EXPLICITLY DEFINED PROTOCOLS AS MEANS OF
INTER-ORGANIZATIONAL COMMUNICATIONS:
HOW THEY ARE ADOPTED AND WIDELY SHARED
Yoko TAKEDA and Jiro KOKURYO
Graduate School of Business Administration, Keio University
The Keio Management Society
Research Papaer Series No. 42 (October 1996)
慶應義塾経営管理学会
リサーチペーパー・シリーズ
No. 42 (1996年10月）
This paper is not to be cited and copied without authors‘ permission.
本研究は、財団法人電気通信普及財団より助成をうけました。
ABSTRACT
This paper explores the mechanism of media choice, i.e., how a particular medium may be chosen among a portfolio of available media in inter-organizational communications.
A particular emphasis is placed on identifying (1) when media that employ “explicitly defined” protocols are adopted and (2) how such explicitly defined protocols may be shared among a wide range of trade partners, thereby creating switching flexible (easy-to-change trade partners) situations. This analysis was implemented based on the authors’ recognition that it is advantageous to use media that employ explicitly defined and widely shared protocols in order to flexibly utilize external resources for management.
The results of a mail survey among 666 users of mold and die/automated machinery components support the following hypotheses: (1) Uncertainty and equivocality affect the choice of inter-organizational media choice in agreement with the information richness theory that traditionally explained intra-organizational communications. (2) Concerning the switching flexibility, a firm that has power over a partner and that has a large installed base of internal protocols tends to impose its own protocol on its trade partners (i.e., protocols whose switching flexibility is high for the focal firm and low for the partner). Notably, even if the power balance was controlled, the installed base of internal protocols significantly reflected on the media choice.
Table of contents
3. Research Framework and Hypotheses
5. Results of the Survey and Analyses
In the high-technology arena, particularly in the rapidly growing IT-electronics industries, there is a growing awareness that excellence in the use of external resources is a vital factor for survival of firms. The emerging philosophies in management emphasize that firms focus on products that they excel in, while forming alliances to meet the needs of the customers that opt to receive integrated services. This is contrasted with traditional strategies, in which companies pursued to the internalization of a full range of capabilities in providing a product line.
The drivers behind such a shift include the following factors: (1) the technologies involved in each product are becoming too complex for a single company to manage; (2) shortening life cycle of products is adding a further burden on R&D capacity; (3) the optimal supply chain is becoming global; and (4) investment in R&D is becoming increasingly large and requires absorption by more than one company’s product line. Such tendencies are expected to strengthen as the computer networks develop to form “cyberspace,” in which any combination of computers in the world can exchange information in multimedia form flexibly, cheaply, and in large volume.
The use of external resources, however, is not a trivial undertaking. To implement cooperative work effectively, firms have to have the capabilities to communicate and coordinate with outside organizations effectively. Without such skills, the vision of “networked organizations,” however large the potential benefits of such innovations may be, will fall short of becoming a reality.
Thus the study of inter-organizational communications is critically important for today’s businesses. This paper attempts to provide some insight into this topic by focusing on the mechanism by which a particular medium is chosen for inter-organizational communications.
The task of inter-organizational communications is relatively easy when we can assume a long-term relationship among the players. The reality in the high-technology arena, however, is that such partnerships are formed and dissolved frequently, requiring firms to deal frequently with new partners. Kokuryo (1995) identified two major forms of inter-firm collaboration on an electronic platform, i.e., electronic alliances and electronic markets. Electronic markets (Malone et al. 1987; Kokuryo 1995) refer to firms making ad hoc transactions with a large number of unpredetermined partners. Electronic alliances refer to partnerships of firms that exchange large volumes of information on computer networks to maintain a tightly coordinated supply chain (Clemons et al. 1993; Kokuryo 1995). Electronic alliances differ from the traditional keiretsu type of long term corporate relations in that the firms opt to form interchangeable alliances that are limited to a particular generation of a particular line of product.
This fluid nature of inter-firm relations in the cyberspace requires that firms maintain an open structure, i.e., use an explicitly defined and widely shared interface to exchange physical objects and information with the outside world. Kokuryo (1995) argues that open interfaces are necessary on three fronts: (1) openness of the product design so that products of many companies can be used integratively, (2) openness of managerial resources, such as personnel, vendors, and distribution channels, so that resources can be assembled and reassembled flexibly, and (3) openness of the logistic systems, which allows companies to form and dissolve operational alliances quickly.
