t) maintained
Mid Knowledge, KM(t) Stock of technology knowledge greater 0
than 5 but less than 10 years old
New Product<1 Products < 1 year old 0
01d Knowledge, K0(t) Stock of knowledge greater than 10 years
old
Perceived Actual performance as perceived by 14
Performance: Actual, management
1’7!)
Performance Goal: Target performance 14
Target, T P(t)
Products Launched, New products launched into the 0
Pzaunchﬂ) marketplace
ProductsI-5 yr, Products 1-5 years old 10
P1 -5yr(t)
Products6-10yr, Products 6-10 years old 10
P6-10yr(t)
Products10+, P 10y,(t) Products 10+ years old 5
Quarter Proﬁts, Q(t) Proﬁts from products 14
R&D Proﬁciencies, R&D proﬁciencies for NPD activities 10
RDP(t)

 

207

 

Table B.2 Stock (Levels) with Initial Values at t=0 (con’t)

 

 

 

 

 

 

 

Resources for Resources allocated to exploration 0
Exploration, RR(t) projects

Resources for Resources allocated to exploitation 0
Exploitation, RD(t) projects

Slack Resources, Slack resources available for projects 0
SR(t)

Technology # or ongoing exploration 0
Exploration, ER(t) proj ects/quarter

Technology # or ongoing exploitation 0
Exploitation, ER(t) projects/quarter

 

 

Table B.3 Auxiliary Variables used in Vensim Modeling

 

Label

Definition

 

AC moderating effects

Adaptive Capacity acts as a ampliﬁer (moderator) to
marketing and R&D proﬁciencies

 

Catalytic Innovations,

number of catalytic innovations per quarter that result

 

PC from exploration (research) projects
discount factor AC Old- ampliﬁer effect of adaptive capacity; when adaptive
Know capacity < 0 then Old and Mid knowledge is not

discounted, but New Knowledge is.

 

discount factor AC
Mid-Know

ampliﬁer effect of adaptive capacity; when adaptive
capacity = 0 then Mid knowledge is not discounted, but
Old and New Knowledge are.

 

discount factor AC 1-
Know

ampliﬁer effect of adaptive capacity; when adaptive
capacity > 0 then New knowledge is not discounted but
old and Mid knowledge are.

 

exploration project
constraint

constraint on new projects; can only undertake new
projects if (1) resources are available, and (2) projects
have been approved

 

exploitation project
constraint

 

constraint on new projects; can only undertake new
projects if (1) resources are available, and (2) projects
have been approved

 

 

208

 

 

Table B.3 (con’t) Auxiliary Variables

 

Label

Definition

 

expenses

 

quarterly expenses incurred by maintaining exploration
projects

 

frac slack resources to
exploration

fraction of slack resources allocated to exploration
projects

 

fraction resource
allocation to
exploration, fR/D

fraction of resources allocated to exploration projects; 1-
fR/D goes to exploitation (development) projects

 

GAP: (Target - Actual),
G:

gap in performance where the target exceeds the actual

 

knowledge limitation

constraint on new projects; can only undertake new
projects if the knowledge stock is available

 

 

 

 

 

 

 

 

 

marketing-R&D constraint on timing of new projects; if marketing-R&D

constraint stock goes below 0, then projects are delayed by 2
quarters

Mktg constraint constraint on success in launch of new products, if
marketing proﬁciencies < 0, then decrease success factor
by 2

profitnew proﬁts from product less than 1 year old

profitl-S proﬁts from product 1-5 years old

profit6-l0 proﬁts from product 6-10 years old

profitl0+ proﬁts from product 10+years old

perception gap gap between actual performance and perceived
performance due to time delays or other ‘misperceptions’

quarterly slack constraint on quarterly slack; only when the actual
exceed the target does slack resources become available

R&D constraint constraint on success in launch of new products, if R&D

proﬁciencies < 0, then decrease success factor by 2

 

 

total approved projects

 

quarterly total number of approved projects (both
research & development)

 

209

 

 

B.1 Differential Equations used in SD Model
8.1.] Resources Equations

Resources are identiﬁed as “slack” resources and “budgeted” resources. Slack
resources, SR(t) , is deﬁned as a function of the performance gap, G(t) , in the overall

program. The gap is deﬁned as target performance — actual performance (see section

B. 1 .7.1 below). Slack resources only become available when the gap is less than
zero G(t) < 0, or actual performance exceeds target performance. The inﬂow to slack

resources, sr is deﬁned as follows:

srm = aSRG(t); aSR = 0 when G(t) 2 0 and am = CSR when G(t) < 0 (Bl)

where cSR represents the percentage of the total slack resources invested back into the NPD

program. Outﬂow of slack resources models the allocation of slack resources to projects:

 

srout-exploitation = (I - 'Cf/D)SR(t) (B2)
SR-D

srout-exploration : ﬁf—Sgﬂ (B3)
SR-R

where the fraction of exploration projects to exploitation projects approved, f R , D , is a

function of the performance gap as well and TSR_D and TSR- R is the time delay in

allocating resources.
“Budgeted” resources are deﬁned as resources allocated to the new product
program under standard operating policies. The amount is typically budgeted annually

by upper management. These standard resources are allocated to exploration and

210

exploitation as 7“,, = ( f R , D)R R and r0 = (I - f R , D)RD . This implies that every quarter,

managers make decisions on how to allocate resources based on the past quarter’s
performance level. The stocks, Resources to Exploration and Resources to Exploitation

are represented as;

 

 

 

dSI'out-explore dI' explore dI'lost-explore drallocated-explore
R R = J + — +

R 0
dt a’t dt dt )3, "( ) (8'4)

 

 

dSI'out-exploit dl' exploit dl'lost-exploit drallocated-exploit
RD 2 I + - +

dt dt dt dt )J'RDw) (13.5)

dI'los . o . . o
—d—'— represents resources “lost” if not used wrthrn the quarter. Thrs pollcy of “don’t use-
t

lose” is standard practice in many organizations. Unused resources indicate too generous

of a budget that may better be used in other ﬁmctional areas of the organization.

B. 1 .2 Exploration and Exploitation Equations
At each quarter, the NPD program undertakes n7 new projects, which is a
function of the resources available such that;

RC +SR t)
"r = (
CR

(B.6)

where C}; is the maximum resources that might be required for a research project. Since
RC (quarterly standard resources) is typically constant, the variability in the number of
projects undertaken in a quarter is a function of the availability of slack resources.

N R (t) , the number of approved research projects, and N D (t) , the number of approved

development projects for the quarter. Each is determined by N R (t) = f R , D(n7) and

N0“) = fR/D(1_n7')-

211

Two stocks, Exploration Projects Requiring Funding, N RH), and Exploitation
Projects Requiring Funding, N 00) represent projects that have received “approval” but
have not received funding. These stocks act as a “limiter” which insures that only
projects that have ﬁmding are actually undertaken. Additionally it allows projects to be

“shelved” until funding becomes available. Projects that aren’t funded can become

drp

outdated after a set time, Toutdate . Projects that do receive ﬁmding, Tm and
t

 

ddpm become the inﬂows for the stocks, Technology Exploration and Technology
Exploitation.
5%- : Min(R(t) / CR , N R (t)) (B-7)
dED-in .
T = Min(R(t) / cD, N D (t), KS, (0 / CK) (3.8)

where cR and cD are constants that determine resources required to undertake an
exploration (research) projects or exploitation project (development), respectively.

KS, (t) represent the total knowledge stock at time = t, and KS7. (t) / cK represents the

knowledge stock required to undertake a new development project. Exploration projects
can only be added to the exploration stock, ER, if there are adequate resources available.
Likewise, exploitation projects can only be added to the exploitation stock, ED, if there
are adequate resources and sufﬁcient knowledge to undertake a new development project.
Only a fraction of projects undertaken are successful; ﬁg = fractional success rate

of research projects, f0 = fractional success rate of development projects. The number of

successful projects is represented by f R E R (t) / TR and f DE D (t) / 2'0 where TR and TD are

212

the average time to complete a project. Unsuccessful projects are represented by

(1— ﬁe)ER(t) and (1- fD)ED(t). In this study, TR = 4 quarters and r0 = 2 quarters.

B. 1 .2.1 Catalytic Innovations
A percentage of the successful research projects are catalytic innovation, the focus

of this study. Catalytic innovations are modeled as

PC = fcaz(fRDER(t)) (B.9)

Catalytic innovations, as previously discussed, are innovations discovered during
research activities that act as a catalyst to instigate new research projects and new product
development projects not originally planned by the ﬁrm. Catalytic innovations also
embody new knowledge that is more valuable to the organization compared to the

standard results from research activities. In this study fan = 0.10, based on the fact that

10% of inventions are considered new-to-the-world innovations (Grifﬁn, 1997).

B. 1 .3 Technology Knowledge Equations
Each successful project results in “learning” or a gain in knowledge. The amount
of knowledge gained (inﬂow) from a project is variant. Catalytic projects have higher

means and variances in knowledge acquisition compared to research projects, where
,u(. > ,uR> ,uD and 01.2 >0R2> 0",,2 (March, 1991). New knowledge is also acquired from

successful development and failed development projects. These are modeled as constant

pr0portions of these types of projects, or as;

Kr = (I — fD)ED(t)cKD_fail (B.lO)

KD = fDED(t)CKD_success (B.l l)

213

where cKD,fail = 0.25 and CKD-success = 1.0 in this study.

