PLACE IN RETURN BOX
to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

DATE DUE

DATE DUE

 

mum;
n t
54000 ”“

 

tmmszmus

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

use WM“

COORDINATION AND CONTROL OF RETAILER-SUPPLIER TRANSACTIONS:
FACTORS INFLUENCING ORGANIZATIONAL ADOPTION AND USE OF
ELECTRONIC INFORMATION NETWORKS

By

Alice Pikyan Chan

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Mass Media Doctoral Program
Departments of Advertising and Telecommunication, and School of Journalism
College of Communication Arts and Sciences

1997

ABSTRACT
COORDINATION AND CONTROL OF RETAILER-SUPPLIER TRANSACTIONS:

FACTORS INFLUENCING ORGANIZATIONAL ADOPTION AND USE OF
ELECTRONIC INFORMATION NETWORKS

By

Alice Pikyan Chan

This dissertation research investigates the factors inﬂuencing organizational
adoption and use of electronic information networks. It focuses speciﬁcally on
retailer-supplier transactions associated with product acquisitions. Derived from
transaction cost analysis, several factors antecedent to electronic network adoption and
use are proposed; they include uncertainty, asset speciﬁcity and complexity of product
description. In addition, three social relational predictor variables are tested, namely
interpersonal relationships, mutual commitment and trust. One of the dependent
variables is an information technology cluster, which measures the adoption of six
information technologies that enable electronic exchange of information. The other
major criterion variable deals with the perceived importance of electronic network use to
supporting transactions associated with product acquisitions.

Empirical observations were obtained from a telephone survey of 143 US.
grocery retailers. Interviews focused on the way in which product buyers acquired key
products from their major suppliers. One main ﬁnding is that signiﬁcant predictors of
technological adoption and network use are different. In particular, adoption is more
likely when technology is viewed as a functional substitute for two social relational

modalities of coordination and control: interpersonal relationships with and trust in the

supplier. Trust and the importance of electronic network use also appear to be
substitutable in managing transactions that involve routine, straightforward
communication of price and order information. Furthermore, in executing these same
types of transactions, using an electronic interconnection with a supplier appears to be a
tangible indication of mutual commitment between the trading parties. When the
transactions involve higher complexity, ambiguity or need for recursive interactions with
the supplier (e.g., negotiations, monitoring and problem solving), a different group of
factors are found to be correlated with electronic network use. First, electronic network
use is perceived as complementary to interpersonal relationships as a coordination and
control mechanism. In addition, electronic network use is strongly associated with high
levels of complexity of product description: by offering sophisticated capabilities for
representing and transmitting information, technologies facilitate the communication of a
product’s critical features, which are otherwise difﬁcult to describe. All analyses
controlled for the impact of organizational size and retailer-supplier integration.

In conclusion, this research shows that social relational factors inﬂuence
organizational adoption and use of electronic networks in ways that are not explained by
transaction cost variables. Future research efforts devoted to the reﬁnement of the
conceptualizations and measurement proposed in this study can further test the relative
utility of transaction cost and social relational analyses. At a practical level, ﬁndings in
this research suggest that retailers should not be concerned with technological
implementation costs alone. Decisions to adopt and use electronic networks need to
factor in the varying complementary and substituting relationships technologies have

with the pre-existing social relational conditions characterizing trade with suppliers.

Copyright by
ALICE PIKYAN CHAN
1997

ACKNOWLEDGMENTS

This dissertation signiﬁes a triumphant—yet humbling—transition in my
intellectual development as a ﬂedgling scholar. More importantly, it will always remind
me of how blessed I am for having crossed paths with the individuals who have leﬁ
memorable imprints in the Michigan State University segment of my academic journey.

First of all, I would like to thank Dean Gerard Crawley and Dr. Les Manderscheid
of the Graduate School for arranging funding for this dissertation project as part of my
University Distinguished Fellowship Award (1994-1998). The support has helped me
achieve the goal of ﬁnishing up my doctoral program one semester ahead of schedule.

Next, there are a number of faculty members whose contributions I would like to
acknowledge. First, I am greatly indebted to Dr. Charles Steinﬁeld, who, inter alia, has
been instrumental in my preparation to be a researcher in information technology and
organizations. As my major professor, Dr. Steinﬁeld gave me many wonderful
opportunities and invaluable experience, all of which will undoubtedly serve me well as I
continue to develop my scholarship. This dissertation is an exemplar of his inﬂuence,
although only I should be faulted for any ﬂaw. I am also particularly grateful to Dr.
David Johnson for his insights and zeal as one of my advisers. He has always managed
to ask me just the “right” questions, skillfully steering me toward ﬁlling the missing links

in my logic and thinking. In addition, a note of gratitude is owed to Dr. Johannes Bauer

and Dr. Thomas Muth for their guidance, and for having graciously taken the time out
from their demanding schedules to serve on my dissertation committee. Last, but not
least, words indeed fail me in my wish to thank Dr. Charles Salmon, who has inspired me
to love my work as a researcher and a teacher. Since very early on in my doctoral
program, he has had tremendous faith in my scholarly abilities and potential, and went
above and beyond the call of duty to help me get situated in an exceptional start-up
faculty position at Cornell University. For these reasons, and more, I shall always be
grateful to him. I

Furthermore, there are a number of special friends and colleagues without whom I
cannot imagine having been able to survive the last three-and-a-half years. Among them
is Teresa Mastin, my very dear friend and cohort-member. She has always been
available, lavish of genuine care, encouragement and support, helping me get through
some of the most trying moments in my life thus far. For similar reasons, a heartfelt
“Thank You!” also goes to Katherine Bradshaw and Kristine Nowak. Moreover, I would
like to extend a warm “Mahalo!”—the residual inﬂuence of having lived in Hawaii for
six years—to all my other friends, who, at one point or another, have been my support
network in Michigan. They include Robert Albers, Rick Busselle, Kenzie Cameron,
Anne Hoag, Denise Mahoney, Vanessa Pollok, and many others. Also, Michael Plummer
will never be forgotten for the role he played in my life, for better or for worse, and for
having believed in my potential to become Dr. Alice someday.

Finally, I cannot possibly conclude without thanking my family in California and
Macau. I am especially grateful to my mother for her faith and support, but, above all,

for being the single person in this world who truly loves me unconditionally.
vi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. ix

LIST OF FIGURES ............................................................................................................. x
Chapter 1

INTRODUCTION ............................................................................................................... 1

Why Do (Should) Organizations Use Information Technology? ............................ 1

Information Technology: Communication for Coordination and Control ............... 2

The Inter-organizational Context ................................................................. 4

Transaction Cost Analysis ........................................................................... 6

Social Relational Analysis ........................................................................... 8

Research Question ................................................................................................... 9

Overview of this Dissertation ................................................................................ 1 1
Chapter 2

REVIEW OF THEORETICAL AND EMPIRICAL LITERATURE ................................ 13

Transaction Cost Factors ........................................................................................ 14

Transaction Cost Theory ............................................................................ 14

Electronic Markets and Electronic Hierarchies .......................................... 18

Social Relational Factors ....................................................................................... 24

Networks of Social and Interpersonal Relationships ................................. 25

Mutuality and Trust .................................................................................... 27

Chapter Summary .................................................................................................. 33
Chapter 3

METHODS AND PROCEDURES .................................................................................... 35

Measurement .......................................................................................................... 36

Use of Perceptive Measures ....................................................................... 37

Dependent Variables: Electronic Network Adoption and Use ................. 39

Independent Variables: Transaction Cost Factors .................................... 44

Independent Variables: Social Relational Factors .................................... 49

Control Variables ....................................................................................... 52

Data Collection ...................................................................................................... 55

Population (Industry) Selection ................................................................. 55

Sampling Procedures ................................................................................. 57

Telephone Survey ...................................................................................... 59

vii

Chapter Summary .................................................................................................. 62

Chapter 4
RESULTS .......................................................................................................................... 63
General Data Preparation ....................................................................................... 64
Data Integrity Checks ................................................................................ 64
Recoding of Scale and Response Categories ............................................. 65
Missing Data Treatment ............................................................................. 68
Construction of Index Measures ............................................................................ 74
Construct Validity and Factor Analysis ..................................................... 76
Reliability ................................................................................................... 84
Hypotheses Testing ................................................................................................ 87
Regression Analysis ................................................................................... 87
Regression Results ..................................................................................... 91
Chapter Summary ................................................................................................ 100
Chapter 5
DISCUSSION .................................................................................................................. 102
Adoption and Use of Electronic Information Networks ...................................... 103
Adoption vs. Usage .................................................................................. 103
Importance of Electronic Network Use to “Static” vs. “Dynamic”
Activities ...................................................................................... 109
Transaction Cost Hypotheses (H1 to H3) ............................................................ 112
Uncertainty (Hypothesis 1) ...................................................................... 113
Asset (Product, Time and Knowledge) Speciﬁcity a-Iypothesis 2) ......... 115
Complexity of Product Description (Hypothesis 3) ................................. 118
Social Relational Hypotheses (H4 to H6) ............................................................ 121
Importance of Interpersonal Relationships (Hypothesis 4) ...................... 121
Mutual Commitment (Hypothesis 5) ....................................................... 124
Trust (Hypothesis 6) ................................................................................ 127
Chapter Summary ................................................................................................ 129
Chapter 6
SUMMARY AND CONCLUSIONS .............................................................................. 130
Summary of Findings ........................................................................................... 130
Limitations ........................................................................................................... 133
Research Design ....................................................................................... 133
Model Speciﬁcations ............................................................................... 136
Implications .......................................................................................................... 138
Conclusions .......................................................................................................... 140
APPENDIX A — Survey Instrument ................................................................................ 141
APPENDIX B - Survey Log ........................................................................................... 158
REFERENCES ................................................................................................................ 159

viii

LIST OF TABLES

Table 1 -— Integration with Major Supplier ........................................................................ 54
Table 2 — Product Speciﬁcity ............................................................................................. 65
Table 3 — Time Speciﬁcity ................................................................................................. 66
Table 4 - Total Annual Sales ............................................................................................. 67
Table 5 - Total Number of Employees .............................................................................. 67
Table 6 — Total Number of Stores ...................................................................................... 68
Table 7 - Comparison of Electronic Network Users and Non-Users ................................ 72
Table 8 — Indices Resulting from Factor Analysis ............................................................. 81
Table 9 — Importance of Electronic Network Use ............................................................. 82
Table 10 — Internal Consistency Reliability of Index Measures ........................................ 86
Table 11 — Pearson Product-Moment Correlations ............................................................ 89
Table 12 - Regression of Information Technology Cluster ............................................... 93

Table 13 — Regression of Importance of Electronic Network Use to “Static”
Activities .................................................................................................... 96

Table 14 - Regression of Importance of Electronic Network Use to “Dynamic”
Activities .................................................................................................... 97

Table 15 — Summary of Findings ..................................................................................... 101

ix

LIST OF FIGURES

Figure 1 - Factors Inﬂuencing Organizational Electronic Network Adoption
and Use ....................................................................................................... 34

Figure 2 — Importance of Electronic Network Use to Various Product Acquisition
Activities .................................................................................................... 83

Chapter 1

INTRODUCTION

Why Do (Should) Organizations Use Information Technology?

The question‘of why organizations ought to use information technology is not a
new area of inquiry. Researchers spanning across several decades have been intrigued
with what the antecedents are to the successful diffusion and adoption of innovations
(e.g., Burns & Stalker, 1961; Rogers, 1962, 1986, 1995). Contemporary literature
suggests that organizations use information technology to achieve greater efﬁciencies in
their internal operations and transactions with external suppliers and customers. This
desire to be more efficient can be traced to two major sources of motivations: (i) attaining
a strategic necessity, e.g., following the technological “norm” in the industry or matching
the actions of competitors; and (ii) pursuing competitive advantage, e.g., attempting to
out-perform industrial rivals or gain a larger market share (Bradley, 1993; Porter &
Millar, 1985; Sabherwal & Vijayasarathy, 1994; Venkatraman & Zaheer, 1994).1

Different theories adopted to study organizations carry different assumptions
about how ﬁrm requirements are met (e. g., managing differentiated but interdependent

units and tasks).2 In turn, predictions and explanations for organizational adoption and

 

See other contributions to the volumes edited by Bradley, Hausman and Nolan (1993)
and Allen & Scott Morton (1994) on reasons for implementing different types of
information technology, networks and systems, and their relationships with intra- and
inter-organizational structure, processes and other issues.

For instance, “organic” (as opposed to “mechanistic”) approaches that recognize
organizations as open systems call attention to the impact of external contingencies,

1

2
use of information technology vary according to the theoretical angle used. For instance,
economic theory is useful in helping to shed light on the impact of costs, such as how the
reduction in the costs of coordination can motivate technology use. Social relational
approaches can help to explain other variables affecting technological use that are less
directly related to costs, such as, how an electronic information network is perceived to ﬁt
the established network of communication and social relationships. What can be
observed is that these different perspectives complement each other in identifying
conditions driving organizational technology adoption and use.

As the introduction to this dissertation, this chapter presents the research question
addressed and how it is derived. Presented ﬁrst is a brief review of how information
technology supports communication for the purposes of coordination and control,
especially in the inter-organizational context. Integrated in the discussion is an
introduction to how transaction cost and social relational factors help to explain the way
these communication objectives can be facilitated by electronic information networks.
Following this review, the research question is stated, and the rationale and signiﬁcance

of the study presented. This chapter closes with an overview of the dissertation.

Information Technology: Communication for Coordination and Control
Many organizations, especially large ones, tend to be highly complex and

differentiated, with many inter-related functional subunits performing tasks and activities

 

(continued from previous page)
such as environmental uncertainty, on managerial decision making. See Morgan
(1986) for a comparative review of “classical” theories of organizations.

3

(Lawrence & Lorsch, 1967). As organizations become more complex and differentiated,
rules, programs and other mechanisms are needed to integrate and coordinate these
interdependent tasks and subunits (Lawrence and Lorsch, 1967; March & Simon, 195 8,
1993; Tushman & Nadler, 1978). It is through coordination that otherwise disparate
activities or events are brought into a relationship, and diversiﬁed functional units
integrated into the overall organizational structure (Allen & Hauptrnan, 1990; Frances,
Levacic, Mitchell, & Thompson, 1991). The more effective organization is one that has a
greater capacity to exchange information among subunits, thereby allowing for higher
coordination of tasks (Hart & Estrin, 1991).

Tushman and Nadler (1978) submitted that coordination is achieved when there is
a match between information processing capacities and requirements within the
organization. Information processing requirements are created by such factors as task
ambiguity and uncertainty. If these requirements could not be reduced, the organization
facing them must be able to increase its information processing capacities. The key is
that there must be a ﬁt between information processing requirements and capacities.
Because information technology effectively expands a ﬁrm’s information processing
capacities, coordination is enhanced. Therefore, facilitating the critical function of
coordinating interdependent tasks among organizational subunits becomes a driving force
for information technology use (Galbraith, 1973; Thompson, 1967).

The role played by communication and information networks in coordination is
also articulated in the conceptualization of organizations as “communication networks in

which actors or subunits recurrently process resources and information” (Dow, 1988, pg.

4
56). Based on this conceptualization, technology deﬁnes the channels and range of
information available to an organization. In addition, the signiﬁcance of communication
in coordination is explicitly stated in the deﬁnition of coordination structure as “a pattern
of decision-making and communication among a set of actors who perform tasks”
(Malone, 1987, pg. 1319).

In order for there to be successful coordination among and integration of highly
differentiated organizational units, appropriate control mechanisms are necessitated
(Beniger, 1986, 1990; Galbraith, 1973; Thompson, 1967). According to Beniger’s
characterization (1990), as control technologies, formal organizations exist “to control
large numbers of elements, excessive scope or complexity of interactions and
contingencies, or an otherwise unmanageable volume of processing” (pg. 34). He also
argued that information constitutes the basis of control, and that every task performed by
information technology is directed towards control. Extending this argument, Nass and
Mason (1990) contended that “all forms of control are based on information-processing

and feedback” (pg. 5 7).

The Inter-organizational Context

In many ways, communication is more important to coordination and control
when dealing with external suppliers or customers than internal organizational subunits
(Rockart & Short, 1989). One major reason stems from the need to reduce information
asymmetries that often exist in a trading situation. For instance, in order to know if a

potential supplier’s offer is attractive, the buyer needs information on the fair market

5
price. Another reason can be traced to the general lack of direct control on an external
trading constituent, relative to an internal department or a division of the company of
which there are ownership ties between the parties doing business with each other.

From a structural perspective, many organizational forms in existence to-date are
not strict hierarchies, operating solely on bureaucratic administrative rules, nor are they
pure markets, in which contracting relies on price competition (see contributions in
Thompson, Frances, Levacic & Mitchell, 1991). Furthermore, few organizations are
self-sufﬁcient, but rather must depend on and interact with external constituents (e. g.,
suppliers of needed raw materials) to meet various resource requirements (Pfeffer &
Salancik, 1978). These observations again point to the fact that managing
inter-organizational relationships and interdependence necessitates coordination and
control. Therefore, it should not be surprising to ﬁnd communication and information
exchanges to be central to the maintenance of inter-organizational relational networks
(F lanagin, Monge & Fulk, 1997).

Furthermore, as much as communication and interactions within the boundaries of
an organization are supported, electronic networks can facilitate the same functions of
information exchange and processing in the inter-organizational context. At an intuitive
level, electronic networks may be more important to inter-ﬁrm coordination and control,
as opportunities for face-to-face interactions are often fewer when working with external
constituents. In fact, inter-organizational networks, most notably electronic data
interchange (EDI) systems, have been recognized for helping to ease the burden of

interacting with suppliers, customers and other groups, and play a key role as a linking

6
element of business relationships (Davidow and Malone, 1992; Rockart & Benjamin,
1991). Well known cases of success include Baxter Intemational’s (previously known as
American Hospital Supply Corporation) ASAP system and McKesson’s Economost
system. These electronic supplier-customer networks have helped to boost overall sales
for these two supplier ﬁrms and improve their positions in the respective markets of
hospital supplies and drugs (Clemons, 1993). In addition, inter-ﬁrm data networks have
been found to enable full-scale coordination in form of customer-supplier partnerships,
bringing both parties beneﬁts beyond cost-savings associated with reducing inventory
levels (Konsynski & McFarlan, 1990; Malone, Yates & Benjamin, 1989). Overall
operations associated with production, delivery and processing of ordering and billing for

both the buying and selling ﬁrms are trimmed substantially.

Transaction Cost Analysis

Over the last ten years, much of the discussion on how electronic information
networks facilitate intra- and inter-organizational coordination and control has focused on
economic variables, adopting the transaction cost analytical framework (Malone et. al.,
1987).3 Electronic networks are evaluated in the context of how they relate to changes in

production and transaction costs that are associated with different coordination structures

 

In the article, Malone et. a1. (1987) focused more explicitly on the beneﬁts associated
with coordination, i.e., the reduction in transaction costs, than the issue of control.
How electronic networks can serve the control function is more implicit, as embedded
in the framework of hierarchical governance and market coordination in transaction
cost theory advanced by Williamson (1975).

7
governing any given set of economic activities (e. g., acquiring a key raw material for the
production of a product).

Coordination places intense requirements on communication and information
processing associated with such factors as uncertainty, asset speciﬁcity and complexity of
product description faced by an organization (Malone et. al., 1987). Uncertainty relates
to the unpredictability of such business environmental conditions as that associated with
the availability a key asset (e.g., a product). Asset speciﬁcity deals with the extent to
which a needed asset is speciﬁc to an organization. Complexity of product description
concerns the difﬁculty in describing the critical attributes of a product that affects buying
decisions. High levels of uncertainty, asset speciﬁcity and complexity of product
description all serve to drive up the costs of coordination.‘ Therefore, by facilitating
communication and information processing associated with these factors, electronic
networks can help to lower coordination costs (Malone, et. al., 1987, 1989).
(Conceptualization and operationalization of these variables will be discussed more in
detail in Chapters 2 and 3, respectively.)

Adopting this applied transaction cost theoretical approach, it would appear that a
key motivation for using electronic networks would be the reduction of coordination
costs. As coordination costs are driven by the levels of uncertainty, asset speciﬁcity and
complexity of product description, it is hypothesized that these three transaction cost

analytic factors will likely affect electronic networks use for coordinating transactions.

 

4

Malone et. al. (1987, 1989) used “transaction costs” to distinguish between external
coordination costs, i.e., those incurred in the market place, and internal coordination
costs, or what Williamson (1975) called “governance costs.”

Social Relational Analysis

Many social theorists posit that economic exchanges are embedded in particular
social structural environments (e.g., Bradach and Eccles, 1991; Granovetter, 1985).
Purposive actions, including those associated with the diffusion of innovations, are
shaped and constrained by the established social structures in an organization (Burt,

1982; Valente, 1995). Therefore, merely focusing on cost considerations would lead to
overlooking the impact of social relations on whether and how information networks are
used for communication, coordination and control. In support of this position, Contractor
and Eisenberg (1990) argued that in order to understand technology, one must ﬁrst
understand social relationships.

Furthermore, Nohria & Eccles (1992) observed that there has been converging use
of the label “networks” to convey both the connotation of organizational structure and the
technological sense of the term.’ Childs (1987) called this phenomenon the “mutuality”
between organizational and information technology design. These observations are
suggestive of the possibility that the established social relational structure and norms of
communication serve as the underlying commonality deﬁning both the organizational
structure of coordination and control, as well as the web of electronic links for supporting

information processing and exchange.

 

They cautioned that such converging use of the term is oﬁen misguided, and that the
structure of a technological network is not necessarily equivalent to the
communication network characterizing the organizational structure. Not to
misrepresent what these researchers wrote, the signiﬁcance of pointing out this
converging use of the term “network” is in highlighting the fact that the conﬁguration
of an electronic network is likely related to and affected by organizational structure.

9

Based on this empirically testable connection between the social relational norms
and information networks, a likely set of key antecedents to adopting and using electronic
linkages would be factors characterizing established relationships. These relationships
can be among divisions of an organization (internal) or trading ﬁrms (external). Some
important social relational variables to be considered include interpersonal relationships,
trust and mutual commitment that exist between the buyer and seller ﬁrms. If electronic
network use is indeed to be consistent with the pre-existing nouns of interactions and
patterns of relationships, these social relational variables ought to have impact on how
electronic networks are used to support these relational norms and patterns. (These
variables and their hypothesized relationships with network adoption and use will be

discussed further in the next chapter.)

Research Question

The above introduction provides the conceptual background to why information
technologies and networks are needed in organizations, and what are likely some key
conditions inﬂuencing adoption and usage. Speciﬁcally, it is proposed that there are
transaction cost and social relational factors in communication and information
exchanges within and between organizations for coordinating and controlling important
tasks. The present research integrates theoretical insights from these two bodies of
literature in the study of electronic information network adoption and use in supporting
important buyer-seller transactions. In particular, the research is aimed at addressing the

following research question:

10

What are the factors inﬂuencing organizational adoption and use of electronic
information networks in retailer-supplier transactions associated with product

acquisitions?

General reports by the trade press have continually suggested that electronic
networks, such as EDI systems, are increasingly perceived to be a signiﬁcant mechanism
for industrial buyer-seller transactions and interactions. This is especially true in the
retail sector, where electronic supplier-customer interconnections can help the ordering
and stocking of products become more efﬁcient and less costly. Anecdotal observations
aside, existing research has also been mainly conceptual and case-based with limited
generalizability. It would appear that there is a lack of empirical, generalizable insights
into the conditions favoring the adoption and use of electronic communication links to
support commercial transactions. The current research aims at ﬁlling this void,
subjecting to empirical scrutiny the hype concerning the importance of electronic
information networks to retailer-supplier transactions.

Based on this rationale, the empirical study of the above stated question entails a
survey of 143 grocery buyers in the United States. It is focused on the importance of
computer-based information networks to the acquisition of key products from major
suppliers. In turn, the signiﬁcance of the research undertaken lies in contributing to the
currently limited generalizable, empirical knowledge about the importance of electronic
interconnections to supporting retailer-supplier transactions. Key factors antecedent to

network use and adoption that are identiﬁed and tested in the current study can aid

11
managerial decisions concerning technologies. In addition, the ﬁndings obtained can
serve to guide future research aimed at continuing the study of electronic interconnections
in retail trade. This includes other industries that share similar characteristics as grocery

retailing (e.g., apparel).

Overview of this Dissertation

In the next chapter (Chapter 2), the theoretical and empirical literature on
transaction cost economics and social relations is examined in more depth, introducing
the key concepts studied and the associated hypotheses tested.

Chapter 3 discusses the methods and procedures used in this project, i.e.,
measurement and data collection. Included in the chapter are the descriptions of how
variables have been operationalized, as well as the rationale for and procedures used in
the telephone survey of grocery retailers.

Data processing and statistical procedures employed to prepare and analyze the
collected data are presented in Chapter 4. Also reported in this chapter are the overall
results, which reveal varying patterns of relationships between the two sets of
independent variables and the different operational measures of electronic network
adoption and use. Factors that are strongly associated with technological adoption do not
appear to have signiﬁcant impact on the perceived importance of retailer-supplier
electronic linkages. Furthermore, signiﬁcant predictors of the importance of electronic
network use to carrying out product acquisition activities of varying nature are different;

these activities differ in their demands on interactive communication between a retailer

12
and its supplier. Theoretical and methodological explanations for these observed ﬁndings
are discussed in Chapter 5.

Finally, Chapter 6 closes the presentation of this dissertation research with
concluding observations. In addition to a summary of ﬁndings, discussion in the chapter
covers the limitations stemming from conceptual caveats and drawbacks associated with
the measurement and research design adopted in the current study. Theoretical and

practical implications are also presented.

Chapter 2

REVIEW OF THEORETICAL AND EMPIRICAL LITERATURE

Introduced brieﬂy in the last chapter is the central proposition of this dissertation
research, that the adoption and use of electronic information networks in organizations is
affected by organizational conditions and requirements, some of which are transaction
cost variables and others are social relational factors. The common underlying link
connecting the utility of economic and social theories in the current inquiry is the role
electronic communication and information networks play in coordination and control.

Economic theory deals mainly with “objective” conditions in that actions are
aimed at responding to contingencies and constraints imposed by the environment.
Following this perspective, technology adoption and use for the purposes of coordination
and control is driven largely by objective needs, as created by the industrial and market
environment in which a ﬁrm carries out its business activities and transactions.

What is also recognized is that, rather than involving completely autonomous,
atomized units in the market place, many economic transactions are often conducted
among the same trading partners with a history of interacting with each other. Given the
existence of recurring transactions, relationship-based factors will likely play a crucial
role in whether and how networks are used to help coordinate and control these
exchanges.