This paper focuses on a subset of the logistic system, i.e., the openness of inter-organizational communications media. An open medium is defined as a medium which have (1) explicitly defined and (2) widely shared protocols.
It is to be noted here that protocols carry meaning only in relation to the context of social interactions in which they are used (Mead 1934). Adopting this line of thought, it is important in cyberspace (in which the firm relations are expected to form and dissolve quickly) to reduce the cost of associating protocols with the context and thereby reduce the cost of changing trade partners.
Such reduction in the cost of context building is accomplished by defining following the protocols prior to the initiation of communications sessions: (1) elements (words, codes, and the like) that express specific meanings, (2) formats that define the relationships among the elements so that elements collectively form a meaningful message, and (3) the procedure in which messages are exchanged. An explicit definition of protocols at these three levels increases the speed in which effective communications are established among partners that have little experience in dealing with each other before.
The sharing of the protocol definition is also an important factor. A company limits the scope of partners that it can deal with when it adopts a set of protocol definitions that is specific to a particular relation. In order to maintain flexibility in dealing with a large number of partners, it is beneficial to share the definition with a large number of partners, including potential ones.
Figure 1 demonstrates the general logic and motivation behind the authors’ research on this topic.
With the above perception of the role of the explicit definition of protocols and the sharing of protocols, the authors have framed the issue in the context of media choice. That is, research was conducted on when media with explicitly defined protocol are adopted and how such protocol definitions are shared by many firms for inter-organizational communications.
To this goal, this paper first looks at the theories of media choice, which have so far focused primarily on intra-organizational issues. Second, we consider additional factors that should be incorporated upon theorizing the mechanisms of inter-organizational media choice. Third, we construct frameworks and hypotheses on media choice to explain the nature of two dependent variables, i.e., (1) the degree of explicitness in the definition of protocols and (2) the degree of switching flexibility of protocols. Finally, the hypotheses are tested using empirical data.
Regarding organizational communications theory, Krone et al. (1987) pointed out the existence of four perspectives: i.e., (1) mechanistic, (2) psychological, (3) interpretive-symbolic, and (4) systems-interaction perspectives. Likewise, the theories of media choice have evolved from (a) mechanistic ones ( e.g., theories that explain the choice by the fit between the inherent media characteristics and the communications requirements), to (b) psychological approach ( e.g., theories that incorporates users’ perceived characteristics of media), then to (c) an interpretive-symbolic perspective that pays attention to symbolic meanings involved in the use of a particular medium, and finally to (d) a systems-interaction perspective that looks, within the community in which the medium is used, at the effects of other people’s actions in the choice of one’s medium. (See Figure 2.)
Information richness
The information richness theory (Daft and Lengel 1986) drawing on information processing theory (Galbraith 1973; 1977; Tushman and Nadler 1978), points at uncertainty and equivocality of information as the determinant of media choice. It demonstrates that media that are good at processing large amounts of information are chosen when high priority is placed on uncertainty reduction (i. e., obtaining additional data, seeking answers to explicitly expressed questions). Likewise, it shows that media that are good at processing rich information (i. e., information that requires a high level of interpretation to comprehend) are chosen in instances in which equivocality reduction (i. e., clarifying research agreement, deciding which questions to ask) is the priority.
In the general application of this theory, face-to-face communication is regarded as the medium that can process the richest information. This is followed by telephone, e-mail, and written documents. In empirical and meta-analyses, the applicability of the media richness theory has been confirmed for traditional media of communications, such as face-to-face, telephone, and written documents. On the other hand, some doubts are cast on the applicability of the theory to new modes of communications, such as e-mail (Rice and Shook 1990; Markus 1994).
The Information richness theory is a mechanistic model, in that it assumes that there is an inherent and fixed fit between a particular medium and a particular content of communications. Trevino et al. (1987) extended this theory by pointing out that the choice of media may be made from interpretive-symbolic points of view instead of purely functional ones. For example, one may choose written documents in order to convey authority while less formal means of communications media may suffice to convey the protocol itself. More recent studies are converging toward a social constructionist’s view, that an organization’s view on how rich a medium is is influenced by the interaction that took place within the organization employing the medium.
Physical constraint
Obviously, situational constraints have a strong influence on media choice. When the partners are far away and/or there is a severe constraint in the availability of time, the desire to use face-to-face meetings for a rich communications session is overridden in favor of telephone or e-mail (Trevino et al. 1987). It is observed that the predictions on media choice based on the media richness theory are often inaccurate in these instances in which there are physical constraints (such as time or geographic location) in conducting the communications session.
