Description
Electronic Data Interchange (EDI) has come to revolutionize in recent years the way in which businesses conduct
their trading activities. This study examines alternative organizational forms for EDI processing within organizations
and evaluates them on several task performance indices including independence, saturation, and satisfaction, and also
relates them to EDI effectiveness as a function of information technology (IT) intensity and length of EDI use. Results
from an extensive, two-phase survey of EDI users indicate that organizational form for EDI processing influenced task
performance indices such that the more decentralized structures (i.e. star and circle structures) exhibited higher independence
and lower saturation than the more centralized structures (i.e. wheel, kite, and chain structures). More
decentralized forms also exhibited higher satisfaction with EDI than the more centralized forms on nonroutine EDI
tasks. EDI task performance indices were also correlated with EDI effectiveness, such that higher independence, lower
saturation, and higher satisfaction were associated with higher EDI effectiveness.
Electronic data interchange: an evaluation of alternative
organizational forms
Vairam Arunachalam*
School of Accountancy, College of Business, University of Missouri, 333 Cornell Hall, Columbia, MO 65211-2600, USA
Abstract
Electronic Data Interchange (EDI) has come to revolutionize in recent years the way in which businesses conduct
their trading activities. This study examines alternative organizational forms for EDI processing within organizations
and evaluates them on several task performance indices including independence, saturation, and satisfaction, and also
relates them to EDI e?ectiveness as a function of information technology (IT) intensity and length of EDI use. Results
from an extensive, two-phase survey of EDI users indicate that organizational form for EDI processing in?uenced task
performance indices such that the more decentralized structures (i.e. star and circle structures) exhibited higher inde-
pendence and lower saturation than the more centralized structures (i.e. wheel, kite, and chain structures). More
decentralized forms also exhibited higher satisfaction with EDI than the more centralized forms on nonroutine EDI
tasks. EDI task performance indices were also correlated with EDI e?ectiveness, such that higher independence, lower
saturation, and higher satisfaction were associated with higher EDI e?ectiveness. IT intensity and length of EDI use
also positively moderated the e?ects of decentralization on EDI e?ectiveness. These results are discussed in terms of the
organizational implications of this strategic IT resource: EDI.
#2003 Elsevier Ltd. All rights reserved.
Corporate America spends billions of dollars
each year on computing and business communi-
cations, with a goal of integrating hardware, soft-
ware, and people into information collection,
processing, and delivery structures (Scala &
McGrath, 1993). Many organizations now include
as part of this process the proactive use of infor-
mation technology (IT) in their arsenal of strategic
weapons (Van Over & Kavan, 1993). Among these
‘‘weapons’’ are cooperative systems for exchanging
information electronically within and across orga-
nizational boundaries. These intra- and inter-
organizational systems have both processing
capabilities and communications links. They
enable organizations to coordinate and share
information when pursuing a common objective,
representing a cooperative information system
that can enhance competitive advantage (Swat-
man & Swatman, 1992). Electronic data inter-
change (EDI) is a prime example of this type of
system. EDI has strong implications for the design
and operation of accounting systems, such as
Accounts Receivable, Accounts Payable, and
Inventory, among others.
EDI has come to revolutionize in recent years
the way in which businesses conduct their trading
activities. Based on the establishment of trading
partner relationships, EDI systems provide for a
speedy, e?cient, and accurate means of electro-
nically exchanging business transactions. Ranging
0361-3682/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0361-3682(02)00101-0
Accounting, Organizations and Society 29 (2004) 227–241
www.elsevier.com/locate/aos
* Tel.: +1-573-882-1935; fax: +1-573-882-2437.
E-mail address: [email protected]
(V. Arunachalam).
from the manner in which purchases are made to
how payments are remitted, they can contribute to
reducing paperwork, decreasing human error,
increasing accuracy, and improving productivity.
In addition, EDI systems, if appropriately inte-
grated into organizational use of IT, can be of
signi?cant strategic value to the organization
(Swatman & Swatman, 1992). To the extent that
they are e?ectively integrated with existing infor-
mation systems and resource management, EDI
systems can also facilitate the reengineering of
some critical business processes (Borthick & Roth,
1992; Hammer & Champy, 1993), wringing the
slack out of business processes such as ordering
that result in excess inventory. Thus, EDI systems
represent a valuable information resource man-
agement philosophy rather than just a technical
systems issue. Additionally, as Elliott (1992)
points out, they present opportunities for accoun-
tants to adapt accounting systems as these
IT-based innovations are creating a wave of
change that is altering the way business is done
and the problems faced by managers.
However, though the technology for EDI has
been available for some time, its full e?ects are still
being learned (Clinkunbroomer, 1991), including
those from an organizational coordination and
control perspective (cf. Arunachalam, 1997;
Birnberg, 1998; Fisher, 1998). Speci?cally, even as
there is growing popularity and potential under-
lying EDI, there appears to be insu?cient insight
into how organizations adapt their organizational
forms in response to the inevitable business process
reengineering that occurs with EDI use. From a
contingency perspective of management accounting
system (MAS) design, the ?t between organiza-
tional forms and EDI tasks and processes is
important as varying organizational forms can
a?ect the realization of EDI bene?ts. Further study
of these aspects of EDI is therefore warranted.
With this motivation, this study reports on the
results of an extensive, two-phase survey of EDI
users. The survey focuses on the intra-organiza-
tional dimensions of EDI use. Speci?cally, draw-
ing from the communication networks literature
(Shaw, 1978), it examines ?ve alternative organi-
zational forms for EDI processing: (1) wheel
structure, (2) chain structure, (3) kite structure, (4)
circle structure, and (5) star (or ‘‘comcon’’) struc-
ture. It looks at the types of communication
network structures that are evolved within
organizations to support their EDI processing and
evaluates them on several task performance indi-
ces including independence, saturation, and satis-
faction, and also relates them to EDI e?ectiveness
as a function of IT intensity and length of EDI
use.
This paper is organized as follows. The next
section contains the background and hypotheses
of the study, structured around a discussion of the
EDI process, its potential merits, organizational
implications, and the communications network
perspective of organizational forms for EDI pro-
cessing. The section following it contains a
description of the method. This is followed by a
section containing the results of the study. The last
section contains a summary and the conclusions of
the study.