As can be readily inferred, it would appear that theoretical insights grounded in

economic and social literature offer thoughts in complementary domains concerning what

13

14
are some key factors driving the adoption and use of information technology to support
transactions. This chapter presents a review of the transaction cost and social relational
literature that helps to shed light on different antecedents to electronic information

network adoption and use in organizations.

Transaction Cost Factors

To date, transaction cost analysis, as advanced by Williamson (1975), is a popular
theoretical approach used to study electronic network use and its relationship to structures
of coordinating economic activities, namely, “markets” and “hierarchies” (e.g., Benjamin
& Wigand, 1995; Gurbaxani & Whang, 1991; Kraut, Steinﬁeld, Plummer, Butler &
Hoag, 1997; Steinﬁeld, Caby, Jaeger & Kraut, 1997). Initiation of this research “trend” is
credited to the seminal précis on “electronic markets” and “electronic hierarchies” by
Malone, Yates and Benjamin (1987). This section traces the roots and key propositions

of this approach, leading to the derivation of the ﬁrst three hypotheses tested in this study.

Transaction Cost Theory

Grounded in the neoclassical economic paradigm, the theory of the ﬁrm posits
that hierarchies in the form of organizations emerge to internalize coordination of
economic activities as a consequence of market failure (Coase, 193 7). Market failure
refers to the inability of price mechanisms to bring together efﬁciently sellers and buyers
in the open market place. One of the chief reasons for prices failing to match supply with

demand is the presence of imperfect information, thereby violating a key assumption of

15
perfect markets. What this means is that, without incurring high costs of search and
negotiations, a procurer of a needed good or service cannot know what the best offer
might be in the market (Hennart, 1986). In this scenario, inefﬁcient coordination by price
mechanisms is replaced by decision making by managerial ﬁat in hierarchies.

Building on this assumption that ﬁrms exist because markets fail, institutional
economist, Oliver Williamson (1975) advanced transaction cost theory to elaborate on
how various conditions associated with an imperfect market affect the costs of
coordination and control. Rational economic actors must choose between the two
alternative structures of hierarchies or markets for coordinating the performance of
economic activities, whichever minimizes costs. Relative to markets, hierarchies
generally incur higher production and internal governance costs associated with all
processes necessary to produce or distribute a product or service. However, when it is
difficth to anticipate all possible contingencies of transacting with an external supplier,
negotiations will likely become cumbersome and complicated. In addition, the costs of
detailing a complete contract to deal with these contingencies will be high. In effect, the
costs of transacting with this party will increase, perhaps to the point of exceeding the
costs of having the needed product made or service performed by a unit owned (fully or
partially) by the organization. When this is the case, incentives exist to internalize
production to minimize costs, thereby substituting complex, costly negotiations of
transacting in the market with internal governance.

According to transaction cost theory, whether it is more costly to practice

hierarchical governance or conduct market transactions depends on the amount of

l6
uncertainty associated with the key economic activities to be performed (Williamson,
1975; 1979). For instance, if it is highly uncertain as to how often, from whom and/or
how much of a key asset can be obtained when needed, it may be very difﬁcult and costly
to detail adequately in a contract all possible contingencies that may arise.

The desire to reduce such uncertainty will lead to internal production of the asset,
rather than relying on external suppliers. Again, a key motivation is to minimize
transaction costs incurred in searching for, negotiating with, monitoring and settling with
an external supplier. Another important underlying motivation for internalization is to
maximize control. Typically, management can exert considerably more control over an
internal division or a corporate subsidiary than an outside trading partner or contractor.

A number of organizational researchers have dealt with this notion that managers
face a high level of difﬁculty making rational decisions when there is high uncertainty
(e.g., March & Simon, 1958, 1993; Lawrence & Lorsch, 1967; Pfeffer & Salancik, 1978).
Nohria and Eccles (1992) argued that successful interactions depend on how well
uncertainty in all exchanges are resolved. In this regard, internalization allows for much
external uncertainty to be absorbed, resulting in savings on communication costs and
permitting the activation of incentive, monitoring and control processes (Hennart, 1986;
Klein, Frazier & Roth, 1990). As a case in point, Anderson (1985) concluded that an
advantage for employing an internal sales force, as opposed to contracting external,

independent agents, is the substitution of surveillance for commission as a control

mechanism.

17

In addition to uncertainty, production and transaction costs are affected by the
characteristics inherent in what a ﬁrm needs in order to perform its business functions, for
example, the distribution of a product. These needs are generally manifested in such
factors as raw materials, equipment and human knowledge necessary to perform the core
activities of an organization. The extent to which these needs are speciﬁc to any given
organization has been termed asset speciﬁcity (Williamson, 1975). An example of a
highly speciﬁc asset is a button that carries the crest or design of a particular apparel
manufacturer, which is useless to other clothes-makers. Exclusive knowledge of
idiosyncratic business philosophies or practices of a client ﬁrm to an advertising agency
is another case in point.

When the speciﬁcity of a needed asset is high, an organization is vulnerable to
being taken advantage of by opportunistic suppliers, who are in the position to capitalize
on the buyer’s dependence (Williamson, 1975, 1979). Furthermore, this vulnerability is
exacerbated when the ex-ante number of suppliers is small. Knowing that the buyer has
few alternatives, an opportunistic supplier can, for instance, charge a higher than fair
market price. In order to avoid being placed in a vulnerable position or being forced into
small-numbers bargaining, the ﬁrm will have incentive to internalize the
production/supply of this speciﬁc asset, since with ownership comes control.
Alternatively, some means of locking in one of the few providers must be devised to
control the likelihood of opportunistic behavior. As an example, due to reasons such as
the time it takes for someone’s learning curve to run its course and the desire to preserve

trade secrets, an organization will likely opt to hire a software engineer internally to

18
develop its proprietary computer-aided manufacturing system than to outsource this job
to an external contractor who has few incentives to be loyal to the ﬁrm. Empirical
observations have generally lent support to this hypothesized relationship between asset
speciﬁcity and coordination structure (e.g., Klein et. al., 1990; Kraut et. al., 1997;

Steinﬁeld et. al., 1997).

Electronic Markets and Electronic Hierarchies

What has precipitated from these transaction cost considerations are some areas in
which electronic communication networks can facilitate the coordination and control of
transactions and business activities. The main effect is in the reduction of all types of
coordination costs, including those associated with search, conveying product
information and quality monitoring (Bakos, 1991; Gurbaxani & Whang, 1991). Bakos &
Treacy (1986) also proposed that electronic networks can reduce the costs of contracting
(with external trading constituents) and monitoring, thus aiding in overcoming
managerial bounded rationality and vulnerability to opportunism.

Although cost savings apply to both hierarchical production activities and market
transactions, Malone et. a1. (1987, 1989) argued that external coordination costs will be
reduced more relative to internal governance costs. As a result, electronic
interconnections will favor the evolution of what they called “electronic markets” over
“electronic hierarchies.” Benjamin and Wigand (1995) concurred with this proposition,
contending that network use has the potential to drive transaction costs down to fairly

close to non-existence.

19

Malone et. al. (1987) suggested three broad types of cost savings associated with
electronic network use. First, there is the electronic communication effect, which deals
with the fundamental improvement in overall communication and information exchanges.
In addition, electronic linkages enhance the tighter coupling of production and
distribution activities; this is the electronic integration effect. An example of this
improved integration is the practice of just-in-time product ordering and delivery in the
retail sector being enhanced by EDI systems. A third type of cost savings is the
electronic brokerage effect, which entails the idea of electronic information networks
serving the functions of a broker in bringing together buyers and sellers. By connecting
buyers with suppliers in a central database, needs and offerings can be matched more
efﬁciently and cost-effectively.

Due to savings in the costs of communication and information processing,
electronic networks can help with coordination and control by reducing uncertainty.
When the uncertainty faced by an organization is high, sound decision making
necessitates a greater amount of information to be gathered, interpreted and synthesized
(Galbraith, 1973; Tushman & Nadler, 1978). In order to control this situation, there must
be either a reduction in the amount of information to be processed or an increase in the
capacity to process information. Electronic information networks effectively increase the
technological capacity available to an organization for meeting the heightened
information processing demands, and, as a result, place the ﬁrm in a better position to

cope with uncertainty. In support of this observation, Sabherwal & Vijayasarathy (1994)

20
found empirical evidence that organizations rely on telecommunication links with
suppliers in the face of high environmental uncertainty.

Since this study is focused on electronic network use in supporting retailer-
supplier interactions related to product acquisitions, the most relevant form of uncertainty
would be that related to the contingencies surrounding those transactions. This is in
contrast to examining uncertainty arising within the retailer ﬁrm, such as that associated
with stafﬁng requirements. Therefore, in this study, uncertainty is conceptually deﬁned
as the predictability of the conditions in which a transaction takes place.

Based on the discussion thus far, electronic network adoption and use would be
motivated by the need to reduce high uncertainty, and thus the following relationship is

hypothesized:

H1: Uncertainty is positively associated with electronic network adoption and

use.

F urtherrnore, Malone et. a1. (1987) argued that electronic information networks
can diminish the constraints on the market coordination of economic activities stemming
from asset speciﬁcity. As deﬁned earlier, asset specificity deals with the extent to which
a needed asset is specific to any given organization (Williamson, 1975). (In the current
study, the asset needed is the product bought and sold.) With electronic interconnections,
producers become more aware of similar product needs faced by multiple buyers, which
may have been previously deemed as highly asset-speciﬁc. Having this information

concerning a larger market than previously known, more ﬁrms will be willing to get

21

tooled up for production, effectively increasing the number of suppliers in the market. A
potential beneﬁt is in electronically gathering and linking up information about suppliers
who may be geographically spread out, thereby reducing the difﬁculty and costs of
search. These effects serve to attenuate the threat of small-numbers bargaining and fear
of opportunism, affording procurers of needed assets more choices in the marketplace. In
turn, ﬁrms are rendered less vulnerable to being taken advantage of by few powerful
suppliers or customers in the market. As such, it would appear that a rational motivation
to adopt and use networks is to overcome the disadvantages associated with high asset
speciﬁcity.l

Similar logic applies to the effect of network use in reducing the impact of the
complexity of product description. Product description deals with the information a
buyer needs to have in order to evaluate the features of an offering that ultimately affects
the purchasing decision (Porter & Millar, 1985). Complexity of product description, then,
refers to the diﬂiculty in describing these critical product attributes.

In the context of coordinating supplier-customer transactions, it is important that
the information about products be usable to the buyer (Malone & Crowston, 1994).
Electronic networks, especially sophisticated systems offering multimedia features, can
help to convey complex information. Moreover, by allowing buyers to have information

about the offerings of alternative trading partners of whom these buyers may have not

 

Of course, the above arguments are based on the assumption that all eligible ﬁrms are
equally well-represented on the networks being used, and that information about
possible trading partners is available electronically, at least in the sense of being
useful to an organization for deciding among alternatives.

22
been aware previously, the relative imbalance of power, dependence and bargaining
positions can be potentially eliminated (Johnston & Vitale, 1988).

Most business and economic researchers using transaction cost analysis have
treated coordination structures as a dependent variable; electronic network use, if studied,
has been modeled as an independent variable or a moderator of the relationship between
asset speciﬁcity and coordination structure. Kraut et. a1. (1997) found that, aside from
predicting the level of internalization of production activities, various types of asset
speciﬁcity and the complexity of product description have meaningful associations with
whether networks have been used and regarded as important. These ﬁndings are
particularly true for the cases in which networks have been used for coordinating internal
activities (as opposed to enhancing transactions with outside contractors and suppliers).
This 250-case empirical study of four American industries extends transaction cost
analysis to a large-scale study of electronic network use within and between
organizations. Similar results were also found in a replication of the study in France
(Steinﬁeld et. al., 1997), suggesting that the utility of the transaction cost analytical
framework applies to another country that has markedly different characteristics in terms
of market conditions and network infrastructure and use.

Derived from the above review are the following hypotheses regarding how asset

speciﬁcity2 and complexity of product description relate to electronic network adoption

and use:

 

2 Following explication in the literature, three types of asset speciﬁcity are measured in

this study, namely, product speciﬁcity, time speciﬁcity and knowledge speciﬁcity.
These measures are discussed in the methods chapter (Chapter 3).

23

H2: Asset speciﬁcity is positively associated with electronic network adoption

and use.

H3: Complexity of product description is positively associated with electronic

network adoption and use.

The above hypothesis concerning the complexity of product description is based
on the ﬁnding reported by Kraut et. al. (1997), which represents the only empirical test of
the construct to date. Contrary to the original conceptualization by Malone et. a1. (1987),
Kraut et. al. (1997) found that the less difﬁcult it is to describe a product, the more likely
electronic interconnections are viewed to be important in producer-supplier transactions.
One possible explanation for the contradiction could be the difference in network
capabilities. When the construct of complexity of product description was ﬁrst
introduced a decade ago, the state of technology did not include multimedia features,
which are supported by today’s systems and networks, such as the Worldwide Web.
Visual images provide an effective and “easy” way to represent product description that
might be difﬁcult to express in other media forms or modalities (e. g., a text-based data
network). The current study takes into consideration the increased sophistication of
information technology today as compared to what was available during the original
inception of the construct of complexity of product description. Hypothesis 3 stated

above is an empirical test of this consideration.

24
Social Relational Factors

Critics of neoclassical and institutional economics call attention to the fact that
these paradigms carry biases and limiting conditions, which must be recognized when
applied to the study of organizational phenomena. Most importantly, limitations exist in
the key assumptions of these economic perspectives in viewing social relations in
transactions as a source of “ﬁiction” (Biggart & Hamilton, 1992). Social relations are
arguably seen as conﬂict to the presumed preferences of economic actors: the pursuit of
self-interests through arms-length transactions in open markets consisting of atomized
buyers and sellers. These perhaps ﬂawed assumptions about the pursuit of self-interests,
frequently at the expense of another party (as often modeled in game theory), belie the
concerns for vulnerability and opportunism (Parkhe, 1993). Even if these assumptions
were accurate reﬂections of reality, as observed by Granovetter (1985), it is unclear how
strategies like internalization can guarantee elimination of opportunism.

There are a host of coordination and control mechanisms, such as trust and
mutuality of interests, that are typically not present in spot-market transactions, but rather
operate on a relational basis (Bradach & Eccles, 1991). In addition, negotiations oﬁen
take place in the context of a relationship rather than ephemeral matching of buyers and
sellers (Sheppard & Tuchinsky, 1996). This section reviews existing literature on social
relations that can help lend some insights into why electronic networks are used to

support transactions.

25
Networks of Social and Interpersonal Relationships

Empirical observations to date do not support the proposition that networks tend
to favor the coordination structural outcome of electronic markets, which are
characterized by ﬂeeting associations between buyers and sellers. The closest set of
ﬁndings to the Malone et. a1. (1987) proposition is the study by Brynjolfsson, Malone,
Gurbaxani & Kambil (1994). Over a two-year period, there is a decrease in ﬁrm size,
suggestive of information technology use leading to more outsourcing of previously
internalized activities. Hess & Kemerer (1994) did not ﬁnd support for the electronic
markets hypothesis in their research on computerized loan systems in the home mortgage
market, even though these electronic data systems have been used in the industry for
more than a decade.

Other research has found that network use has been associated more with
furthering pre-existing trading relationships, or that, at the minimum, organizations with
electronic interconnections are likely to have already been doing business with one
another (Steinﬁeld, Kraut & Plummer, 1995). These observed results lead to the question
of whether established relationships between organizations necessarily serve as a
pre-condition for electronic networks to be implemented and used.

Furthermore, many researchers have called attention to the importance of social
relationships to coordination and control efforts, particularly in the inter-organizational
context. One reason is that many buyer-supplier relationships do not ﬁt the traditional
market-hierarchy dichotomy (Sheppard & Tuchinsky, 1996). Furthermore, it has been

argued that inter-ﬁrm communication systems increase the range of possible relationships

26
between organizations beyond the traditional structures of hierarchies and markets
(N ohria and Eccles, 1992). This proposition is consistent with proponents of the network
approach in viewing organizational structure and processes (e.g., Johanson & Mattson,
1987). Coordination in this network context involves exchange relationships that take
time and efforts to establish and develop. Consequently, these relationships are more
stable than market contracts (although organizations are free to choose business
counterparts), but are more ﬂexible than traditional hierarchies.

While markets and hierarchies rely on prices and routines, respectively, for
coordination and control, “hybrid” inter-organizational arrangements depend on
relationships, mutual interests and reputation to manage exchanges that tend to be more
social than mundane or bureaucratic (Powell, 1990).3 As observed by Eisenberg, Farace,
Monge, Bettinghaus, Kurchner-Hawkins, Miller and Rothman (1985), information is
usually exchanged at the personal level. In practice, organizations are indeed represented
by individual members, for example, sales force or upper management. Besides,
coordination success and other perceptual assessments in inter-organizational systems are
often largely determined by personal linkages (Eisenberg et. al., 1985).

In discussing the strengths and limitations of electronic media use in interpersonal
negotiations and interactions, Nohria & Eccles (1992) argued that it is a fallacy to
conceive electronic networks as substitutes for the face-to-face communication modality.

According to these researchers, electronic networks are good for mobilizing action but

 

3 Some examples of hybrid forms include different types of information partnerships,

value-added partnerships, the network organization and relational contracting
(Johnston & Lawrence, 1988; Konsynski & McFarlan, 1990; Powell, 1990).

27
face-to-face interactions are required to take actions themselves. One of the main reasons
is that electronic networks are simply not robust enough for these actions. Finding
empirical evidence consistent with this argument, Kraut et. al. (1997) concluded that
electronic networks have not been regarded as substitutes for interpersonal links between
ﬁrms; instead network linkages have been viewed predominantly as complements to
personal relationships, contacts and interactions.

Based on the above review, interpersonal relationships are conceptualized as
interactions, contact and knowledge at the personal level (between members of trading
relationships). In addition, adopting the testable view that interpersonal relationships and
electronic information networks complement each other in supporting the execution of

inter-ﬁrm transactions, the following hypothesis is derived:

H4: (Importance of) interpersonal relationships is positively associated with

electronic network adoption and use.

Mutuality and Trust

Examination of existing research shows that there is general recognition on the
part of ﬁrms in a trading relationship that, rather than remaining separate and
autonomous, their interests and welfare are interdependent and their fate intertwined with
one another (Hart & Estrin, 1991; Johanson & Mattsson, 1987; Johnston & Lawrence,
1988; Powell, 1990). For instance, perspectives such as the resource dependency view
(Pfeffer & Salancik, 1978) points to the fact that few organizations are fully

self-sufficient with respect to their critical resources (or assets, in transaction cost terms).

28
This reality means that many ﬁrms will likely need external constituents to fulﬁll their
resource needs, and, in turn, a resource-based dependency is created.

Recognizing the presence of resource control and dependence, Johanson &
Mattson (1987) argued that “an interﬁrrn relationship is a mutual orientation of two ﬁrms
toward each other” (pg. 37). Adopting a social exchange theoretical angle, they posited
that an inter-organizational relationship evolves through a process of members learning
about and inﬂuencing each other. Powell (1990) deﬁnes mutual orientation as
“knowledge which the parties assume each has about the other and upon which they draw
in communication and problem solving” (pg. 272). These perspectives posit that the key
to the social exchanges is mutuality, that is, demonstration of respect for each other’s
needs. Overtime, interactions help to develop a mutual orientation between members of
the relationship, showing increased interests and knowledge about, as well as adaptations
and commitment to one another.

This sort of mutuality of interests and commitment faced by organizations in the
above sketch bears resemblance to the central tenet of coorientation theory (N ewcomb,
1953). When the two individuals in a dyadic relationship like one another, their attitudes
towards an object, person or event would likely be congruent with each other, thereby
demonstrating mutual orientation. Should dissonance exist in their views toward the
common object, person or event—especially when it is highly important to one party or
has joint relevance to both parties—the two individuals engage in communication to
restore cognitive balance and consistency in their coorientation. Willingness to work

toward eliminating dissonance, instead of severing relational ties, implies the existence of

29
commitment to the relationship. Although coorientation theory mainly deals with
interpersonal dyads, rather than business associations, individual actors are central to
inter-ﬁrm relationships, as discussed earlier. Consistent with this perspective, Johanson
& Mattson (1987) argued that mutuality among organizations is primarily the orientation
and actions of individuals members.

The notion of mutuality as manifested in mutual commitment to a relationship is
increasingly evident in what has been called information partnerships, which are
characterized by organizations sharing crucial information resources and costs in their
relationships (Konsynski & McFarlan, 1990). Computerized databases and other
information technologies are among the critical, but expensive, resources that motivate
these types of partnerships. This is why electronic networks are less likely to be
associated with spot-market than relationship-based transactions. In many cases, the
promise of a large amount of recurring business and transactions overtime is required to
justify and recover the large sunk costs. Without the guarantee of a long-term business
relationship over which to spread the substantial initial technological costs, ﬁrms will be
reluctant to invest in electronic networks. This is especially true in cases where the
electronic link is proprietary to the parties connected. Having invested in an electronic
interconnection raises the costs of switching to a different partner, and effectively binds

the trading ﬁrms to the business relationship.‘ This effect can be positive, as a lock-in

 

‘ For instance, in a series of in-depth interviews with a few large magazine publishing

ﬁrms, one reason for hesitating to establish an electronic connection with an external
supplier is that the network cannot be “moved” if the publisher wishes to terminate
the relationship and work with a different supplier. The network referenced here,
again, is proprietary in nature, since the interviewee was not referring to an

30
situation tends to raise the likelihood of the trading partners being committed to a quality
relationship (Dwyer, Schurr & Oh, 1987).

Given the above discussion, it appears logical to expect electronic network
adoption and use to be an indication of, and favored by, the existence of mutual
commitment in a pair of trading ﬁrms. With mutual commitment (toward the business
relationship) conceptually deﬁned as pledge to each other ’s business needs and the

success of the relationship, the following hypothesis is derived:

H5: Mutual commitment is positively associated with electronic network

adoption and use.

Cooperation in a committed business relationship necessarily places ﬁrms in a
vulnerable position (Powell, 1996). Vulnerability develops, in part, because of reliance
on one other (such as resource-based dependency) as the relationship grows. In order to
off-set the unsettling feelings of vulnerability and interdependence associated with
cooperation, the governance structures regulating a trading relationship must allow for
constant monitoring and consultation.

It has been proposed that constant monitoring and consultation can be substituted
by the presence of trust. For instance, Bradach and Eccles (1991) suggested the view of

trust as a general control mechanism when the nature of inter-ﬁrm associations extends

 

(continued ﬁom previous page)
interconnection built on “public” infrastructure or backbone, such as the Internet.

(These interviews were conducted during the summer of 1995 and supported by an
NSF grant, IRI-9408271.)

3]
beyond ephemeral transactions, and when specialized control mechanisms like price and
authority are insufﬁcient. Ouchi (1991) proposed the notion of “clans” as a form of
governance structure that depends on trust as a means of coordination and control.
According to this view, trust is embedded in the networks of social relationships. Rather
than relying on explicit auditing and monitoring for coordination and control, task
performance is evaluated through more subtle reading of signals among members of the
clan; these signals cannot easily be translated into explicit, veriﬁable measures of
performance.

Moreover, recall from earlier arguments that it is highly unlikely that direct
control of an external trading partner can be enforced. Therefore, when two organizations
are bound to an interdependent trading relationship, there are few alternatives to
managing the business liaison than trusting each other. This perspective is precisely what
Kipnis (1996) argued, that out of increased reliance arises the need for partners to trust
each other. Developing trust would be a necessary strategy to control the relationship, if
dependence cannot be reduced. This is because trust conveys the acknowledgement—
and, more importantly, expectations—of each other’s ability to manage resources valued
by the parties involved.

Owing to the fact that trust is not a state with which human actors feel naturally
comfortable, the intended effects of trust often rely on the use of explicit procedures
(Kipnis, 1996). Consistent with this logic, Granovetter (1985) argued that institutional
arrangements can be devised to discourage malfeasance. An example of an institutional

arrangement to guard against the breach of trust is a contract detailing terms of an

32
agreement. Besides punishing deviating actions by formal, legal sanctions, other more
subtle safeguards entail covert threats to such valued considerations as organizational
reputation. Capturing the notions described above, Creed and Miles (1996) proposed that
trust deals with expectations of equitable treatment and reputational concerns.

Based on the discussion thus far about vulnerability, interdependence and the need
for trust, it would appear that a network connection between two trading partners would
be more likely when there is already some measure of trust between them. The
underlying reasoning begins with the observation that inter-organizational electronic
linkages have the effect of shifting the nature of interdependence among organizations
(Hart and Estrin, 1991). It is argued that “new” costs and vulnerability are created as the
ﬁrms come to depend on the electronic interconnections in their transactions with one
another. Therefore, it would be reasonable to expect that before two organizations are
willing to open up themselves to such new risks, there needs to be a certain level of
conﬁdence and assurance that they would not take advantage of each other’s vulnerability
and dependence on the relationship.

In addition, recall from Chapter 1 that information forms the basis of control, and
that information technology serves as a control mechanism (Beniger, 1986, 1990; Nass
and Mason, 1990). According to this perspective, electronic interconnections serve to
provide an infrastructure for two trading companies to monitor and manage task
interdependence. Linking this observation to the role of trust as a means of managing

trading a relationship, it is possible (and empirically falsiﬁable) that electronic

33
information networks and trust complement each other in supporting business
transactions.
Conceptualizing trust as the belief that one ’s trading partner will reﬁain from
malfeasance, the extent to which this belief affects the likelihood of electronic network

use and adoption is hypothesized as follows:

H6: Trust is positively associated with electronic network adoption and use.

Chapter Summary

This chapter presented a review of the transaction cost and social relational
literature, guiding the formation of the hypothesized relationships presented above.
Figure 1 illustrates a diagrammatic representation of these relationships. Restated below

are the propositions driving the hypotheses in this study.

1. Transaction cost hmtheses (H1-H3 ). High levels of uncertainty, asset speciﬁcity
and complexity of product description drive up the costs of coordination. Ad0ption
and use of electronic information networks are motivated by the desire to reduce these

costs.

2. Social relational hyp_otheses (H4-H6). Electronic retailer-supplier linkages adopted
and used for the purposes of coordination and control are affected by the existing
characteristics of a business relationship. In particular, network adoption and use
complements interpersonal relationships, and indicates the existence of mutual

commitment and trust.