Total knowledge is modeled as three stocks; Innovative Knowledge, Mid
Knowledge and Old Knowledge. Innovative Knowledge is new technological knowledge
acquired within the past four quarters. Knowledge more than two years old becomes Mid
Knowledge, and Old Knowledge is more than 5 years old. Modeling knowledge as three
different stocks is required as not all technological knowledge is treated equally by the

organization. The knowledge stocks are modeled as:

 

 

 

 

 

dK dK dK dK dK_ dK_ -n
K,(t): I dtR+ dtD+ dtf+ dtC— dtlm — Id? 3 +K,(O) (B.12)
dK_ . dK _ . dK
KM“): {_L;ﬂ&—{ Mdtagmg + 1:;lost]+KM(O) (8.13)
dK _ .n dK _ .
K00): ] Mdt“g'g— Odfg'"g+1<o(0) (3.14)

The outﬂows represent the fractional knowledge lost by the departure of key
employees, the shelving of technology for future use, or even new technology replacing
old technology.

It is important to note that ‘dated’ knowledge is not necessarily useless or
ineffective. In some cases it may more useﬁrl as proven knowledge since new knowledge
may still have some application uncertainties. However, keeping in mind Moore’s Law
(which predicts that the amount of information storable on a silicon chip will roughly
double every), aging knowledge may indeed quickly be ineffectual. The greater the
percentage of old knowledge, the greater the need exists to restock innovative
technological knowledge through exploration activities. Thus, technically oriented

organizations must balance the need to “renew” aging knowledge through research

214

activities and “appropriate rents” from proven older knowledge. In order to model this
utilization of different ages of knowledge, Adaptive Capacity is modeled as a moderator
to the three knowledge stocks. This moderator is a modeled as a inverse s-shaped lookup

function (see sec B.l.7.2).

B. l .4 Proﬁciencies Equations

Both Marketing Proﬁciencies and R&D Proficiencies are a function of the gains
in knowledge. It is also moderated by adaptive capacity. The higher the adaptive
capacity the higher the use of marketing and R&D proﬁciencies in the building of new

products.

MP," =K,m *cP_K *AC(t) (B.lS)
RDPm =K,m *cP_K *AC(t) (8.16)
where c,,_K represents the increase in marketing/R&D proﬁciencies based on one new

knowledge unit. c,,_K =0.20 in this study based on conversations with NP managers of

the case studies ﬁrms.

B. 1 .5 Marketing-R&D Integration Equations

Marketing-R&D integration is a simple linear function of the number of
successfully completed exploitation projects the ﬁrm has achieved. The assumption is
that is the project has been successfully completed and handed off to marketing for

launch, there has been some interaction between the two departments.

MRD(t) = f (successful exploit projects) * CMRD (8.17)

215

where cMRD =0.25 in this study, which indicates that for every successful exploitation

project, marketing and R&D will have interacted with each other at 25% of capacity for
the two departments. This number would be higher if the ﬁrm was more marketing

oriented.

8.1 .7 Products/Proﬁt Equations

Product Stocks are also modeled as an aging chain. Products enter the aging

chain through Products Launched stock;

 

 

 

dE dP dP, h- l
P t = D—success __ launch-success + “WC fa' W e + P 0 B18
launch ( ) I dt { dt dt launch ( ) ( )
dPlaunch— failure Home}, (I )
Not all products launched are successﬁil so that = where

dt

TIaunch— fail
r,mmch_fm, = 0.05 . Successfully launched products become New Products<1, Pnewpmda).

Products that are successful in the market place after 1 year enter the stock, Products]-
5 yr, P1-5y,(t), which then become Products6-10, P6-,0y,(t) and then Products 10+, P 10”“)-

A similar model is applicable for all four of these stocks;

 

derr - dPW" — I
Pl_5v,,,(t)=J. [ pod success _ n pod failure

dt dt ]+ I)l—Syr (O) (819)

The stock of Quarterly Proﬁts is increased by the successful launch of new

products into the marketplace and sales of existing products. Quarterly Profits is
represented by Q(t) = am ,where m = number of products in marketplace and It is a

relative proﬁt level. Actual performance cannot be measured instantaneously.

216

B. 1 .7 Performance Gap & Adaptive Capacity Equations
B. 1 .7.1 Performance Gap
The performance gap, G, , is a ﬁmction of the target performance and actual

performance. As an auxiliary variable changes with each quarter but does not act as an

in/outﬂow to a stock variable, it is represented by a subscripted t.
G, = (Target Performance — Actual Performance) (820)

Actual performance is determined by quarterly proﬁts realized. A time delay

exists from realization of proﬁts to the actual reporting of proﬁts. Thus, allocation

Q0)

’9

 

decisions are based on perceived performance, PP(t), or , where IQ is the time it

takes to transfer the information Target Performance, TP(t) , is typically set by upper

management although it may be inﬂuenced by exogenous factors such as competition. In
this study, target performance is modeled here as a “ﬂoating goal” (Sterman, 2000). If a
NPD program performs below expectations, the target goal will be adjusted downward; if
performance is above expectation, the target will be adjusted upward. Since changes in
target performance cannot be implemented instantaneously after being set by top

management, it may take several time periods for new goals to become reality within the

NPD program. The performance gap, G, , is thus represented by:

G. =—TP(" —P”(t) (3.21)

76
where 70 is also a time constant to factor in the delay of receiving actual performance

information and setting of new target performance goals.

217

As previously noted in Chapter 4, the fraction of exploration (research) projects
compared to exploitation (development) projects approved, fR, D , is a function of the

performance gap. It is represented by:

 

fR/D : (322)

At the inﬂection point, G, = 0 and fR,D=0.5. The constant, g, is set at 0.053 in this

model. The constant, g, was calculated based on the nonlinear function that was

described by the case ﬁrms during the interview sessions, see Figure B]. The starting
point for all simulations is set at TP(t) = AP(t) , or 00 = 0 so that equal number of

research and development projects are approved for the initial quarters.

Figure 8.1 f R , D as a Function of Performance Gap

,———.—2_1_—_-—‘2#. ‘—#—_—“"T—L’ 2 if _'

 

 

 

 

 

-10 ‘ i '5 ‘ "1‘0

218

B .1 .7 .2 Adaptive Capacity

Levinthal and March (1981) proposed that ﬁrms will adapt their behavior based
on past performance, thus, Adaptive Capacity is modeled as a linear function of the
performance gap. As the gap increases, adaptive capacity also increases (the worse the

ﬁrm is performing, the more the openness to new ideas) .
AC," (t) = G, * cAC (B23)
This linear relationship is based on the theoretical model and empirically supported by
Lant (1992). It was found that ﬁrms base their aspiration levels on the deviation between
the targeted goal and the actual performance achieved in the last period. The constant in
the Lant study ranged between 0 and 1. In this study cAC was set at 0.10.
The linear relationship described here for Adaptive Capacity is then used as a

moderator for Marketing and R&D Proficiencies as described in section 8.1 .4.

219

(01)

(02)

(03)

(04)

(05)

(06)

(07)

(08)

B.2 Vensim Code
Legend:

Stocks noted in Italics

Flows noted in Underlined
Auxiliary variable in Bold
Constants listed in Table 8.1

8.2.1 RESOURCES MODULE (see Figure B.2)

allocation of exploit resources = funded exploitation projects * exploit resources/
projects

Units: resources/Quarter

- Resources allocated to exploitation projects

allocation of explore resources = funded exploration projects *explore
resources/projects

Units: resources/Quarter

- Resources allocated to exploration projects

delay for distribution explore = WITH LOOKUP (Slack Resources)
Units: Quarter
- Delay for distributing available slack resources 2Q

frac slack resources to exploration = l-"ﬁac resources w/equil"
Units: Dmnl
- Slack resources allocated to exploration projects

frac resource allocation to exploration = WITH LOOKUP (.“GAP: ( TARGET-
ACTUAL”)

Units: Dmnl

- Fraction of resources allocated to exploration activities. l-frac goes to
exploitation activities\

quarterly slack = WITH LOOKUP (-"GAP: ( TARGET-ACTUAL)" *

"perform~ - Slack conversion")

Units: resources/Quarter

- Quarterly slack resources only if the performance gap is negative or actual
performance exceed target performance\

resources exploitation = (1-"frac resources w/equil")*resources/quarter
Units: resources/Quarter
- Standard resources to development projects

resources exploration = resources*("ﬁ'ac resources w/equil")/quarter

220

(09)

(10)

(11)

(12)

(13)

(14)

Units: resources/Quarter
- Standard resources to research projects

Resources for Exploitation = INTEG (+slack to exploit+resources exploitation-
allocation of exploit resources- "don't use-lose", 0)

Units: resources

- Total resources available for exploitation (development projects),INITIAL = 0

Resources for Exploration = INTEG (slack to explore+resources exploration-
allocation of explore resources-"don't use-lose explor", 0)

Units: resources

- Resources allocated to new research (exploration) projects, INITIAL = 0

Slack Resources = INTEG (slack resources in-slack to exploit-slack to explore, 0)
Units: resources
- Resources not already allocated to other projects

slacﬂesources in = quarterly slack
Units: resources/Quarter
- Resources not previously allocated to other projects

slack to exploit = Slack Resources*(1-frac slack resources to exploration)/delay
exploit distribution

Units: resources/Quarter

- Slack resources committed to exploitation

slack to explore = Slack Resources*(frac slack resources to explorationydelay for
distribution explore

Units: resources/Quarter

- Slack resources committed to exploration projects

822 EXPLORATION & EXPLOITATION MODULES (See Figure 3.3)

B.2.2.1 Exploration

(15)

(16)

catalytic innovations = funded exploration projects * 0.1 *"mktg—r&d
constraint"

Units: projects/Quarter

- Number of catalytic innovations per quarter that result from research projects

exploration project constraint = min(Resources for Exploration/"explore
resources/ projects", Exploration Projects Requiring Funding)

Units: projects

- Can only undertake projects if (1) resources are available and(2) projects have
been approved

221,

(17)

(18)

(19)

(20)

(21)

(22)

(23)

(24)

(25)

Exploration Projects Requiring Funding = INTEG (new exploration projects-
outdated exploration projects-funded exploration projects, 0)