Background and hypotheses
EDI: process and potential merits
EDI is the electronic, computer-to-computer
exchange of business information in a structured
format between business trading partners or
between various units within an organization
(Ferguson, Hill, & Hansen, 1990). It is a high-
speed method of electronic communication that
facilitates the exchange and processing of high
volumes of business data from one computer to
another. EDI is being used by many companies to
order and pay for goods from suppliers, to
arrange transportation with carriers, to receive
orders from customers, to invoice customers, and
to collect payments from customers.
The application of EDI involves the conversion
of written documents into structured, machine-
readable formats so that a computer in one com-
pany or functional unit within a company can
receive and process data from another company’s
or unit’s computer. These documents relate to
business events such as purchasing, sales, inven-
tory management, accounts receivable, and
accounts payable.
228 V. Arunachalam/ Accounting, Organizations and Society 29 (2004) 227–241
EDI can present an adopting organization with
many advantages (Scala & McGrath, 1993). First,
the time needed to exchange information is greatly
reduced (Muller, 1994). Secondly, EDI systems
can contribute to reduced clerical error (Scala &
McGrath, 1993) due to the reduced need to re-key
data. Thirdly, EDI can contribute to lowered
administrative costs (Solis, 1993; Stamper, 1991).
Fourthly, as paper-based systems are slow and
order lead times are higher, companies without
EDI systems may need to maintain higher inven-
tory levels (Sadhwani & Sarhan, 1987).
Organizational implications
The use of EDI has several implications for
organizations. Primarily, it can be a change agent
for traditional business processes and organization
structures—and how they may have been adapted/
reengineered to accommodate EDI. Such adapta-
tion/reengineering is likely to be re?ected not only
in the business processes, but also in the organi-
zational structures underlying such processes.
More generally, as Lucas and Baroudi (1994)
point out, new organizational designs are made
possible through IT such as EDI. These include
structural, work process, and communications
variables, which may be used to characterize four
new organization structures: virtual, negotiated,
traditional, and vertically integrated (Lucas &
Baroudi, 1994). Boynton, Victor, and Pine (1993),
in contrast, note that two new designs are emer-
ging: mass customization (dynamic product
change and stable process change) and continuous
improvement (stable product change and dynamic
process change). They also note that there is
potential for a new form: dynamic stability, a
synergy between mass customization and con-
tinuous improvement. Miles et al. (1997) observe
that each major era in business history has featured
a particular form of organization: early hier-
archical, vertically integrated organizations have
largely given way to network organizations. They
identify cellular organizations as a new way of
organizing based on the principles of entrepreneur-
ship, self-organization, and member ownership.
Combined with the organization’s decision pro-
cesses, examination of business processes can
characterize the range of systems that will best ?t
the organization—IT will then either support or
transform an organization’s business and decision
processes (Fedorowicz & Konsynski, 1992). Addi-
tionally, as Kulkarni and Heriot (1999) note, one
needs to consider costs of acquiring, storing, pro-
cessing, and disseminating information in addition
to transaction costs to be able to predict an
organizational form.
Overall, given the reality that organizations are
continually faced with changing requirements due
to technological and internal as well as external
environmental changes, various organizational
adaptations are necessitated in order to maintain
e?ectiveness (Lederer & Mendelow, 1990; Macy &
Arunachalam, 1995). The focus of this study is on
one particular aspect of this change dimension
related to the interaction between technology and
organizational structure
1
in the context of EDI
technology: the evolution of alternative intra-
organizational forms to support EDI processing.
Speci?cally, drawing from the communication
networks literature from social psychology, this
study examines several types of intra-organiza-
tional forms for EDI processing adopted by EDI
users and relates them to task performance indices
as well as EDI e?ectiveness and some other key
organizational context variables. The next section
describes the communication networks perspective
and outlines the hypotheses of this study.
Organizational forms for EDI: the communication
networks perspective
As Sheombar (1992) notes, in order to use
EDI optimally, organizational forms need to be
redesigned—and EDI can have an impact on all
three determinants of such organizational design:
1
Organizational structure represents the patterns and rela-
tionships that exist among organization or work unit elements.
A variety of structural dimensions has been used in previous
research to characterize such relationships, including central-
ization, decentralization, size, age, etc. (Grinyer & Yasai-
Ardekani, 1980; Macy & Arunachalam, 1995; Reimann, 1973).
This study focuses on one aspect of this concept: organizational
forms for EDI processing, i.e. alternative intra-organizational
con?gurations to support the transmission and processing of
EDI transactions.
V. Arunachalam/ Accounting, Organizations and Society 29 (2004) 227–241 229
product-market combinations, structure, and pro-
cess. The ensuing coordination that is needed
among organizational units is achieved by
communication between them (Sheombar, 1992).
The study of organizational forms to support EDI
processing, therefore, stands to gain insights from
research on communication networks in social
psychology.
This is because EDI has many important intra-
organizational applications and aspects of use
such as the intra-organizational processes and
arrangements to support inter-organizational
transaction processing, and intra-organizational
transaction exchange (e.g. purchase requisitions,
payment authorizations, etc.). Thus, while EDI
may be perceived from the outside as an inter-
organizational system with respect to areas such as
transaction exchange or supplier-customer rela-
tionships (cf. Sriram, Arunachalam, & Ivancevich,
2000), EDI also has many important aspects of
intra-organizational use that are worthy of further
study. This study deals with one such important
intra-organizational aspect of EDI use: alternative
intra-organizational forms to support EDI pro-
cessing. Communication networks research is used
to provide a context to characterize these organi-
zational forms (i.e. communication networks)
adopted for EDI processing within the respondent
organizations.
Bavelas (1948), then Leavitt and colleagues (e.g.,
Leavitt, 1951), and subsequently many other
researchers (for review, see Shaw, 1964,1978)
compared communication networks in which only
certain channels were possible. A diagrammatic
representation of these structures is provided in
Fig. 1. One key comparison network has been the
wheel, a highly centralized pattern in which all
other members can communicate only to one cen-
tral member, hence, all messages must pass
through the person at the ‘‘hub’’ position in the
wheel. Communication links a–c, b–c, d–c, and
e–c are the operative links (leading to the term
wheel because c is a hub and all the connections
are spokes). Other centralized structures include
the kite (with a–b, b–c, c–d and c–e the operative
links) and the chain (with a–b, b–c, c–d and d–e
the operative links). A decentralized comparison
pattern has been the circle, in which each member
is connected to two other members, but no one
member is more central than any other. That is,
each member is connected to two others such that
the communication links a–b, b–c, c–d, d–e, and
e–a are the operative links. Another decentralized
structure is the star or ‘‘comcon’’ (in which each
member/unit is connected to every other member/
unit). The patterns of communication within these
structures represent the basis for classi?cation as
centralized or decentralized. While all these pat-
terns have been used in various studies, the wheel,
kite, and chain structures de?ne one extreme of
the continuum of network variables—centraliza-
tion—while the circle and star structures de?ne the
other (decentralization).