34

 

 

83 ES Etna—5.
xueEo Z ﬁsts—H

 

 

 

 

 

 

 

, .ﬁesﬁuhagm. a enmeﬁaﬁé Sega . ,_ enemy

 
 
 

on: can 5593?. €0,502 3.5.502“,— EEEQNEQEO ”52.2.5.— Eeuoah .n 95”:—

Chapter 3

METHODS AND PROCEDURES

Having stated the research question and hypotheses to be tested (in Chapters 1 and
2, respectively), the current chapter reports the research design used in this dissertation
project, including measurement, population selection, sampling and data collection. To
recapitulate the research task at hand, it is to identify and test some of the key antecedent
factors to organizational adoption and use of electronic information networks in
retailer-supplier transactions associated with product acquisitions. With this research
objective of seeking explanations for network adoption and usage and testing
hypothesized relationships with different transaction cost and social relational factors, a
cross-sectional survey method has been selected (Babbie, 1990).

More speciﬁcally, the research design includes a telephone survey of 143 product
buyers in the US. grocery retail industry. They were asked a series of questions
concerning two broad areas: (i) how they deal with a major supplier in acquiring a key
product; and (ii) how important electronic networks are in working and interacting with
that supplier. These questions have been created from the operationalization of the major
concepts in the hypotheses to be tested.

Presentation in this chapter is organized under two major sections: measurement
and data collection. In the measurement section, the development and operationalization
of the key constructs in this study is discussed. As for the data collection section, issues

and procedures associated with the survey conducted are reported, including the rationale

35

36
for having chosen the grocery retail industry as the population of interest, sampling,

design and execution of the telephone survey.

Measurement

As observed by Emmert (1989b), social science research often deals with the
measurement of hypothetical constructs. These are concepts of which the presence or
absence cannot be “seen” directly but can only be inferred, such as in the case of
assessing attitudes or perceptions. Further complicating the observation of these
hypothetical constructs is the fact that, unlike physical science research, there are no
established “objective” measurement instruments (e.g., yardsticks or measuring
cylinders). Due to the absence of these pre-established instruments, one of the chief
reasons for the need to discuss measurement in any given study is to specify the set of
procedures used for inferences to be made about the hypothetical constructs and
relationships among them. This is the only way social science researchers can
communicate to each other about standardized procedures of observation and description,
and arrive at inter-subj ective agreement about the appropriateness of these procedures.
Furthermore, these standardized measurement speciﬁcations can serve the useful function
of guiding later research efforts investigating similar, if not identical, hypothetical
constructs.

In the context of the measurement procedures employed for the research project
reported in this dissertation, few empirical studies of organizational electronic network

use have attempted to integrate cross-disciplinary perspectives. As referenced in the last

37
chapter, the study conducted and reported by Kraut et. a1. (1997)——and replicated in
France (Steinﬁeld et. al., 1997)—represents foremost efforts in drawing on insights based
on inter—disciplinary theoretical guidance.‘ As an attempt to minimize unanticipated
problems associated with design and execution, this study adopted some of the
operational procedures employed by Kraut et. al. (1997). For instance, operationalization
of the electronic network use measure in this study is based on adaptations from that used
by Kraut et. al. (1997), which yielded a highly reliable measure (Cronbach’s Alpha =
.94). Other “successful” design strategies, such as the measurement the importance of
interpersonal relationships, have also been adopted. (Further discussion of these

operational adaptations is included in the description of individual variables below.)

Use of Perceptive Measures

In a meta-analysis of seventy-ﬁve innovation studies, Tomatzky and Klein (1982)
defended the predominant use of perceptive measures in studying technological adoption
and implementation. It was submitted that, even given the presence of “objective”
conditions surrounding the decision to adopt an innovation, such as the actual
technological costs, these conditions have subjective meaning to an adopter. For
instance, one manager may view a set of costs as prohibitive, while another technology

decision maker may consider the same ﬁgures as reasonable. In addition, a potential

 

There are research studies that have employed transaction cost analysis along with
other organizational theories, such as those dealing with international business,
marketing channel choice and strategic management. Reference to the dearth of
empirical research integrating transaction cost theory and literature on social relations
is restricted to the study of organizational information technology use.

38
adopter’s perceptions are likely shaped by organizational attributes, for example, the
ﬁnancial resources available for investments in the company’s information technology
infrastructure. Therefore, recognizing the subjectivity of human actors in innovation
adoption and use, it is appropriate to seek the perceptions of individuals, especially those
in the organization who are most affected by the technology. Based on this argument by
Tomatzky and Klein (1982), the design of the current research project lends itself to the
recommended research strategy for surveying grocery buyers, who are most affected by
the use of electronic networks to support transactions with suppliers.

Measurement of the dependent and most of the independent variables in this study
mainly entails having survey respondents rate numerically a series of survey questions
intended for the construction of Likert-type response scales. As described in the literature
(Babbie, 1990; Emmett, 1989a), Likert-type scales are constructed by assigning
numerical scores along a range of answer categories that indicate different levels of
agreement with a question or statement. However, Likert scaling is not restricted to
indicating the level of agreement, but has also been widely used with other response
categories, such as approval (Babbie, 1990). In the current study, the electronic network
use measures entails the scaling of perceived importance (see discussion on the
“importance of network use” measure below).

Reﬂecting a characteristic of Likert scaling, as implied in another name for the
technique—the method of summated ratings (Emmert, 1989a), multiple items were
created a priori for most variables in this study. Sharing identical response categories,

scores on items written to measure the same construct are summed to create composite,

39
index measures. Single-item measures cannot be subject to reliability and validity tests,
and thus effectively carry the assumption that they are perfectly reliable and valid for
assessing the attitudes and perceptions being assessed. When dealing with the
measurement of hypothetical constructs that cannot be observed directly, indices are

especially more desirable because of these reliability and validity considerations.

Dependent Variables: Electronic Network Adoption and Use

The central construct in this study is the implementation of electronic information
networks to support transactions with the supplier of a key product. Adopting the
operational deﬁnition by Kraut et. a1. (1997), electronic network use deals with any
computer-based communication that allows for the electronic exchange of information
with a major supplier. (See Appendix A for the telephone survey instrument and the
exact language respondents were read during the interviews.)

In this research, empirical assessment of the adoption and usage of electronic
information networks to support transactions is performed separately with two variables,
namely information technology cluster and importance of electronic network use,
respectively. This measurement decision takes into consideration the recommendation by
Tomatzky and Klein (1982). They contended that it is a fallacy commonly committed by
researchers in assuming that organization-wide implementation or usage of a technology
necessarily comes with adoption. One reason why adoption may not guarantee
widespread usage in an organization is arguably that the individuals most affected by

using the technology are often not the ones directly responsible for technological

40
adoption decisions. As such, it is recommended that research aimed at studying diffusion
of innovations in organizations needs to distinguish between these two phenomena of

adoption and practice/use, and incorporate separate measures for both constructs.

Information Technology Cluster

This variable of electronic network adoption aims at assessing whether a number
of technologies that can enable electronic information exchange with a supplier are
available for use in an organization. This is measured by the question, “Does your
company use...,” with a dichotomous response scale of “yes” or “no.” The cluster of

technologies surveyed include:

1. the Worldwide Web;
2. electronic data interchange (EDI) systems;
3. electronic point-of-sale (POS) systems;

4. computer-based on-line services (e.g., America On—line, Compuserve and
Prodigy);

5. electronic mail for internal communication; and

6. electronic mail for communication with external suppliers.

Incorporated in the survey instrument are descriptions of these technologies, in
anticipation that respondents might need clariﬁcations (see Appendix A). For instance, if

a respondent did not know what an EDI system is, the description, “a computer-based

41
system of exchanging data, such as pricing and billing information,” would have been
provided.

Selection of these six computer-based information technologies and networks
were based on reported or potential use in organizations. Electronic mail, especially for
internal communication, is a case in point (Barnes & Greller, 1994). As for the inclusion
of the Worldwide Web in the technology cluster, it is due to the increasing discussion of
the potential role of the lntemet in supporting commercial transactions, promotional and
marketing activities, as well as other business-to-business information exchanges (Raphel
& Raphel, 1996). Computer-based on-line services are incorporated in the survey
because they provide, among other features, electronic mail capabilities and gateway
access to the Internet for some businesses. As for EDI and POS technologies, they have
been widely reported as information systems and networks commonly used for
supporting operations and transactions with suppliers and customers, especially in the
retail sector (Frozen Food Age, August, 1996; Progressive Grocer, April, 1996).

Scores on these six technologies form the information technology cluster variable.
Speciﬁcally, the cluster score is the total count of how many of the six technologies

has/have been adopted. Therefore, the possible scores for the information technology

measure range from 0 to 6.

Importance of Electronic Network Use
This variable is constructed to gauge computer-based information network use in

supporting transactions. Respondents were asked to consider how important electronic

42

networks are to the execution of a number of activities. These activities are associated

with various stages of a transaction—that is, the acquisition of a key product in the

current research—as explicated in existing literature (e. g., Johnston & Vitale, 1988;

Kambil 1993). This operationalization strategy was used by Kraut et. a1. (1997), which

appeared to be a fairly useful and reliable method of tapping the perceived importance of

electronic networks in supporting core business activities.2

In operationalizing the importance of electronic network use variable, the

perceived importance of electronic links to the following activities associated with the

acquisition of a key product from a major supplier is measured:

1.

developing the speciﬁcations of an order to acquire a key product;
seeking price information;

negotiating the terms of an agreement to purchase a key product;
ordering a key product;

monitoring the quality of the orders received;

ﬁxing problems after-sales errors or problems; and

practicing just-in-time inventory management.

Each of these activities is rated with a ﬁve-point, Likert-type scoring scale, with “1”

being “not important at all” to “5” being “very important.” For those who do not use

 

2

As mentioned earlier, the items constituting the electronic network use variable yield

a fairly reliable composite measure (Cronbach’s Alpha = .94).

43
electronic networks for any or all of these activities, they were given a score of “9”
(“don’t use”) to represent non-usage.

With the exception of #2 and #7 listed above, these items make up the highly
reliable network use composite index in the study by Kraut et. al. (1997).3 “Seeking price
information” (#2) has been included in the current study due to the reason that this is
reportedly an activity conducted electronically in the grocery industry (Lewis, 1996). As
for the practice of just-in-time inventory management (#7), it has increasingly been
discussed as a practice greatly facilitated by the use of electronic supplier-customer
networks and systems in the retail sector, including grocers (Barnes & Greller, 1993;
Gomes & Mentzer, 1991).4 This demand-driven inventory management principle deals
with the idea that better sales forecast resulting from the use of electronic data enable
more efﬁcient ordering and shipping on an as-needed basis (Clemons, 1993; Ratliff,
1996). Consequently, retailers are able to keep less in stock and minimize inventory and
operating costs. Due to such reported beneﬁts associated with using electronic
information systems, the practice of just-in-time inventory management has been

included in the assessment of the importance of network use in this study.

 

Kraut et. a1. (1997) investigated the acquisition of key production input, while the
current research focused on buying products. As such, “input” was replaced by
“product” in the operationalization in this study.

There are other labels used to refer to this just-in-time inventory management
principle, for example, “continuous replenishment program” (CRP) or “computer
assisted ordering” (Barnes & Greller, 1993; Ratliff, 1996). In fact, these terms,
especially CRP, were mentioned by some of the survey respondents when asked
about just-in-time inventory management.

44
Independent Variables: Transaction Cost Factors
Derived from transaction cost literature are a number of independent variables
hypothesized to have an impact on electronic network use (H1-H3). They are
uncertainty, asset specificity and complexity of product description. With two exceptions
(product and time speciﬁcity to be discussed below), all independent variables described
in this section are measured by multiple items, each using the ﬁve-point agree-disagree

scale described earlier.

W

Williamson (1975, 1978) posited that varying types and number of contingencies
are usually embedded in the environment in which economic activities are coordinated.
These contingencies create uncertainty, which serves to drive up transaction costs, in part,
through complicating the contracts necessarily created to address the uncertainty.
Similarly, it has been argued by communication scholars that organizations must consider
and attempt to control uncertainty through planning (Beniger, 1986; Nass & Mason,
1990).

Measurement of uncertainty has been challenging as it is a complex,
multi-dimensional construct. Much of the research dealing with the use of information
technologies in supporting commercial activities and inter-organizational transactions has
focused on what has been labeled as environmental uncertainty (e. g., Gomes & Mentzer,
1991; Sabherwal & King, 1992; Sabherwal & Vijayasarathy, 1994). Distinguished from

internal uncertainty, this construct deals with contingencies arising outside the

45
organizational boundaries from diverse sources such as the political volatility of the
market, unpredictability of the actions on the part of suppliers and customers, and
complexity of production and marketing orientations. As such, it can be readily observed
that the “sub-concept” of environmental uncertainty still entails multiple dimensions.
Failure to distinguish among these dimensions has led to inconsistent results in existing
literature that are confusing and difﬁcult to interpret (Milliken, 1987).

As discussed in Chapter 2, the current research is focused on the importance of
networks to the acquisition of a key product from a major supplier. Rather than focusing
on unpredictable circumstances associated with internal task coordination and control, a
more relevant dimension of uncertainty would be the contingencies associated with the
efﬁcient and cost-effective ordering and receiving of inventory as replenishment needs
arise. Therefore, the survey items written for uncertainty are aimed at assessing the
perceived predictability of changes in the availability and price of a product when it is

needed.

Asset (ProductI Time and Knowledge) Speciﬁcity

Three types of asset speciﬁcity are measured in the current study, namely product
specificity, time specificity and knowledge specificity. Recalling the discussion in Chapter
2, this construct deals with the notion of the uniqueness of an asset needed by an
organization which renders the same asset useless to another ﬁrm. Inserting these
variables into the hypothesis, that asset speciﬁcity is positively associated with electronic

network use (HZ), yields the following “sub-hypotheses:”

46

H2a: Product speciﬁcity is positively associated with electronic network

adoption and use.

H2b: Time speciﬁcity is positively associated with electronic network adoption

and use.

H2c: Knowledge speciﬁcity is positively associated with electronic network

adoption and use.

Product Specificity. This variable is derived from the “physical asset speciﬁcity”
concept advanced by Williamson (1975) to refer to the unique machinery, equipment or
any other type of physical materials speciﬁc to a transaction. In their adoption of
transaction cost analysis to study the electronic networks and coordination, Malone et. a1.
(1987) also mentioned physical asset speciﬁcity as a key variable. Kraut et. a1. (1997)
labeled it “object speciﬁcity” in the empirical investigation of the extent to which a
production input is useful only to a speciﬁc organization needing it. Adopting this
deﬁnition to the current research, product speciﬁcity is deﬁned as the extent to which a
product is useful only to a speciﬁc organization needing it.

Product speciﬁcity is operationally measured by the variations in the label/brand
name of a product in three levels: private/store, regional or national. A product with a
private or store label (e.g., the Kroger label) represents high speciﬁcity, relative to
regional or national labels, since only a particular store can use that product without
repackaging. Similarly, a regional label is more speciﬁc than national brands because

stores outside of the region do not carry it. An example is the Spartan brand that is only

47
available in a few mid-western states. (Treatment of product speciﬁcity as it relates to
sampling procedures and the execution of the telephone survey is discussed more below
in the data collection section.)

Time Specificity. This type of speciﬁcity deals with the time pressure involved in
completing the transfer of an asset to the intended user (Malone et. al., 1987). The
underlying logic is that due to the short time window within which a product must be
delivered, there may be only one or a small number of suppliers capable of meeting the
time pressure. For example, because of the highly perishable nature of produce, a grocery
retailer may choose to order from a local supplier, rather than having inventory shipped
from a distant farm.’ Kraut et. a1. (1997) attempted to measure this construct by
observing the extent to which production is rushed and subject to rigid deadlines.
However, using three survey items, the composite measure was not highly reliable
(Cronbach’s Alpha = .68).

In the current study, time speciﬁcity is operationally deﬁned as the perishability
of a product. It is varied by the type of product on which a respondent was asked to focus
throughout the survey. “Fresh fruits and vegetables” and “dairy products” are considered
high in perishability, while “breakfast cereals” and “confectionery goods” are coded as
low in time speciﬁcity. (As in the case of product speciﬁcity, more discussion on the

treatment of the time speciﬁcity measure will be presented later in this chapter.)

 

’ This is holding constant other factors affecting the choice of supplier, such as the

quality of services, price, etc.

48

Knowledge Specificity. As another type of asset speciﬁcity, this has been
explicated in the literature as human asset speciﬁcity (Williamson, 1975, 1979; Malone
et. al., 1987) or human capital speciﬁcity (Joskow, 1985) that deals with skill transfer,
learning-by-doing and other types of human capital development that are speciﬁc to a
trading relationship. Kraut et. al. (1997) measured this construct with four items
assessing the degree to which employees of trading ﬁrms possess knowledge and skills
for dealing with each other and the industry as a whole.

Focusing on the dimension of relationship (rather than the industry as a whole),
survey items constructed to measure knowledge speciﬁcity in the current study is based
on this deﬁnition: the extent to which knowledge required to coordinate business

activities is perceived to be useful only to a speciﬁc pair of trading organizations or units.

Complexity ofProd_uct Description

Malone et. a1. (1987) deﬁned this concept as “the amount of information needed
to specify the attributes of a product in enough detail to allow potential buyers (whether
producers acquiring production inputs or consumers acquiring goods) to make a
selection” (pg. 486). At a practical level, it is rather difficult to implement “the amount
of information needed” in a way that reﬂects how product description is conveyed in
reality. The only attempt to measure this construct empirically was undertaken by Kraut
et. al. (1997). Unfortunately, the results yielded two single-item measures tapping the

ease and quickness with which a key input could be described.

49

The current study deﬁnes complexity of product description as the extent to which
it is perceived to be difﬁcult to describe a product. By this operationalization, complexity
is gauged by the perceived adequacy of different forms of representing the product to
convey its key attributes, such as pictures or multimedia representations (see Appendix A
for questionnaire). In addition, the use of standardized codes provide “short-hand”
descriptions of products that are simple to communicate, and hence effectively involve
less complex product information than goods that do not come with a coding system.
Some of the commonly used standardized coding systems in grocery retailing include

uniform product codes (UPC), stock keeping units (SKU) and universal look-up numbers

(ULN)

Independent Variables: Social Relational Factors

With respect to the impact of social relational factors, three variables are
considered in this study, namely importance of interpersonal relationships, mutual
commitment and trust. 6 All three variables entail Likert-type response ratings, with
mutual commitment and trust using the ﬁve-point agree-disagree scale, while responses
to the items dealing with interpersonal relationships vary along the continuum of

importance, like the importance of network use measure described above.

 

6 Originally, a fourth variable was developed, i.e., relative bargaining position (Bakos

& Treacy, 1986; Johnston & Vitale, 1988) and hypothesized to have a positive
association with electronic network use. However, the four survey items written for
this variable did not load together in post-hoe factor analyses. Therefore, this variable
was dropped from further consideration in the study.

50

lmportcmce of Interpersonal Relationﬂigs

Following the deﬁnition by Kraut et. al. (1997), this variable assesses the extent to
which relationships that have already been established among individual members of the
trading pair are perceived to be important to the two parties. This operationalization
strategy is adopted in the current study, in part, due to the fact that the treatment led to a
fairly reliable index measure in the Kraut et. a1. study (Cronbach’s Alpha = .85). For the
survey respondents, an interpersonal relationship is deﬁned as “any type of personal
connection, interactions and knowledge between members of your company and the
supplier” (see Appendix A).7

Furthermore, the measurement of the interpersonal relationship variable
incorporates the same seven activities associated with the acquisition of a key product on
which the importance of network use are rated. This design strategy provides for a
common base for the comparison of the modalities of personal vs. electronic connections,
and aids in providing a richer depiction of how interpersonal relationships affect
perceptions toward the importance of network use. Kraut et. al. (1997) found the two
modes of coordinating transactions to be complementary.

Also, like the importance of network use measure, the importance of interpersonal
relationships is rated with the same ﬁve-point importance scale. This treatment of the

concept as a continuous rather than dichotomous variable is consistent with Johnson’s

 

7 Knowledge referenced here is at the personal level, rather than the

institutional/organizational level, which is related to the knowledge speciﬁcity
construct.

51
(1993) observation that relationships in organizational networks are not a matter of

whether they exist or not, but that they are manifested in gradations.

Mutual Commitment

 

Recalling the literature review presented in Chapter 2, this construct is derived
ﬁ'om the notions of mutuality of interests, respect and orientation as manifested in the
commitment two parties to a relationship have toward each other (Johanson & Mattson,
1987; Powell, 1990). Speciﬁcally, inferred from existing literature is the argument that
the adoption and use of an electronic interconnection between two trading units is an
explicit demonstration of such mutual commitment (e.g., as indicated by substantial sunk
costs).

The construct of mutual commitment has not been previously deﬁned
conceptually or operationally. It is proposed in the current research that mutual
commitment gauges the extent to which decisions and actions of the two parties in a
retailer-supplier relationship are reﬂective of concerns for each others’ business interests
and pledge to maintaining a successful business relationship. Survey items written to

measure mutual commitment are based on this deﬁnition.

Trust
Trust is another elusive, multi-dimensional construct that makes measurement a
complex task. Creed and Miles (1996) found this concept to be related to such

sentiments as reputation and expectations of equitable treatment. Zaheer, McEvily and

52
Perrone (1996) argued that trust is a multi-level concept, at least to be distinguished
between the interpersonal and inter-organizational levels. In order to avoid the potential
problem of collinearity with the importance of interpersonal relationships variable, this
research focuses on trust at the inter-organizational level. Inter-organizational trust deals
with “the extent of trust placed in the partner organization by the members of a focal
organization” (Zaheer et. al., 1996, pg. 5). To avoid deﬁning a concept with itself, an
alternative deﬁnition is adopted, that is, the extent to which an organization believes that
its trading partner will refrain from malfeasance.8 With modiﬁcations in wording to
eliminate potential ambiguity and to ﬁt the current study, survey items measuring

inter-organizational trust were adopted from the Zaheer et. al. study (1996).9

Control Variables

Organizational Size

Organizational theories, such as the resource dependency perspective (Pfeffer &
Salancik, 1978), call attention to the fact that the size of an organization often has
implications for the slack resources available for investments in technology. Moreover,
size is often associated with the power a ﬁrm possesses in bargaining situations. In the

context of a trading relationship, especially when the business counterpart is relatively

 

This conceptual deﬁnition is adopted from Kraut et. a1. (1997). While the conceptual
deﬁnition is useful, the operational index was not reliable, with an alpha score of .43.

Therefore, the survey constructed for the current study did not follow those reported
by Kraut et. al. (1997).

The ﬁve items assessing inter-organizational trust in that study yielded a composite
measure of moderate reliability (Cronbach’s Alpha = .77).

53
smaller, a larger organization may be able to control how transactions should be
conducted (e.g., delivery schedule, pricing, etc.). Similarly, the decision to adopt a
supplier-customer electronic linkage may be dictated by the larger and more powerful
party.l0 Furthermore, as a speciﬁc consideration in this study, private store labels (i.e.,
high product speciﬁcity) are more likely to be available in large grocery chains than small
independent stores. As such, an observed relationship between product speciﬁcity and
electronic network use, or the magnitude of the association, might be affected by ﬁrm
size. Therefore, care should be taken to control for the impact of company size in testing
the hypothesized relationships.

As a control variable in this study, organizational size is measured by an index
comprised of annual sales, total number of employees and total number of stores. Figures
on annual sales are obtained from the 1997 Ward Business Directory of US. Private and
Public Companies, while the total number of employees and stores are provided by the

survey respondents.

Integration with Maior Supplier
Previous research has led to the conclusion that electronic networks use tends to
be associated more with internal suppliers or trading ﬁrms with whom there is a close,

pre-established relationship (Kraut et. al., 1997; Steinﬁeld et. al., 1995). Besides,

 

'° As a case in point, in the interviews referenced earlier, one executive of a large

magazine publishing ﬁrm contended that if his company should decide to transact
with color separators and printers electronically, these suppliers would have to
comply in order to continue getting business from the publishing ﬁrm.

54

whether a major supplier is internal or external is theoretically and empirically linked to

the transaction cost predictors of asset speciﬁcity and uncertainty. Therefore, to avoid the

speciﬁcation error of failing to account for an important confound, the level of integration

with the major supplier is included as a control variable.

Integration with major supplier is measured by the level of ownership the retailer

has in the most important supplier dealt with in the last 12 months to acquire a key

product. Three levels ofthis control measure, i.e., no ownership, partial ownership and

full ownership, are recoded from ﬁve response categories for the major supplier question,

as illustrated in Table 1.

Table l - Integration with Major Supplier

 

Original survey response categories
for major supplier question

“Integration with Major Supplier”
(level of ownership)

 

1. independent manufacturer or producer

independent wholesaler or distributor

1 (no ownership)

 

a partially-owned wholesaler or distributor

a joint-venture with another company

2 (partial ownership)

 

 

9‘95”!"

a wholly-owned wholesaler or distributor

 

3 (full ownership)

 

 

55
Data Collection
This section covers the issues and procedures related to data collection for the
current study. Presented ﬁrst is the rationale for the population (industry) selection,
which is followed by the discussion of sampling procedures, as well as the design and

execution of the telephone survey.

Population (Industry) Selection

The population selected for the empirical study undertaken is the US. grocery
retail industry (SIC 5411). According to the standard industrial classiﬁcations, this
category of grocery stores includes “supermarkets, food stores and grocery stores,
primarily engaged in the retail sale of all sorts of canned goods and dry goods, such as
tea, coffee, spices, sugar, and ﬂour, fresh fruits and vegetables, and fresh and prepared
meats, ﬁsh and poultry” (Hillstrom & Ruby, 1994).

The grocery retail industry has been selected for its empirical appeal as an
interesting population for addressing the research question of understanding motivations
for network adoption and use. For a number of years, electronic information systems and
networks have reportedly led to some transformation in terms of strategies and business
practices in the grocery industry. For instance, computers have extended the reach of
retail managers to stores and allowed them to exercise more control (The Economist,
March 4, 1995). Furthermore, the use of EDI and POS systems have been credited for
leading to the adoption of more efﬁcient and cost-saving operational principles and

practices, including different variations of the just-in-time inventory management

56
concept, such as “quick response” or “continuous replenishment programs” (CRP)
mentioned earlier (Barnes & Greller, 1994; Clemons, 1993). In the case of a large
grocery chain, EDI also facilitates category management by eliminating the need to
process paperwork associated with new or promotional items (Frozen Food Age, August,
1996)

At a industry-relational level, information technologies have also been found to be
associated with changing the bargaining dynamics between retailers and suppliers in
favor of network users over non-users (Sealey, 1994). Furthermore, as a mature,
“classic” retail industry of over 100 years old, it provides a suitable context for observing
how established orientations and patterns of doing business, based on such factors as
interpersonal contact and trust, may affect perceptions toward the use of electronic
networks.