Units: projects

- Exploration projects that have received approval but have not yet received
funding. this is based on the total number of approved projects / quarter INITIAL
= 0

failed proiects = frac failed exploration‘Technology Exploration
Units: projects/Quarter
-Failed research projects

funded exploration proiects = exploration project constraint/quarter

Units: projects/Quarter

- Can only undertake projects if (1) resom'ces are available and(2) projects have
been approved

inﬂow integration = successﬁrl exploitation projects*0. l l""converter integration/
proj ects"

Units: integration/Quarter

- Increase in marketing R&D integration as a result of success in exploitation
projects

"Marketing-RD Integration"= IN TEG (inﬂow integration-outﬂow integration, 0)
Units: integration
- Marketing & R&D integration stock INITIAL = 10

“mktg-r&d constraint” = WITH LOOKUP ("Marketing-RD
Integration"*mktgrd switch)

Units: Dmnl

- If integration <0 increase delay time by factor of 2\

new exploration proiects= catalytic innovations+(' ‘frac resources w/equil")*total

approved pr0j ects/quarter

Units: projects/Quarter
- Research projects approved per quarter

outﬂow integtion = "Marketing-RD Integration"""ﬁac. decrease integration"
Units: integration/Quarter
- 1% Outﬂow of integration per quarter

outdated exploration proiacts = Exploration Projects Requiring Funding‘f’rac

outdated exploration
Units: projects/Quarter

222

 

(26)

(27)

(23)

- DELAY FIXED projects that have not received funding in 4 quarters are expired
DELAY FD(ED(Exploration Projects Requiring Funding,frac outdated
exploration , 0 )

successful explore projects = Technology Exploration/delay time
Units: projects/Quarter
- % of research projects that are successful

Technology Exploration = INTEG (ﬁmded exploration projects-failed projects-
successﬁtl explore projects, 0)

Units: projects

- Number of projects being undertaken that focus on research efforts, INITIAL=0

total approved projects = 6 + ((Slack Resources)/"explore resources/
projects")*STEP(l,8)

Units: projects

- 6 total projects (exploration & exploitation) projects approved quarterly + slack
resources available

8.2.2.2 Exploitation

(29)

(30)

(31)

(32)

exploitation projects constraint = min(Resources for Exploitation/"exploit
resources/ projects", Exploitation Projects Requiring Funding)
Units: projects
Exploitation projects can only be undertaken if they have enough resources. this
acts as a limiter.

Exploitation Projects Requiring Funding = INTEG (new exploitation projects-
outdated exploitation projects-funded exploitation projects, 0)
Units: projects
- Exploitation projects that have received approval but have not yet received
funding. this is based on the total number of approved projects / quarter INITIAL
= 0

failed exploitation proiects = (Technology Exploitation*"frac. fail")
Units: projects/Quarter
- Development projects that fail

funded exploitation proiects = min(knowledge limitation/quarter, exploitation
projects constraint/quarter)

Units: projects/Quarter

- # of development (exploitation) projects that have received resource allocations

223

(33)

(34)

(35)

(36)

new exploitation proiects = (1-"frac resources w/equil")*total approVed projects/
quarter

Units: projects/Quarter

- Exploitation projects approved for the quarter

outdated exploitation proiects = Exploitation Projects Requiring Funding*ﬁ'ac
outdated exploitation/quarter

Units: projects/Quarter

- DELAY F D(ED Projects that do not receive funding within 4Q are expired

successful exploitation projects = Technology Exploitation/delay on exploit
Units: projects/Quarter
- Successful development projects

Technology Exploitation = INT EG (ftmded exploitation projects-failed
exploitation proj ects-successful exploitation projects, 0)

Units: projects

- Number of project underway concerned with exploiting existing technology
INITIAL = 0

B.2.3 KNOWLEDGE MODULE (see Figure B4)

(37)

(33)

(39)

(40)

(29)

AC moderating effects = WITH LOOKUP (Adaptive Capacity)
Units: adaptcap
- AC acts as a moderator for R&D & Mktg Proﬁciencies\

agiirg innov l_<now = max(0,ZIDZ(Innov Knowledge, aging new to mid))
Units: knowledge/Quarter
- Outdating of knowledge

agiag mid knowledga = (Mid Knowledge)/aging mid to old
Units: knowledge/Quarter
- Mid-knowledge aging to old knowledge

"discount factor AC Mid-Know" = WITH LOOKUP (Adaptive Capacity)
Units: Dmnl

- When AC = 0 then both new and old knowledge are discounted but not mid
knowledge. New knowledge is too risky - not yet proven & old knowledge is
outdated - uncompetitive.

"discount factor AC Old-Know" = WITH LOOKUP (Adaptive Capacity)
Units: Dmnl

- When AC = 0 then no discount, when AC>0 then discount mid and old
knowledge. When AC<O then do not discount old and mid, d-mid & d-old

224

(30)

(31)

(41)

(42)

(43)

(44)

(45)

(46)

(47)

"discount factor AC on I-Know" = WITH LOOKUP (Adaptive Capacity)
Units: Dmnl

- When AC is >0 then no discount factor d-new =1, if AC<O then discount new
knowledge

f_or_'getting innov knowledga = max(0,forgetting factor new knowledge*1nnov
Knowledge)

Units: knowledge/Quarter

- Forgetting factor for innovative knowledge

mgetting mid knowledge = Mid Knowledge*"forgetting factor - mid"
Units: knowledge/Quarter
- Knowledge lost

Innov Knowledge = INTEG ("knowledge acquired: failed proj"+new knowledge-
aging innov know-forgetting innov knowledge,5)

Units: knowledge

- A ﬁrm's knowledge base will determine if it can build new products. Assumes
no loss of knowledge only outdating. INITIAL = 5

"knowledge acquired: failed proj" = failed exploitation projects""know1edge~
failed projects"

Units: knowledge/Quarter

- Amount of knowledge gained from failed exploitation projects

knowledge limitation = ((Innov Knowledge*"discount factor AC on I-Know") +
(Mid Knowledge*"discount factor AC Mid-Know")+(Old Knowledge * "discount
factor AC Old-Know"))/"knowledge/ project constant"

Units: projects

- Total stock available. Can't undertake projects unless knowledge is available

Mid Knowledge = INTEG (aging innov know-aging mid knowledge-forgetting
mid knowledge, 0)

Units: knowledge

- Knowledge of mid-age, INITIAL = 0

new knowledga =(catalytic innovations*"lmowledge ~catalytic projects") +
("know explore projects“ (successful explore projects-catalytic innovations)) +
("knowledge~ exploit projects"*successful exploitation projects)

Units: knowledge/Quarter

- Knowledge gained from projects

01d Knowledge = IN TEG (aging mid knowledge-outdated knowledge, 0)

Units: knowledge
- Knowledge greater than 20Q,INITIAL = 0

225

(48)

outdated knowledga = ZIDZ(Old Knowledge,aging old )+Old Knowledge/aging
old/5

Units: knowledge/Quarter

- Knowledge that is outdated and no longer of use to the NPD process

B.2.4 PROFICIENCIES MODULES (see Figure 3.5)

(49)

(50)

(51)

(52)

(53)

(54)

(55)

(56)

"in R&D prof“ = new knowledge*"proﬁciencies~ knowledge converter"*AC
moderating effects

Units: proﬁciencies/Quarter

- Proﬁciencies increased by new knowledge

Marketing Proﬁciencies = INTEG (mktg proﬁciency gain-outﬂow mktg, 0)
Units: proﬁciencies
- Mktg Proﬁciencies necessary to launch products

mktg constraint = WITH LOOKUP (Marketing Proﬁciencies)
Units: l/Quarter

- If Mktg Proﬁciencies <0, decrease success of launch by a factor of 2\

m_ktgproﬁciencv gain = new knowledge*"proﬁciencies~ knowledge converter" *
AC moderating effects

Units: proﬁciencies/Quarter marketing proﬁciencies gained ﬁom new knowledge
- Gain in marketing proﬁciencies based on moderating factor of adaptive capacity

outﬂow mktg = loss factor mktg*Marketing Proﬁciencies
Units: proﬁciencies/Quarter
- Loss of Mktg Proﬁciencies over time

"out R&D" = "R&D Proﬁciencies"""loss factor R&D"
Units: proﬁciencies/Quarter
- Loss of R&D proﬁciencies /quarter

"R&D constraint" = WITH LOOKUP ("R&D Proﬁciencies")

Units: products/proj ect

-If R&D proﬁciencies<0, projects to products conversion is decreased by a factor
of2\

"R&D Proﬁciencies" = INTEG ("in R&D prof“-"out R&D",10)
Units: proﬁciencies
- R&D Proﬁciencies necessary in order to conduct successful projects

B.2.5 RODUCT/PROF ITS MODULE (see Figure B6)

226

(57)

(58)

(59)

(60)

(61)

(62)

(63)

(64)

(65)

(66)

(67)

(68)

 

expenses ="quarterly cost /eprore project"*Technology Exploration
Units: proﬁts
- Expenses incurred by overhead

"New Products <1" = INTEG (+product to market-inﬂowZ-obsolete1,7)
Units: products
- Products less than 1 year old, INITIAL = 0

obsolete = Products Launched*(1-frac success launch ratel)
Units: products/Quarter
- Products that do not make it to launch

 

obsoletel = "New Products <1 "fracdeathl
Units: products/Quarter
- Aging of new products

obsolete2 = "Products 1-5yr"*fracdeath2
Units: products/Quarter
- Products obsolescing after 5 years

obsolete3 = "Product 6-10 yr""ﬁ‘acdeath3
Units: products/Quarter
- Obsoleted projects in 3rd period of time

"outﬂow Prod10+" = "Product lO+"/"aging prod10+"
Units: products/Quarter
aging of products to obsolescence

"Product 10+"= INTEG (+ inﬂow4-"outﬂow Prod10+", 5)
Units: products
- Products older than 10years, INITIAL = 5