In an EDI transaction processing context, the
organizational units involved may include, for
example, Accounting, Purchasing, Receiving,
Shipping/Distribution, Reporting/Budgeting, Pay-
ments (Financial EDI), and Production Planning.
For purposes of a brief illustration of the alter-
native con?gurations described above, consider
the following units: Materials-requisitioning unit,
Purchasing, Receiving, Accounts Payable, and
Treasury. In a wheel structure, a central EDI unit
would be the hub of all EDI activity and would
coordinate all the EDI document processing
involved in requisitioning, purchasing, receiving,
vendor invoicing, authorizing payment, and issu-
ing payment. In a chain structure, the materials-
requisitioning unit would transmit the requisition
to purchasing, purchasing would transmit the
purchase order (with quantities blanked out) to
receiving, receiving would transmit the receiving
report to accounts payable, and accounts payable
would authorize cash disbursement to treasury,
which would issue payment electronically to the
vendor. In a kite structure, a similar process would
be used, except that accounts payable would not
issue payment authorization directly on vendor
invoice (and receiving report), but also upon
reconciliation with purchase order, the kite units
being purchasing and receiving. In a circle struc-
ture, each unit would be able to communicate with
its immediate predecessor or successor in the chain
of transaction processing outlined above. In a star
structure, all units involved would be able to
communicate/transmit documents to each other,
230 V. Arunachalam/ Accounting, Organizations and Society 29 (2004) 227–241
as would be feasible with a database or enterprise
resource planning (ERP) system such as Oracle,
PeopleSoft, or SAP R/3.
Shaw (1964, 1978) has reviewed many of the
studies in this area, and has clari?ed their ?ndings.
He suggests that the idea of degree of centralization
of nets, or positions in nets, is not as useful as the
ideas of independence and saturation of the positions.
Independence of a position has to do with the
degree to which the member/unit in a given posi-
tion can carry out their part of the task autono-
mously, rather than depend on (one or more)
others in the group. Saturation of a position refers
to the total load-task and role demands as well as
message transmission demands that the member/
unit in that position must handle. In principle,
Fig. 1. Organizational forms for EDI processing.
V. Arunachalam/ Accounting, Organizations and Society 29 (2004) 227–241 231
positions can vary independently between inde-
pendence and saturation, although in reality they
often vary in opposite directions. Both are a?ected
by the nature of the task, and by the pattern of the
network.
Satisfaction derives primarily from high inde-
pendence, but also from moderate saturation.
Task performance may be aided by independence,
and is seriously hindered by saturation. Indepen-
dence is generally higher in decentralized nets,
regardless of tasks. Saturation is lower in cen-
tralized nets for simple tasks (few messages need
be sent); higher in centralized nets for complex
problems (because the task requires more mes-
sages, and they ?ow through a centralized posi-
tion, while peripheral members tend to demand
more information feedback on such problems).
Hence, centralized nets should be more e?cient
on routine tasks and less e?cient on nonroutine
ones, compared to decentralized nets. Networks
should not di?er in satisfaction on routine tasks
(though central and peripheral members may dif-
fer). On nonroutine tasks decentralized networks
(all of whose members are fairly independent, but
not saturated) should have higher satisfaction
than centralized nets (whose peripheral members
have low independence, and all of whose members
are likely to be saturated). The evidence supports
these expectations (Shaw, 1978).
This suggests that on routine EDI processing
tasks—which are characterized by low variety,
high analyzability, standardized procedures, and
few exceptions (Marcella & Chan, 1993; Lee, Han,
& Kym, 1998) e.g. transmitting purchase order or
receiving vendor invoice — more centralized net-
works will not di?er signi?cantly from more
decentralized networks (i.e. progressing across
wheel, kite, chain, circle, and star structures,
respectively) on satisfaction with EDI, even as they
have lower independence and higher saturation. In
contrast, on more nonroutine EDI processing
tasks—which are characterized by high variety, low
analyzability, non-standardized procedures, and
more exceptions, e.g. error reconciliation such as
with reference to vendor invoices, receiving
reports, and purchase orders, or exception report-
ing — more centralized networks will exhibit
lower satisfaction with EDI, and also have lower
independence and higher saturation.
2
These dif-
ferences will translate into concomitant di?erences
in their respective perceptions of EDI e?ective-
ness.
3
This discussion is represented diagram-
matically in Fig. 2 and stated in the form of the
following hypotheses.
H1. Centralization of EDI is related to lower
independence and higher saturation.
H2. Centralization of EDI and EDI task routi-
neness interact to a?ect satisfaction with EDI,
such that centralization results in lower satisfac-
tion than decentralization on nonroutine tasks
and in not signi?cantly di?erent satisfaction than
decentralization on routine tasks.
H3. EDI e?ectiveness will be positively corre-
lated with EDI task performance indices. Speci?-
cally, lower levels of saturation and higher levels
of independence and satisfaction will be asso-
ciated with higher EDI e?ectiveness.
Overall, factors related to the extent of usage
and integration of IT can also play a role in EDI.
Prior research has not empirically examined these
factors, but suggests that length of EDI use
(Rouland, 1991) and IS resources devoted to this
function (cf. Barua, Kriebel, & Mukhopadhyay,
1995; Mukhopadhyay, Kekre, & Kalathur, 1995;
O’Callaghan, Kaufmann, & Konsynski, 1992;
Weill, 1992) such as IT-related investment and
number of employees can a?ect EDI e?ectiveness.
This is because EDI use cannot yield bene?ts over-
night. Rather, a more valid presumption may be
that the greater the length of time the organization
has had to use and integrate EDI technology and to
develop a su?ciently robust EDI customer base
and a population of acclimatized EDI users, the
greater will be the contribution of EDI technol-
ogy. As Hitt and Brynjolfsson (1997) indicate,
increased IT investment could also be associated
2
EDI transaction categories under ANSI/EDIFACT stan-
dards for EDI were used in developing this classi?cation.
3
This study uses the bene?t dimensions of improved custo-
mer service, reduced clerical error, improved control of data,
increased sales, and decreased administrative cost to evaluate
EDI e?ectiveness. Such user-based perceptual criteria have
been demonstrated to be robust indicators of information sys-
tems e?ectiveness (Baroudi & Orlikowski, 1988; Ives & Olson,
1984).