Critical mass
Network externality is another factor that is known to influence media choice. The utility of a particular communications medium is strongly affected by whether or not there is a critical mass of users that already adopt the medium (Granovetter 1978; Oliver et al. 1985; Markus 1987). The critical mass theory is reported as effective in analyzing the adoption of emergent interactive media that have not become ubiquitous (Markus 1994).
Social construction
Fulk et al. (1990) point out that “media perceptions are, in part, subjective and socially constructed” (pp. 121) and suggest that perception and usage patterns for a particular medium are born in an organization by its members’ interexchange of actions and/or perceptions.
The irrational choice of media is possible through the mechanism of social influence. Fulk et al. (1987) also reported that members of a group converged on a media choice pattern while another group converged on a different usage pattern. Lea et al. (1995) further employed a case study to demonstrate how media usage pattern in a local network may disperse to become a global pattern. The social constructionist view is considered applicable in cases in which the perception of a medium is not established in the group in which it is used.
Existing literature on media choice primarily deals with phenomena inside of a firm. Most notable is the chain of studies since Mintzberg (1973) that analyze how managers choose among available media.
By contrast, studies on inter-organizational media choice are relatively uncultivated. Whether or not theories on intra-organizational media choice have power in explaining inter-organizational media choice is yet to be known.
Exploratory studies on this issue by Takeda and Kokuryo (1995) and Takeda (1996) suggest that while factors that affected media choice in an intra-organizational environment are also important in inter-organizational settings, an analysis of inter-organizational media choice requires attention to several other factors. The list includes firms’ resourcefulness, inter-organizational context, and corporate policies.
Resourcefulness alludes to the capabilities of organizations to bear the cost of acquiring communications channels, building internal information systems, and creating interface to the external systems. Takeda and Kokuryo (1995) found that these factors play major roles in EDI (electronic data interchange) adoption.
An example of inter-organizational context is the balance of power between firms. The interdependence of operations offers another example. Balance of power is relevant in that it is easier for powerful firms to impose their own protocols on their partners. In the case of operational interdependence, Takeda and Kokuryo (1995) found that adoption of EDI may be positively related to increased face-to-face communications in high interdependence environments.
Third, there is the corporate policy factor, i.e., boundary-spanning strategy on what a firm decides to procure from outside, who it chooses as trade partners, and how the relations are to be managed. When a firm wants to lock-in a particular set of trading relations, it may attempt to have the trade partners acquire data terminals that employ protocols proprietary to the firm’s transaction system. This increases trade partners’ loyalty by increasing the switching cost involved in abandoning the firm (McFarlan 1984; Clemons and Row 1991; Mata et al. 1995).
On the other hand, when the strategy is to deal with a large number of heterogeneous trade partners and/or when flexibility in changing trade partners is important, a firm would opt to adopt a medium that the potential trade partners find easy to adopt. In the case of electronic transaction systems, this implies the adoption of systems that employ widely accepted standard protocols (Kokuryo 1995).
3. Research Framework and Hypotheses
3.1 Explicit Definition of Protocols
A decision to adopt a medium often determines the structure of protocols to be conveyed through the medium. The extent to which a medium requires the explicit definition of protocols depends on how much message elements, formats (that define the relation among elements), and communications procedures should be defined prior to the initiation of a communications session.
The degree to which protocol is explicitly defined varies from one medium to another. For example, EDI requires detailed definition of all aspects of communications, i.e., protocol element, format and message exchange procedures. Without such structuralization, it is impossible for the receivers of protocols to interpret the meaning of messages. Protocols used to order goods by fax are generally less explicitly defined than EDI. They nevertheless often involve standardization of product codes, account ID, and protocol formats. A salesperson can take almost free format orders from customers upon face-to-face meetings.
A note of caution here is that, as the fax example demonstrates, the extent of explicit definition is variable even for a particular medium depending on the instances of their use. With such reservations, we nevertheless present Table 1, in which we classified various uses for order taking depending on their typical level of explicitness.
To express this in an alternative way, explicit definition of protocols means the reduction of the possible interpretation of messages. Huber and Daft (1987) pointed to the following as the criteria for determining the richness of the media: (1) the opportunity for timely feedback, (2) the ability to convey multiple cues, (3) the tailoring of protocols to personal circumstances, and (4) language variety. Looking at the matter from this angle, the degree of explicitness in the definition of protocol structure is an inverse function of media richness. Media with highly explicitly defined protocols are typically not equipped to provide feedback or cues to eliminate equivocality.