"Product 6-10 yr"= INTEG (+inﬂow3-inﬂow4-obsolete3, 10)
Units: products
- Products 6-10 years old, NITIAL = 10

product to market = Products Launched“ frac success launch ratel
Units: products/Quarter
- Projects converted to products and launched into marketplace

"Products 1-5yr"= INTEG (+inﬂow2-inﬂow3-obsolete2, 10)

Units: products
- Products 1-5 years old, INITIAL = 10

products from proiects = (successful exploitation projects*"conver products/
projects")

227

(69)

(70)

(71)

(72)

(73)

(74)

(75)

(76)

Units: products/ Quarter
- Projects are converted into products that are launched into the marketplace

Products Launched = INTEG (products ﬁom projects-product to market-obsolete,
0)

Units: products

- Total number of products in the marketplace, INITIAL = 0

proﬁt outﬂow = Quarter Proﬁts/quarter

Units: proﬁts/Quarter

drain for proﬁts so that proﬁt does not accumulate on a quarterly basis. Assume -
- Allow proﬁts are used in some fashion

"profitl-S" = "Products 1-5yr""‘"proﬁtl-5 /product"*"disc pulse 1-5"
Units: proﬁts
- Proﬁts from products l-Syr old

"profit10+" = "Product 10+""‘"proﬁt10+ /product"*"discount pulse 10+"
Units: proﬁts
- Proﬁt from products 10+ yrs old

"profit6-10" = "Product 6-10 yr"*"proﬁt6-10/product"*"disc pulse 6—10"
Units: proﬁts
- Proﬁts from products 6-10yr old

profitnew = "New Products <1 "*"proﬁtnew /product"*disc pulse new
Units: proﬁts
- Proﬁts from new products < lyr old

proﬁts increase = (("proﬁtl—5"+"proﬁt10+"+"proﬁt6-l0"+proﬁtnew)*0.67-
expenses)/quarter

Units: proﬁt/Quarter

- Total proﬁts per quarter after expenses of 33% and discounting because of bad
products

Quarter Proﬁts = IN TEG (proﬁts increase-proﬁt outﬂow, 14)
Units: proﬁts
proﬁts from products, INITIAL = 14

8.2.6 PERFORMANCE GAP & ADAPTIVE CAPACITY MODULE (See Figure 8.7)

(77) Adaptive Capacity = INTEG ((change in adaptive capacity-drain adaptive)*AC

switch, 0.001)

228

(73)

(79)

(80)

(81)

(82)

(83)

(84)

(85)

Units: adaptcap
- NPD programs willingness & ability (ﬂexibility & agility) to change strategic
direction - 'not broken, don't ﬁx it mentality' INITIAL = 1

ﬂange in adaptive capacity = GAP: ( TARGET-ACTUAL) * unit conversion
Units: adaptcap/Quarter

- Change in AC based on gap: target-actual. The higher the gap, the greater the
increase. Can be negative if gap is less than 0.

drain adaptive =Adaptive Capacity*frac decrease in AC
Units: adaptcap/Quarter
- AC that drains over time through forgetting

change in perfioal = (Perceived Performance: actual - Performance Goal:
target)/delay for target adjust
Units: performance/Quarter

- Change in performance goal based on the size of the gap 1:1 change

"GAP: (TARGET-ACTUAL)" = ("Performance Goal: target"-"Perceived
Performance: actual")* STEP(l ,8)

Units: performance

- "target goal - actual goal/ target goal" - the target goal is the performance goal +
the aspiration level being driven internally based on stock of old and new
knowledge

perceived performance change = perception gap/perception time
Units: performance/Quarter
- Change in perceived performance based on size of gap

"Perceived Performance: actual"= IN TEG (+ perceived performance change,l4)
Units: performance

- Perceived performance is perception time delay of actual performance, INITIAL
= 14

perception gap = Quarter Proﬁts*"converter performance /proﬁt"-"Perceived
Performance: actual"

Units: performance

- Gap between actual performance & perceived performance

"Performance Goal: target"= INTEG (change in perf goal,14)

Units: performance
- Floating goal, initial value INITIAL = 14

229

Admeeoow
.omﬁv

\

mmoatom: u.aov

3050mm... ‘ :0 mun—£526

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

:2on $2on %e _ov
mo seen 62:... C039}: 93% $
a
mu ‘ D MOM “mo—93 on Jon—m
m OOHSOmDMH F 4N £03m b59336
5583—93 A .5 ts SUV
30.5 0 me.—
"a
moon: 0 mac
80h DOmDh Gomwdh 0— 0060 GomuﬂHO—QXO m QOHQOWQNH v— om— m
Ow €038 02m 11" OH UUuSOmOH ON I m G
00.5 Oman . ooh,“ low—m 09¢ M mm
: 0930—93
80.5 Owe
much: 0 mo.—
0.5.93
.«0 nova ooam Gomwmhoﬁ Qvnmm 0.. o. 93 9 v— 03 m
9+ N mom ‘ll d.
m condemned
Pac— Xo GO San—Cummﬁ
~2on no.“ NM 3%
0m 070 m:
rcou

 

8850p deb

0:602 mooBOmom Nd Emmi

230

38 FE Eozﬁwoﬁm

 

 

    

 

 

 

     

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.3.—0280 «Enhance
EMqu—E sewage—88
3 segue—E
5583:: 23..on 8382 08m
Boas 8:835 3.33:0 . £85535:
Gﬂmﬂ' am 3558
.wéecmz . a
$8 8
8:835 sewage—88 .8389on
838% dab Bangui—5 out 26:
$8 .85
538235 mﬁﬁcsm
£885 uBmWSo wﬁhsvom
08: >28; 8298 DO/ 3005.5
53825 couﬁoaxm
$8.33 3:3 :omueoaxm
_ - ao_o§ooh aspen
corms—axe BEE
corms—93
wozﬂ 2%

Aﬁousomom
xoﬂmv

388E

c3895 _88

300.8
Enema»: Bog

 

385:8
8%ch sewage—98

0:602 :osﬁoaxm m.m oSmE

231

 

 

 

 

 

GA

 

 

 

 

 

 

 

 

 

 

 

 

 

A3863
Bataan _SBV
ﬂoomohu
“ﬂamingo? “moves 3
so 0
_ammooozm s u ﬂ £8332on
$8.85 8 83823
cowwwwas $8.85 08: 88 ome
m. 20:86—me Summee w.
$22.50 0% t 38.3on
Gouge—98
mayhem 339:0
Ba .23 .
wcﬂsgm «4 g
ﬂoowoa Ahowaswv 300 Bum
53393 . cosgoaxm Asgav
82a . .
sagasgxo
3398 03

2:82 nousmoaxm E 2&3

232

ﬂoobo

mpﬁﬂotoﬁ U<

“gnu—CO

26ch :0 803a

U< Baum.“ H5536

Abﬁmamv

305732
03:??? [V

U< hOHUMuv Hgoommmv
/’ /' COUME

30:! .20
U< 88% 35086

 

    
   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

    

 

 

omnsaoﬁ
wﬁuuwuom 26: “Scam
098—305— 95838
>25
28 wEuowuom BE EtowSM
9 Bo: maﬁa .85
owvo—Boﬁ 8:8 ”UBBcow
owEqum yawn? $62399 . M. owBqum &Bmsi &
20 ﬁg Dl/ >05: ‘H
>65. >85 mama A300 .33
So 1! $8.85 enigma—98
:c 8
&3325 om“? &BES. zowwwmoi B :ﬂv
339:0 w . acumen >5:
f/Eo mama :298 Iowvmaocx A800 .3.—Q 8298
. A3835 3.338sz
astute—98 a0 a 80—98
38. ._
3,3300va $0035 oubmﬁoz 2 035—305

ownsaoﬁ

2252 ego—Bog mm 035

233

  

HOUOEOO

&BEE
zmoﬁcoﬁcoa

men
a: E

   
 

 

 

mosses—MBA
Gag

 

 

5 ECU
£388 a? :5 Es

“Sow.“ amo— Saw

 

 

 

 

 

 

 

wEE xaomumoa
Dana :5 309:0 mom ocom 299$ mm”: mm
waging + A“, J
/\ £88280
9mm owuozpoﬁ
H88.“ 32

zmomoﬁxom oav

2252 86:22.“on 9m oSmE

234

 

 

 

 

 

 

 

 

 

 

 

 

  

 

 

 

 

 

 

  

 

 

 

 

 

 

   

 

3E :25.» .5833 no Q msauuuwc
$083 ope 2. 8n:
+o _ Ea may“
a 225a V
Bug—Owns / Ego—Owns 30””; NOE—8&0 w- _ U95 mﬁwn mono—8&0 maﬁa
. $52;
$8.35 8x38 a /< / \V >5ch
a 9 .89: u 95 S
We wave U 35:33 V _v 32605 NBMEV in _ a: W i o To , ESE +2 8:85
326th 302 \u ‘ 32605 \c 6.68;
$8.85 \1/7 ( F F h
20:85 52.8 Bonanza WZLEJ «2623 +2an .2629
33.23 29:2; 2 32g 1/ $an
Bonanza 8323 ll

 
      

 

 

 

 

>655
Goa

ASHE?

WEE

   

    

Acouﬂoixm
Boﬁcﬁﬂv

88098‘.’