232 V. Arunachalam/ Accounting, Organizations and Society 29 (2004) 227–241
with decentralized authority as ?rms that adopt a
complementary set of organizational practices
including decentralization of decision authority,
emphasis on subjective incentives, and a greater
reliance on skills and human capital tend to
use IT more. Additionally, the greater the IS
resources that have been devoted to EDI (i.e.
IT investment), the greater the likelihood that
bene?ts have been predicted to materialize and
the higher the exposure and support provided to
EDI users.
These factors suggest that length of EDI use and
IT intensity (i.e. ratio of total IT investment to
total number of employees) could positively mod-
erate the e?ects of decentralization of EDI pro-
cessing on EDI e?ectiveness, such that the positive
relationship between greater decentralization and
higher EDI e?ectiveness is enhanced as IT inten-
sity and length of EDI use increase. This suggests
the following hypothesis.
H4. The impact of decentralization on EDI
e?ectiveness will be positively moderated by IT
intensity and length of EDI use.
Method
Design and procedures
A two-phase survey methodology was employed
to test the hypotheses proposed in this study and
to gain deeper insights into EDI. In the ?rst phase,
surveys were mailed to 1400 organizations, all
registered EDI users, in the manufacturing,
wholesale, and retail sectors. These organizations
were selected from a database of EDI users pro-
vided to the author under a special agreement with
Phillips Business Information, publisher of the
EDI Yellow Pages. The surveys were addressed to
the EDI coordinators identi?ed in the database as
the current—and key—contact persons for EDI
and information systems within the respondent
organizations.
The initial (phase I) survey was made up of six
parts. The ?rst part asked for several details
regarding the organization’s background, speci?-
cally annual revenues, principal line of business,
and number of stores/outlets. The second part
asked about the organization’s perceived barriers
to EDI adoption. The third part sought to ascer-
tain the reasons for EDI adoption within the
respondent organization. Questions about length
of EDI use, pre-implementation justi?cation ana-
lysis, type of IT platform for EDI (e.g. database
implementation, enterprise resource planning
(ERP) system, etc.), investment in IT, total num-
ber of employees, number of employees in IT
support, and EDI document volume were also
contained in this section. The fourth part elicited
ratings of bene?ts actually obtained from EDI use,
along with some information regarding customer/
supplier use of EDI. Speci?cally, respondents were
asked to rate several provided items (improved
Fig. 2. Organizational form e?ects on EDI e?ectiveness and EDI task performance indices.
V. Arunachalam/ Accounting, Organizations and Society 29 (2004) 227–241 233
customer service, reduced clerical error, improved
control of data, increased sales, and decreased
administrative cost) on a 10-point scale in terms of
bene?ts that their organization had actually
obtained from EDI use.
The ?fth part asked the respondents to describe
EDI processing. Speci?cally, they were asked to:
(1) specify the names of the organizational units/
departments involved in EDI processing (e.g.
Accounting, Purchasing, Receiving, Shipping/Dis-
tribution, Reporting/Budgeting, Payments/Finan-
cial EDI, Production Planning), (2) outline the
steps in EDI transaction processing, and (3) trace
their ?ow across the various organizational units/
involved in EDI processing. They were then pre-
sented with a list of ?ve organizational forms
(wheel, chain, kite, circle, or star structures) along
with brief descriptions of each form, and were
asked to indicate which one most closely corre-
sponded with the EDI processing in their organi-
zation (if none corresponded, they could choose
none of the above). They were then asked to indi-
cate, for each of two types of EDI processing
(routine and nonroutine tasks, as described ear-
lier), their overall perceptions of the extent of
independence, saturation, and satisfaction experi-
enced among the participants as a whole in EDI
processing. The last part asked respondents about
the extent to which the bene?ts from EDI had ful-
?lled their overall expectations. The last question in
this part requested the respondents to indicate the
names, mailing addresses, and phone numbers of
the managers of each of the organizational units
identi?ed in the previous part. They were also
requested to provide a copy of their business card
in case there were further questions and to enable
them to receive a copy of the research results.
The second (phase II) survey was made up of
only four parts and was addressed to each of the
organizational unit/department managers identi-
?ed in the phase I surveys. The ?rst part requested
information about the nature of their unit’s invol-
vement in EDI processing, EDI document volume,
and sta?ng levels in their units. The second part
asked them to rate, on a 10-point scale, the same
EDI bene?ts contained in part 4 of the phase I
survey. The third part of the survey replicated part
5 of the phase I survey. Speci?cally, they were
asked to: (1) outline the steps in EDI transaction
processing, and (2) trace their ?ow across the var-
ious organizational units/involved in EDI proces-
sing. They were then presented with a list of ?ve
organizational forms (wheel, chain, kite, circle, or
star structures) along with brief descriptions of
each form, and were asked to indicate which one
most closely corresponded with the EDI processing
in their organization (if none corresponded, they
could choose none of the above). They were then
asked to indicate, for each of two types of EDI pro-
cessing (routine and nonroutine tasks, as described
in previous section), their overall perceptions of the
extent of independence, saturation, and satisfaction
experienced among the participants in EDI proces-
sing within their own unit. The last part asked
respondents about the extent to which the bene?ts
from EDI had ful?lled their overall expectations.
Independent variables
Organizational Form. Organizational forms for
EDI processing were classi?ed into one of ?ve
types: (1) wheel structure, (2) chain structure, (3)
kite structure, (4) circle structure, and (5) star (or
‘‘comcon’’) structure. This classi?cation was done
by means of a three-part veri?cation process: (1) if
the organizational forms indicated among an
organization’s respondents (i.e. EDI coordinator,
unit managers) were congruent, the coding was
obviously straightforward (79% of initial respon-
dent organizations ?t this category), (2) if there
was any incongruence within an organization’s
selections, then two independent raters (both doc-
torally quali?ed) independently coded the written
descriptions provided by the respective respon-
dents and came to agreement on the organiza-
tional form (15% of initial respondent
organizations ?t this category),
4
and (3) if there
4
A coe?cient of agreement generally referred to as Cohen’s
K of 0.93 was computed for the coding assignments. K
measures the proportion of agreement between two scorers in
placing items into a set of K unordered categories, and is indica-
tive of the proportion of judgements in which there is agree-
ment after pure random chance is excluded. As no sampling
distribution exists for K when it is transformed into z scores, it
is signi?cant at P
Electronic Data Interchange (EDI) has come to revolutionize in recent years the way in which businesses conduct
their trading activities. This study examines alternative organizational forms for EDI processing within organizations
and evaluates them on several task performance indices including independence, saturation, and satisfaction, and also
relates them to EDI effectiveness as a function of information technology (IT) intensity and length of EDI use. Results
from an extensive, two-phase survey of EDI users indicate that organizational form for EDI processing influenced task
performance indices such that the more decentralized structures (i.e. star and circle structures) exhibited higher independence
and lower saturation than the more centralized structures (i.e. wheel, kite, and chain structures). More
decentralized forms also exhibited higher satisfaction with EDI than the more centralized forms on nonroutine EDI
tasks. EDI task performance indices were also correlated with EDI effectiveness, such that higher independence, lower
saturation, and higher satisfaction were associated with higher EDI effectiveness.