Thus, if we follow the information richness theory (Daft and Lengel 1986), when uncertainty reduction has the priority, i.e., when a large amount of information is required, media with a high degree of explicit definition in the protocol are chosen. On the other hand, if equivocality reduction has priority, i.e., when the interpretation is important, media with a low degree of explicit definition in the protocol are chosen. (See Figure 3.)
This leads us to our first set of hypotheses.
H.1.1 In inter-organizational transactions, given that the level of equivocality is constant, there is a positive relation between the degree of uncertainty and the choice of media that employs highly explicitly defined protocols.
H1.2 In inter-organizational transactions, given that the level of uncertainty is constant, there is a negative relation between the degree of equivocality and the choice of media that employs highly explicitly defined protocols.
3.2 Sharing of Protocol Definitions
When we assume explicitly defined protocols, the next major factor is the extent to which the definition is shared by a large scope of players. This directly affects the switching flexibility, i.e., the ease of changing one’s trade partners while using the same set of protocols and thus the same infrastructure for communications. A transaction-specific asset, such as machinery specially designed for a particular customer’s needs, increases the trade partner’s bargaining power and consequently the risk of its taking opportunistic actions against you (Williamson 1975; 1985). This is applicable to communications protocols.
Communications protocols are directly related to business processes. Thus, attempts to change communications protocols often involve sizable costs, which are generally called the switching cost. It is possible to take advantage of this phenomenon if a firm wants to lock-in their trade partners. This is accomplished by raising the trade partners’ switching costs through the use of proprietary communications protocols with them (McFarlan 1984: Clemons and Row 1991; Clemons 1991; Mata et al,. 1995).
Switching flexibility in inter-organizational communications protocols can be classified along two dimensions: (1) switching flexibility for the focal company and (2) switching flexibility for the trade partners. (See Figure 4.)
I. (Specific protocols for a peer-to-peer relationship)
This quadrant represents a situation when a couple of trade partners employ protocols that are entirely specific for their relationship. The protocol cannot be used for trade with other firms at all.
II. (Proprietary protocols of a focal firm)
A company might enjoy switching flexibility for itself, while forcing the trade partners to accept low switching flexibility. This happens when a company succeeds in having a large number of potential companies adopt its proprietary protocol.
III.(Proprietary protocols of a partner)
A firm might have to endure low switching flexibility while its trade partner enjoy high switching flexibility. This is the other side of dimension II, i.e., when a powerful trade partner imposes its proprietary protocol on you.
IV. (Socially shared protocols “standard”)
When there are socially accepted and adopted standard protocols, both sides of a trade partnership can enjoy high switching flexibility.
Two factors are identified as potential variables for the basis for the selection of media with various switching flexibility for trade partners. They are (1) resource-dependency and (2) the installed base of protocol of partner companies embedded in their communications infrastructure.
Resource-dependency
Type I (specific protocols for a certain peer-to-peer relationship: low switching flexibility for both partners) in Figure 4 represents a situation in which both of the trade partners face high switching costs. Strong interdependence and long-term commitment have to be present for such a selection of media to be made.
Type II and III relations, in which one party in a relation accepts low switching flexibility while its partner does not, tend to occur when there is a strong power imbalance, typically based on resource dependence. The relationship between a large retail chain and small prepared food manufacturers is a typical example.
The adoption of socially shared protocols by both parties as represented in Type IV tends to happen when both parties want flexibility in transacting with a large number of firms. Resource-dependency in such cases is weaker than in other instances. (See Figure 5.)
Installed base of systems based on internal protocols
Most firms have internal systems, such as control systems and production management systems, that employ firm specific protocols. Such protocols are translated into protocols that trade partners agree upon. The costs for such translation (the largest of which is the development costs of the software) can be quite substantial. Thus, when a company has a large installed base of mechanisms that employ proprietary protocols, it has strong incentives to impose its protocols on its trade partners.
The extent to which companies have the incentive to impose such proprietary protocols depends on the size of the installed base of ( in other words, the large sunk costs in) transaction systems that are constructed using the companies’ proprietary protocol. When a company is equipped with the capability to communicate using widely shared standard protocols (we subsequently call such systems “open” systems), then the firm has little incentive for the imposition of its own protocols. We hypothesize that such companies tend to be firms that have a small installed base of old systems. (See Figure 6.)