83.9; BQQQ
moo €3.55

2:82 gem mm 2:5

235

am 89 w 0< E
mam :83ng ooggotoa 3&2on 08m
€328
033w?
£5 a.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

    
  

 

 

 

 

 

owcwno ooggoﬁoq @893
2: an m
QABZMBQQ ' Hahn S mwﬁwo
”mongoﬁom mu‘ . m. v @3530
@038qu o>ﬂmmv<
osﬁ 8:90.59
£€o>coo
_aow .tom HE:
E «was 0 “owns“ ”BOO
71 ooagomuom
up?
Swag pom mica
A123“?
-593; v “5:5
..z\‘

2352 bmomamu o>cmmc< a. 95 ouggoﬁom w.m 2:me

236

APPENDIX C

Longitudinal Study Supporting Documents

237

C.l Measures used in the Longitudinal Study (and sources)

 

 

 

 

 

Survey 1 scale: Likert l — 7.
Surveys 2-3 change score: Likert -3 — +3

 

 

 

Adaptive Capacity x {=1 )7? =2 .7 =3

1. Our ﬁrm can quickly change strategic direction based on
new technological advances made in our industry. (new) 5.04 5.47 5.84
Survey 1 scale: Likert 1 — 7. (1.74) (1.96) (2.29)
Surveys 2-3 change score: Likert -3 — +3

2. Our ﬁrm can quickly change strategic direction based on
new marketing information (i.e., information regarding 5.43 6.31 6.96
customers, competitors, or other market changes). (new) (1.32) (1.23) (2.01)
Survey 1 scale: Likert 1 — 7.
Surveys 2-3 change score: Likert -3 — +3

3. If a major competitor were to launch an intensive
campaign targeted at our customers, we would be able to 5.37 5.98 6.35
implement a responsive campaign immediately. (Jaworski (1.36) (1.52) (2.19)
and Kohli, 1993)
Survey 1 scale: Likert l — 7.
Surveys 2-3 change score: Likert -3 — +3

4. Our ﬁrm can quickly change strategic direction based on
signiﬁcant changes in our competitor’s pricing structure. 4.96 5.71 6.35
(Jaworski and Kohli, 1993) (1.54) (1.89) (2.14)

 

*calculated by adding timel score to time2 change score;
** calculated by adding time2 score to time3 change score

238

 

 

 

 

 

 

 

 

 

 

 

Slack Resources x F] i =2 )7 =3

1. Capital is a scarce resource for new product development
activities in my ﬁrm. (Calantone, Garcia, and Droge, 3.59 3.20 2.63
forthcoming) (2.04) (2.02) (2.72)
Survey 1 scale: Likert l — 7.
Surveys 2-3 change score: Likert -3 — +3

2. Material supplies are a scarce resource for new product
development activities in my ﬁrm. (Calantone, Garcia, and 3.14 3.10 2.96
Droge, forthcoming) (1.85) (2.03) (2.41)
Survey 1 scale: Likert l — 7.
Surveys 2-3 change score: Likert -3 — +3

3. R&D resources are a scarce in my ﬁrm. (Calantone,
Garcia, and Droge, forthcoming) 3.92 3.65 3.29
Survey 1 scale: Likert 1 - 7. (2.02) (2.0) (2.24)
Surveys 2-3 change score: Likert -3 — +3

4. Advertising and promotion resources for new product
launches are scarce in my ﬁrm. (Calantone, Garcia, and 3.55 3.33 2.90
Droge, forthcoming) (1.70) (1.77) (2.10)

Survey 1 scale: Likert l — 7.
Surveys 2-3 change score: Likert -3 — +3
*calculated by adding timel score to time2 change score;
** calculated by adding time2 score to time3 change score
Marketing-R&D Integration )7 F1 )7 =2‘ )7 ___3"

1. When developing new or next generation products,
information about customers’ needs is fully considered. 5.73 6.29 6.90
(Li and Calantone, 1998) (1.19) (1.50) (1.92)

 

Survey 1 scale: Likert 1 - 7.

Surveys 2-3 change score: Likert -3 - +3

 

 

 

 

239

 

 

 

 

When developing new or next generation products,
information about competitors’ products and strategies are
fully considered. (Li and Calantone, 1998)

Survey 1 scale: Likert l — 7.

Surveys 2-3 change score: Likert -3 - +3

5.39
(1.40)

5.80
(1.45)

6.35
(1.75)

 

When developing new or next generation products, market
conditions are fully considered in establishing new product
development goals and properties. (Li and Calantone,
1998)

Survey 1 scale: Likert l — 7.

Surveys 2-3 change score: Likert -3 - +3

5.06
(1.27)

5.51
(1.73)

6.27
(2.04)

 

 

Market conditions are fully considered starting from the
very early phases in new or next generation product
development. (Li and Calantone, 1998)

Survey 1 scale: Likert 1 — 7.

Surveys 2-3 change score: Likert -3 — +3

 

4.82
(1.42)

 

5.33
(1.80)

 

5.80
(2.04)

 

*calculated by adding timel score to timeZ change score;
** calculated by adding time2 score to time3 change score

Adoption of Innovations

it

HI
‘11)

 

1.

This innovation is currently being used in new product
designs. (new based on Tomatzky and Klein, 1982)
Survey 1 scale: Likert 1 — 7.

Surveys 2-3 change score: Likert -3 — +3

6.37
(0.91)

7.63
(2.0)

8.61
(2.18)

 

 

Implementing this innovation in our ﬁrm has increased
efﬁciency in designing new products. (new based on
Tomatzky and Klein, 1982)

Survey 1 scale: Likert 1 -— 7.

Surveys 2-3 change score: Likert -3 - +3

 

5.49
(1.29)

 

6.24
(2.0)

 

6.80
(2.37)

 

240

 

 

 

3. Implementing this innovation in our ﬁrm has increased
technical performance of new products. (new based on
Tomatzky and Klein, 1982)

Survey 1 scale: Likert 1 — 7.
Surveys 2-3 change score: Likert -3 — +3

5.99
(0.87)

7.31
(1.64)

8.27
(2.18)

 

4. Implementing this innovation in our ﬁrm has increased
the cost effectiveness of new products. (new based on
Tomatzky and Klein, 1982)

Survey 1 scale: Likert 1 — 7.
Surveys 2-3 change score: Likert -3 — +3

 

 

5.49
(1.14)

 

6.49
(1.96)

 

7.35
(2.67)

 

*calculated by adding timel score to time2 change score;
** calculated by adding time2 score to time3 change score

Knowledge Exploitation

 

1. Improving our existing manufacturing skills is more

important than searching out state-of-the-art manufacturing

capabilities. (new, based on March, 1991)
Survey 1 scale: Likert 1 — 7.

Surveys 2-3 change score: Likert -3 — +3

4.29
(1.44)

5.76
(2.05)

 

2. Building competences in our product design process is
more important than testing out new methods of
developing products. (new, based on March, 1991)
Survey 1 scale: Likert 1 — 7.

Surveys 2-3 change score: Likert -3 - +3

4.82
(1.24)

5.78
(1.43)

6.53
(1.82)

 

 

3. Exploiting our existing in-house technology base is more
important than acquiring new, yet unproven technology.
(new, based on March, 1991)

Survey 1 scale: Likert 1 — 7.

Surveys 2-3 change score: Likert -3 - +3

 

4.76
(1.45)

 

5.71
(1.62)

 

6.53
(2.19)

 

*calculated by adding timel score to time2 change score;
*" calculated by adding time2 score to time3 change score

241

 

 

 

 

Marketing Proficiencies

How well does your ﬁrm perform the following marketing

 

 

 

 

 

 

 

 

 

 

activities? x F. )7 =2 )7 =3
1. Conducting market studies or market research — (i.e.,
detailed studies of market potential, customer preferences, 3.24 3.69 4.33
purchase process, etc.) (Song & Parry, 1997) (1.28) (1.75) (2.03)
Survey 1 scale: Likert 1 — 7.
Surveys 2-3 change score: Likert -3 — +3
2. Designing and implementing effective advertising
campaigns. (new) 4.35 4.76 5.22
Survey 1 scale: Likert l — 7. (1.51) (1.88) (2.15)
Surveys 2-3 change score: Likert -3 — +3
3. Distributing new product announcements. (new)
Survey 1 scale: Likert 1 — 7. 4.76 5.27 5.65
Surveys 2-3 change score: Likert -3 — +3 (1.49) (1.73) (1.98)
*calculated by adding timel score to time2 change score; D
** calculated by adding time2 score to time3 change score
R&D Proficiencies
How well does your ﬁrm perform the following engineering/
R&D activities? 3? 1:1 3? =2“ 3? =3”
1. Conducting preliminary engineering and technical
assessments. (Song & Parry, 1997) 4.61 5.47 6.24
Survey 1 scale: Likert 1 — 7. (1.38) (1.57) (1.85)
Surveys 2-3 change score: Likert -3 — +3
2. Designing of products to meet designated or revised
speciﬁcations. (Song & Parry, 1997) 5.16 6.02 6.78
Survey 1 scale: Likert 1 — 7. (1.26) (1.52) (1.69)
Surveys 2-3 change score: Likert -3 — +3

 

 

 

 

242

 

 

 

 

Conducting tests to determine basic performance against
speciﬁcations. (Song & Parry, 1997)

Survey 1 scale: Likert 1 - 7.

Surveys 2-3 change score: Likert -3 — +3

5.02
(1.53)

5.82
(1.74)

6.41
(2.11)

 

. Continuously designing for cost reduction and quality

control. (Song & Parry, 1997)
Survey 1 scale: Likert 1 — 7.
Surveys 2-3 change score: Likert -3 — +3

 

4.92
(1.66)

 

5.80
(1.70)

6.59
(1.88)

 

 

*calculated by adding timel score to timeZ change score;
** calculated by adding time2 score to time3 change score

New Product Program Success

 

1.

Relative to your ﬁrm’s maior competitors. how successful
has your new product development (N PD) programs been
in terms of proﬁts? (Calantone, Garcia, and Droge,
forthcoming)

Survey 1 scale: Likert 1 — 7.

Surveys 2-3 change score: Likert -3 — +3

1.06
(1.21)

2.67
(2.29)

 

Relative to your ﬁrm’s maior competitors, how successful
has your ﬁrm’s NPD programs been in terms of sales?
(Calantone, Garcia, and Droge, forthcoming)

Survey 1 scale: Likert 1 — 7.