Electronic data interchange: an evaluation of alternative
organizational forms
Vairam Arunachalam*
School of Accountancy, College of Business, University of Missouri, 333 Cornell Hall, Columbia, MO 65211-2600, USA
Abstract
Electronic Data Interchange (EDI) has come to revolutionize in recent years the way in which businesses conduct
their trading activities. This study examines alternative organizational forms for EDI processing within organizations
and evaluates them on several task performance indices including independence, saturation, and satisfaction, and also
relates them to EDI e?ectiveness as a function of information technology (IT) intensity and length of EDI use. Results
from an extensive, two-phase survey of EDI users indicate that organizational form for EDI processing in?uenced task
performance indices such that the more decentralized structures (i.e. star and circle structures) exhibited higher inde-
pendence and lower saturation than the more centralized structures (i.e. wheel, kite, and chain structures). More
decentralized forms also exhibited higher satisfaction with EDI than the more centralized forms on nonroutine EDI
tasks. EDI task performance indices were also correlated with EDI e?ectiveness, such that higher independence, lower
saturation, and higher satisfaction were associated with higher EDI e?ectiveness. IT intensity and length of EDI use
also positively moderated the e?ects of decentralization on EDI e?ectiveness. These results are discussed in terms of the
organizational implications of this strategic IT resource: EDI.
#2003 Elsevier Ltd. All rights reserved.
Corporate America spends billions of dollars
each year on computing and business communi-
cations, with a goal of integrating hardware, soft-
ware, and people into information collection,
processing, and delivery structures (Scala &
McGrath, 1993). Many organizations now include
as part of this process the proactive use of infor-
mation technology (IT) in their arsenal of strategic
weapons (Van Over & Kavan, 1993). Among these
‘‘weapons’’ are cooperative systems for exchanging
information electronically within and across orga-
nizational boundaries. These intra- and inter-
organizational systems have both processing
capabilities and communications links. They
enable organizations to coordinate and share
information when pursuing a common objective,
representing a cooperative information system
that can enhance competitive advantage (Swat-
man & Swatman, 1992). Electronic data inter-
change (EDI) is a prime example of this type of
system. EDI has strong implications for the design
and operation of accounting systems, such as
Accounts Receivable, Accounts Payable, and
Inventory, among others.
EDI has come to revolutionize in recent years
the way in which businesses conduct their trading
activities. Based on the establishment of trading
partner relationships, EDI systems provide for a
speedy, e?cient, and accurate means of electro-
nically exchanging business transactions. Ranging
0361-3682/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0361-3682(02)00101-0
Accounting, Organizations and Society 29 (2004) 227–241
www.elsevier.com/locate/aos
* Tel.: +1-573-882-1935; fax: +1-573-882-2437.
E-mail address: [email protected]
(V. Arunachalam).
from the manner in which purchases are made to
how payments are remitted, they can contribute to
reducing paperwork, decreasing human error,
increasing accuracy, and improving productivity.
In addition, EDI systems, if appropriately inte-
grated into organizational use of IT, can be of
signi?cant strategic value to the organization
(Swatman & Swatman, 1992). To the extent that
they are e?ectively integrated with existing infor-
mation systems and resource management, EDI
systems can also facilitate the reengineering of
some critical business processes (Borthick & Roth,
1992; Hammer & Champy, 1993), wringing the
slack out of business processes such as ordering
that result in excess inventory. Thus, EDI systems
represent a valuable information resource man-
agement philosophy rather than just a technical
systems issue. Additionally, as Elliott (1992)
points out, they present opportunities for accoun-
tants to adapt accounting systems as these
IT-based innovations are creating a wave of
change that is altering the way business is done
and the problems faced by managers.
However, though the technology for EDI has
been available for some time, its full e?ects are still
being learned (Clinkunbroomer, 1991), including
those from an organizational coordination and
control perspective (cf. Arunachalam, 1997;
Birnberg, 1998; Fisher, 1998). Speci?cally, even as
there is growing popularity and potential under-
lying EDI, there appears to be insu?cient insight
into how organizations adapt their organizational
forms in response to the inevitable business process
reengineering that occurs with EDI use. From a
contingency perspective of management accounting
system (MAS) design, the ?t between organiza-
tional forms and EDI tasks and processes is
important as varying organizational forms can
a?ect the realization of EDI bene?ts. Further study
of these aspects of EDI is therefore warranted.
With this motivation, this study reports on the
results of an extensive, two-phase survey of EDI
users. The survey focuses on the intra-organiza-
tional dimensions of EDI use. Speci?cally, draw-
ing from the communication networks literature
(Shaw, 1978), it examines ?ve alternative organi-
zational forms for EDI processing: (1) wheel
structure, (2) chain structure, (3) kite structure, (4)
circle structure, and (5) star (or ‘‘comcon’’) struc-
ture. It looks at the types of communication
network structures that are evolved within
organizations to support their EDI processing and
evaluates them on several task performance indi-
ces including independence, saturation, and satis-
faction, and also relates them to EDI e?ectiveness
as a function of IT intensity and length of EDI
use.
This paper is organized as follows. The next
section contains the background and hypotheses
of the study, structured around a discussion of the
EDI process, its potential merits, organizational
implications, and the communications network
perspective of organizational forms for EDI pro-
cessing. The section following it contains a
description of the method. This is followed by a
section containing the results of the study. The last
section contains a summary and the conclusions of
the study.