In addition, ﬁom a practical, methodological perspective, there must be assurance
for variability in the key measures, especially the dependent variables of network
adoption and use. In this regard, academic ﬁndings and trade-press reports have
conﬁrmed varying levels of existing electronic network use, as well as continued interests
in exploring such potential beneﬁts as increased efﬁciency and enhanced relational
marketing (e.g., McGee, 1991; Sealey, 1994; Progressive Grocer, April, 1996). For
instance, according to a survey conducted by Frozen Food Age (July, 1996), retailers and
wholesalers use EDI for an average of 63% of their orders and transactions, including
interactions with manufacturers. Growing recognition for the strategic role played by

information technologies is also reﬂected in the report that industrial participants consider

57
meeting challenges of changing technological innovations and capabilities as a top-rated

problem to be tackled (Progressive Grocer, April, 1996).

Sampling Procedures

Ward ’s business directory of US. private and public companies (1997), along
with the supplement published in April, was used as the sampling ﬁame. Subsidiaries
and divisions of larger grocery chains, as well as unclassiﬁed companies, were excluded
ﬁ'om the sample population, based on the assumption that it is unlikely for technology
decisions to be made at the store level.” In order to avoid effectively surveying the same
company more than once, only corporate headquarters or holding companies of grocery
stores were targeted in this study.

A total of 230 ﬁrms were contacted, with the hope of targeting a sample size of
120. Of these 230 potential companies, 43 were unusable due to either disconnected
numbers or ﬁrms which are no longer in the grocery retail business, reducing the sample
population to 197. There were 44 refusals to participate, including “implied” rejections
from ﬁrms given unsuccessful attempts to set up interviews. One of the most commonly
provided reasons for refusal to participate was company policy, especially among the

larger grocery chains. Due to the competitive conditions in the grocery industry,

 

” For the unclassiﬁed listings, there was no information about whether these companies

were divisions or subsidiaries of a larger grocery chain. Furthermore, the annual sales
ﬁgures were not provided either. As was discovered later during pretests, some of
these unclassiﬁed companies are actually restaurants. It is likely that they were
classiﬁed under SIC 5411 because they practice food retailing within the deﬁnition of
the industrial classiﬁcation.

58
employees were instructed not to respond to any inquiries, including academic research,
to safeguard against the leaking of “trade secrets” to industrial rivals through the media.
Upon termination of data collection, a total of 143 ﬁrms were surveyed, representing a
response rate of 73%.

Sampling of organizations was stratiﬁed by size, based on the reported annual
sales of each ﬁrm in the Ward’s Business Directory. Large, medium and small ﬁrms
have annual sales of $200 million and above, $25 to $199.99 million and below $25
million, respectively. This research decision was based on the consideration of the likely
positive association between organizational size and network use, as addressed earlier in
the measurement section. At the early design stage of this research (as speciﬁed in the
research proposal), the intention was to have 30 ﬁrms representing each level of
organizational size, thereby yielding a total of sample 120; this was to ensure equal
representation of each category. Out of the total sample of 143, there are 48 large ﬁrms
in both the large- and medium-sized categories and 47 small grocery retailers.

Sampling was also conducted such that variability in the measures of product
speciﬁcity and time speciﬁcity would be ensured. As described earlier, product
speciﬁcity varies with the brand names/labels of products, from national (non-speciﬁc) to
regional to local/store (highly speciﬁc). Completed surveys were monitored in an on-
going basis as an attempt to achieve a somewhat proportional distribution of ﬁrms along
the three levels of the product speciﬁcity variable. As for time speciﬁcity, it was treated
by selecting four product categories, two highly perishable (i.e., fresh fruits and

vegetables and diary products) and two relatively low in perishability (i.e., breakfast

59
cereals and confectionery goods, e.g., potato and chips, candy, etc.). Each potential

respondent was assigned one of the four product types on a random basis.

Telephone Survey

A survey research design was chosen for the empirical portion of this study
because it is an effective method for measuring perceptions (Babbie, 1990). (The
rationale for collecting perceptive data was discussed in the measurement section earlier.)
Response rate is a critical factor in the present study, and as such, telephone interviewing
was chosen over self—administered survey by mail, since real-time interviewing normally
yields a considerably higher response rate (Babbie, 1990). Besides, telephone interviews
bring timely results and are effective when the survey is limited in length.

Recalling an earlier discussion on identifying key informants based on how they
are likely to be affected by technology use (Tomatzky and Klein, 1982), respondents
targeted in each organization is “the person who is most responsible for acquiring” a key
product. This strategy led to subjects who are maj or/head buyers, product or category
managers for one of the four product categories pre-assigned to their companies. These
are the people most knowledgeable of the suppliers with whom their companies do
business. These informants also have ﬁrst-hand experience with whether and how
electronic networks are used to support transactions with major suppliers.

As an attempt to minimize unexplained error variance arising from unspeciﬁed
product variations, respondents were asked to think of and focus on “one of the best

selling type” of the pre-assigned category of product throughout the interview (see

60
Appendix A). Similarly, they were requested to consider the same major supplier, that is
“the most important supplier you’ve worked with within the past 12 months,” in the
entire questionnaire. This is especially important to helping ensure that the relational

factors assessed are based on the same set of retailer-supplier relationship.

melt

Three interviewers were hired to help with conducting the telephone interviews.
Each of them were given a two-hour training to introduce them to the survey and explain
the ﬂow of questions, particularly in parts where they are required to complete a
questionnaire item with information provided by respondents. For instance, the code
{TOP PRODUCT} in the survey protocol was to be replaced with the best selling product
reported by each respondent. Also covered in the training sessions were the pay structure

and scheduling issues. '2

Pretesting

Ten companies selected from the unclassiﬁed sub-listing of ﬁrms in the Ward’s
Business Directory were contacted for pretesting the survey instrument. The pretest

sample was not drawn from the actual sample population due to the concern of “wasting”

 

‘2 The training sessions also included the use of the Survey System, a computer-assisted

telephone interviewing (CATI) system in the Information Technology and Services
Laboratory. The survey was originally intended to be executed with the use of the
CAT] system. However, due to problems related to system and network instability, it
was quickly determined to be too risky to rely on the CATI system, and the data
collection was conducted using paper-and-pencil questionnaires.

61
potential cases, given the small population. Besides, the main objective of the pretest was
to check for ambiguity of wording, which was accomplished with the unclassiﬁed ﬁrms.
As a result of pretesting, some items were revised, such as two of the questions written to

measure operationally knowledge speciﬁcity.

AM Telephone Interviews

The telephone interviewing took place during the ﬁrst two weeks of September,
1997. Each sampled grocery retailer was contacted at the main number provided in the
Ward ’s Directory and the key respondent identiﬁed.” Then, the interviewer would ask to
be transferred to the potential informant, who would then be given the introduction to the
project, promised anonymity and conﬁdentiality, and invited to participate in the
research. As an attempt to minimize rejection arising out of the coincidence of reaching
potential respondents at a bad time, appointments were made for the survey to be
executed at a more convenient time. In addition, in some cases, the key informant was
not reachable during the ﬁrst call. In those situations, messages were either left in their

voice-mail or with their assistants, requesting for an appointment. When speaking to the

 

'3 A survey log sheet was used to facilitate the management of calls and the recording of

crucial information (see Appendix B). Interviewers were given these survey call logs
with the telephone number and pre-assigned product category (measuring product
speciﬁcity) prior to the ﬁrst attempt to contact each sampled ﬁrm. They were
instructed to use the log for recording the names and contact information of
respondents, as well as appointment times, if necessary. As a project management
tool, the log sheets were colored-coded: light pink, lavender and dark pink to
represent small, medium and large ﬁrms. In addition, the call logs also served as a
record of how many calls each interviewer was able to complete in an hour; this is
related to the incentive portion of their pay.

62
assistants, they were also asked as to what day and time might be best for reaching the
identiﬁed buyers/ product managers. In order to accommodate these contingencies of not
being able to speak to a potential respondent during the initial call, up to seven callbacks
were built into the design to follow up on appointments and messages left.”

Interview calls were made in the Information Technology and Services
Laboratory of the Department of Telecommunication. Dillrnan (1978) recommended the
use of a centralized interviewing site to maximize the opportunity for monitoring of calls,
especially with inexperienced interviewers. It also enabled the interviewers to seek

clariﬁcations, as needed, throughout the data collection.

Chapter Summary

This chapter began the description of the empirical portion of this dissertation by
reporting the methods and procedures employed in this dissertation. In doing so, issues
and procedures adopted in the measurement of key variables and data collection—
population selection, sampling and the execution and management of the telephone
survey—were described above. Together with the results chapter to follow (Chapter 4),

the presentation of the “technical” aspects of the empirical portion of this study will be

completed.

 

” Towards the end of the data collection, a number of potential respondents actually

returned my call after six or seven messages have been left with them. In one case,
the product manager interrupted a meeting he was having with his assistant to take
my call, telling her that “this lady has been trying to reach me for over a week!”

Chapter 4

RESULTS

Continuing the description of the empirical portion this dissertation research
begun in Chapter 3, the current chapter reports the results obtained from the telephone
survey. Included in the presentation are procedures employed to interpret the data
collected and to evaluate their ﬁt with what has been conceptualized and proposed.
Empirical evidence is furnished to reveal the presence (and absence) of signiﬁcant
relationships between the various transaction cost and social relational factors and
electronic network adoption and use. Research ﬁndings also uncover differential patterns
of association between the two sets of antecedent factors and the two criterion variables,
i.e., adoption of electronic information networks and their perceived importance to
supporting transactions with major suppliers.

This chapter is organized into three major sections. First, general steps taken to
prepare the data for further statistical analysis are presented. In the second section,
techniques used for the construction of index measures are described, including the
discussion of validity and reliability issues associated with index construction. This is
followed by the last section of the chapter, reporting the results of regression analysis

conducted to assess the hypothesized relationships.

63

64

General Data Preparation

As commonly practiced in quantitative studies, before the results from the
telephone survey could be used for testing the research hypotheses in the current research,
a number of data processing steps needed to be performed. These procedures were
necessary for “cleaning” the “raw” survey data and transforming them into appropriate
formats for use in further statistical analysis. Procedures conducted to fulﬁll these data
cleaning and transformation necessities were associated with the following major
objectives: (i) data integrity checks; (ii) recoding of scale and response categories; and

(iii) missing data treatment.l

Data Integrity Checks

An important data processing task was checking for entry errors, regardless of
how much care has been taken to ensure accuracy when the data were entered. One data
check procedure employed involved randomly selecting cases in the data set and
comparing the scores with the original responses recorded in the paper questionnaires for
those companies. Another technique used for detecting errors was generating univariate
frequencies, an efﬁcient statistical routine for checking data integrity that is widely
practiced. Any unusual scores would have been revealed in the frequency tables, such as

a score of “6” appearing in the distribution of a questionnaire item measured with a ﬁve-

 

In many cases, another common reason for conducting data checks is for identifying

the presence of outliers, which would lead to false estimation of relationships. In the
current research, since most variables are measured with Likert-type response scales

with pre-assigned score ranges, the threat of outliers is not a concern.

65
point response scale. F urtherrnore, histograms were also consulted as aids for visualizing
anomalies in the data. Using these techniques, several errors were identiﬁed and

corrected.

Recoding of Scale and Response Categories

Most variables in this research have been assessed by the scaling method,
particularly Likert-type. Scaling is a technique of validating the existence of a
hypothetical construct, the relative magnitude of which is inferred by a range of
numerical scores assigned to different levels of the construct (Arnold, 1989). This
technique also aids in the development of operational indices, which will be discussed in
the next major section on the “construction of index measures.”

As introduced in the research design (Chapter 3—Methods and Procedures), the
independent variables of product speciﬁcity and time speciﬁcity entail a scaling
procedure that involved assigning scores to different types of product labels and products,

as summarized in the following Tables 2 and 3.

Table 2 - Product Speciﬁcity

 

 

 

 

 

Product Label Type Product Speciﬁcity (Scale of l - 3)
National 1 (low)

Regional 2 (medium)

local/store (private) 3 (high)

 

 

 

66

Table 3 - Time Speciﬁcity

 

 

 

 

Product Type Time Speciﬁcity (Scale of 1 - 2)
Breakfast cereals 1 (low)

Confectionery goods

Fresh fruits and vegetables 2 (high)

Dairy products

 

 

For a number of reasons, it is not uncommon to ﬁnd that originally assigned or
recorded scale scores need to be recoded before further data processing and analysis can
be applied meaningfully. One purpose served by data recoding is the grouping of cases
that are similar in some observed dimension to facilitate a more efﬁcient way of
describing the sample along that characteristic. In the current study, a case in point is the
questionnaire item gauging how integrated a retailer is with its major supplier, introduced
in the last chapter (see Table 1 in Chapter 3). For the purposes of analyzing the level of
integration, the key dimension of variation to be considered is whether a major supplier is
an independent trading partner, a partially-owned operation or fully owned by the retailer.
Whether this supplier produces the key product needed by the grocery buyer or is merely
a distributor is not a critical distinction to be made. Therefore, the response categories of
“independent manufacturer or producer” and “independent wholesaler or distributor”
could be collapsed into one group, “no ownership.” Following the same logic, the

recoded category of “partial ownership” was created by grouping the two original

 

67
response categories of “partially-owned wholesaler or distributor” and “joint-venture with
another company.”

Furthermore, each company’s total annual sales, total number of employees and
total number of stores were all originally measured with real numbers spread out in large
ranges. These raw numbers were collapsed into more manageable and interpretable
number of scale categories for describing these items. The grouping criteria used are

shown in the following tables (Tables 4, 5 and 6).

Table 4 - Total Annual Sales

 

 

 

 

 

Original Figures Percentage of Cases Grouped Categories and Scores
less than $25 million 32.4 % 1 (small)

$25 million to $199.99 million 34.4 % 2 (medium)

$200 million and above 33.2 % 3 (large)

 

 

 

Table 5 - Total Number of Employees

 

Original Figures Percentage of Cases Grouped Categories and Scores

 

l to 99 20.3 % 1

 

100 to 299 20.4 %

 

 

2

300 to 999 18.6 % 3
1000 to 4999 18.7 % 4
5

 

 

5000 and above 22.0 %

 

 

 

 

 

68

Table 6 - Total Number of Stores

 

 

 

 

 

 

 

Original Figures Percentage of Cases Grouped Categories and Scores
1 16.9 % 1
2 to 4 20.6 % 2
5 to 14 22.1 % 3
15 to 70 19.8 % 4
71 and above 20.6 % 5

 

 

 

Another reason for data recoding is to ensure consistency in the directional
ordering of scale scores for all items measuring the same operational construct.
Expressed differently, a high score for all of the items constituting the same construct
should indicate a high level of its presence or absence. For instance, scores of “5” on a
ﬁve-point agree-disagree response scale across items should all indicate strong
agreement. This is a concern associated with the construction of multi-item operational
‘ indices, which will be discussed in more detail below in the section on the construction of

index measures.

Missing Data Treatment
Another data preparation necessity is associated with the identiﬁcation and
treatment of missing data. With survey research, missing values often result from

respondents not being willing or able to answer one or more questions. A commonly

 

69
discovered reason for the lack of willingness to respond to a survey has to do with the
wording of questionnaire items, which might be considered too sensitive or value-loaded
by some respondents. In addition, a lengthy, laborious questionnaire can potentially
exhaust the patience of some participants, causing them to abandon the survey before it
has been completed. Missing data can also result from situations in which respondents
are asked questions seeking knowledge they do not have. F urtherrnore, the inability to
answer a question can arso be traced to the item not being understood by a respondent.
These potential pitfalls of survey research are discussed in the survey methodology
literature (e.g., Babbie, 1990; Dilhnan, 1978; Schuman & Presser, 1981).

In order to minimize both the number of survey items that would be difﬁcult to
answer and the number of survey participants who would decline to respond to any given
question, care was taken with the design of the questionnaire. As discussed in the
previous chapter, one chief reason for pretesting the survey instrument was, in fact, to
help to ensure that the above stated problems would be avoided.

None of the 143 respondents in the sample abandoned the telephone interview
before it was completed. There were hardly any missing values in the items measuring
the independent variables of this study. However, this is not the case with the dependent
variables, especially the importance of electronic network use measure. Because of the
different considerations associated with operationalization and measurement decisions,
missing data from the telephone survey were not treated with the same method. Missing

data treatment for the different variables is discussed below.

70

Dependent Measures

For a reason other than the common “don’t know” and “no answer” problems
discussed earlier, a “special” case of missing data was associated with the importance of
electronic network use variable. Out of the 143 grocery buyers who participated in the
survey, 43 indicated that they were not using any computer-based data networks with
their major suppliers at the time.2 These non-usage responses were coded as “9” (“don’t
use”) in each of the seven items asking respondents about the importance of using
electronic networks for carrying out various activities associated with the acquisition of a
key product. In turn, these 43 cases were excluded from later analysis performed to test
the relationships among the different independent variables and the importance of
electronic network use. Because the missing information associated with this variable
has special meaning, replacing them with other scores (such as the series mean value),

would be inappropriate.3

 

A number of respondents qualiﬁed their non-usage by adding that they deal with their
major suppliers via telephone and fax.

One possibility considered was that the absence of network use could be interpreted
as being equivalent to networks not being important at all to the acquisition of key
products. (After all, the ﬁrms not using networks have been acquiring products via
other means!) This interpretation would mean replacing the non-usage scores of “9”
with “1” (“not important at all”). This data recoding strategy was adopted by Kraut
et. al. (1997). However, there does not seem to be a strong enough theoretical or
conceptual justiﬁcation for treating the non-users as equivalent to the users who
consider network use to be not important at all in this study. Furthermore, there is
little to be gained empirically, given that these ﬁrms constitute almost one-third of the
sample. Recoding this variable would cause the distribution to be censored below,
i.e., a high concentration of responses clustering in the low end of the score range.

71

With the exclusion of this large pr0portion of cases, there is the concern that the
sub-sample included in later analysis might be different from the excluded cases along
some critical dimensions. Should such systematic differences exist, it would have
implications on the representativeness of the group of cases analyzed and the
generalizability of the research ﬁndings.4 Therefore, the included cases are compared
against the excluded ﬁrms, by observing whether the mean scores for the independent and
control variables are different. As can be seen in Table 7, the two groups are different
with respect to three variables: (1) The excluded non-users group rated complexity of
product description to be lower than the 100 cases included in subsequent analysis. (2)
The 43 buyers who did not use networks with their major suppliers considered
interpersonal relationships to be more important than the network users. (3) The non-
users also considered their suppliers to be relatively more trustworthy. As will be seen
later in this chapter and Chapter 5, these group differences are consistent with the
observed relationships between these three independent variables and the various
criterion measures.

Listwise exclusion of cases with missing values were also applied to the measure
for electronic network adoption, that is, the information technology cluster variable.
Three respondents did not answer the questions gauging the adoption of six technologies

that would allow for electronic exchange of information with suppliers. Recall from the

 

‘ In examining reporting of survey research involving organizations, the issue of

potential differences between ﬁrms included in the analyses and those in the non-
response category, researchers have predominantly compared ﬁrm characteristics.
Common attributes compared include size, as indicated by annual sales, number of
employees.

Table 7 -— Comparison of Electronic Network Users and Non-Users

72

 

 

 

Mean scores Difference

Users Non-users Prob.

Variable (N = 100) (N = 43) t-score Level
Uncertainty 2.35 2.17 -1.41 .17
Product Speciﬁcity“ 1.90 2.07 1.19 .24
Time Speciﬁcity" 1.46 1.56 1.28 .21
Knowledge Speciﬁcity 3.74 3.86 .97 .34
Complexity of Product Description 2.67 2.17 -2.67 .01
Importance of Interpersonal Relationships 3.79 4.06 2.03 .05
“static” activities 3.79 4.10 2.09 .04
“dynamic” activities 3.80 4.03 1.68 .10
Mutual Commitment 4.18 4.19 .12 .91
Trust 3.91 4.19 2.32 .02
Organizational Size 2.82 2.51 -l.73 .09
Integration with Major Supplier“ 1.55 1.46 -.83 .41

 

 

 

 

 

* Based on 3-point scale.

** Based on 2-point scale.

Note: The remaining items are based on a 5-point scale. Mean scores for multi-item

indices are obtained from taking the mean of the summed, average score. For example,

the mean score for mutual commitment of 4. 1 8 is the average of the mean score for the

four items measuring it, i.e., 16.72/4.

73
previous chapter that this information technology cluster measure reﬂects the total count
of the six technologies adopted by a respondent’s organization. Without a theoretically
justiﬁable ﬁgure to be used for substituting the missing data in this criterion measure, the
more conservative treatment of excluding these three cases from hypothesis testing was

adopted.

Independent Measures

For the independent variables, among the common alternatives for treating
missing data, the strategy adopted was that of replacing them with the mean scores of the
items. Listwise exclusion of cases was ruled out as too “strict” of a missing data
treatment for these variables due to a number of reasons. First, no survey question
measuring any of the independent variables is critical to the extent of warranting the
entire case to be excluded due to missing values. Secondly, there was not a single
respondent who had consistently refused or failed to answer questions in the survey.
Finally, given the listwise exclusion treatment of missing data in the two dependent
measures, the analytic sample sizes of 100 and 140 do not lend much statistical power to
hypothesis testing. Deleting more cases due to a small number of missing values in the
independent variables would further attenuate the already limited statistical power in the
study. Similarly, because of this consideration of statistical power, the option of pairwise

deletion of scores was also deemed less desirable than replacing them.

74

Control Measures

For the two control variables, missing data were replaced after the raw scores
have been grouped (discussed earlier). Missing values for the total number of employees
and total number of stores were replaced by the average scores in the associated level of
organizational size as stratiﬁed by annual sales. For instance, a missing value for the
number of employees in a ﬁrm with medium-level annual sales was replaced by the mean
score of cases in the same sales category. Due to the large ranges associated with these
two items, this method of missing value replacement has been deemed a relatively more

accurate “approximation” of the missing data than using the series means.

Construction of Index Measures

Babbie (1990) distinguished between “scales” and “indices,” even though these
labels have tended to be used interchangeably to refer to multi-item, operational
measures.’ A scale assumes a more rigid intensity structure than an index in the scaling
of responses across items constituting the composite measure. One manifestation of this
intensity structure is that the agreement with one item implies agreement with other
items, as in the case of a Guttman scale. In other types of scales, the difference between
any pair of scores must measure the same relative magnitude of the presence/absence of a

construct across all items constituting the same scale. As such, scales are normally

 

The term “scale” described here in the context of composite measures should not be
confused with scaling, the assignment of numerical scores to an operational construct,
referenced earlier. An index, though not a scale in the strictest sense, involves scaling
in form of pre-assigned numerical response categories.

75
created with items measured by at least interval-level scaling. Index measures deal with
more relaxed intensity requirements, as long as the scaling patterns are the same across
items forming the intended composite measure. Unlike scales, indices can be created
with ordinal-level measures, a characteristic of Likert-type scaling. Following these
“technical” speciﬁcations as described by Babbie (1990), the composite measures in the
present research are best attributed as indices.

Because of the requirement that an index must consist of items with the same
scaling patterns, the need for scores to be recoded sometimes arise. This is generally the
case when items have been worded in Opposite directions. In the context of the ﬁve-
point, agree-disagree response scales used in this study, this wording direction issue is
manifested in relatively high scores (e.g., 4 or 5) on some items being associated with
relatively low scores of other items measuring the same construct. For instance, in
examining the content of two of the items measuring uncertainty, a high score on (i.e.,
strong agreement with) these two statements would suggest the perception of low
uncertainty.6 However, for the other three uncertainty items, strong agreement would
indicate a high level of uncertainty. Therefore, even assuming that these ﬁve items were
unidimensional indicators of uncertainty, the response scales of the ﬁrst two items

mentioned must be reverse-coded (reﬂected), before the uncertainty index could be

 

6

The two items are: “You can easily get a replenishment of {TOP PRODUCT}
anytime you need it;” and “Your company has a great deal of control over the
ﬂuctuations of price you pay to acquire {TOP PRODUCT}” (Appendix A).

76
constructed from summing all these items.7 Another example is the reﬂection of the item,
“you will not hesitate to switch to a different supplier,” to convey a high (instead of low)

level of mutual commitment when the score is high.

Construct Validity and Factor Analysis

Validity is the extent to which a variable measures what it purports to measure. In
working with hypothetical constructs, it is particularly important that construct validity be
established, that is, the extent to which the operationalized indicators indeed assess the
theoretical constructs. In the context of the current study, the operationalized indicators

are the survey questions.

Convergent and Discriminant Valiplja

Examining the wording of survey items represents an attempt to check construct
validity through face validation, i.e., the extent to which the questions read or sound as if
they measure the intended variables. A “stricter” construct validation method involves
the statistical assessment of two types of validity evaluated in conjunction with each
other. These two types of validity are convergent and discriminant validity (Bollen,

1989; Campbell and Fiske, 1959; Nunnally & Bernstein, 1994).8

 

7 Unidirnensionality is discussed in the next section on construct validity and factor

analysis. As it turned out, the two reﬂected uncertainty items did not load on the
same factor as the other three items when factor analysis was run, and were excluded
from the construction of the uncertainty index.

8 Discriminant and convergent validities are often discussed in association with the use

of multi-trait, multi-method matrices, introduced by Campbell and Fiske (1959). In

77

Convergent validity deals with the extent to which the operational measures of a
hypothetical construct have strong associations with one another. Statistically, an index
with high convergent validity means that the inter-item correlations among its indicators
are both positive and signiﬁcant.9 In the context of the current research, the items
measuring trust, for instance, must be highly and positively correlated with one another.

Discriminant validity is established when it can be conﬁrmed that the indicators
of one construct is not actually tapping a different underlying trait. In empirical terms,
the inter-item correlations among operational measures of the same constructs ought to be
higher than the associations between these measures and the indicators of other
constructs.lo Following the above trust example, the inter-item correlations among the

trust items must be higher than their correlations with other survey questions.

 

(continue ﬁom previous page)
these types of construct validation designs, at least two types of methods, e.g.,
Guttman scales and Likert-type indices, are used to measure each construct (trait), and
two or more traits are assessed. Although this study only employed one method, the
principles of comparing relative associations among items measuring the same
construct vs. those tapping different underlying hypothetical traits are still applicable
to the evaluation of construct validity here.

If multiple methods were used, the correlations between the different methods
measuring the same trait must be statistically signiﬁcant.