Surveys 2-3 change score: Likert -3 — +3

0.96
(1.22)

1.67
(1.85)

2.22
(2.36)

 

 

Relative to your ﬁrm’s maior competitors, how successful
has your ﬁrm’s NPD programs been in terms of market
share? (Calantone, Garcia, and Droge, forthcoming)
Survey 1 scale: Likert 1 — 7.

Surveys 2-3 change score: Likert -3 — +3

 

0.94
(1.27)

 

1.57
(1.86)

2.10
(2.31)

 

 

243

 

 

 

 

 

 

 

 

 

4. Relative to your ﬁrm’s objectives, how successful has the
NPD program been in the last year in terms of market 0.69 1.04 1.41
share? (Calantone, Garcia, and Droge, forthcoming) (1.40) (1.89) (2.65)
Survey 1 scale: Likert 1 — 7.
Surveys 2-3 change score: Likert -3 — +3
*calculated by adding timel score to time2 change score;
** calculated by adding time2 score to time3 change score
Table C.1 Examples of Innovations Cited by Respondents
. self-cleaning wet vacuum ﬁlter . riﬂe scope eye piece
. swing type charger valve . heat tape
. parallel plate technology . lithography techniques
. new molding machine design . computer-aided design
. machine for manufacturing of cores . current sensor assembly
. die system for high speed stamping . appliance leakage sensors
. comparison software . archive applications

. leather drum dye

244

 

   

 

 

mm~o_-m-m-

mm_o_-m-m-

mm_o_-m-m-

.mwBoSﬁB 00:0:
-5 wqumcs 00.0 0.00:
30: :o mammsoom .m

000005
$600 Saga
$030095 :0 0000“— .m

0:30 $980358

washes
@3895 so 900& ._

 

00000005
50600000

050m 00000009
30560005

001203035...

00000.83 050m
0860200

00000009
3880005

gal—3:03:50

00000005
00560000

050m vomwouoon—
30560600

.. 91:03—03:50

200305 0053000
Q03: weigﬁwﬁ
w5>5=£ 05 8.8.0000
8 .3250 0.80m 50»
00: >5: £03035:
mlmﬂ $5300 000%

 

 

mZOF<>OZZH

 

NZOHH<>OZZH

 

_ ZO~H<>OZZH

 

 

 

05¢ 00000.04 00x00; maﬁa 2020055: 00303.a 03 0080.5 0 couﬁmoaxm 0w00§>ocv~ mo 0aﬁ0xm #0 8&5

.5503 £000 00m 0003000 0000 05 20:0 0005

245

BIBLIOGRAPHY

Abernathy, William J. and James M. Utterback (1978), "Patterns of Industrial
Innovation," Technology Review, 80(7), 41-47.

Aiken, Michael and Jerald Hage (1971), "The Organic Organization and Innovation,"
Sociology, 5, 63-82.

Alange, Sveker, Staffan J acobsson and Annika J arnehammar (1998), "Some Aspects of
an Analytical Framework for Studying the Diffusion of Organizational Innovations,"
Technology Analysis & Strategic Management, 10 (1), 3-21.

Anderson, James C. and David W. Gerbing (1988), "Structural Equation Modeling in
Practice: A Review and Recommended Two-step Approach," Psychological Bulletin, 103
(3), 411-23.

Arbuckle, James L. (1996), "Full Information Estimation in the Presence of Incomplete

Data," in Advanced Structural Equation Modeling Issues and Techniques. George A.
Marcoulides and Randall E. Schumaker ed. Mahwah, New Jersey: Lawrence Erlbaum
Associates.

Armour, H. and David J. Teece (1978), "Organizational Structure and Economic
Performance: A Test of the M-Form Hypothesis," Bell Journal of Economics, Spring.

Arthur, W. Brian (1990), "Positive F eedbacks in the Economy," Scientiﬁc American, 92-
99.

Ash, Robert, and Dwight E. Smith-Daniels (1999), The Effects of Learning, Forgetting,
and Relearning on Decision Rule Performance in Multiproject Scheduling. Decision
Sciences, 30 (1), 47-82.

Bak, Per (1996), How Nature Works, the Science of Self-Organized Criticality. New
York: Copernicus.

Baldridge, J. Victor and Robert A. Burnham (1975), "Organizational innovation;
individual, organizational, and environmental impacts," Administrative Science
Quarterly, 20, 165-76.

Bass, Frank (1969), "A New Product Growth Model for Consumer Durables,"
Management Science, 15 ((January)), 215-27.

Bentler, Peter M. (1995), EQS Structural Equations Program Manual. Encino, CA:
Multivariate Software.

246

Berthon, Pierre, James M. Hulbert, and Leyland F. Pitt (1999), "To Serve or Create?
Strategic Orientations Toward Customers and Innovation," California Management
Review, 42 (1), 37-58.

Bijleveld, Catrien OJ. and Leo J. Th. van der Kamp with Ab Mooijart, Willem A. van
der Kloot, Rien van der Leeden, & Eeke va der Burg (1998), Longitudinal Data Analysis
- Designs, Models and Methods. London: Sage Publications.

Black, Laura J. and Nelson P. Repenning (2001), Why ﬁre ﬁghting is never enough:
preserving high-quality product development. Systems Dynamics Review, 17 (1), 33-62.

Blau, Judith R. and William McKinley (1979), "Idea, Complexity and Innovation,"
Administrative Science Quarterly, 24, 200-19.

Blau, Peter M. and Richard Schoenherr (1971), The Structure of Organizations. New
York: Basic Books.

Bollen, Kenneth A. (1989), Structural Equations with Latent Variables. New York: John
Wiley & Sons.

Bourgeois, L. J ., III and Kathleen M. Eisenhardt 1988, "Strategic Decision Processes In
High Velocity Environments." Management Science, 34(7) 816-835.

Browne, Michael W. and Robert Cudeck (1993), "Alternative Ways of Assessing Model
Fit," in Testing Structural Equation Models, Kenneth Bollen and J. Scott Long, Eds.
Newbury Park, CA: Sage.

Brown, Shona L, and Kathleen M. Eisenhardt (1995), "Product Development: Past
Research, Present Findings, and Future Directions," Academy of Management Review, 20
(2), 343-78.

Burgelman, Robert A. (1983) "Corporate Entrepreneurship and Strategic Management:
Insights from a Process Study." Management Science, 29, 1349-63.

Burns, Tom and GM. Stalker (1961), The Management of Innovation. Oxford: Oxford
University Press.

Calantone, Roger, Rosanna Garcia, Cornelia Droge (forthcoming), "The Effects of
Environmental Turbulence on New Product Development Strategy Planning," Journal of
Product Innovation Management.

Carroll, John S., John D. Sterman, and Alfred Markus (1994), "Playing the maintenance
game: how mental models drive organization decisions," Debating Rationality:

Nonrational Elements of Organizational Decision Making, Robert. Stern and Jennifer J.
Halpem, Ed. Ithaca, NY: ILR Press.

247

Chan, David (1998), "The conceptualization and analysis of change over time: An
integrative approach incorporating longitudinal mean and covariance structures analysis
(LMAC S) and multiple indicator latent growth modeling (MLGM)", Organizational
Research Methods, 1.

Christensen, Clayton, M (1997), The Innovator 's Dilemma: When New Technologies
Cause Great Firms to Fail. Boston, MA: Harvard Business School Press.

Cohen, Wesley M. and Daniel A. Levinthal (1990), "Absorptive Capacity: A New
Perspective on Learning and Innovation," Administrative Science Quarterly, 35, 128-52.

Colarelli O'Connor, Gina (1998), "Market Learning and Radical Innovation: A Cross
Case Comparison of Eight Radical Innovation Projects," Journal of Product Innovation
Management, 15 (2), 151-66.

Cooper, Robert.G. (1979), “The Dimensions of Industrial New Product Success and
Failure”, Journal of Marketing 43, (Summer), 93-103(1979)

---- (1984a), "New Product Strategies: What Distinguishes the Top Performers?," Journal
of Product Innovation Management, 2, 151-64.

---- (1984b), "New Product Strategies Impact on Performance," Journal of Product
Innovation Management, 2 (5-18).

---- (1993), Winning at New Products: Accelerating the Process ﬁ-om Idea to Launch
(2nd ed.). Reading, MA: Addison-Wesley Publishing Company.

---- and Elko J. Kleinschmidt (1995), "Benchmarking the F irm's Critical Success Factors
in New Product Development," Journal of Product Innovation Management, 12, 374-91.

Coyle, R.G. (1996), System Dynamics Modeling: A Practical Approach. London, UK:
Chapman and Hall Publishers.

Creswell, John W. (1998), Qualitative Inquiry and Research Design: Choosing Among
Five Traditions. Thousand Oaks: Sage Publications.

Curran, Patrick J. (2000) "A Latent Curve Framework for Studying Developmental
Trajectories of Adolescent Substance Use," in Multivariate Applications in Substance
Use Research. L. Chassin J. Rose, C. Presson, and J. Sherman ed. Hillsdale, New Jersey:
Lawerence Erlbaum Associates.

Czepiel, John. A. (1974), "Patterns of Interorganizational Communications and the

Diffusion of a Maj or Technological Innovation in a Competitive Industrial Community,"
Academy of Management Journal, 18 (1), 6-24.

248

 

Cyert, Richard M. and James G. March (1963), A Behavioral Theory of the Firm.
Englewood Cliffs, NJ: Prentice Hall.

Daﬁ, Richard L. and Selwyn W. Becker (1978), The Innovation Organization:
Innovation Adoption in School Organizations. New York: Elsevier.

Damanpour, Fariborz (1991), "Organizational Innovation: A Meta-Analysis of Effects of
Determinants and Moderators," Academy of Management Journal, 34 (3), 555-90.

---- and Evan, W.M. (1984), "Organizational Innovation and Performance: The Problem
of Organizational Lag," Administrative Science Quarterly, 29, 392-409.