Background and hypotheses
EDI: process and potential merits
EDI is the electronic, computer-to-computer
exchange of business information in a structured
format between business trading partners or
between various units within an organization
(Ferguson, Hill, & Hansen, 1990). It is a high-
speed method of electronic communication that
facilitates the exchange and processing of high
volumes of business data from one computer to
another. EDI is being used by many companies to
order and pay for goods from suppliers, to
arrange transportation with carriers, to receive
orders from customers, to invoice customers, and
to collect payments from customers.
The application of EDI involves the conversion
of written documents into structured, machine-
readable formats so that a computer in one com-
pany or functional unit within a company can
receive and process data from another company’s
or unit’s computer. These documents relate to
business events such as purchasing, sales, inven-
tory management, accounts receivable, and
accounts payable.
228 V. Arunachalam/ Accounting, Organizations and Society 29 (2004) 227–241
EDI can present an adopting organization with
many advantages (Scala & McGrath, 1993). First,
the time needed to exchange information is greatly
reduced (Muller, 1994). Secondly, EDI systems
can contribute to reduced clerical error (Scala &
McGrath, 1993) due to the reduced need to re-key
data. Thirdly, EDI can contribute to lowered
administrative costs (Solis, 1993; Stamper, 1991).
Fourthly, as paper-based systems are slow and
order lead times are higher, companies without
EDI systems may need to maintain higher inven-
tory levels (Sadhwani & Sarhan, 1987).
Organizational implications
The use of EDI has several implications for
organizations. Primarily, it can be a change agent
for traditional business processes and organization
structures—and how they may have been adapted/
reengineered to accommodate EDI. Such adapta-
tion/reengineering is likely to be re?ected not only
in the business processes, but also in the organi-
zational structures underlying such processes.
More generally, as Lucas and Baroudi (1994)
point out, new organizational designs are made
possible through IT such as EDI. These include
structural, work process, and communications
variables, which may be used to characterize four
new organization structures: virtual, negotiated,
traditional, and vertically integrated (Lucas &
Baroudi, 1994). Boynton, Victor, and Pine (1993),
in contrast, note that two new designs are emer-
ging: mass customization (dynamic product
change and stable process change) and continuous
improvement (stable product change and dynamic
process change). They also note that there is
potential for a new form: dynamic stability, a
synergy between mass customization and con-
tinuous improvement. Miles et al. (1997) observe
that each major era in business history has featured
a particular form of organization: early hier-
archical, vertically integrated organizations have
largely given way to network organizations. They
identify cellular organizations as a new way of
organizing based on the principles of entrepreneur-
ship, self-organization, and member ownership.
Combined with the organization’s decision pro-
cesses, examination of business processes can
characterize the range of systems that will best ?t
the organization—IT will then either support or
transform an organization’s business and decision
processes (Fedorowicz & Konsynski, 1992). Addi-
tionally, as Kulkarni and Heriot (1999) note, one
needs to consider costs of acquiring, storing, pro-
cessing, and disseminating information in addition
to transaction costs to be able to predict an
organizational form.
Overall, given the reality that organizations are
continually faced with changing requirements due
to technological and internal as well as external
environmental changes, various organizational
adaptations are necessitated in order to maintain
e?ectiveness (Lederer & Mendelow, 1990; Macy &
Arunachalam, 1995). The focus of this study is on
one particular aspect of this change dimension
related to the interaction between technology and
organizational structure
1
in the context of EDI
technology: the evolution of alternative intra-
organizational forms to support EDI processing.
Speci?cally, drawing from the communication
networks literature from social psychology, this
study examines several types of intra-organiza-
tional forms for EDI processing adopted by EDI
users and relates them to task performance indices
as well as EDI e?ectiveness and some other key
organizational context variables. The next section
describes the communication networks perspective
and outlines the hypotheses of this study.
Organizational forms for EDI: the communication
networks perspective
As Sheombar (1992) notes, in order to use
EDI optimally, organizational forms need to be
redesigned—and EDI can have an impact on all
three determinants of such organizational design:
1
Organizational structure represents the patterns and rela-
tionships that exist among organization or work unit elements.
A variety of structural dimensions has been used in previous
research to characterize such relationships, including central-
ization, decentralization, size, age, etc. (Grinyer & Yasai-
Ardekani, 1980; Macy & Arunachalam, 1995; Reimann, 1973).
This study focuses on one aspect of this concept: organizational
forms for EDI processing, i.e. alternative intra-organizational
con?gurations to support the transmission and processing of
EDI transactions.
V. Arunachalam/ Accounting, Organizations and Society 29 (2004) 227–241 229
product-market combinations, structure, and pro-
cess. The ensuing coordination that is needed
among organizational units is achieved by
communication between them (Sheombar, 1992).
The study of organizational forms to support EDI
processing, therefore, stands to gain insights from
research on communication networks in social
psychology.
This is because EDI has many important intra-
organizational applications and aspects of use
such as the intra-organizational processes and
arrangements to support inter-organizational
transaction processing, and intra-organizational
transaction exchange (e.g. purchase requisitions,
payment authorizations, etc.). Thus, while EDI
may be perceived from the outside as an inter-
organizational system with respect to areas such as
transaction exchange or supplier-customer rela-
tionships (cf. Sriram, Arunachalam, & Ivancevich,
2000), EDI also has many important aspects of
intra-organizational use that are worthy of further
study. This study deals with one such important
intra-organizational aspect of EDI use: alternative
intra-organizational forms to support EDI pro-
cessing. Communication networks research is used
to provide a context to characterize these organi-
zational forms (i.e. communication networks)
adopted for EDI processing within the respondent
organizations.
Bavelas (1948), then Leavitt and colleagues (e.g.,
Leavitt, 1951), and subsequently many other
researchers (for review, see Shaw, 1964,1978)
compared communication networks in which only
certain channels were possible. A diagrammatic
representation of these structures is provided in
Fig. 1. One key comparison network has been the
wheel, a highly centralized pattern in which all
other members can communicate only to one cen-
tral member, hence, all messages must pass
through the person at the ‘‘hub’’ position in the
wheel. Communication links a–c, b–c, d–c, and
e–c are the operative links (leading to the term
wheel because c is a hub and all the connections
are spokes). Other centralized structures include
the kite (with a–b, b–c, c–d and c–e the operative
links) and the chain (with a–b, b–c, c–d and d–e
the operative links). A decentralized comparison
pattern has been the circle, in which each member
is connected to two other members, but no one
member is more central than any other. That is,
each member is connected to two others such that
the communication links a–b, b–c, c–d, d–e, and
e–a are the operative links. Another decentralized
structure is the star or ‘‘comcon’’ (in which each
member/unit is connected to every other member/
unit). The patterns of communication within these
structures represent the basis for classi?cation as
centralized or decentralized. While all these pat-
terns have been used in various studies, the wheel,
kite, and chain structures de?ne one extreme of
the continuum of network variables—centraliza-
tion—while the circle and star structures de?ne the
other (decentralization).