In the multi-trait multi-method context, the correlations among items measuring the
same trait ought to be higher than: (i) the correlations with any other item with which
these items do not share a common method or trait; and (ii) the correlations of
different traits measured by the same method.

78

Unidimenpionalitv

In addition to establishing the convergent and discriminant validity of operational
indices, construct validation also involves verifying the unidimensionality of composite
measures. Recalling the discussion in Chapter 3, like all hypothetical constructs, the
variables in the current study are complex and multi-dimensional. Examples include the
concepts of uncertainty and trust. A priori operationalization was aimed at tapping a
single dimension of each of these constructs. The need to establish construct validity
deals, in part, with the a posterior conﬁrmation that an index measure is consisted of

unidimensional indicators.

Factor Analysis

Factor analysis is a statistical technique used for establishing the various
considerations associated with construct validation described above. This method
identiﬁes underlying factors that are accountable for different “pools” of common
variance in the data (Nunnally & Bernstein, 1994). In the (unlikely) absence of
measurement, sampling and other types of research design errors, these empirically
extracted factors should match with the theoretical constructs of a study.

In factor analysis, validity is established, in part, by examining factor loadings,
which are the correlations between operational items and extracted factors. For
convergent and discriminant validity to be established, the factor loading of each item on
the underlying construct it is supposed to measure must be signiﬁcantly higher than its

cross-loadings on other factors. Unidirnensionality is veriﬁed if survey items written for

79
one construct do not load on more than one factor separately. A widely used decision
rule for interpreting factor analytic results is that the absolute value of the factor loading
of an item must be at least .60 and that of cross-loadings must not be higher than .40
(Nunnally & Bernstein, 1994).

To establish convergent and discriminant validity, as well as unidimensionality,
the survey items constructed for the independent variables of this study were factor
analyzed using SPSS for Windows 6.1 (Norusis/SPSS, Inc., 1993). The principal
components method was used to extract factors sequentially, such that each additional
factor accounts for the largest amount of common variance left to be explained in the
data.

Furthermore, to help with the interpretation of factors extracted, a varimax rotated
solution was requested. Varimax rotation aims at maximizing the variance of the squared
loadings in the identiﬁcation of factors (hence the name “varimax” for “variance
maximization”). In addition, this method of rotating extracted factors is intended to
produce a solution with orthogonal factors, i.e., ones that are uncorrelated with each
other. As can be seen, these features of varimax rotation are desirable for maximizing
construct validity. Moreover, having orthogonal indices would also help to reduce the
potential problem of multicollinearity among the predictor variables in regression
analysis.

Based on the above techniques, procedures and criteria for evaluating factor
analytic results, survey items measuring the ﬁve theoretical constructs of mutual

commitment, trust, complexity of product description, uncertainty and knowledge

80
speciﬁcity were factor analyzed. Items with low factor loadings and/or high cross
loadings were eliminated, resulting in a ﬁve-factor solution (see Table 8). Items with
high factor loadings on each of the ﬁve factors were summed for the construction of
indices. For instance, due to low factor loadings on the intended construct and high cross
loadings, four of the six items intended to measure knowledge speciﬁcity were excluded.
Similarly, both the measurement indices for mutual commitment and trust each excluded
an item that did not load together with the other items constituting the composite
measure.

Aside from the independent variables, the items measuring the importance of
electronic network use were also factor-analyzed. Somewhat surprisingly, the seven
activities associated with the acquisition of a key product loaded on two separate factors
(see Table 9 for factor loadings). These results indicate that the responses to the seven
items did not follow a consistent pattern. Indeed, as can be seen in Figure 2, it would
appear that networks were considered to be more important to some activities than others.
Speciﬁcally, electronic linkages have been rated to be more important for activities that
do not seem to require much interaction with the suppliers than tasks that would likely
require more back-and-forth exchanges. In other words, higher perception of network
importance appears to be associated with product acquisition activities that are more
“static” than “dynamic” in nature. Based on these observations, two separate importance
of network use indices were created, which will be labeled as “importance of network use
to ‘static’ activities” and “importance of network use to ‘dynamic’ activities.” (Further

elaboration on the conceptual differences between these two measures will be presented

81

Table 8 - Indices Resulting from Factor Analysis

 

 

 

 

 

 

 

 

 

 

 

Factor
Factor (variance explained) Loadings
Mutual Commitment (27.9 %)
You believe that this supplier is as committed as you are to the success of
your business relationship .84
You believe that this supplier cares about the success of your business .80
The success of your supplier’s business is important to you .74
This supplier is very responsive to your company’s needs .74
Trust (15 %)
You have complete conﬁdence that this supplier will not take advantage of
you, even if given the opportunity to do so .79
Overall, you would consider this supplier to be trustworthy .79
You have complete conﬁdence that this supplier will always keep promises
made to your company .79
Complexity of Product Description (10.1%)
Multimedia representation of {TOP PRODUCT}, that is some combination
of text, picture or sound, is necessary to describe the features of the .90
product
Only a picture or other visual representation of {TOP PRODUCT} can .85
describe adequately the features of the product
Uncertainty (8.6 %)
The availability of {TOP PRODUCT} changes a great deal over the course
of a year .75
The amount of {TOP PRODUCT} changes a great deal from one order to .66
the next
The price you pay for {TOP PRODUCT} changes a great deal from one .65
order to the next
Knowledge Specificity (8.1 %)
Having only general knowledge of grocery suppliers is not enough for .81
doing business with this supplier effectively
Having only general knowledge of grocery retailing is not enough for .80

doing business with your company effectively

 

82

Table 9 - Importance of Electronic Network Use

 

 

 

 

 

Factor
Factor (variance explained) Loading
3
Importance of Network Use to “Static” Activities (17.1 %)
Developing speciﬁcations of an order .82
Ordering a key product .79
Seeking price information .68
Importance of Network Use to “Dynamic”Activities (43.4 %)
Fixing after-sales errors or problems .81
Monitoring the quality of the orders received .75
Practicing just-in-time inventory management .74
Negotiating the terms of an agreement to purchase a key product .60

 

 

Figure 2 - Importance of Electronic Network Use to Various Product Acquisition Activities

 

1
I
V!
M

83

—
"l
M

1

l
M In
N

mean scores

1.5 a—

ﬁxing after-sales
errors

just-in-time
r inventory
management

_ monitoring of
order quality

r ordering

_ negotiating an
agreement

b seeking price
information

 

specifying an order

84
in Chapter 5—Discussion.) These two indices were analyzed separately in the
hypothesis-testing phase of the data analysis.

Recall that the same seven activities have been used to gauge the importance of
interpersonal relationships, and that the rationale of this design strategy was such that
there would be a common base for the comparison of the modalities of personal vs.
electronic connections. For this reason, the same treatment of constructing two separate
variables for the importance of interpersonal relationships was applied when analyzing
the two importance of electronic network use measures, producing “the importance of
interpersonal relationships to ‘static’ activities” and “the importance of interpersonal
relationships to ‘dynamic’ activities.” The seven items, which load on the same factor,
were added to form one composite measure when the criterion variable was the
information technology cluster.

Also factor-analyzed are the three indicators of the organizational size index, i.e.,
annual sales, total number of employees and total number of stores. The scores of these
three measures had to be standardized ﬁrst, because they were based on different scale
ranges, i.e., three (“1” to “3”) for annual sales and ﬁve (“1” to “5”) for the other two

items.

Reliability
Aside from validity concerns, it is important to check the reliability of the index

measures created. Reliability in general deals with the extent to which measurement of a

85
construct is consistent. A reliable measure should produce almost identical results across
different measurements in the same research settings.

In the context of index construction, the most relevant type of reliability is that of
internal consistency, i.e., how well the items summed to form an index “hang together”
(Babbie, 1990). Statistically, this is observed by examining the size of inter-item
correlations. High reliability is indicated by strong associations among items constituting
the index. As a commonly used measure of internal consistency reliability, Cronbach’s
Alpha is computed based on inter-item correlations (“standardized item alpha” in SPSS
output). Table 10 shows the alpha scores for all the indices in this study.

As can be seen, the survey yielded moderately reliable index measures. These
reliability results are not uncommon for indices based on only two to three items.
Overall, the social relational variables have been measured more reliably than the
transaction cost factors. In particular, caution needs to be called especially to the lack of
internal consistency reliability in the indices of uncertainty and knowledge speciﬁcity.
Unfortunately, the lack of reliability in measurement attenuates the strength of
associations, which, in turn, has negative implications on hypothesis testing. Absence of
a relationship cannot be conﬁdently attributed to the “failure” of theoretical predictions or

conceptual reasoning.

86

Table 10 - lntemal Consistency Reliability of Index Measures

 

 

Index Cronbach’s Alpha (or)
Importance of Network Use (based on 100 cases)1
“static” activities .71
“dynamic” activities .75

 

Transaction Cost Factors

 

 

 

Uncertainty .51
Knowledge speciﬁcity .52
Complexity of product description .76

 

Social Relational Factors

 

 

 

Mutual commitment .84
Trust .82
Importance of interpersonal relationships2 .84
“static” activities .83
“dynamic” activities .72

 

Control variable

 

 

Organizational size .94

 

 

 

‘ When the “don’t use” responses for the 43 excluded cases were replaced with “1” (as

discussed in an earlier footnote), the seven items loaded on the same factor, and
Cronbach’s Alpha came out to be .92. This data structure is likely a result of having
one-third of the sample scoring in a consistent pattern in all seven items constituting
the index. Again, this is a procedure that is hard to defend, although it yields a more
respectable alpha score.

2 This is the alpha score when the seven items were summed to form one index.

 

87
Hypotheses Testing
Regression Analysis

Given that the current research is aimed at examining the relationships among a
number of independent variables and each of the dependent variables in this study,
multiple regression analysis was deemed the most appropriate statistical technique (Hair,
Anderson, Tatharn & Grablowsky, 1979). This method entails estimating the
mathematical best-ﬁt plane that summarizes the relationship between the dependent
(criterion) and independent (predictor) variables. The key is to minimize the error of
estimation. In particular, the estimation method of ordinary least squares (OLS) was
employed, which minimizes the sum of squared deviations of the criterion variable.

Like most statistical techniques, for OLS regression analysis to function
“properly,” a number of assumptions must be met. Among them, the distribution of the
observed variables must be linear and normal. In general, regression analysis is fairly
robust with respect to violations of these two assumptions. In other words, even if the
distributions are not perfectly linear or normal, the OLS estimates will not be affected
very much. However, OLS is less tolerant with violations of the assumption of
homoscedasticity, i.e., the variance of the criterion variable must be constant across
different levels of the independent variables. Another key assumption is that errors must
not be correlated because for some systematic, common cause failed to be speciﬁed in the
regression model. These assumptions were checked by examining descriptive statistics,
histograms, scatterplots and residual plots; there are no serious violations of these

assumptions.

88

Another important consideration when running regression analysis is that
predictors are not highly correlated with each other. In the event that two regressors are
highly correlated with one another, the problem of collinearity is present
(multi-collinearity, if more than two highly correlated regressors). When this is the case,
the magnitude of the impact a particular predictor has on the criterion is really attributable
to the predictor’s high correlation with another strong regressor. As discussed in
methodological literature, there is no ideal way for treating the problem of collinearity.
One method is to enter the two (or more) highly correlated regressors into separate
regression equations and observe their effects in isolation from one another. As can be
seen in the Pearson Product-Moment correlations (Table 11), mutual commitment
is highly correlated with trust, raising the possibility that collinearity might be a problem.
This was noted when regression analysis was conducted.

In interpreting regression results for the purposes of testing hypotheses, a number
of statistics are meaningful. First, there is the regression coefficient, which is statistically
the partial correlation between a predictor variable and the criterion variable. This
coefficient indicates the strength of association between the predictor and the criterion,
while controlling for the impact of other regressors. When standardized, regression
weights, also known as betas (B), allow for the relative magnitude of impact among
different regressors to be compared. Since the measurement in this study involves

different levels of scaling, standardized beta weights were examined when evaluating the

research hypotheses. For instance, product speciﬁcity was measured with three levels

89

37. Reese-£580 u oﬁmwco 305027. 8355 u comm—83

 

5:395 3qu 53> coup—wax: n 98.8.: 3650on 2:: u 332::
25:58:89 332: u 803:8 b58325 n :35.
mqimcozﬂﬂ 3586985 Co 00509:: n 3:85 .6820 329502 actuate-*5 n.- .5820:
533.8% 6.69:. we bio—9:8 n onEou 8:338 :omEmSAP. 8 8: x33»: 2.55020 .3 35:23: N N830:
.9850on omen—Bozo. u nonmaocx 823:8 :888: 2 om: x556: 2:280? mo cognac: n 3856:
3238a :o_Em§P. I
8:333 .238: .-
mo. V Q a
a. V a .3
tam.- S. 8. m5.- :.- ask-N. 2. no: 2.- 2. Not 2.9.. onmmwco
S. 2. 8. 8. 8.- 8.- :. 8. 8.- 8.- 8. 8%.:
ion. 0—. 8. oo. 5. .2. 8. 3. not 2 .- 8.5
.1 m. m _ .- m _. S. 2.5m. 3.- we. *2. Not 808:8
.2. at. E .- No. 2. 1:3. 3mm. 3 .- BEBE
8.- 8.- 8.- :8. :8. .2. 8. 5388
8. 8. 8. .8. .2. 2. cameos—
8- .2; m _. 2. 8. 88828
:8. 8.- 8.- 2 .- 0888:
2. 3.- No. :35
.5820:

 

95.35 8.5 808::— 3:25 5358 00%265 03305 common—5 too—5 women: .332. .8820:

macaw—oboo Eon—02.832; 588m I 2 038-

90
(“1” to “3”) while a series of 5-point Likert-type indices were used to measure other
predictor variables.

Due to the fact that the research hypotheses in this study are directional, two
pieces of information about the beta weights are needed for conﬁrrning/rej ecting the
hypothesized relationships. The ﬁrst point to note is the mathematical signs of the beta
weights, i.e., whether they are positive or negative. Because all relationships have been
hypothesized to be positive, a negative beta will result in the rejection of the associated
hypothesis. In addition to the direction of association, the beta weights must be
statistically signiﬁcant, at least at the probability level of .05. In other words, there must
be less than a 5% chance that the observed relationship is due to random error.

Another important statistic in regression analysis is R-square, or the “coefficient
of determination.” One commonly held interpretation of R-square is that it indicates the
proportion of variation of a criterion about its mean that is explained by the regressors
(Hair et. al., 1979). Therefore, the higher the value of R-square, the more variance in the
data is said to be explained, and, in turn, the better the estimation or prediction is
considered. Another reading of R-square is that the statistic conveys the extent to which
the model of predicted associations among the predictors and criterion ﬁt the data
empirically (Norusis/SPSS, Inc., 1993).

When there is a large number of predictors, as in the case of the current study, it is
useful to observe the adjusted R-square value. As the name implies, this statistic adjusts
for the number of predictors used in the estimation of the mathematical best-ﬁt plane. By

taking into consideration the degrees of freedom in the regression equation (given the

91

number of predictors and sample size), this adjusted coefﬁcient of determination provides

some indication of the parsimony of the estimation model. As such, it is not uncommon

for the size of adjusted R-square to increase, while that of R-square drops, when poor

predictors are removed from the regression equation.

Regression Results

This section presents the regression results of the current study. To restate the six

hypotheses tested, they are:

H1:

H3:

H4:

H5:

H6:

Uncertainty is positively associated with electronic network adoption and

1186.

Asset (product, time and knowledge) speciﬁcity is positively associated

with electronic network adoption and use.

Complexity of product description is positively associated with electronic

network adoption and use.

(Importance of) interpersonal relationships is positively associated with

electronic network adoption and use.

Mutual commitment is positively associated with electronic network

adoption and use.

Trust is positively associated with electronic network adoption and use.

The following presentation will be organized by the separate regression runs associated

with each of the three dependent measures.

92

1nf0rma_tion Technology Cluster

Table 12 shows the multiple regression results for predicting the ﬁrst dependent
measure of the study, that is, the information technology cluster. Note that the two
control variables are among the regressors that are statistically signiﬁcant at the alpha
level of .05. In particular, organizational size has a strong positive relationship with the
information technology cluster (0 = .43), suggesting that the larger an organization is, the
more likely it is that a larger share of the technology cluster has been adopted. With
respect to the measure of integration with the major supplier, the positive beta weight (13
= .15) indicates that the higher the retailer-supplier integration is, a grocery retailer’s
technological adoption level is likely to be higher as well.

None of the transaction cost variables, i.e., uncertainty, asset (product, time and
knowledge) speciﬁcity and complexity of product description, appears to have notable
impact on the dependent variable. Given the insigniﬁcant betas, the observed patterns of
associations between these predictors and the criterion could have appeared by chance or
due to random error alone. Consequently, the three transaction cost hypotheses (H1 to
H3), as they relate to predicting adoption of the information technology cluster, must be
rejected.

As for the social relational factors, two of them are found to be statistically
signiﬁcant regressors, namely the importance of interpersonal relationships (13 = - .17, p <
.05) and trust (13 = - .18, p < .05). However, contrary to what was hypothesized (H4 and
H6), both of them are shown to have a negative relationship with the dependent variable.

These empirical results suggest that when interpersonal relationships are considered

93

Table 12 - Regression of lnforrnation Technology Cluster

 

 

 

 

Independent Variable Standardized Beta Probability Level
(B)
Origina Modiﬁed Origina Modiﬁed
1 model model 1 model model

Transaction Cost Factors

uncertainty .06 .44

product speciﬁcity -.05 .53

time speciﬁcity -.11 .20

knowledge speciﬁcity .06 .49

complexity of product description .12 .15
Social Relational Factors

importance of interpersonal relationships -.17 -.1 l .04 .17

mutual commitment .12 .23

trust -.18 -.14 .05 .08
Control Factors

organizational size .43 .44 .00 .00

integration with major supplier .15 .16 .05 .04
Original Model Modiﬁed Model
N = 140
Multiple R = .50 Multiple R = .47

R-square = .25 (F = 4.35, p = .00)

Adjusted R-square = .19

Adjusted R-square = .19

R-square = .22 (F = 9.36, p = .00)

94

highly important to dealing with suppliers, the number of information technologies
adopted is smaller. Regarding trust, the more a grocery retailer trusts its major supplier,
the less likely it is to have adopted a large share of the information technology cluster.
As far as the mutual commitment variable is concerned, although the regression
coefﬁcient sign is positive, as hypothesized (H5), the beta weight is not signiﬁcant;
hence, H5 must be rejected.

As far as the overall model ﬁt is concerned, R-square is .25, while adjusted R-
square is .19. Dropping the regressors that are not statistically signiﬁcant (at p < .05)
reduces R-square to .22 and adjusted R-square remains unchanged. These results suggest
that the original regression model was not very parsimonious (given that adjusted R-
square does not change). However, in the “modiﬁed” model, the interpersonal
relationships variable no longer appears as a signiﬁcant regressor (B = -.11, p = .17), and
the beta for trust becomes marginally signiﬁcant (13 = -.14, p = .08). This can be
attributed to the association these two variables have with eliminated regressors which
appear to share a small portion of common variance explained in the data. In particular,
interpersonal relationships is signiﬁcantly correlated with knowledge speciﬁcity (r = .17,
p < .05), while trust is highly associated with mutual commitment and time speciﬁcity (r

= .18 and .56, respectively) (see Table 10 for zero-order correlations).

Importance at Electronic Network Use
When the dependent variable is one of the two network usage measures, the

pattern of relationships observed are quite different from the results associated with the

95
adoption variable. First, the two control variables, namely organizational size and
integration with major suppliers, do not seem to have much impact on the perceived
importance of electronic network use to various transactional activities associated with
the acquisition of a key product. This can be seen in the regression results shown in
Tables 13 and 14. Beta weights for these two variables in both sets of runs are not
statistically signiﬁcant.

Importance of Electronic Network Use to “Static” Activities. When the
regression criterion is the importance of electronic network use to the three “static”
activities of specifying an order, seeking price information and ordering a key product,
the transaction cost hypotheses (H1 to H3) must again be rejected (see Table 13). Despite
this, it is noteworthy that several of the transaction cost variables demonstrate the
hypothesized direction of association with the importance of electronic network use.
These variables are product speciﬁcity, knowledge speciﬁcity and complexity of product
description, which all have positive regression weights, although they were only
approaching statistical signiﬁcance at the probability level of .05. Attention should be
called to the relatively small sample size of 100, which does not lend much statistical
power to the regression analysis. Given a larger N value, these beta weights would have
been statistically signiﬁcant.

As for the effects of social relational factors, the perceived importance of
interpersonal relationships to the “static” activities has a positive, albeit weak, association
with the extent to which electronic network use has been viewed as important to the same

set of activities (B = .17, p > .05). Therefore, H4 must be rejected. As for the factors of

96

Table 13 - Regression of Importance of Electronic Network Use to “Static” Activities

 

 

 

Independent Variable Standardized Beta Probability Level
(B)
Origina Modiﬁed Origina Modiﬁed
1 model model 1 model model
Transaction Cost Factors
uncertainty -.10 .32
product speciﬁcity .18 .19 .07 .05
time speciﬁcity -.04 .72
knowledge speciﬁcity .18 .18 .07 .08
complexity of product description .17 .14 .11 .16
Social Relational Factors
importance of interpersonal relationships
to “static” activities .17 .17 .12 .12
mutual commitment .28 .28 .03 .03
trust -.25 -.25 .03 .03
Control Factors
organizational size .09 -.09 .38 .41
integration with major supplier .08 -.08 .43 .42
Original Model Modiﬁed Model
N = 100
Multiple R = .44 Multiple R = .42

R-square = .19 (F = 2.10, p = .03)

Adjusted R-square = .10

Adjusted R-square = .11

R-square = .18 (F = 2.46, p = .02)

97

Table 14 - Regression of Importance of Electronic Network Use to “Dynamic” Activities

 

 

Independent Variable Standardized Beta Probability Level
(13)
Original Modiﬁed Origina Modiﬁed
model model 1 model model

Transaction Cost Factors

uncertainty .08 .45
product speciﬁcity .11 .12 .23 .19
time speciﬁcity -.08 .43
knowledge speciﬁcity .09 .10 .38 .31
complexity of product description .21 .24 .05 .01

Social Relational Factors

importance of interpersonal relationships

 

to “dynamic” activities .33 .31 .00 .00
mutual commitment -.09 .47
trust .08 .49
Control Factors
organizational size .09 .08 .56 .44
integration with major supplier .01 -.03 .93 .75
Original Model Modiﬁed Model
N = 100
Multiple R = .50 Multiple R = .49
R-square = .25 (F = 3.03, p = .00) R-square = .24 (F = 4.88, p = .00)

Adjusted R-square = .17 Adjusted R-square = .19

98
mutual commitment and trust, they are found to be statistically signiﬁcant regressors. As
hypothesized (HS), regression analysis ﬁnds mutual commitment to have a positive
relationship with the perception of how important electronic network use is to the “static”
activities associated with acquiring a key product (13 = .28, p < .05). With respect to trust,
although this variable shows a strong relationship with network use (13 = -.25, p < .05),
the association is not in the hypothesized direction. This result leads to the rejection of
H6. What is suggested here is that the more a grocery retailer trusts a major supplier, the
less likely electronic networks will be viewed as important for these more “static”
activities of doing business with this supplier.

As far as the overall ﬁt of the regression model is concerned, R-square is .19
(adjusted R-square = .10). Eliminating insigniﬁcant regressors (uncertainty and time
speciﬁcity) leads to a .02 drop in R-square and a .01 increase in adjusted R-square.
Again, what can be observed here is that the original conceptual model is not
parsimonious, as a number of predictors do not contribute to the overall estimation of the
best-ﬁt regression equation. With respect to changes in the effects of individual
regressors, the beta weight of product speciﬁcity (13 = .19) is found to be signiﬁcant at the
.05 probability level. Dropping uncertainty from the modiﬁed regression equation is
likely a major cause, as this variable has a signiﬁcant negative relationship with product
speciﬁcity (r = -.l7, p < .05). At the same time, the impact of complexity of product
description is attenuated, which can also be attributed to the absence of uncertainty in the

regression model (r = .29, p < .01).

99

Importance of Electronic Network Use to “Dynamic” Activities. Regression
results for predicting the perceived importance of electronic network use to the more
“interactive” activities associated with product buying are shown in Table 14. To
recapitulate these “interactive” activities, they are those associated with: (i) negotiating
the terms of an acquisition agreement; (ii) monitoring the quality of an order; (iii)
practicing just-in-time inventory management; and (iv) ﬁxing after-sales errors.
Interestingly, a different set of predictors are found to be signiﬁcantly associated with the
perceived importance of electronic network use to these activities than the more “static”
ones discussed above.

First, the complexity of product description variable is found to be a statistically
signiﬁcant predictor of the perceived importance of network use ([3 = .21, p < .05). Also,
as hypothesized (H3), the more complex the product description is perceived to be, the
more important respondents view network use to be in supporting transactions with a
major supplier of the product. As for the other transaction cost variables, they do not
yield signiﬁcant results, leading to the rejection of H1 and H2 (see Table 14).

With respect to the social relational factors, unlike earlier results, neither trust nor
mutual commitment is found to have any meaningful association with the perceived
importance of network use outside of chance alone. Consequently, H5 and H6 must be
rejected. Nevertheless, the importance of interpersonal relationships to the same set of
“dynamic” activities yields a beta weight that is both positive and strong (13 = .33, p <

.01). In other words, the more important interpersonal relationships are thought to be in

100
interacting with a major supplier, the more likely the retailers are to consider electronic
network use to be important to help in carrying out the same activities as well.

As shown in Table 14, the regression model shows an overall ﬁt at the magnitude
of R-square = .25 (adjusted R-square of .17). When only the variables that have
signiﬁcant correlations with the criterion variable are included in the regression equation,
R-square drops to .24, and adjusted R-square increases by .02 to .19. Complexity of
product description and the importance of interpersonal relationships remain strong
regressors. These negligible changes suggest that the other transaction cost and social
relational factors do not contribute to the explanation of why electronic networks have
been perceived to be important to supporting interactions with major suppliers of key

products.

Chapter Summary

Detailed in this chapter were the techniques used to prepare and analyze the data
collected to test the proposed relationships between the adoption and use of electronic
information networks and the various transaction cost and social relational variables.
Results of hypotheses testing are summarized in Table 15. Together with the description
of research design procedures and issues (Chapter 3—Methods and Procedures), this
chapter completes the report on the “technical” aspects of the empirical portion of this
dissertation. More discussion on linking the research ﬁndings with existing theoretical

and empirical literature will be presented in the next chapter (Chapter 5—Discussion).