Daneke, Gregory A. (1999), Systemic Choices: Nonlinear Dynamics and Practical
Management. Ann Arbor: The University of Michigan Press.

Day, George (1994), "The Capabilities of Market Driven Organizations," Journal of
Marketing, 58 (October), 37-52.

Dewar, Robert D. and Jane E. Dutton (1986), "The Adoption of Radical and Incremental
Innovations: An Empirical Analysis," Management Science, 32 (11), 1422-33.

Dillman, Don A. (1978), Mail and Telephone Survey: The Total Design Method. New
York: John Wiley & Sons.

Dosi, Giovanni (1990), "Finance, Innovation, and Industrial Change," Journal of
Economic Behavior and Organization, 13, 229-319.

Dougherty, Deborah and Cynthia Hardy (1996), "Sustained product innovation in large,
mature organizations: Overcoming innovation-to-organization problems," Academy of
Management Journal, 39(5), 1120-53.

Drucker, Peter Ferdinand (1985), Innovation and Entrepreneurship: Practice and
Principles. New York: Harper & Row.

Duncan, Terry E, S. Duncan, L. Strycker, F. Li, and A. Alpert. (1999), An Introduction to
Latent Variable Growth Curve Modeling. Mahwah New Jersey: Lawrence Erlbaum
Associates.

Eisenhardt, Kathleen M. 1989. "Building Theories From Case Study Research." The
Academy of Management Review, 14(4), 532.

---- and Behnam N. Tabrizi (1995), "Accelerating Adaptive Processes: Product

Innovation in the Global Computer Industry," Administrative Science Quarterly, 40, 84-
110.

249

Evan, William, and Guy Black (1967), "Innovations in Business Organizations: Some
factors associated with success or failure," Journal of Business, 40, 519-30.

Fenell, Mary L. (1984), "Synergy, Inﬂuence, and Information in the Adoption of
Administrative Innovations," Academy of Management Journal, 27 (1), 113-29.

Fomell, Claes and David F. Larker (1981), "Evaluating Structural Equation Models with
Unobservable Variables and Measurement Error," Journal of Marketing Research, XVIII
((February)), 39-50.

Forrester, Jay. W. (1968), Principles of Systems. Cambridge: Massachusetts: Productivity
Press.

---- (1987), "Fourteen 'Obvious Truths'," System Dynamics Review, 3 (no. 2), 156-59.

---- (1994), Policies, decisions, and information sources for modeling. Modeling for
Learning Organizations, J. D.W. Morecroft and J. D. Sterman, Ed. Portland, OR:
Productivity Press.

---- and Peter Senge (1980), Tests for building conﬁdence in system dynamics models.
System Dynamics: T IMS studies in the Management Sciences, J. W. Forrester, A.
Legasto, and J. Lyneis, Ed. Vol. 14. New York: North-Holland.

Freeman, Christopher (1988), Diffusion: The Spread of New Innovation Technology and
Finance. Oxford, England: Basil Blackwell.

---- (1994), "Critical Survey: The Economics of Technical Change, " Cambridge Journal
of Economics, 18 (5), 463-514.

Garcia, Rosanna and Roger Calantone (2001), “A Critical Look at Technological
Innovation Typology and Innovativeness Terminology: A Literature Review”, Journal of
Product Innovations Management.

---, Roger Calantone, and Ralph Levine (2002), “The Role of Knowledge in Resource
Allocation to Exploration vs. Exploitation in Technologically Oriented Organizations”,
working paper.

Goodman, Michael R. (1974), Study Notes in System Dynamics. Cambridge:
Massachusetts: Wright-Allen Press, Inc.

Grant, Robert M. (1996), "Toward a knowledge-based theory of the F irm," Strategic
Management Journal, 17 (Winter Special Issue), 109-22.

Grifﬁn, Abbie (1997), "PDMA research on new product development practices:

Updating trends and benchmarking best practices," Journal of Product Innovation
Management, 14 (6), 429-58.

250

 

 

---- and Albert L. Page (1993), "An interim Report on Measuring Product Development
Success and Failure," Journal of Product Innovation Management, 10, 291 -308.

---- and John R. Hauser (1996), "Integrating R&D and Marketing: A Review and
Analysis of the Literature," Journal of Product Innovation Management, 13, 191-215.

---- and Albert L. Page (1996), "PDMA Success Measurement Project: Recommended
Measures for Product Development Success and Failure." Journal of Product Innovation
Management, 13, pp. 478-96.

Gujarati, Damodar N. (1995), Basic Econometrics. New York: McGraw-Hill, Inc.

Han, J in K., Namwoon Kim, and Rajendra K. Srivastava (1998), "Market Orientation and

Organizational Performance: Is Innovation a Missing Link?," Journal of Marketing, 62
(October), 30-45.

Hannan, Michael T. and John Freeman (1984), "Structural Inertia and Organizational
Change," American Sociology Review, 49, 149-64.

Hu, Li-tze and Peter M. Bentler (1995), "Evaluating Model Fit," in Structural Equation
Modeling Concepts, Issues, and Applications. Rick H. Hoyle ed. Thousand Oaks: Sage
Publications.

Hurley, Robert and G. Tomas M. Hult (1998), "Innovation, Market Orientation, and
Organizational Learning: An Integration and Empirical Examination," Journal of
Marketing, 62 ((July)), 42-54.

J anszen, Felix (2000), The Age of Innovation. London: Prentice Hall.

J ap, Sandy D. (1999), "Pie-Expansion Efforts: Collaboration Processes in Buyer-Supplier
Relationships," Journal of Marketing Research, XXXVI (November), 461-75.

J aworski, B. and Ajay Kohli (1993), "Market Orientation: Antecedents and
Consequences," Journal of Marketing, 57(3), 53-70.

Johnson, Gerry (1988), "Rethinking Incrementalism," Strategic Management Journal, 9
(1), 75-91.

Kamien, Morton I. and Nancy L. Schwartz (1982), Marketing Structure and Innovation.
Cambridge: Cambridge University Press.

Kimberly, John R. (1981), "Managerial Innovation," in Handbook of Organizational

Design, Paul C. Nystrom and William H. Starbuck, Ed. Vol. 1. New York: Oxford
University Press.

251

Klein, Katherine J. and Sorra, Joann Speer (1996) "The Challenge of Innovation
Implementation" The Academy of Management Review, 21(4), 1055 -71.

Knight, Kenneth (1967), "A Descriptive Model of the Intra-ﬁrm Innovation Process," The
Journal of Business, 40, 478-96.

Kogut, Bruce and Udo Zander (1992), "Knowledge of the Firm, Integration Capabilities,
and the Replication of Technology," Organization Science, 3, 383-97.

Kohli, Ajay K. and Bernard J. Jaworski (1990), "Market Orientation: The Construct,
Research Propositions, and Managerial Implications," Journal of Marketing, 54 ((April)),
1-18.

Kuczmarski, Thomas D (1988) "Success Isn't Always its Own Reward -- Big Bucks
Help," Marketing News, 22(24), 10.

Lant, Theresa K. (1992), "Aspiration Level Adaptation: An Empirical Exploration,"
Management Science, 38 (5), 623-44.

Layton, E. (1977), "Conditions of Technological Development," in Science, Technology,
and Society, I. and de Solla Price Spiegel-Rosing, D., Ed. Beverly Hills, CA: Sage.

Leonard-Barton, Dorothy (1988), "Implementation as Mutual Adaptation of Technology
and Organization," Research Policy, 17, 251-67.

---- (1992), "Core Capabilities and Core Rigidities: A Paradox in Managing New Product
Development," Strategic Management Journal, 13, 11 1-25.

---- (1995), Wellsprings of Knowledge: Building and Sustaining the Sources of
Innovation. Boston, Massachusetts: Harvard Business Press.

Levine, Ralph L. (2002), personal correspondence.

----, Mary Van Sell, and Beth Rubin (1992), "System Dynamics and the Analysis of
Feedback Processes in Social and Behavioral Systems," in Analysis of Dynamic
Psychological Systems, Ralph L. and Hiram E. Fitzgerald Levine, Ed. Vol. 1. New York:
Plenum Press.

----, A. Leholm, and R.Vlasin (2001), "Come be a Leader in a Self-Directed Work Team:
The Dynamics of the Transition from being a Supervisor to a Team Leader," 2001
International Systems Dynamics Conference Proceedings. Atlanta, GA.

Levinthal, Daniel A. and James G. March (1981), " A model of adaptive organizational
search." Journal of Economic Behavior and Organization, 2, pp. 307-33.

252

---- and James G. March (1993), "The Myopia of Learning," Strategic Management
Journal, 14, 95-112.

Lewin, Arie Y., Chris P. Long and Timothy N. Carroll (1999), "The Coevolution of New
Organizational F orms," Organization Science, 10 (5, September-October), 535-50.

Li, Tiger and Roger J. Calantone (1998), "The Impact of Market Knowledge Competence

on New Product Advantage: Conceptualization an Empirical Examination," Journal of
Marketing, 62 (October), 13-29.

Little, Roderick J. A. and Donald B. Rubin (1987), Statistical Analysis with Missing
Data. New York: Wiley.

Maani, K.E. and R.Y. Cavana (2000), Systems Thinking and Modeling: Understanding
Change and Complexity. Auckland, New Zealand: Prentice Hall.

Madhavan, Ravindranath and Raj iv Grover (1998), "From Embedded Knowledge to
Embodied Knowledge: New Product Development as Knowledge Management," Journal
of Marketing, 62, 1-12.

Mansﬁeld, Edwin (1968), Economics of Technological Change. New York: Norton.

March, James G. (1991), "Exploration and Exploitation in Organizational Learning,"
Organization Science, 2, 71-87.

Meredith, W. and J. Tisak (1990), "Latent Curve Analysis." Psychometrika, 55, pp. 107-
22.