In an EDI transaction processing context, the
organizational units involved may include, for
example, Accounting, Purchasing, Receiving,
Shipping/Distribution, Reporting/Budgeting, Pay-
ments (Financial EDI), and Production Planning.
For purposes of a brief illustration of the alter-
native con?gurations described above, consider
the following units: Materials-requisitioning unit,
Purchasing, Receiving, Accounts Payable, and
Treasury. In a wheel structure, a central EDI unit
would be the hub of all EDI activity and would
coordinate all the EDI document processing
involved in requisitioning, purchasing, receiving,
vendor invoicing, authorizing payment, and issu-
ing payment. In a chain structure, the materials-
requisitioning unit would transmit the requisition
to purchasing, purchasing would transmit the
purchase order (with quantities blanked out) to
receiving, receiving would transmit the receiving
report to accounts payable, and accounts payable
would authorize cash disbursement to treasury,
which would issue payment electronically to the
vendor. In a kite structure, a similar process would
be used, except that accounts payable would not
issue payment authorization directly on vendor
invoice (and receiving report), but also upon
reconciliation with purchase order, the kite units
being purchasing and receiving. In a circle struc-
ture, each unit would be able to communicate with
its immediate predecessor or successor in the chain
of transaction processing outlined above. In a star
structure, all units involved would be able to
communicate/transmit documents to each other,
230 V. Arunachalam/ Accounting, Organizations and Society 29 (2004) 227–241
as would be feasible with a database or enterprise
resource planning (ERP) system such as Oracle,
PeopleSoft, or SAP R/3.
Shaw (1964, 1978) has reviewed many of the
studies in this area, and has clari?ed their ?ndings.
He suggests that the idea of degree of centralization
of nets, or positions in nets, is not as useful as the
ideas of independence and saturation of the positions.
Independence of a position has to do with the
degree to which the member/unit in a given posi-
tion can carry out their part of the task autono-
mously, rather than depend on (one or more)
others in the group. Saturation of a position refers
to the total load-task and role demands as well as
message transmission demands that the member/
unit in that position must handle. In principle,
Fig. 1. Organizational forms for EDI processing.
V. Arunachalam/ Accounting, Organizations and Society 29 (2004) 227–241 231
positions can vary independently between inde-
pendence and saturation, although in reality they
often vary in opposite directions. Both are a?ected
by the nature of the task, and by the pattern of the
network.
Satisfaction derives primarily from high inde-
pendence, but also from moderate saturation.
Task performance may be aided by independence,
and is seriously hindered by saturation. Indepen-
dence is generally higher in decentralized nets,
regardless of tasks. Saturation is lower in cen-
tralized nets for simple tasks (few messages need
be sent); higher in centralized nets for complex
problems (because the task requires more mes-
sages, and they ?ow through a centralized posi-
tion, while peripheral members tend to demand
more information feedback on such problems).
Hence, centralized nets should be more e?cient
on routine tasks and less e?cient on nonroutine
ones, compared to decentralized nets. Networks
should not di?er in satisfaction on routine tasks
(though central and peripheral members may dif-
fer). On nonroutine tasks decentralized networks
(all of whose members are fairly independent, but
not saturated) should have higher satisfaction
than centralized nets (whose peripheral members
have low independence, and all of whose members
are likely to be saturated). The evidence supports
these expectations (Shaw, 1978).
This suggests that on routine EDI processing
tasks—which are characterized by low variety,
high analyzability, standardized procedures, and
few exceptions (Marcella & Chan, 1993; Lee, Han,
& Kym, 1998) e.g. transmitting purchase order or
receiving vendor invoice — more centralized net-
works will not di?er signi?cantly from more
decentralized networks (i.e. progressing across
wheel, kite, chain, circle, and star structures,
respectively) on satisfaction with EDI, even as they
have lower independence and higher saturation. In
contrast, on more nonroutine EDI processing
tasks—which are characterized by high variety, low
analyzability, non-standardized procedures, and
more exceptions, e.g. error reconciliation such as
with reference to vendor invoices, receiving
reports, and purchase orders, or exception report-
ing — more centralized networks will exhibit
lower satisfaction with EDI, and also have lower
independence and higher saturation.
2
These dif-
ferences will translate into concomitant di?erences
in their respective perceptions of EDI e?ective-
ness.
3
This discussion is represented diagram-
matically in Fig. 2 and stated in the form of the
following hypotheses.
H1. Centralization of EDI is related to lower
independence and higher saturation.
H2. Centralization of EDI and EDI task routi-
neness interact to a?ect satisfaction with EDI,
such that centralization results in lower satisfac-
tion than decentralization on nonroutine tasks
and in not signi?cantly di?erent satisfaction than
decentralization on routine tasks.
H3. EDI e?ectiveness will be positively corre-
lated with EDI task performance indices. Speci?-
cally, lower levels of saturation and higher levels
of independence and satisfaction will be asso-
ciated with higher EDI e?ectiveness.
Overall, factors related to the extent of usage
and integration of IT can also play a role in EDI.
Prior research has not empirically examined these
factors, but suggests that length of EDI use
(Rouland, 1991) and IS resources devoted to this
function (cf. Barua, Kriebel, & Mukhopadhyay,
1995; Mukhopadhyay, Kekre, & Kalathur, 1995;
O’Callaghan, Kaufmann, & Konsynski, 1992;
Weill, 1992) such as IT-related investment and
number of employees can a?ect EDI e?ectiveness.
This is because EDI use cannot yield bene?ts over-
night. Rather, a more valid presumption may be
that the greater the length of time the organization
has had to use and integrate EDI technology and to
develop a su?ciently robust EDI customer base
and a population of acclimatized EDI users, the
greater will be the contribution of EDI technol-
ogy. As Hitt and Brynjolfsson (1997) indicate,
increased IT investment could also be associated
2
EDI transaction categories under ANSI/EDIFACT stan-
dards for EDI were used in developing this classi?cation.
3
This study uses the bene?t dimensions of improved custo-
mer service, reduced clerical error, improved control of data,
increased sales, and decreased administrative cost to evaluate
EDI e?ectiveness. Such user-based perceptual criteria have
been demonstrated to be robust indicators of information sys-
tems e?ectiveness (Baroudi & Orlikowski, 1988; Ives & Olson,
1984).