101

3328:» Eugene: 05 5:5 338%: 39680: on 8 38852:: 26>» 33:53 3020209: =<

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a guano: 3 032%»: 2.35:»:
388.3— .:au=_:w_m “:5 magi-oy— Env eoaoﬂvm 8pc. 6
Bauer-om cow-8:95 woaooaom EoEHMEEoU 33:2 .m
3:398 <2 Sz 8388 .3825.
$2 coughed $2 8238: :ocﬁm:
a Piano: “585:9”:
$2 $2 2:: 683.3— mQEmgzﬂom 8:082:85 mo oo:S.8:E_ .v
a “585:“?-
c~tama=m 3:532: «=5 concomom nogoohom :oumtomoa 8.60:; we bio—9:00 .m
a 88:8»:
Bacon-cm EBB-:2: 2.5 339.3% vacuumed bmocmooqm ammo—38M .om
coaoovm 088.3% :38qu bUEooqm oEE- EN
8 2.35:»:
voaoomom ESE-:2: 2.5 vogue-no: 880.8% 86$”:on 825.5 .mm
860.8,; :Boohum uoaoomom 35380:: _
$33.3» :oEEAQ: 8:353: :uusm.
8820
3 cm: x5382 25:85 me 85:09:: 32058..- Etna-.85
488.50%: 98
838:3, «5:539 8328> Eugoqncvﬂ

 

 

852E a 5858 - 2 29:.

Chapter 5

DISCUSSION

Reported in the previous chapter are the ﬁndings of this research and whether the
data are consistent with a priori conceptualization. The current study ﬁnds that a sizeable
portion of the surveyed retailers were not using electronic information networks to
support their product acquisitions from major suppliers (43 out of 143 cases). This
indicates that the level of electronic network use to support retailer-supplier transactions
may not be as high as that suggested by what appears to be inﬂated general perception.

In addition, lending mixed support to hypothesized relationships, this study reveals
differential patterns of associations between the adoption and use of electronic
information networks and the various transaction cost and social relational factors.
Predictors of adoption are not the same ones that are found to be accountable for
electronic networks being perceived to be important to aiding the coordination and
control of supplier-retailer interactions. Furthermore, depending on the nature of the
transactional activities, different “mixes” of transaction cost and social relational
considerations are found to be associated with networks being viewed as important to
facilitating the acquisition of key products.

This chapter extends the discussion of the results reported in Chapter 4, linking
the ﬁndings back to existing literature. Insights and guidance are sought to qualify the
empirical observations and explain why they are or are not consistent with the hypotheses

of the study. Presentation in this chapter begins with the general observations of

102

103
analyzing separately the different measures of electronic network adoption and use. This
is followed by the discussion of the two sets of research hypotheses tested in the study,
i.e., those derived from the literature on transaction cost analysis and social relations in
organizations. In discussing whether the data collected are consistent with each
hypothesis, conceptual and methodological considerations are incorporated, factoring in
the limiting conditions associated with such issues as measurement and analytic

speciﬁcations.

Adoption and Use of Electronic Information Networks

In the current study, due to a number of factors related to a priori
conceptualization and post hoc analysis, multiple measures of electronic network
adoption and use have been constructed. Before proceeding to the discussion related to
speciﬁc hypotheses, this section presents some general observations about the different

results associated with these various dependent measures.

Adoption vs. Usage

Recall from earlier discussion that the design of this study followed the strategy of
separating the conceptual and operational deﬁnition of adoption from that of usage, as
proposed by Tomatzky and Klein (1982). Indeed, this research decision appears to have
been sound, as evident in the differential results generated that have interesting
conceptual and practical implications. As the regression analysis shows, the adoption of

the information technology cluster is associated with the independent variables in

104
different ways than the two variables measuring the perceived importance of electronic
network use.

One possible explanation for these different patterns of associations could be due
to the different operational treatment of the adoption and usage variables. As described
before, the adoption variable is more general, measuring at the overall organizational
level the availability of a cluster of information technologies that enable electronic
exchange of information. As for the importance of electronic network use variable, this
notion is tied to the interactions associated with a speciﬁc trading pair, that is, the grocery
buyer interviewed and a major supplier. Therefore, predictor variables that are strongly
associated with the information technology cluster may not be related to the perceived
importance of electronic network use in the same way or magnitude. It is possible that
while electronic linkages adopted may not be rendered important to supporting one
speciﬁc set of buyer-supplier transactions, other buyers managing different product
categories in the same company may ﬁnd electronic exchange of information with their
suppliers to be important. In addition, technological adoption may be motivated by other
priorities than facilitating the coordination and control of product acquisition activities
(such as various reasons associated with attaining competitive advantage or meeting
strategic necessities referenced in Chapter l—Introduction).

Aside ﬁom these operationalization considerations, there are likely some other
conceptual reasons that can help shed light on the results. Some of the most readily

observable differences between adoption and usage are seen in their different

105
relationships with the two control variables in the study, i.e., organizational size and the

level of integration with a major supplier. These are discussed in turn below.

Organizational Size

Organizational size is found to have a signiﬁcant association with the adoption of
the cluster of six information technologies that can enable electronic exchange of
information (B = .43, p < .01; see Table 11). In fact, more than any other contingencies
faced by the organization that are associated with its needs and environment (reﬂected by
the transaction cost and social relational factors), size accounts for the strongest empirical
reason for the number of technologies adopted to enable electronic exchange of
information. (Organizational size, measured by the composite of a ﬁrm’s total annual
sales, total number of stores and employees, has the largest beta weight). Recall that the
information technology cluster is made up of the WWW, EDI, POS, on-line services,
internal and external email systems. Given this operationalization of adoption, the
regression result indicates that the number of information technology functions and
systems adopted increases with the size of the grocery retailer. This ﬁnding is consistent
with predictions based on the resource contingency perspective, which posits that the size
of an organization has direct implications on how much slack resources are available for
investment in technologies.

Along these cost considerations, the results of the current study also show that
once technologies have been adopted, organizational size no longer appears to be an

important determinant of usage. (Beta weights for the regression of the two importance

106
of electronic network use criterion variables on organizational size were not statistically
signiﬁcant; see Tables 13 and 14). This empirical observation is consistent with the
ﬁndings of Kraut et. a1. (1997): Organizational size in four U.S. industries (advertising,
magazine publishing, pharmaceutical and women’s apparel) studied was also found to be
not signiﬁcantly related to how important network use is perceived to be in transacting
with a major supplier. It would appear, then, that this relationship between ﬁrm size and
technological use is fairly stable, even in the presence of varying contextual conditions
associated with distinct norms and characteristics of different industrial settings.

At an intuitive level, the differential associations of organizational size with
adoption and perceived importance of usage appear quite logical. Consider that the
highest technological costs are generally incurred during initial implementation.
Relatively speaking, the operating costs thereafter are usually signiﬁcantly smaller,
especially with regard to the six types of technologies used to gauge network adoption in
the current project. As such, it is not surprising to ﬁnd organizational size to be a critical
determinant of technological adoption, but that it becomes less of a factor inﬂuencing
later usage, that is, the perceived importance of network use to dealing with a major

supplier to acquire a key product.

Integration with Maiar Sypplier
As for the other control variable in the current study, empirical observations show
that a larger portion of the information technology cluster is more likely to be adopted by

grocery retailers who are integrated with their major suppliers ([3 = 15, p < .05; see Table

107
11). This ﬁnding is also consistent with existing research concluding that electronic
linkages are more likely to be considered important in situations where the major supplier
is an internal unit or a trading partner with whom a ﬁrm has a pre—established, long-term
business relationship (Kraut, et. al., 1997; Steinﬁeld et. al., 1995).

Moreover, this observed positive relationship between technological adoption and
the level of retailer-supplier integration is also suggestive of the possible perception that
electronic interconnections are beneﬁcial to supporting smoother interfacing with major
suppliers. This appears to be a manifestation of what Malone et. al. (1987) proposed to
be the electronic integration effect (described in Chapter 2—Review of Theoretical and
Empirical Literature).

Adding further insights into the advantages of electronic integration is the
conceptual analysis of inter-organizational connectivity in grocery distribution by Clark

and Schiano (1996). In particular, they made the following observation (pg. 285):

“EDI ordering is only faster or more efﬁcient when the data transmission process
is more tightly integrated with the order-processing information systems within
the manufacturer, which enables EDI orders to bypass some steps of the less

automated phone or fax ordering process.”

This position is offered to explain why the factor of speed alone is not sufficient for
explaining why EDI is not as attractive as it was originally thought to be; this is because

ordering via telephone and fax has been rendered to be just as fast and efﬁcient. In fact,

108
this point is consistent with anecdotal comments offered by some survey respondents in
the current study (from the non-usage portion of the sample) that the reason why they are
not using networks is that they still rely on telephone and fax communications. It would
appear that for electronic information networks to be adopted, such advantages beyond
faster transactions as those associated with the tighter coupling of activities need to be
available and recognized. For instance, potential beneﬁts to be considered may include a
reduction in the amount of paperwork processed and data entry errors.

Similar to the ﬁndings associated with organizational size, the level of retailer-
supplier integration has also been found to have no effect on the importance attributed by
adopters to network use for supporting activities associated with buying key products
from these major suppliers. (Regression analysis shows that the beta weights for the
integration variable are not statistically signiﬁcant; see Tables 12 and 13.) These results
are contrary to those reported by Kraut et. a1. (1997), which found network use to be
perceived as more important to the acquisition of a key production input from an
integrated supplier than from an external contractor. A possible explanation for these
contradictory ﬁndings may be the extent to which value is added after the object of a
transaction is obtained. In the current study, the object traded between supplier and
retailer is a ﬁnished product ready to be purchased by an end-customer. This is in
contrast to intermediate, industrial products or services studied by Kraut et. al. (1997).
After acquisition, value is added to the input before ﬁnally being sold to consumers, e. g.,
fabric is sponged, cut and sewn before a dress is made. What appears to be suggested is

that electronic networks are perceived to be important to enhancing tighter coupling and

109
interfacing of production activities internal to an organization where the value-adding
intensity is higher. When little value is to be added, electronic linkages are considered to
be equally effective (or not effective) in supporting interactions with internal units as well

as external trading partners.

Importance of Electronic Network Use to “Static” vs. “Dynamic” Activities

As reported in the results chapter, the importance of electronic network use to the
seven activities associated with the acquisition of a key product were found to load on
two separate factors. To recap, the “static” activities include specifying an order, seeking
price information and ordering; the “dynamic” activities include negotiating the terms of
an acquisition agreement, monitoring the quality of orders received, practicing just-in-
time inventory management and ﬁxing after-sales errors or problems. What is suggested
in these ﬁndings is that there is (are) some abstract dirnension(s) concerning electronic
communication and information exchange in which these activities vary qualitatively.
One possible source of difference is the likely amount, frequency and complexity of
interactions between the retailer and supplier necessitated in carrying out these
transactional tasks. It is quite logical to expect more frequent and complex interactions
with suppliers to be associated with contract negotiations, quality monitoring and
problem ﬁxing (the “dynamic” activities) than ordering or seeking price information (the
“static” activities).

Conceptually, there are a number of reasons why communication and interactions

conducted via electronic networks would be deemed more appropriate for the “static”

110

tasks and activities than the “dynamic” ones. (Recall that the mean scores on the “static”
items are higher than those on the “dynamic” ones; see Figure 2, Chapter 4.) This is
especially true when electronic network use is considered in comparison and contrast
with face-to-face interactions. Negotiations and problem solving likely rely on real-time
back-and-forth exchanges between two parties. Face-to-face meetings provide by far the
highest level of interactivity among all communication modalities, including many
electronic networks. It follows logically that the execution of “dynamic” activities would
still rely heavily on face-to-face interactions and less so on electronic networks.
Moreover, relative to the “dynamic” transactional tasks, the “static” ones incur relatively
less demand on interactive communication. Therefore, it is not surprising to ﬁnd
computer-based networks to be considered more important to the “static” activities than
the “dynamic” ones.

Nohria and Eccles (1992) offered some additional insights on when electronic

networks may substitute for face-to-face interactions (pg. 299):

“...electronically mediated exchange can substitute for face-to-face interaction
only when the identities of the interactants are not very important, when the
circumstances at hand are certain and unambiguous, when the actions necessary

are standard and routine, and when on-going interaction does not depend on a

robust structure of relationships.”

It is likely that the identity of the person in the supplier ﬁrm is less important to the tasks

of seeking price information or specifying an order than to situations involving the

111
negotiations of a contract or trying to correct an order problem. Likewise, compared to
the “dynamic” activities, the “static” ones also appear to be relatively less ambiguous,
more standard and routine, as well as less reliant on robust relationships.

Furthermore, having coordination routines, patterns and rules can help to
eliminate the need to treat each situation as new and provide stability to an organization’s
operations (Galbraith, 1973; March & Simon, 1958, 1993). Borrowing these insights
from organizational theories, it can be inferred why networks may be instrumental to
automating some standard and routinized tasks (i.e., the “static” ordering activities) but
not other less routine ones (i.e., the “dynamic” interactions). For instance, once a retailer-
supplier EDI connection is implemented, a system and routine for performing such
standard activities as price check and ordering can be developed and followed. On the
other hand, activities involving negotiations or problem solving are likely less routine,
but may instead require different treatment across situations. As such, it would be more
difficult to routinize or automate the execution of the “dynamic” activities electronically
than the “static” ones.

In addition, as has been alluded to in the earlier discussion, there may be
pre-established patterns of communication that have proven to be effective for dealing
with some uncertain, ambiguous or complex tasks (Chapter 2). A case in point, again,
relates to the anecdotal reports on the interactions with suppliers being done on the phone
or by fax. These established communication strategies and routines might preclude the

tendency and motivation to perform the same tasks electronically via computer networks.

112

Related to this point of established patterns of interactions is the notion of power
structure. For instance, Weisband, Schneider and Connolly (1995) argued that
information technologies have the potential of equalizing status or power in bargaining
and negotiations. Other writings have offered complementary propositions (e. g., Bakos
& Treacy, 1986). Powerful buyers or sellers may not want to relinquish their more
advantageous position. As such, although network use might be allowed to automate
some of the ordering activities, electronic exchange of information that has implications
on negotiations, bargaining and other more sensitive types of interactions would be
discouraged. This would further explain the differential perceptions of importance
attributed to the coordination and control of different transactional activities.

At the methodological level, it is very likely that some of the insigniﬁcant results
found in this study are partly due to the less than perfectly reliable indicators of network
use. Having these seven items split up into two separate variables is less desirable than if
the items constituted a single composite index. The two criterion measures gauging the
importance of network use yield moderate internal consistency reliability (or = .71 and

.75). If summed to form one index, Cronbach’s Alpha would be .92.

Transaction Cost Hypotheses (H1 to H3)

As described in Chapters 2 and 3, there are a number of transaction cost variables
that are proposed to inﬂuence the adoption and use of electronic information networks.
Results reported in Chapter 4 show that the hypotheses associated with these variables

have received mixed support from the empirical observations undertaken in this study.

1 13
This section provides a discussion of these transaction cost variables and their

relationships with electronic network adoption and use.
Uncertainty (Hypothesis 1)

H1: Uncertainty is positively associated with electronic network adoption and

use.

In the current study, uncertainty has not been found to have signiﬁcant impact on
electronic network adoption or use, leading to the rejection of Hypothesis 1. This ﬁnding
could be attributable to a number of conceptual reasons, measurement problems, or a
combination of both of these effects.

At the conceptual level, it is not unreasonable to expect the adoption and use of
electronic networks to be encouraged when there is a high level of certainty (low
uncertainty) associated with product availability and price. Recalling the earlier quote
from Nohria and Eccles (1992), electronically mediated exchanges are likely to be more
effective when situations involve higher certainty. Also discussed previously, some
organizational theorists (e.g., Galbraith, 1973; March & Simon, 1958, 1993) would argue
that lower uncertainty can facilitate the development and implementation of routines and
programs to manage standard tasks. Again, the reason is that the need to treat each
electronic order or price seeking activity as a new, unique situation is eliminated.

In addition, as discussed in Chapter 3, it has been widely recognized that

uncertainty is highly complex and multi-dimensional. As such, while it is an important

114
construct to consider in research dealing with information technologies, it is a concept
that is difficult to measure directly, and is often manifested or embedded in other
hypothetical constructs. For instance, it has been argued that the complexity and
compatibility variables in diffusion of innovations research “are merely indices of the
degree and type of ‘uncertainty-arousing’ potential” (Tomatzky & Klein, 1982, pg. 41).

Moreover, in the present study, it can be observed that uncertainty is correlated
with other transaction cost variables, speciﬁcally complexity of product description,
product speciﬁcity and time speciﬁcity (see Table 11, Chapter 4). The presence of these
associations appears to be consistent with what Tushman and Nadler (1978)
conceptualized, that conditions related to task complexity and the environment faced by
an organization are sources of uncertainty and, in turn, information processing needs.
However, the empirical ﬁndings in the current study appear to suggest that uncertainty
does not increase the need for coordination and control through the use of electronic
information networks.

Unfortunately, due to methodological caveats, it cannot be ascertained as to
whether the insigniﬁcant results associated with uncertainty indeed refute theoretical
predictions. After all, the uncertainty index suffers from a lack of reliability; the three-
item measure yields a Cronbach’s Alpha score of only .51 (Table 10, Chapter 4). In
addition, it is possible that the dimension of uncertainty operationally measured—
(perceived) predictability of changes in the availability and prices of a product when it is
needed—might not have been the “correct” one, as it relates to motivation for network

adoption and use.

115

Asset (Product, Time and Knowledge) Specificity (Hypothesis 2)

H2: Asset @roduct, time and knowledge) specificity is positively associated

with electronic network adoption and use.

As shown in the regression results (Table 12, Chapter 4), in the presence of other
antecedent and control variables, none of the asset speciﬁcity measures are found to have
any noteworthy impact on the adoption of the information technology cluster. It would
appear that factors related to the social relational conditions (interpersonal relationships
and trust) and organizational characteristics (size and integration with major supplier)
could better explain the conditions that favor technological adoption. These results
suggest that considerations of reduction in coordination costs are relatively less important
than viewing technology as a functional substitute for interpersonal relationship and trust
in managing retailer-supplier transactions. This ﬁnding has practical implications for
managers trying to decide whether to ad0pt an electronic link with a trading partner.
Rather than focusing solely on potential increase in cost efﬁciencies, how electronically
mediated interactions ﬁt in with the established reliance on social relational ties need to
be considered.

One possible reason for the insigniﬁcant results for the asset speciﬁcity measures
could be traced to the problem of measurement. Operationally deﬁning product and time
speciﬁcity by the brand name and product types, respectively, takes advantage of

objective attributes rather than relying on subjective rating. While this strategy avoids

116
the methodological pitfalls associated with perceptive measurement (e. g., uncontrolled
sources of variance stemming from individual differences in respondents), and appears
logical for the grocery industry, these are nonetheless single-item measures.
Consequently, reliability cannot be assessed. As for knowledge speciﬁcity, the two-item
measure yields an internal consistency reliability score of only .52. This lack of
reliability has likely contributed to attenuating the strength of associations observed.

As reported in the previous chapter, some measures of asset speciﬁcity have been
found to have some effects on the perceived importance of network usage for some
transactional activities. Speciﬁcally, both product and knowledge speciﬁcity have been
shown to be partly, albeit moderately, accountable for grocery buyers viewing electronic
network use to be important in supporting the “static” activities associated with acquiring
a key product. (The two beta weights approach statistical signiﬁcant at the .05 level; see
Table 13). These observations are similar to ﬁndings by Kraut et. al. (1997), that the
more speciﬁc an object is to the organization and the more knowledge about the supplier
and industry needed in general, the higher is the perception concerning the importance of
electronic network use.

In addition, these results concerning product and knowledge speciﬁcity are
consistent with the predictions of transaction cost theory. They lend moderate support to
the proposition that network use is partly motivated by the desire to reduce the costs of
coordinating and controlling transactions involving highly asset-speciﬁc products
(discussed in Chapter 2). A product bearing a retailer’s private label (high product

speciﬁcity) is unlikely to be produced by more than one or a few suppliers. Given the

117

need to engage in recurring transactions with this supplier (or small number of suppliers),
automating the ordering process has the appeal of reducing coordination costs.
Moreover, as long as the retailer needs to replenish a store-label product supplied by a
particular trading partner, there is a guarantee for long-term business over which to
spread technological investment and operational costs. After all, the supplier cannot sell
the highly asset-speciﬁc product to another store without repackaging. Likewise, the
retailer cannot easily switch to a different supplier without requiring the latter to make
relationship-speciﬁc investments in producing and/or distributing private-label products.

Similar logic applies to the ﬁnding concerning knowledge speciﬁcity
characterizing retailer-supplier transactions. Having highly speciﬁc knowledge about
each other can facilitate the joint effort to automate pricing and order processing
electronically. Furthermore, the average respondent in this study attributed fairly high
importance to the need for both retailers and suppliers to know each other beyond basic
industrial practice in order to do business effectively (mean score = 3.74). This
observation implies the existence of idiosyncrasies in the highly complex and intricate
grocery retail industry. It also suggests a considerable amount of strategizing being
practiced by the industry participants. These two points have received wide support from
trade press reports. In addition, many survey participants in the current study were
adamant about not letting the information they provided “leak” to competitors. Such
volatility characterizing the industry, in turn, renders it difﬁcult to develop and
implement standard programs and routines to manage many complex transactions. This

difﬁculty likely applies to automation as well. Therefore, it is not surprising to ﬁnd that

118
negotiations, monitoring and problem ﬁxing (the “dynamic” transactions) activities

cannot be readily automated and carried out electronically.

Complexity of Product Description (Hypothesis 3)

H3: Complexity of product description is positively associated with electronic

network adoption and use.

In the current study, complexity of product description is a signiﬁcant predictor
for computer-based networks being viewed as important to the transactional activities that
are more “dynamic” in nature (see Table 13). In other words, the more difficult it is to
describe the critical attributes of a key product needed, the more a grocery buyer is likely
to rely on electronic networks for pre- and post-sales negotiations and interactions with
their suppliers. This ﬁnding also receives support from the fact that the average rating for
the complexity of product description was signiﬁcantly higher among the network users
than the non-users (see the comparison of the two groups reported in Table 7, Chapter 4).

This observation can be traced partly to the fact that variations of products are
often associated with various forms of standardized codes used in the grocery industry
(e.g., UPCs, ULNs, SKUs, etc.). It would also appear that the more complex a product is,
the more efﬁcient these codes tend to be in describing the grade or quality that is desired
in the product. For instance, one respondent reported that there are different numbers
associated with different grades of bananas. While it would be difﬁcult to describe in

words precisely the shade of yellow that an order of bananas should have, the

119
standardized codes serve as an efﬁcient short-hand description for this otherwise highly
complex communication task. Since these product codes can be easily transmitted over
networks, whenever there is a problem with an order, or when an urgent replenishment is
needed (just-in-time), interactions with the supplier can be expedited.

Consistent with the ﬁnding by Kraut et. al. (1997), this empirical observation
rejects the hypothesis advanced by Malone et. al. (1987) that electronic linkages are
associated with lower complexity of product description. A possible explanation is that,
if a set of product attributes could be easily described, existing communication
mechanisms might sufﬁce (e. g., faxing and telephoning). As such, it would be less
worthwhile or justiﬁable to invest in a state-of-the-art electronic network. Moreover,
recall from the conceptual discussion in Chapter 2 that, compared to a decade ago, the
technologies available today are more sophisticated; they are more capable of supporting
multi-media representation of product features that may be otherwise challenging to
describe verbally or in writing alone. Empirical observations in this study concerning the
construct of complexity of product description appear to conﬁrm this proposition. In
turn, this “revelation” calls for a change in the seemingly out-dated conceptualization
originally proposed by Malone et. al. (1987).

Conceptual reasons aside, certain characteristics about grocery retailing may offer
some pragmatic explanations for the observed patterns of associations between
complexity of product description and the two network use measures. As a case in point,
many grocery retailers rent shelf space in their stores to product vendors. The suppliers

themselves are responsible for monitoring the inventory levels and stocking the shelves.

120
A famous example is F rito-Lay chips. Sales people scan shelves with hand-held
electronic devices, and the codes associated with different types of chips requiring
replenishments are transmitted electronically back to distribution centers. In these types
of arrangements, regardless of the level of product description complexity, there would be
minimal interaction between the retailers and suppliers after a contract has been
negotiated and signed. Communication needs arise when there are “exceptions” to the
agreement, such as when there are order problems to be resolved. Under these
arrangements, it is likely that the grocery stores would not interact with the suppliers on a
regular basis for the purposes of ordering (the “static” activities). This fact was
conﬁrmed by a number of respondents in the survey. For instance, one buyer said that
the only time the supplier of a candy product would be contacted is if a store customer
complained that a certain type of candy was not on the shelf. As a result, electronic
network use to support such sparse communication would not be important. Again, in
these contexts, the complexity of product description appears to be irrelevant (hence,
explaining the insigniﬁcant relationship with the importance of electronic network use to
“static” activities).

An electronic linkage with a supplier may become more useful when a supplier
needs to be informed of a problem or a special product need that involves highly complex
product description. Rather than having to relay verbally problems with an order that is
difficult to describe, the product code and other order information stored electronically in
an EDI system can be recalled to facilitate problem ﬁxing. Similar logic also applies to

negotiations, when the exchange of complex product speciﬁcations can be enhanced by

121
the use electronic data networks. Networks in this context would likely play a
supplemental or complementary role to other modalities of coordination and control, e.g.,
interpersonal contact. Indeed, as will be discussed below, networks have been found to

supplement/complement interpersonal relationships in executing the “dynamic” activities.

Social Relational Hypotheses (H4 to H6)
Reported in Chapter 4 are the different associations between the social relational
variables proposed (interpersonal relationships, mutual commitment and trust) and the

various network use and adoption measures. This section provides a discussion on these

ﬁndings.

Importance of Interpersonal Relationships (Hypothesis 4)

H4: (Importance of) interpersonal relationships is positively associated with

electronic network adoption and use.

Perceptions toward interpersonal relationships present another interesting case in
which the associations are different with technological adoption and the importance
attributed to electronic network use. As far as adoption of the information technology
cluster is concerned, it is found to have a negative relationship with the perceived
importance of interpersonal relationships (B = -.17, p < .05; Table 12). In other words,
the more a grocery buyer views “personal connections, interactions and knowledge” with

members of the supplier ﬁrm to be important to their transactions, the less likely his/her

122
company would have adopted a large portion of the information technology cluster.
What can be inferred are signs of substitutability between the two modalities of dealing
with suppliers in the acquisition of a key product. When interpersonal contact is not
considered crucial to the coordination and control of interactions, electronically mediated
communication can sufﬁce.