Meyer, Alan D. (1982), "Adapting to Environmental Jolts," Administrative Science
Quarterly, 27, 515-37.

---- and James B. Goes (1988), "Organizational Assimilation of Innovations: A
Multilevel Contextual Analysis," Academy of Management Journal, 31 (4), 897-923.

Miles, Raymond E. and Charles C. Snow (1978), Organizational Strategy, Structure &
Process. New York: McGraw-Hill Publishing Company.

Mileti, Dennis S., David F. Gillespie, and J. Eugene Haas (1977), "Size and Structure in
Complex Organizations," Social Forces, 56, 208-17.

Miller, Danny (1990), The Icarus Paradox. New York: Harper Business.

---- (1991), "Stale in the Saddle, CEO Tenure and the Match between Organizational
Strategy," Management Science, 37 (1), 34-53.

253

---- and Peter H. Friesen (1982), "Innovation in Conservative and Entrepreneurial Firms,"
Strategic Management Journal, 3, 1-25.

Milling, Peter M. (1996), "Modeling Innovation Processes for Decision Support and
Management Simulation", Systems Dynamics Review, 12 (3), 211-34.

Moch, Michael K. and Edward V. Morse (1977), "Size, Centralization and
Organizational Adoption of Innovations," American Sociology Review, 42, 716-25.

Moenaert, Rudy K. and William E. Souder (1990), "An Information Transfer model for
integrating market and R&D Personnel in New Product Development Projects," Journal
of Product Innovation Management, 7 (2), 91-107.

Mohr, Lawrence B. (1969), "Determinants of Innovation In Organizations," American
Political Science Review, 63, l 1 1-26.

Montya-Weiss, Mitzi M. and Roger Calantone (1994), "Determinants of New Product
Performance: A Review and Meta-Analysis," Journal of Product Innovation
Management, 11, 1-21.

Morecroft, John D. W. (1985), “Rationality in the analysis of behavioral simulation
models”, Management Science, 31 (5), 900-16.

----- and Sterman J .D., Eds. (1994), Modeling for Learning Organizations. Portland, OR:
Productivity Press.

Narver, John, Stanley F. Slater, and Douglas L. MacLachlan (2000), "Total Market
Orientation: Proactive and Responsive Behavior". Cambridge, MA: Marketing Science
Institute Working Paper Series.

---- and Stanley F. Slater (1990), "The Effect of a Market Orientation on Business
Proﬁtability," Journal of Marketing, 54 (October), 20-35.

Nelson, Richard R. and Sidney G. Winter (1974, 1982), An Evolutionary Theory of
Economic Change. Cambridge, MA: The Belknap Press.

Nonaka, Ikujiro (1994), "A Dynamic Theory of Organizational Knowledge Creation,"
Organization Science, 5 (1), 14-36.

----and Hirotaka Takeuchi (1995), The Knowledge Creating Company. New York, NY:
Oxford University Press.

Nunnally, Jum C. (1978), Psychometric Theory. New York: McGraw-Hill Book
Company.

254

Perez, Carlota (1983), "Structural Change and Assimilation of New Technologies in the
Economic and Social Systems," Futures, 15 (4), 357-75.

Pettigrew, A. (1988), "Longitudinal ﬁeld research on change: Theory and Practice," in
National Science Foundation Conference on Longitudinal Research Methods in
Organizations. Austin, TX.

Phillips, Fred and Sugandha D. Tuladhar (2000), "Measuring Organizational Flexibility:
An Exploration and General Model," Technology Forecasting and Social Change, 64,
23-38.

Plotkin, Henry C. (1994), Darwin Machines and the Nature of Knowledge, Cambridge,
MA: Harvard University Press.

Quinn, James Brian (1992), Intelligent Enterprise. New York: Free Press.

Repenning, Nelson P. (2000), "A Dynamic Model of Resource Allocation in Multi-
project Research and Development Systems," System Dynamics Review, 16 (3), 173-212.

---- (2001). Understanding Fire Fighting in New Product Development. Journal of
Product Innovation Management, 18 (5), 285-300.

---- (2001), "A Simulation-based Approach to Understanding the Dynamics of Innovation
Implementation,": MIT System Dynamics Group, working paper.

Rice, RE. and Rogers, Everett (1980), "Reinvention in the Innovation Process,"
Knowledge, 1, 499-514.

Robertson, Thomas S. and Hubert Gatignon (1986), "Competitive Effects on Technology
Diffusion," Journal of Marketing, (July), 1 -1 3.

Rogers, Everett M. (1983, 1995), Diffusion of Innovations. New York: The Free Press.
Romanelli, Elaine and Michael L. Tushman (1994), "Organizational Transformation as
Punctuated Equilibrium: An Empirical Test," Academy of Management Journal, Vol. 37
(No. 5), 1141-66.

Rosner, M.M. (1968), "Economic Determinants of Organizational Innovation,"
Administrative Science Quarterly, 12, 614-25.

Rothwell, Roy and Paul Gardiner (1988), "Reinnovation and Robust Designs: Producer
and User Beneﬁts," Journal of Marketing Management, 3 (3), 372-87.

Rowe, Lloyd A and William B. Boise (1974), "Organizational Innovation: current
research and evolving concepts," Public Administration Review, 34 (284-293).

255

Ruekert, Robert W. and Orville C. Walker (1987), "Marketing's Interaction with Other
Functional Units: A conceptual Framework and Empirical Evidence," Journal of
Marketing, 51, 1-19.

Sastry, Anjali (1997), "Problems and Paradoxes in a Model of Punctuated Organizational
Change," Administrative Science Quarterly, 42, 237-75.

Schumpeter, John A. (1934), The Theory of Economic Development ((English translation
from 1912 German edition, Leipzig) ed.). Cambridge, MA: Harvard University Press.

Senge, Peter M. (1990), The Fifth Discipline: The Art and Practice of the Learning
Organization. New York: Doubleday Currency.

Slater, Stanley F. and John C. Narver (1995), "Market Orientation and the Learning
Organization," Journal of Marketing, 59 (July), 63-74.

Smith, Bruce LR. and Claude E. Barﬁeld Ed. (1996), Technology, R&D, and the
Economy. Washington, DC: The Brookings Institution and American Enterprise
Institute.

Song, X. Michael and Mark E. Parry (1997), "The Determinants of Japanese New
Product Successes, " Journal of Marketing, 34(February), 64-76.

Spender, J .C. (1996), "Making Knowledge the Basis of a Dynamic Theory of the Firm,"
Strategic Management Journal, 17 (Winter Special Issue), 45-62.

SPSS, Inc. (1999), SPSS Base 10.0.7 Applications Guide, Chicago, IL: SPSS Inc.

Sterman, John D. (1987), "Testing behavioral simulation models by direct experiment"
Management Science, 33 (12), 1572-92.

---- (2000), Business Dynamics: Systems Thinking and Modeling for a Complex World.
Boston: Irwin McGraw Hill.

Stoolmiller, Mike (1995), "Using Latent Growth Curves to Study Developmental
Processes," in The Analysis of Change. John M. Gottrnan ed. Cambridge, MA:
Cambridge University Press, pp. 103-38.

Sultan, Fareena, John U. Farley, and Donald R. Lehmann (1990), "A Meta-Analysis of
Applications of Diffusion Models," Journal of Marketing Research, Vol. XXVH

((February)), 70-7.

Tomatzky, Louis, E. Fergus, J. Avellar, and G.W. F airweather (1980), Innovation and
Social Process. Elmsford, NY: Pergamon Press.

256

Tomatzky, Louis and Katherine J. Klein (1982), "Innovation Characteristics and
Adoption-Irnplementation: A Meta-Analysis of Findings," IEEE Transactions on
Engineering Management, EM-29 (1), 28-45.

Tushman, Michael L and Elaine Romanelli (1986), "Convergence and Upheaval:
Managing the unsteady pace of organizational evolution," California Management
Review, 29 (1), 1-16.

Tyre, Marcie J. and Wanda J. Orlikowski (1994), "Windows of Opportunity: Temporal

Patterns of Technological Adaptation in Organizations," Organization Science, 5 (N o. 1),
98-118.

Utterback, James M. (1996), Mastering the Dynamics of Innovation. Boston, MA:
Harvard Business School Press.

---- and William J. Abernathy (1975), "A Dynamic Model of Process and Product
Innovation," Omega, 33, 639-56.

Van de Ven, Andrew H. and Everett Rogers (1988), "Innovations and Organizations -
Critical Perspectives," Communication Research, 15, 632-51.

---- and MS. Poole (1990), "Methods for studying Innovation Development in the
Minnesota Innovation Research Program," Organization Science, 1 (3), 313-35.

---- D. Polley, R. Garud, and S. Venkatararnan (1999), The Innovation Journey, New
York: Oxford University Press.

Ventana Systems, Inc (2001), "Vensim PLE,". Harvard, MA: Ventana Systems, Inc.

Walker, Jack L. (1969), "The diffusion of Innovation among the American States,"
American Political Science Review, 63, 880-99.

Weick, Karl E. (1998), "Improvisation as a mindset for organizational analysis,"
Organization Science, 9 (5), 543-55.

Willet, John B. and Aline G. Sayer (1994) "Using Covariance Structure Analysis to
Detect Correlates and Predictors of Individual Change over Time." Psychological
Bulletin, 116:2, pp. 263-381.

Workman, John P. Jr. (1993), "Marketing's Limited Role in New Product Development in
one Computer Systems Firm," Journal of Marketing Research, XXX, 405-21.

Yin, Robert K. and Donald T. Campbell (1989), Case Study Research: Design & Method.
Newbury Park, CA: Sage.

257

Zaltman, Gerald, Robert Duncan, and J onny Holbek (1973), Innovations and
Organizations. New York, NY: Wiley.

Ziman, John Ed. (2000), Technological Innovation as an Evolutionary Process.
Cambridge: University Press.

258