232 V. Arunachalam/ Accounting, Organizations and Society 29 (2004) 227–241
with decentralized authority as ?rms that adopt a
complementary set of organizational practices
including decentralization of decision authority,
emphasis on subjective incentives, and a greater
reliance on skills and human capital tend to
use IT more. Additionally, the greater the IS
resources that have been devoted to EDI (i.e.
IT investment), the greater the likelihood that
bene?ts have been predicted to materialize and
the higher the exposure and support provided to
EDI users.
These factors suggest that length of EDI use and
IT intensity (i.e. ratio of total IT investment to
total number of employees) could positively mod-
erate the e?ects of decentralization of EDI pro-
cessing on EDI e?ectiveness, such that the positive
relationship between greater decentralization and
higher EDI e?ectiveness is enhanced as IT inten-
sity and length of EDI use increase. This suggests
the following hypothesis.
H4. The impact of decentralization on EDI
e?ectiveness will be positively moderated by IT
intensity and length of EDI use.
Method
Design and procedures
A two-phase survey methodology was employed
to test the hypotheses proposed in this study and
to gain deeper insights into EDI. In the ?rst phase,
surveys were mailed to 1400 organizations, all
registered EDI users, in the manufacturing,
wholesale, and retail sectors. These organizations
were selected from a database of EDI users pro-
vided to the author under a special agreement with
Phillips Business Information, publisher of the
EDI Yellow Pages. The surveys were addressed to
the EDI coordinators identi?ed in the database as
the current—and key—contact persons for EDI
and information systems within the respondent
organizations.
The initial (phase I) survey was made up of six
parts. The ?rst part asked for several details
regarding the organization’s background, speci?-
cally annual revenues, principal line of business,
and number of stores/outlets. The second part
asked about the organization’s perceived barriers
to EDI adoption. The third part sought to ascer-
tain the reasons for EDI adoption within the
respondent organization. Questions about length
of EDI use, pre-implementation justi?cation ana-
lysis, type of IT platform for EDI (e.g. database
implementation, enterprise resource planning
(ERP) system, etc.), investment in IT, total num-
ber of employees, number of employees in IT
support, and EDI document volume were also
contained in this section. The fourth part elicited
ratings of bene?ts actually obtained from EDI use,
along with some information regarding customer/
supplier use of EDI. Speci?cally, respondents were
asked to rate several provided items (improved
Fig. 2. Organizational form e?ects on EDI e?ectiveness and EDI task performance indices.
V. Arunachalam/ Accounting, Organizations and Society 29 (2004) 227–241 233
customer service, reduced clerical error, improved
control of data, increased sales, and decreased
administrative cost) on a 10-point scale in terms of
bene?ts that their organization had actually
obtained from EDI use.
The ?fth part asked the respondents to describe
EDI processing. Speci?cally, they were asked to:
(1) specify the names of the organizational units/
departments involved in EDI processing (e.g.
Accounting, Purchasing, Receiving, Shipping/Dis-
tribution, Reporting/Budgeting, Payments/Finan-
cial EDI, Production Planning), (2) outline the
steps in EDI transaction processing, and (3) trace
their ?ow across the various organizational units/
involved in EDI processing. They were then pre-
sented with a list of ?ve organizational forms
(wheel, chain, kite, circle, or star structures) along
with brief descriptions of each form, and were
asked to indicate which one most closely corre-
sponded with the EDI processing in their organi-
zation (if none corresponded, they could choose
none of the above). They were then asked to indi-
cate, for each of two types of EDI processing
(routine and nonroutine tasks, as described ear-
lier), their overall perceptions of the extent of
independence, saturation, and satisfaction experi-
enced among the participants as a whole in EDI
processing. The last part asked respondents about
the extent to which the bene?ts from EDI had ful-
?lled their overall expectations. The last question in
this part requested the respondents to indicate the
names, mailing addresses, and phone numbers of
the managers of each of the organizational units
identi?ed in the previous part. They were also
requested to provide a copy of their business card
in case there were further questions and to enable
them to receive a copy of the research results.
The second (phase II) survey was made up of
only four parts and was addressed to each of the
organizational unit/department managers identi-
?ed in the phase I surveys. The ?rst part requested
information about the nature of their unit’s invol-
vement in EDI processing, EDI document volume,
and sta?ng levels in their units. The second part
asked them to rate, on a 10-point scale, the same
EDI bene?ts contained in part 4 of the phase I
survey. The third part of the survey replicated part
5 of the phase I survey. Speci?cally, they were
asked to: (1) outline the steps in EDI transaction
processing, and (2) trace their ?ow across the var-
ious organizational units/involved in EDI proces-
sing. They were then presented with a list of ?ve
organizational forms (wheel, chain, kite, circle, or
star structures) along with brief descriptions of
each form, and were asked to indicate which one
most closely corresponded with the EDI processing
in their organization (if none corresponded, they
could choose none of the above). They were then
asked to indicate, for each of two types of EDI pro-
cessing (routine and nonroutine tasks, as described
in previous section), their overall perceptions of the
extent of independence, saturation, and satisfaction
experienced among the participants in EDI proces-
sing within their own unit. The last part asked
respondents about the extent to which the bene?ts
from EDI had ful?lled their overall expectations.
Independent variables
Organizational Form. Organizational forms for
EDI processing were classi?ed into one of ?ve
types: (1) wheel structure, (2) chain structure, (3)
kite structure, (4) circle structure, and (5) star (or
‘‘comcon’’) structure. This classi?cation was done
by means of a three-part veri?cation process: (1) if
the organizational forms indicated among an
organization’s respondents (i.e. EDI coordinator,
unit managers) were congruent, the coding was
obviously straightforward (79% of initial respon-
dent organizations ?t this category), (2) if there
was any incongruence within an organization’s
selections, then two independent raters (both doc-
torally quali?ed) independently coded the written
descriptions provided by the respective respon-
dents and came to agreement on the organiza-
tional form (15% of initial respondent
organizations ?t this category),
4
and (3) if there
4
A coe?cient of agreement generally referred to as Cohen’s
K of 0.93 was computed for the coding assignments. K
measures the proportion of agreement between two scorers in
placing items into a set of K unordered categories, and is indica-
tive of the proportion of judgements in which there is agree-
ment after pure random chance is excluded. As no sampling
distribution exists for K when it is transformed into z scores, it
is signi?cant at P