However, the results are altogether different in the assessment of the importance
of electronic network use. Perceived importance of interpersonal relationships does not
appear to have a signiﬁcant association with that attributed to electronic network use for
supporting transactional activities that are more “static” in nature (13 = .17, p > .05; Table
13). It would appear that this ﬁnding is quite logical, especially when considering that a
lack of trust in the supplier motivates network use to perform these routine activities
(trust has a signiﬁcant negative beta weight of -.25; Table 13). Intuitively, interpersonal
relationships as a coordination and control mechanism would not be perceived important
when there is a lack of trust.

With respect to the more non-routine, “dynamic” activities that are likely to
require more recursive interactions between the grocery retailer and supplier, a signiﬁcant
positive relationship has been observed (B = .33, p < .01; Table 14). In other words, it
would appear that electronic networks have been regarded as complements to
interpersonal contact in these transactional contexts. This is consistent with a previous
ﬁnding that there is a complementary relationship between the importance of electronic
network use and interpersonal relationships (Kraut et. al., 1997). When interactions are

complex and ambiguous, redundancy in communication channels is often desired. For

123
instance, Sitkin, Sutcliff and Barrios-Choplin (1992) reported that the use of multiple
media and multiple communicators is especially associated with ambiguous tasks. This
argument is also consistent with the information processing perspective that task
ambiguity would contribute to the need to increase an organization’s information
processing capacity. In this regard, electronic networks appear to provide additional
capacity as an additional modality for supporting “dynamic” interactions. Furthermore,
as described by a respondent, when there is a problem with an order, he would call the
supplier as well as reference the electronic order information. In this regard, the use of an
electronic network with the supplier supplements the interpersonal communication
necessitated to resolve the problem.

Other existing literature can also help to explain the differential patterns of
associations with the two measure of network use. From a media richness perspective,
the “dynamic” tasks that are more complex and ambiguous than the “static” tasks
necessarily require the richest communication media, i.e., face-to-face interactions (Daft
& Lengel, 1984; Trevino, Daft & Lengel, 1990). Therefore, in supporting these more
media-rich tasks, the more likely function of electronic networks is to complement or
supplement face-to-face interactions, rather than substituting the latter in serving as the
sole medium of business exchanges. F urtherrnore, according to what has been reported
as of the present, electronic interconnections, such as EDI, are useful for automating
invoicing and exchanging straightforward pricing and shipping information; many
ordering and payment processing activities beyond simple information exchange still

require human intervention (Clark and Schiano, 1996). In this regard, technological

124
capabilities are merely complements to and supplements for improved communications
and efﬁciency, and are not regarded or used as substitutes for interpersonal connections

and interactions.

Mutual Commitment (Hypothesis 5)

H5: Mutual commitment is positively associated with electronic network

adoption and use.

Powell (1996) contended that organizations in a trading relationship are
sometimes willing to forego fears of vulnerability and collaborate when they have
complementary resources; these resources could be anything, from materials to labor, and
even information technology. Coupled with the discussion so far about varying electronic
network use in different types of transactional activities, this observation helps to explain
the patterns of relationships between mutual commitment and the various technology and
network use measures. In particular, mutual commitment has been found to have a
signiﬁcant, positive relationship with the perceived importance of electronic network use
to “static” activities ([3 = .28, p = .03; Table 13). However, this is not the case with the
“dynamic activities” ([3 = -.O9, p = .47; Table 14).

The above results appear logical and credible, as one would expect that
committing to a trading relationship and forgoing the feeling of vulnerability would
proceed progressively, starting with interactions that are the least “threatening.” As such,

automating the coordination and control of more routine, mundane activities, such as

125
those associated with getting and receiving standard pricing and ordering information,
would be indicative of early signs of mutual commitment to the relationship. With
respect to the more intricate types of the interactions involving negotiations and problem-
solving, they may be deemed too important to be handled electronically but rather better
handled by other means, especially face-to-face interactions. In this regard, it would
appear that whether there is a high level of mutual commitment should have little direct
effect on how important an electronic linkage is to supporting the “dynamic” activities.

Consistent with these observations about vulnerability and mutual commitment,
Hart and Estrin (1991) observed that electronic networks have the effect of shifting the
nature of interdependence among organizations. They concluded that while information
processing is enhanced, new vulnerability is created as the ﬁrms come to depend on the
electronic interconnections in their transactions with one another. This argument appears
to help explain why electronic networks are perceived to be important to supporting
“static” activities between trading partners highly committed to one another, but not
interactions that are more ambiguous and complex, i.e., the “dynamic” activities.

F urtherrnore, it has been argued that electronic networks can facilitate the
establishment of mutuality by providing the infrastructure for “reliable, prompt, and
relatively low-cost information” (Parkhe, 1993, pg. 821). In turn, this translates into
reduction in coordination efforts and safeguards against opportunism. It is likely that at
the early stages of transacting with one another, mutual commitment is manifested in the
form of “credible, signiﬁcant non-recoverable investments on both sides” (pg. 821). It

would appear that investing in an electronic interconnection is a sign of mutual

126
commitment of the more tangible nature. As a history of cooperation accumulates,
mutuality that develops will likely become less reliant on the tangible investments made;
rather the more affective aspects of commitment to each other (such as trust) may become
more dominant.

On this note, anecdotal comments provided by a number of surveyed buyers in the
current study may help to explain why mutual commitment is not a signiﬁcant predictor
of electronic network use to the “dynamic activities.” These respondents commented that
their willingness to cooperate with their suppliers was strictly because that was the nature
of conducting business. To the extent that a trading partner’s success is tied to their own,
they are committed to the business relationship. As such, they would not want to
relinquish the desire to be assured that they are not being cheated by their suppliers.
Therefore, it is not surprising to ﬁnd that negotiations, monitoring and problem solving
did not rely heavily on electronically mediated interactions. While the (perhaps
superﬁcial) display of mutual commitment allows for routine, non-threatening activities
to be coordinated and controlled electronically, the “dynamic” activities are too important
to the protection of self-interests to be performed electronically. The most robust
modality of interaction would still be required, even if networks are used to exchange
information to supplement the face-to-face negotiations and bargaining. This ﬁnding
adds practical insights into the extent to which retailer-supplier transactions can be

automated, based on how vulnerable the parties would likely feel as a result.

127

Trust (Hypothesis 6)

H6: Trust is positively associated with electronic network adoption and use.

Perhaps the most “surprising” ﬁnding of the current study is the negative
association between trust and technological adoption and use. Particularly, a larger
portion of the information technology cluster has been adopted in ﬁrms that do not
necessarily consider their trading partners to be trustworthy (B = -.18, p = .05; Table 12).
Similarly, electronic network use has been perceived to be more important for supporting
the more “static” activities when the level of trust in the supplier is low ([3 = -.25, p = .03;
Table 12). Also, compared to the network users, non-users reported signiﬁcantly higher
trust in their suppliers (see Table 7, Chapter 4, for mean score difference). As argued in
Chapter 2, one would expect intuitively that before two ﬁrms trading with one another
would invest in an electronic interconnection and open up their information base to one
another, trust should be present as a necessary condition. But as can be seen in the results
of this research, this need not be the case.

One possible explanation for the negative relationships observed is that electronic
networks help to increase the span of control (Beniger, 1986; Kipnis, 1996). Recall from
the methods chapter that information is the basis of control and makes processes easier to
monitor (Nass & Mason, 1990). Also, as discussed earlier, electronic networks increase
the bandwidth of human interaction and help to reduce uncertainty involved in a
situation. It would appear that electronic supplier-retailer interconnections in this context

might serve as a functional substitute for trust as a mechanism for control, coordination

128
and coping with the conditions (e.g., uncertainty) surrounding the transactions that occur
with the acquisition of a key product.

Another possible explanation for the negative association is in the
conceptualization of trust, given the different dimensions associated with the concept.
(To recap, trust in the current study is conceptualized as the belief that one’s trading
partner will refrain from malfeasance.) More speciﬁcally, the manifestations of two
parties appearing to trust each other may be the result of deliberate actions rather than the
reality of truly feeling the presence of affective bonds with one another. This is highly
plausible, and has empirical support from the current study (trust and mutual commitment
are highly correlated, r = .56, p < .01). Clark and Schiano (1996) articulated the logic

behind this argument (pg. 284):

“[trust]...must not be constructed through nebulous psychological trust-building
exercises...but through the deliberate undertaking of irreversible actions that

commit both parties equally and irrevocably to the success of the relationship.”

Given this characterization, while a respondent might not perceive a high level of trust in
a supplier in the psychological sense, contractual and other mechanisms are able to
substitute for this lack of affective sense of conﬁdence and security. Expressed
differently, even if a grocery retailer is reportedly not trusting of its major supplier, the
presence of a contract would be a sufﬁcient indication of commitment for routine
activities to be automated with the use of networks. Another safeguard against

opportunism disguised as trust is that of reputation. Several respondents in the current

129
study contended that their suppliers would unlikely take advantage of them because of
negative reputational ramiﬁcations. These observations are consistent with Granovetter’s
(1985) argument concerning the need for institutional arrangements to enforce trust,
which was reviewed in Chapter 2. The gap between conceptual reasoning and empirical
observations stems from viewing trust and electronic information networks as

complementary, rather than substitutable, control and coordination mechanisms.

Chapter Summary

In this chapter, conceptual reasons and methodological caveats were traced to help
explain the results obtained. Arguments presented in existing literature were cited to
qualify the “story” on why the adoption and use of electronic networks are associated in
different ways with the proposed transaction cost and social relational variables. In
addition, the lack of reliability in some of the measures and how that attenuates the
strength of associations has also been discussed. Summary of the ﬁndings discussed
above, implications of these observations for future research, along with other concluding

remarks, will be presented in the next—and last—chapter of this volume.

Chapter 6

SUMMARY AND CONCLUSIONS

As the concluding chapter, this section of the dissertation focuses on tying
together the different pieces of the study, i.e., conceptualization, research design and
ﬁndings. The chief objective is to close the discussion by addressing the following
issues: What are the answers to the research question? What are the limiting conditions
due to conceptual caveats and drawbacks related to various aspects of the research
design? What are the implications of the current study for future research and
“real-world” organizational practice?

In response to the above questions, this chapter is divided into the following
sections: First, a summary of the main ﬁndings of the current study is provided. Then,
conceptual and methodological limitations are discussed. Finally, the chapter concludes

with theoretical and practical implications of this dissertation.

Summary of Findings
Reiterating the chief purpose of the research undertaken, it is to investigate the

following research question (stated in Chapter 1):

What are the factors inﬂuencing organizational adoption and use of electronic
information networks in retailer-supplier transactions associated with product

acquisitions?
1 30

131

Derived from transaction cost and social relational analyses, several factors antecedent to
organizational adoption and use of electronic information networks are proposed and
tested.

The current study ﬁnds that the level of electronic network use to support
retailer-supplier transactions may not be as high as generally perceived. After all, a
sizeable portion of the surveyed retailers were not using electronic information networks
to support their product acquisitions from major suppliers (43 out of 143 cases). With
respect to hypotheses testing, empirical observations show that signiﬁcant factors
inﬂuencing the adoption and use of electronic information networks are different.
Furthermore, associations between predictor factors and the perceived importance of
electronic network use vary across two sets of retailer-supplier transactions. To recap,
they are the “static” and “dynamic” transactional activities/tasks associated with the
acquisition of a key product from a major supplier. In particular, controlling for the
impact of organizational size and retailer-supplier integration, the following results were

obtained:

1. Adoption of the information technology cluster is motivated by the role played by
technology as a ﬁmctional substitute for trust and interpersonal relationships as

mechanisms of coordination and control.

132
2. In conducting the “static” transactions, i.e., those that involve routine, straightforward
communication of price and order information, electronic network use is perceived as

important when:

a. the supplier is not considered trustworthy; again, this result suggests that
electronic information networks and trust are viewed as alternative coordination

and control mechanisms.

b. the retailer-supplier electronic interconnection is regarded as a tangible indication

of mutual commitment between the trading parties.

3. In conducting the “dynamic” transactions, i.e., those that involve higher complexity,
ambiguity or need for recursive interactions with the supplier, electronic network use

is perceived as important when:

a. networks are considered to be complementary/supplemental to interpersonal

relationships as a coordination and control mechanism.

b. networks are perceived to facilitate the representation and communication of a
product’s critical features, which are otherwise difficult to describe (high

complexity of product description).

133
Limitations
Like most empirical research, the current study is tempered with a number of
conditions constraining the interpretations and generalizations of the ﬁndings. This
section discusses the limitations associated with the research design and model

speciﬁcations in this dissertation research.

Research Design

Measurement

Empirical tests of hypothesized relationships and how they can be interpreted are
constrained by the way constructs have been operationally deﬁned and measured. For
instance, it is possible that the negative relationships observed between trust and
electronic network adoption and use may be attributed to the affective dimension of trust
measured (discussed in Chapter 5). A different pattern of associations might be found if
another dimension of trust were to be tapped, such as that related to conﬁdence derived
from and enforced by the existence of a legal contract. Another case in point is the
construct of uncertainty. Preceding discussions (Chapter 5) pointed out the fact that the
“wrong” dimension of uncertainty might have been explicated in the current study.
Therefore, the absence of statistically signiﬁcant relationships with the various
information technology and electronic network variables may be due to the use of
inappropriate operational procedures.

Aside from limitations related to operationalization, reliability of measures

developed in this study is likely another source of measurement caveats. Unreliable

134
measures attenuate the strength of associations observed. Therefore, the absence of a
hypothesized relationship may be due to measurement problems. Speciﬁc cases include
the composite indices for uncertainty and knowledge speciﬁcity, which suffer from low
internal consistency reliability (reported in Chapter 4). Because of this methodological
problem, it cannot be concluded conﬁdently that, in reality, uncertainty and knowledge
speciﬁcity indeed have no inﬂuence on organizational adoption and use of electronic
information networks (discussed in Chapter 5).

Furthermore, because the current study used a cross-sectional design to test the
presence and strength of hypothesized relationships, causality cannot be overstated.
When considering the issue of temporal sequence, interpersonal relationships have
existed long before electronic communication networks have been introduced to support
trade. However, it would not be surprising to ﬁnd that electronically mediated
interactions may alter the way interpersonal contact is viewed in day-to-day business
trading. Therefore, to tease apart explained variance attributable to the two different
directions of association between interpersonal relationships and electronic networks,
overtime data would be desired. Similar logic applies to other variables proposed to be

the antecedent factors in this study.

Population Selection
When designing the current study, caution was taken to ensure that the population
selected maximizes the generalizability of ﬁndings. Bearing this objective in mind, the

grocery retail industry was chosen, in part, because it shares many common

135
characteristics with other industries, particularly in the retail sector (e.g., apparel). For
instance, standardized product coding (e. g., SKUs, ULNs, etc.), which is associated with
the complexity of product description construct, is practiced in other retail businesses as
well. Furthermore, it seems reasonable to expect interpersonal relationships to be viewed
as important in varying extent to negotiations and problem-solving situations across
different industrial settings.

Notwithstanding these commonalities, unique characteristics associated with
grocers limit the extent to which ﬁndings of this research can be generalized to
organizations in other industries. As a case in point, the notions of product and time
speciﬁcity are likely to be qualitatively different across businesses. In turn, whether an
inter-organizational electronic link is viewed as a complementary or substituting
communication modality to different networks of social and interpersonal relations may

vary across industries.

Sample Sizes

Of the 143 completed surveys, 100 and 140 were included in analyzing the
importance of electronic network use measures and the information technology cluster,
respectively. These sample sizes do not yield much statistical power for the analyses
conducted, and constrain the extent to which the results obtained can be generalized.
Therefore, a limiting condition of the current study stems from the relatively small

sample sizes available for the empirical tests of the hypotheses developed in this research.

136
Model Speciﬁcations

As reported in preceding chapters, there is inadequate ﬁt between empirical
observations and hypothesized relationships among variables. R-square values range
from .19 to .25, indicating that the regression models fail to account for the entire pool of
variance found in the data. This is despite the fact that these R-square values are within
the general range of those found in similar studies. For instance, Kraut et. al. (1997)
reported an R-square score of .29 (adjusted R-square = .24) with 17 predictors of the
importance of electronic network use. Sabherwal and Vijayasarathy (1994) studied three
predictors of the use of telecommunication links with suppliers, yielding an R-square of
.14 (with no control variables). Aside from measurement and other research design
problems discussed above, the lack of ﬁt between propositions and empirical ﬁndings
may be attributable to missing variables and/or relationships among them.

Among possible missing explanatory factors, the power dynamics between
retailers and suppliers in the grocery industry may not have been adequately accounted
for in the current study.1 One missing link in the current study could be the failure to
measure directly the possible presence of powerful retailers or suppliers dictating the
implementation and use of electronic interconnections. As qualitative interviews
referenced earlier suggest (see Footnote 10, Chapter 3), a ﬁrm may be forced to interact

with a more powerful trading partner electronically as a prerequisite to continuing the

business relationship.

 

‘ A case in point is the construct of relative bargaining position. Had the measurement

of this concept been successful, the regression results might have been improved.

137

Furthermore, as alluded to brieﬂy in the last chapter, there may be other forces
inﬂuencing the use of electronic networks with major suppliers that have no direct
connection with the retailer-supplier relationship or transaction cost considerations. It is
possible that technology was adopted for other reasons (e. g., to capture customer
purchase patterns); using it for electronic transactions might have come later as a spin-off
or “after-thought” to facilitate the coordination and control of product acquisition
activities. Therefore, considering other organizational attributes may contribute to
investigating the factors inﬂuencing electronic information network use. In this regard,
explanatory/predictor factors derived from the theory of diffusion of innovations (Rogers,
1962, 1986, 1995) may be highly useful. For instance, by looking at a variable such as
compatibility, it is possible to explore how the existing orientation towards technology in
an organization may impact on using networks to transact with trading partners.

In addition, more rigorous application of the information processing approach
(Galbraith, 1973; Thompson, 1967; Tushman & Nadler, 1978) may be useful to
identifying motivating factors for network adoption and use. This theoretical perspective
was used in the current research to guide the general conceptualization of electronic
networks as a mechanism for coordination and control. As a “ﬁt” theory arguing for a
match between information processing capacity with needs, it can offer helpful guidance
in explicating the contingencies and task ambiguities surrounding various
retailer-supplier transactions that create information processing needs. In turn, how
different roles and functions served by electronic networks can satisfy these various needs

faced by organizations may be identiﬁed and tested.

138

Implications

Previous conceptual work grounded in transaction cost analysis offers valuable
insights into how electronic networks may help to reduce costs associated with
buyer-seller transactions. The current study shows that a number of social relational
factors inﬂuence organizational adoption and use of electronic networks in ways that are
not explained by transaction cost variables. Empirical support for this observation is
provided by the ﬁndings summarized above. In particular, the lack of trust in a trading
partner has been found to be a signiﬁcant factor associated with technological adoption
and electronic network use. In addition, technology is seen as a functional alternative to
interpersonal relationships in coordinating and controlling “static” transactions, while
playing a complementary role in “dynamic” ones. Mutual commitment has also been
found to be a crucial factor inﬂuencing organizational network use. The only transaction
cost factor that is signiﬁcantly correlated with network use is the complexity of product
description. Future research eﬁorts devoted to the reﬁnement of the conceptualizations
and measurement proposed in this study can further test the relative utility of transaction
cost and social relational analyses.

It would appear that, in constructing a theoretical model of factors antecedent to
electronic network adoption and use, social relational conditions should be considered in
conjunction with transaction cost variables. In addition, the consideration of other
constructs, such as those related to the retailer-supplier power structure and other

organizational attributes, may be useful to increasing the explanatory and predictive

139
power of a model of organizational network adoption and use. As discussed above, it
would be valuable to consider and incorporate insights derived from other theories, such
as the diffusion of innovations and the information processing approach. Last, but not
least, the incorporation of overtime analysis of a set of key factors would also be
meaningful in distinguishing between the antecedents to and the effects of technology use
on inter-organizational coordination and control.

At a practical level, this research shows that electronic network use in commercial
transactions is not as prevalent as existing hype would suggest. After all, one-third of the
grocery retail companies surveyed did not consider electronic networks to be important to
their interactions with major product suppliers. Therefore, the level at which technology
can actually facilitate trade should not be overstated.

Moreover, ﬁndings in this research also lead to the recommendation that retailers
should not be concerned solely with technological implementation and operational costs.
Decisions to adopt and use electronic networks need to factor in the varying
complementary and substituting roles technologies play in relation to the pre-existing
social relational conditions characterizing trade with suppliers. For instance, as a
mechanism for coordinating and controlling transactions, how does an electronic link ﬁt
in with established interpersonal relationships with the trading partner? To what extent is
the supplier trustworthy? Do both the retailer and supplier feel sufﬁciently committed to
the business relationship to open themselves up to new interdependencies and

vulnerabilities associated with electronic trading?

140
Conclusions

In conclusion, this dissertation represents a beginning attempt to integrate
complementary bodies of literature to investigate organizational adoption and use of
electronic information networks. While a number of social relational and transaction cost
factors have been identiﬁed as inﬂuential to network use in supporting commercial
transactions, the current study could not exhaust all critical factors to be considered.
Possible theories to explore have been suggested for future research efforts in continuing
the investigation of the factors antecedent to electronic network adoption and use to
support commercial transactions.

It would appear that much excitement and momentum has been gathered over the
emerging phenomenon of electronic commerce. This is, in part, fuelled by the
exponential growth of the lntemet and its popularity. Empirical observations in this
research show that the actual use of electronic buyer-seller information networks falls
short of the level inferred by the prevailing hype. Continued research in the actual roles
and functions served by information technology in commercial transactions can help to

separate the “true” value of using inter-ﬁrm electronic networks from sheer hype.

APPENDIX A

APPENDIX A

SURVEY INSTRUMENT

 

 

Information Technology Cluster — Questions 36-43
Importance of Electronic Network Use - Questions 44-51

Uncertainty — Questions 62-66

Product Specificity - Question 13

Time specificity — Question 1

Knowledge Speciﬁcity — Questions 17-22

Complexity of Product Description — Questions 67-70

Importance of Interpersonal Relationsh ips — Questions 52—60
Mutual Commitment — Questions 31-35
Trust — Questions 23-26

Organizational Size - Questions 72-73 (and annual sales from Ward’s Business

Directory)

Integration with Major Supplier — Question 14-15

 

Questionnaire #

1. Hello, I'm calling from Michigan State University. We're doing a
survey of grocery buying across the country. Could you please tell me the
name of the person most responsible for buying {PRODUCT CATEGORY}.

[WRITE DOWN NAME IN SURVEY LOG.]

fresh fruit and vegetables..1
dairy products .............. 2
breakfast cereals ........... 3
confectionery goods ......... 4

141

 

142

. May I have his/her telephone number please?
[ENTER PHONE NUMBER INTO SURVEY LOG.]
. Could you please connect me to him/her?

. [IF SPEAKING TO BUYER'S SECRETARY2]

Hello, I'm calling from Michigan State University. May I speak to
?

[IF ASKED WHAT THIS IS REGARDINGI]

We're doing a survey of grocery buyers as part of a doctoral thesis project to
understand how retailers do business with their suppliers, and how computer
networks support trade.

[BE PREPARED TO TAKE DOWN FAX NUMBER FOR ME TO FAX
LETTER ABOUT THE PROJECT]

. [TO BUYERz]

Hello, I'm calling from Michigan State University. We're doing a
survey of grocery buyers as part of a doctoral thesis project to understand how
retailers do business with their suppliers, and how computer networks support
trade. Your organization has been selected at random, and your participation
is voluntary. This survey will take only about 15 minutes, and your
contribution will be very valuable to the project. Of course, any information
you give will be kept conﬁdential, and will not be attributed to you or your
company. Would you be willing to participate?

yes..1
no...2

[IF THE ANSWER IS 1, THEN SKIP TO QUESTION 7]

. [IF ANSWER TO QUESTION 5 IS "NO":]

Your expert input will be very important to this study. Perhaps, I could call
back at a more convenient time for you?

7.

10.

11.

143

Before I proceed with the survey, I would like to conﬁrm that you are
responsible for acquiring {PRODUCT CATEGORY}. Is that correct?

yes..1
no...2

[IF THE ANSWER IS 1, THEN SKIP TO QUESTION 9]
[IF ANSWER TO QUESTION 7 IS "NO":]
May I have the name and phone number of the right contact person?

What is your title please?

 

This interview will focus on the way you acquire a key product from a major
supplier. What would you say is one of the best selling type of {PRODUCT
CATEGORY} for your ﬁrm?

[USE THE SURVEY LOG TO RECORD THIS PRODUCT AND
SUBSTITUTE INTO {TOP PRODUCT} IN SUBSEQUENT QUESTIONS]

[IF THE ANSWER To QUESTION 1 Is 1, THEN SKIP To QUESTION
14.]

Does your ﬁrm carry your own label of {TOP PRODUCT} or a brand only
sold in a local market but not else where in the country?

[IF MORE THAN ONE MARKET, ASK TO THINK OF MOST
IMPORTANT MARKET]

yes ................... 1
no .................... 2

don't know/no answer..9

12. Does your ﬁrm carry a regional label or brand of {TOP PRODUCT}?

yes ................... 1
no .................... 2

don't know/no answer..9

144

 

13. For the product questions in this survey, please think of of
{TOP PRODUCT}.
the local or store label or brand.. 1
a regional label or brand .......... 2
a national label or brand .......... 3
don't know/no answer ............... 9

14. Now, I would like to ask you some questions about how you deal with
suppliers. Please think of the most important supplier you've worked with in
the past 12 months to acquire {TOP PRODUCT}.

15. Is this supplier: an independent manufacturer or producer.. 1
an independent wholesaler or distributor.. 2

a partially-owned wholesaler or distributor.. 3

a joint-venture with another company ..... 4

a wholly-owned wholesaler or distributor..5

don't know/no answer ..................... 9
16. For the following questions, please indicate how much you agree or disagree
with them, on a scale of 1 to 5. "1" means "strongly disagree" and "5" means

"strongly agree." You may use any number in between.

17. This supplier is very knowledgeable of your company's speciﬁc product
needs?

strongly disagree ..... 1

strongly agree ........ 5
don't know/no answer..9

18. This supplier knows more about the way your company acquires {TOP
PRODUCT} than any other suppliers?

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9

145

19. Having only general knowledge of grocery retailing is insufﬁcient for doing
business with your company effectively?

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9
20. Your company is very knowledgeable of the products this supplier provides?

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9
21. Your company knows more about the way this supplier conducts business
than any other retailers?

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9
22. Having only general knowledge of grocery suppliers is insufﬁcient for doing
business with this supplier effectively?

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9

146
23. Overall, you would consider this supplier to be trustworthy.

strongly disagree ..... 1

strongly agree ........ 5
don't know/no answer..9

24. You have complete conﬁdence that this supplier will always keep promises
made to your company.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9
25. You have complete conﬁdence that this supplier will not take advantage of
you even if given the opportunity to do so.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9
26. You will be hesitant to transact with this supplier unless detailed terms of a
contract are clearly speciﬁed.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9

147

27. When negotiating the terms of an acquisition with this supplier, they generally
accept your terms without much rebuttal.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9
28. When negotiating the terms of an acquisition with this supplier, you often feel
that they have more leverage over your company.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9
29. Your business is critical to this supplier.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9
30. This supplier is a critical supplier to your company.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9

148

31. You believe that this supplier is as committed as your company is to the
success of your business relationship.

strongly disagree ..... 1

strongly agree ........

don't know/no answer..9
32.You will not hesitate to switch to a different supplier.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9
33. This supplier is very responsive to your company's needs.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9
34. You believe that this supplier cares about the success of your business.

strongly disagree ..... 1

strongly agree ........

don't know/no answer..9

149

35. The success of your supplier's business is important to you.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9

36. Now, I'd like to ask for your input on how important electronic networks are

37.

38.

39.

in working with this supplier to acquire {TOP PRODUCT}. By an electronic
network, we mean a connection between computers of your ﬁrm and your
supplier, which allows you to exchange information electronically with each
other. This includes modem connections, local area networks, electronic mail,
electronic data interchange, Lotus Notes or the lntemet.

Does your company use:

the Worldwide Web

[IF NEED TO EXPLAIN: the portion of the lntemet that supports pictures
and sound]

yes ................... 1
no .................... 2

don't know/no answer..9

an electronic data interchange system

[IF NEED TO EXPLAIN: a computer-based system of exchanging data, such
as pricing and billing information]

yes ................... 1
no .................... 2

don't know/no answer..9

40.

41.

42.

43.

45.

150
an electronic point-of-sale system

[IF NEED TO EXPLAIN: a system that scans and records credit and debit
card and sales information electronically]

yes ................... 1
no .................... 2

don't know/no answer..9

a computer-based on-line information service, such as CompuServe, America
Online and Prodigy

yes ................... 1
no .................... 2

don't know/no answer..9
electronic mail for communication within your company

yes ................... 1
no .................... 2

don't know/no answer..9

electronic mail for communication with external suppliers

yes ................... 1
no .................... 2

don't know/no answer..9

. On a scale of 1 to 5, "1" being "not important at all," "5" being "very

important," please tell me how important electronic networks are to the
following activities:

developing speciﬁcations of an order

not important at all..1

151

don't use ............. 9
46. seeking price information

not important at all..1

...................... 2

...................... 3

...................... 4

very important ........ 5

don't use ............. 9

47. negotiating the terms of an agreement to purchase {TOP PRODUCT}

not important at all..1
...................... 2
...................... 3
...................... 4
very Important ........ 5
don't use ............. 9
48. ordering {TOP PRODUCT}

not important at all..1
...................... 2
...................... 3
...................... 4
very important ........ 5
don't use ............. 9

49. monitoring the quality of the orders received
not important at all..1
...................... 2
...................... 3
...................... 4
very Important ........ 5

don't use ............. 9

152
50. practicing just-in-time inventory management
[IF NEED TO EXPLAIN: keeping less in stock & reordering more often]

not important at all..1

...................... 2
...................... 3
...................... 4

very important ........ 5
don't use ............. 9

51. ﬁxing after-sales errors or problems

not important at all..1

very important ........ 5

don't use ............. 9
52. Next, I'd like to ask you about the importance of interpersonal relationships in
working with this supplier. By interpersonal relationships, we're including

any type of personal connection, interactions and knowledge between
members of your company and the supplier.

53. Again, on a scale of 1 to 5, "1" being "not important at all," "5" being "very

important," please tell me how important interpersonal relationships are to the
following:

54. developing speciﬁcations of an order

not important at all..1

very important ........ 5

don't use ............. 9

153

55. seeking price information

not important at all..1
...................... 2
...................... 3
...................... 4

very important ........ 5
don't use ............. 9

56. negotiating the terms of an agreement to purchase {TOP PRODUCT}

57. ordering {TOP PRODUCT}

58. monitoring the quality of the orders received

not important at all..1
...................... 2
...................... 3
...................... 4

very important ........ 5
don't use ............. 9
not important at all..1
...................... 2
...................... 3
...................... 4

very important ........ 5
don't use ............. 9
not important at all..1
...................... 2
...................... 3
...................... 4

very important ........ 5

don't use ............. 9

154
59. practicing just-in-time inventory management
[IF NEED TO EXPLAIN: keeping less in stock & reordering more often]

not important at all..1

very important ........ 5
don't use ............. 9

60. ﬁxing after-sales errors or problems

not important at all..1

don't use ............. 9

61. As a ﬁnal set of questions, I'm going to read you a number of statements
{TOP PRODUCT}. Please tell me how much you agree or disagree with
them, again on a scale of 1 to 5, "1" being "strongly disagree" and "5" being
"strongly agree:"

62. The availability of {TOP PRODUCT} changes a great deal over the course of
a year.
63.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9

155

63. You can easily get a replenishment of {TOP PRODUCT} anytime you need
it.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9
64. The amount of {TOP PRODUCT} to be acquired changes a great deal from
one order to the next.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9
65. The price you pay for {TOP PRODUCT} changes a great deal from one order
to the next.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9
66. Your company has a great deal of control over the ﬂuctuations of price you
pay to acquire {TOP PRODUCT}.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9

156

67. Is there a standard coding system, such as UPC, for describing variations of
{TOP PRODUCT}?

yes ................... 1
no .................... 2
don't know/no answer..9

68. When ordering {TOP PRODUCT}, it is difﬁcult to create a verbal description
for it.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9
69. Only a picture or other visual representation of {TOP PRODUCT} can
describe adequately the features of the product.

strongly disagree ..... 1

strongly agree ........ 5
don't know/no answer..9

70. Multimedia representation of {TOP PRODUCT}, that is some combination of

text, picture and sound, is necessary to describe adequately the features of the
product.

strongly disagree ..... 1

strongly agree ........ 5

don't know/no answer..9
71. To wrap up this interview, I'd like to ask you just a few general questions
about your ﬁrm.

72. How many stores do you have around the country?

73. How many employees do you have?

157

74. How many employees are in charge Of buying activities?

75. These are all the questions I needed to ask you. Thank you very much for
your time and cooperation!
[IF RESPONDENT WANTS A COPY OF THE REPORT, ASK FOR
ADDRESS]
send report ............. 1
no need to send report..2

76. [ENTER THE 3-DIGIT COMPANY NUMBER, AS PRINTED IN THE
SURVEY LOG]

APPENDIX B

APPENDIX B

SURVEY LOG

(ALL FIELDS IN BOLD MUST BE COMPLETED.)

 

 

 

 

 

 

Main number:* Company number:*
Completed (date & time): or Refused (date & time):
Interviewer’s name: Questionnaire number:
Product Category:*

 

Company Name:*

 

Buyer’s name:

 

Phone number: Fax number:

 

 

Secretary’s Name:

 

 

 

 

 

 

 

 

 

 

 

Appointment (date & time):

lst attempt (date & time): 5th attempt (date & time):
2nd attempt (date & time): 6th attempt (date & time):
3rd attempt (date & time): 7th attempt (date & time):
4th attempt (date & time): 8th attempt (date & time):

 

 

Address (if they want report):

 

 

Comments/Notes (include date):

* Information in these ﬁelds was available prior to the ﬁrst call.

158

REFERENCES

REFERENCES

Anderson, E. (1985). The salesperson as outside agent or employee: A transaction cost
analysis. Marketing Science, 4(3), 234-253.

Allen, T. J ., & Hauptrnan, O. (1990). The Substitution of Communication Technologies
for Organizational Structure in Research and Development. In J. Fulk & C.
Steinﬁeld (Eds.), Organizations and communication technology (pp. 275-294).
Newbury Park, CA: Sage Publications, Inc.

Allen, T. J ., & Scott Morton, M. S. (Eds) . (1994). Information technology and the
corporation of the 1990s: Research Studies. New York: Oxford University Press.

Arnold, W. E. (1989). Scaling techniques. In P. Emmert & L. L. Barker (Eds.),
Measurement of communication behavior (pp. 117-133). New York: Longman.

Babbie, E. (1990). Survey research methods (2nd. ed.). Belmont, CA: Wadsworth
Publishing Company.

Bakos, J. Y. (1991). A strategic analysis of electronic marketplaces. MIS Quarterly,
September, 295-310.

Bakos, J. Y., & Treacy, M. E. (1986). Information technology and corporate strategy: A
research perspective. MIS Quarterly, June, 107-119.

Barnes, 8., & Greller, L. (1994). Computer-mediated communication in organization.
Communication Education, 43(April), 129-142.

Beniger, J. R. (1990). Conceptualizing information technology as organization, and vice
versa. In J. Fulk & C. Steinﬁeld (Eds.), Organizations and communication
technology (pp. 29-45). Newbury Park, CA: Sage Publications, Inc.

Beniger, J. R. (1986). The control revolution: Technological and economic origins of the
information economy. Cambridge, MA: Harvard Univeristy Press.

159

160

Biggart N. W., & Hamilton, G. G. (1992). On the limits of a ﬁrrn-based theory to explain
business networks: The bias of neoclassical economics. In N. Nohria & R. Eccles
(Eds.), Networks and organizations: Structure, form, and action (pp. 471 -490).
Boston, MA: Harvard Business School Press.

Benjamin, R., & Wigand. R. (1995). Electronic markets and virtual chains on the
information superhighway. Sloan Management Review, Winter, 62-72.

Bollen, K. A. (1989). Structural equations with latent variables. New York: John Wiley
& Sons.

Bradach, J. L., & Eccles. R. G. (1991). Price, authority and trust: From ideal types to
plural forms. In G. Thompson, J. Frances, R. Levacic & J. Mitchell (Eds.),
Markets, hierarchies and networks: The coordination of social life (pp. 277-292).
London, UK: Sage Publications Ltd.

Bradley, S. P. (1993). The role of IT networking in sustaining competitive advantage. In
S. P. Bradley, J. A. Hausman & R. L. Nolan (Eds.), Globalization technology and
competition: The ﬁtsion of computers and telecommunications in the 1990s (pp.
113-142). Boston, MA: Harvard Business School Press.

Bradley, S. P., Hausman, J. A., & Nolan, R. L. (Eds). (1993). Globalization technology
and competition: The ﬁtsion of computers and telecommunications in the I990s.
Boston, MA: Harvard Business School Press.

Brousseau, E. (1995). Contracts as modular mechanisms: Some propositions for the

study of “hybrid forms.” International Journal of the Economics of Business,
2(3), 409-439.

Brynjolfsson, E., Malone, T., Gurbaxani, V. & Kambil, A. (1994). Does information
technology lead to smaller ﬁrms? Management Science, 40(12), 1628-1644.

Burns, T., & Stalker, G. M. (1961). The management of innovation. London, UK:
Tavistock Publications Limited.

Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant validation by the
multitrait-multi-method matrix. Psychological Bulltetin, 56, 81-105.

Child, J. (1987) Information technology, organization, and the responses to strategic
challenges. California Management Review, 30(1), 33-50.

Clark, T. H., & Schiano, W. T. (1996). Seven levels of interorganizational connectivity:
An examination of the US. grocery distribution channel. Proceedings of the 49th
Annual Conference on System Sciences (pp. 281-289). Hawaii.

161

Clemons, E. K. (1993). Information technology and the boundary of the ﬁrm: Who wins,
who loses, who has to change. In S. P. Bradley, J. A. Hausman & R. L. Nolan
(Eds.), Globalization technology and competition: The fusion of computers and
telecommunications in the 1990s (pp. 219-242). Boston, MA: Harvard Business
School Press.

Coase, R. (1937). The nature of the ﬁrm. Economica, 4, 64-73.

Contractor, N. S., & Eisenberg, E. M. (1990). Communication networks and new media
in organizations. In J. Fulk & C. Steinﬁeld (Eds.), Organizations and
Communication Technology (pp. 143-172). Newbury Park, CA: Sage
Publications, Inc.

Creed, W. E. D., & Miles, R. E. (1996). Trust in organizations: A conceptual framework
linking organizational forms, managerial philosophies, and the opprotunity costs
of controls. In R. M. Kramer & T. R. Tyler (Eds.), Trust in organizations:
Frontiers of theory and research (pp. 16-38). Thousand Oaks, CA: Sage
Publications, Inc.

Daft, R. L., & Lengel, R. H. (1984). Information richness: A new approach to managerial
information processing and organization design. In L. L. Cummings & B. M.
Staw (Eds.), Research in organizational behavior, 6 (pp. 191-134). Greenwich,
CT: JAI Press.

Davidow, W. H., & Malone, M. S. (1992). The virtual corporation. New York:
HarperCollins Publishers.

Dillrnan, D. A. (1978). Mail and telephone surveys: the total design method. New York:
Wiley.

Dow, G. K. (1988). Conﬁgurational and coactivational views of organizational structure.
Academy of Management Journal, 13(1), 53-64.

Dwyer, f. R., Schurr, P. H., & Oh, S. (1987). Developing buyer-seller relationships.
Journal of Marketing, 51 (April), 11-27.

The Economist (1995, March 4). A survey of retailing: Change at the check-out.

Eisenberg, E. M., Farace, R. V., Monge, P. R., Bettinghaus, E. P., Kurchner-Hawkins, R.,
Miller, K. 1., & Rothman, L. (1985). Communication linkages in
interorganizational systems: Review and synthesis. In B. Dervin & M. J. Voight
(Eds.), Progress in communication sciences (pp. 231-261). Norwood, NJ: Ablex.

162

Emmert, P. (1989a). Attitude measurement. In P. Emmert & L. L. Barker (Eds.),
Measurement of communication behavior (pp. 134-153). New York: Longman.

Emmert, P. (1989b). Philosophy of measurement. In P. Emmert & L. L. Barker (Eds.),
Measurement of communication behavior (pp. 87-116). New York: Longman.

Flanagin, A. J ., Monge, P., & Fulk, J. (1997, May). The eﬂects of formative
interorganizational relationships on subsequent interaction. Paper presented at
the 47th annual conference of the International Communication Association,
Montreal, Canada.

Frances, J ., Levacic, R., Mitchell, J ., & Thompson, G. (1991). Introduction. In G.
Thompson, J. Frances, R. Levacic & J. Mitchell (Eds.), Markets, hierarchies and
networks: The coordination of social life (pp. 277-292). London, UK: Sage
Publications Ltd.

Frozen Food Age (1996,- August). Chain’s consumer focus draws praise of brokers, 20.

Frozen Food Age (1996, July). PRWs upgrading tech savvy; value-added services soar,
1, 19 & 24.

Galbraith, J. (1973). Designing complex organizations. Reading, MA: Addison-Wesley
Publishing Company.

Gomes, R., & Mentzer, J. T. (1991). The inﬂuence ofjust-in-time systems on
distribution channel performance in the presence of enviromnental uncertainty.
Transportation Journal, Summer, 36-48.

Granovetter, M. (1985). Economic action and social structure: The problem of
embeddedness. American Journal of Sociology, 91(3), 481-510.

Grover, V. (1992). An Empirically Derived Model for the Adoption of Customer-based
Interorganizational Systems. Decision Sciences, 24(3), 603-640.

Gurbaxani, V., & Whang, S. (1991). The impact of information systems on organizations
and markets. Communications of the ACM, 34(1), 59-73.

Hair, Jr., J. R, Anderson, R. E., Tatham, R. L., & Grablowsky, B. J. (1979). Multivariate
data analysis. Tulsa, OK: Petroleum Publishing Company.

Hammond, J. H. (1993). Quick response in retail/manufacturing channels. In S. P.
Bradley, J. A. Hausman & R. L. Nolan (Eds.), Globalization technology and
competition: The ﬁtsion of computers and telecommunications in the 1990s (pp.
185-214). Boston, MA: Harvard Business School Press.

163

Hart, P., & Estrin, D. (1991). Inter-organizational networks, computer integration, and
shifts in interdependence: The case of the semiconductor industry. ACM
Transactions on Information Systems, 9(4), 370-398.

Hennart, J. (1986). What is internalization? Review of World Economics, 122(4), 791-
804.

Hess, C. M., & Kemerer, C. F. (1994). Computerized loan origination systems: An
industry cse study of the electronic markets hypothesis. MIS Quarterly,
September, 251-275.

Hillstrom, K., & Ruby, M. K. (Eds). (1994). Encyclopedia of american industries.
Volume two: Service & non-manufacturing industries. Detroit, MI: Gale
Research.

Johanson, J ., & Mattson, L. G. (1987). Inter-organizational relations in industrial
systems: a network approach compared with transaction-cost approach.
International Studies of Management and Organization, 1 7(1), 34-48.

Johnston, H. R., & Lawrence, P. R. (1988). Beyond vertical integration: The rise of the
value-adding partnership. Harvard Business Review, July-August, 94-101.

Johnston, H. R., & Vitale, M. R. (1988). Creating competitive advantage with inter-
organizational information systems. MIS Quarterly, June, 153-165.

Joskow, P. L. (1985). Vertical integration and long-term contracts: The case of coal-

buming electric generating plants. Journal of Law, Economics, and Organization,
1(1), 33-80.

Keen, P. G. W. (1990). Telecommunications and organizational choice. In J. Fulk & C.
Steinﬁeld (Eds.), Organizations and communication technology (pp. 295-312).
Newbury Park, CA: Sage Publications, Inc.

Kipnis, D. (1996). Trust and Technology. In R. M. Kramer & T. R. Tyler (Eds.), Trust
in organizations: Frontiers of theory and research (pp. 39-50). Thousand Oaks,
CA: Sage Publications, Inc.

Konsynski, B. R., & McFarlan, E. W. (1990). Information partnerships--Shared data,
shared scale. Harvard Business Review, September-October, 114-120.

164

Kraut, R., Steinﬁeld, C., Plummer, A. C., Butler, 3., & Hoag, A. (1997). Coordination
modes and producer-supplier integration: Empirical evidence from four
industries. Manuscript submitted for publication in the Organization Science and
the Journal of Computer-Mediated Communication Special Issue on Virtual
Organizations.

Lawrence, P. R., & Lorsch, J. W. (1967). Organization and environment: Managing
diﬂerentiation and integration. Boston, MA: Harvard Business School.

Lewis, L. (1996). Adding ABC, ECR & EDI to distribution formula. Frozen Food Age,
44(9), 1, 16 & 42.

Malone, T. (1987). Modeling Coordination in Organizations and Markets. Management
Science, 33(10), 1317-1332.

Malone, T. & Crowston, K. (1994). The interdisciplinary study of coordination. ACM
Computing Surveys, 26(1), 87-120.

Malone, T. W., & Rockart, J. F. (1993). How will information technology reshape
organizations? In S. P. Bradley, J. A. Hausman & R. L. Nolan (Eds.),
Globalization technology and competition: The fusion of computers and
telecommunications in the 1990s (pp. 37-56). Boston, MA: Harvard Business
School Press.

Malone, T., Yates, J ., & Benjamin, R. (1989). The logic of electronic markets, Harvard
Business Review, (May-June): 166-171.

Malone, T., Yates, J., & Benjamin, R. (1987). Electronic markets and electronic
hierarchies: effects of information technology on market structure and corporate
strategies. Communications of the ACM, 30 (6), 484-497.

March, J. G., & Simon, H. A. (1993). Organizations (2nd ed.). Cambridge, MA:
Blackwell Publishers.

March, J. G., & Simon, H. A. (1958). Organizations. New York: John Wiley.

Markus, M. L., & Robey, D. (1988). Information technology and organizational change:
Causal structure in theory and research. Management Science, 34(5), 583-598.

McGee, J. V. (1991). Implementing systems across boundaries: Dynamics of
information technology and integration. Unpublished doctoral dissertation,
Harvard University, Boston, Massachusetts.

 

165

Milliken, F. J. (1987). Three types of perceived uncertainty about the environment:
State, effect, and response uncertainty. Academy of Management Review, 12(1),
133-143.

Morgan, G. (1986). Images of organization. Newbury Park, CA: Sage Publications, Inc.

Nass, C., & Mason, L. (1990). On the study of technology and task. In J. Fulk & C.
Steinﬁeld (Eds.), Organizations and communication technology (pp. 46-68).
Newbury Park, CA: Sage Publications, Inc.

Newcomb, T. (1953). An approach to the study of communicative acts. Psychological
Review, 60, 393-404.

Nohria, N., & Eccles, R. (1992). Face-to-face: Making network organizations work. In
N. Nohria & R. Eccles (Eds.), Networks and organizations: Structure, form, and
action (pp. 288-308). Boston, MA: Harvard Business School Press.

Norusis, M. J./SPSS Inc. (1993). SPSS for Windows. Release 6.0. Chicago, IL: SPSS
Inc.

Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). New York:
McGraw-Hill, Inc.

Ouchi, W. G. (1991). Markets, bureaucracies and clans. In G. Thompson, J. Frances, R.
Levacic & J. Mitchell (Eds.), Markets, hierarchies and networks: The
coordination of social life (pp. 246-255). London, UK: Sage Publications Ltd.

Parkhe, A. (1993). Strategic alliance structuring: A game theoretic and transaction cost

examination of interﬁrrn cooperation. Academy of Management Journal, 36(4),
794-829.

Pfeffer, J. & Salancik, G. R. (1978). The external control of organizations: A resource
dependence perspective. New York: Harper and Row.

Porter, M. E., & Millar, V. E. (1985). How information gives you competitive advantage.
Harvard Business Review, July-August: 149-160.

Powell, W. W. (1996). Trust-based forms of goverance. In R. M. Kramer & T. R. Tyler
(Eds.), Trust in organizations: Frontiers of theory and research (pp. 51-67).
Thousand Oaks, CA: Sage Publications, Inc.

Powell, W. W. (1990). Neither market nor hierarchy: Network forms of organization.
Research in Organizational Behavior, 12, 295-336.

166

Progressive Grocer (1996, April). 63rd Annual Report of the Grocery Industry
Supplement.

Raphel, M., & Raphel, N. (1996). What you can do on the intemet. Progressive Grocer,
May, 31-32.

Ratliff, M. (1996). A glossary of food industry terms. Department of Marketing and
Supply Chain Management, Michigan State University, East Lansing, Michigan.

Rogers, E. M. (1995). Diffusion of innovations (4th ed.). New York: Free Press.
Rogers, E. M. (1983). Diffusion of innovations (3rd ed.). New York: Free Press.
Rogers, E. M. (1962). Diﬁiision of innovations. New York: Free Press.

Sabherwal, R., & King, W. R. (1992). Decision processes for developing strategic

applications of information systems: A contingency approach. Decision Sciences,
23(4), 917-938.

Sabherwal, R., & Vijayasarathy, L. (1994). An empirical investigation of the antecedents
of telecommunication-based interorganizational systems. European Journal of
Information Systems, 3(4), 269-284.

Schuman, H., & Presser, S. (1981). Questions and answers in attitude surveys:
Experiments on question form, wording and context. Orlando, Fl: Academic
Press, Inc.

Sealey, P. S. (1994). The strategic impact of information technology on the relationship
between the supplier and the retailer in the grocery industry. Unpublished
doctoral dissertation, The Claremont Graduate School, Claremont, California.

Sheppard, B. H., & Tuchinsky, M. (1996). Interﬁrm relationships: A grammar of pairs.
Research in Organizational Behavior, 18, 331-3 73.

Sitkin, S. B., Sutcliffe, K. M., & Barrios-Choplin J. R. (1992). A dual-capacity model of

communication media choice in organizations. Human Communication Research,
18(4), 563-598.

Steinﬁeld, C., & Caby, L. (1993). Strategic organizational applications of videotex

among varying network conﬁgurations. Telematics and Informatics, 10(2), 119-
129.

167

Steinﬁeld, C., Caby, L., Jaeger, C., & Kraut, R. (1997, June). Electronic data networks
and inter-ﬁrm relations: A French and US. comparative analysis. Paper
presented at Global Networking ’97, Calgary, Canada.

Steinﬁeld, C., Kraut, R., & Plummer, A. C. (1995). The impact of interorganizational
networks on buyer-seller relationships. Journal of computer mediated
communication, 1 (3). Available
http://www.usc.edu/dept/annenberg/vol1/issue3/vol1no3.html.

Thompson, J. D. (1967). Organizations in action. New York: McGraw-Hill.

Tomatzky, L. G. & Klein, K. J. (1982). Innovation characteristics and innovation
adoption-implementation: A meta-analysis of ﬁndings. IEEE Transactions on
Engineering Management, EM-29(l), 28-45.

Trevino, L. K., Daft, R. L., & Lengel, R. H. (1990). Understanding managers’ media
choices: A symbolic interactionist perspective. In J. Fulk & C. Steinﬁeld (Eds.),
Organizations and communication technology (pp. 71-94). Newbury Park, CA:
Sage Publications, Inc.

Tushman, M. L., & Nadler, D. A. (1978). Information processing as an integrating
concept in organizational design. Academy of Management Review, 3, 613-624.
(Reprinted in D. A. Nadler, M. L. Tushman, & N. G. Hatvany (Eds.), Managing
organizations: Readings and cases (pp. 291-301) Boston, MA: Little, Brown and
Company.)

Valente, T. W. (1995). Network models of the diffusion of innovations. Cresskill, NJ:
Hampton Press, Inc.

Ward’s business directory of US. private and public companies. (1997). Detroit, MI:
Gale Research.

Williamson, 0. E. (1981). The economics of organization: The transaction cost
approach. American Journal of Sociology, 87(3), 548-575.

Williamson, 0. (1979). Transaction-cost Economics: The Governance of Contractual
Relations. Journal of Law and Economics, 22(October), 233-262.

Williamson, 0. (1975). Markets and Hierarchies: Analysis and Antitrust Implications.
New York: Free Press.

168

Zaheer, A., McEvily, B., & Perrone, V. (1996). Does trust matter? Exploring the effects
of interorganizational and interpersonal trust on performance. Unpublished
manuscript. University of Minnesota, Carlson School of Management,
Minneapolis.

Zucker, L. G. (1986). Production of trust: Institutional sources of economic structure,
1840-1920. In B. M. Straw & L. L. Cmnmings (Eds.), Research in organizational
behavior (V 01. 8, pp. 53-111). Greenwich, CT: JAI.

"1111111111111111111111“