We thus reach the following hypotheses:
H2.1 In inter-organizational transactions, the tendency for the selection of media that give high switching flexibility to the powerful and lower switching flexibility to the weaker, is greater when there is a large power imbalance between the trade partners,.
H2.2 In inter-organizational transactions, a firm that has a large installed base of systems employing the firm’s proprietary protocol tends to choose media that allows small switching flexibility for its partners.
Data for this research were obtained from a mail survey of the customers of Misumi Corp., a wholesaler of mold and die/automated machinery components, on order placement methods.
We chose this site because Misumi’s customers include almost all segments of the mechanical manufacturing industry, including automobiles, electric equipment, precision machinery, and office equipment. The sample also covers a wide range of companies in terms of size. It also covers various types of manufacturers including parts suppliers, assemblers, dye and cast manufacturers, and tool manufacturers. Thus, we have unbiased data along these dimensions. As a footnote, we have excluded from our sample some distributors in Misumi’s customer base.
The second reason for the choice of the site was that Misumi employs a wide range of media for receiving orders, including (1) fax using the Misumi format, (2) fax using customers’ formats, telephone, fax OCR (uses Misumi format), host computer based EDI, EDI using the NIFTY-Serve PC communications system, and ordinary mail. Such a wide coverage offers an ideal environment to look into the media choice question.
The third reason is that Misumi represents all of its products by alphanumeric characters. It also clearly states price and delivery time in its catalogs. This highly structured nature of its operations makes it easy to distinguish normal operations, which are conducted with only highly defined protocols, and those exceptional cases in which equivocality presumably requires the use of rich media.
We stratified Misumi customers into the following five categories and conducted random sampling to select a total of 1,300 firms to send questionnaires to.
Group 1: Firms that purchased between \120K and \190K from Misumi in the preceding year (250 samples).
Group 2: Firms that purchased between \190K and \359K from Misumi in the preceding year (250).
Group 3: Firms that purchased from between \360K and \5990K from Misumi in the preceding year (250).
Group 4: Firms that purchased more than \6,000K in the preceding year (250).
Group 5: Firms that employ NIFTY-Serve to place orders regardless of purchase amount (300).
The rationale behind the stratification was to monitor and control for the effect of transaction volume, firm size, and bargaining power vis-a-vis Misumi. In addition, Group 5 was chosen because the number of NIFTY-Serve adopters remains a small proportion of the total population and therefore the number of samples that could be obtained from a simple random sampling was be too small.
The respondents play one of the following functions in placing orders for die/mold parts and/or factory automation parts to Misumi: (1) make decisions on where to place the orders, (2) design product specifications, (3) choose which media to use for orders, or (4) actually place orders. In many cases these four overlap. The questions asked were relatively straightforward, so, we could assume a person that belonged to any one of the four had a total of accurate information.
1,300 questionnaires were mailed in March 1996; 51.2% (666) of the questionnaires were returned.
4.2 Variables and Analytical Method
In this research we used two models. Both were tested using logistic regressions.
Model 1 uses the degree of explicitness in defining protocol as dependent variable, while uncertainty and equivocality are independent variables. This model was used to test hypotheses 1.1 and 1.2. The equation representing the models can be specified as follows:
Model 1 (for hypotheses 1.1 and 1.2):
Explicitness in the definition of protocols (high or low)
=α1 + β11 UNCERTAINTY + β12 EQUIVOCALITY
Model 2 uses the degree of switching flexibility of protocols as dependent variable to test hypotheses 2.1 and 2.2. Independent variables were the balance of resource-dependency and installed base of internal protocols
Model 2 (for hypotheses 2.1 and 2.2):
SWITCHING FLEXIBILITY (high or low)
=α2 + β21 BALANCE OF RESOURCE-DEPENDENCY
+ β22 INSTALLED BASE
Operational definitions of the variables in each model are as follows:
Model 1
Dependent Variable
We have categorized the media that Misumi uses to receive orders into two, i.e., (1) media with a high level of explicitness in the definition of protocols and (2) media with a low level of explicitness in the definition of protocol.
In this particular case, Misumi receives all of its orders with explicitly defined elements. Thus, the primary difference in the level of explicitness in protocol definition exists at the format level.
Order taking by the telephone has almost no protocol definition at the format level. Fax messages either using Misumi or customer forms have explicit formats but allow relatively strong flexibility in that exceptional requests (e.g., on specifications, price, and delivery) can be added easily on the form. On the other hand, fax OCR, NIFTY-Serve-based EDI, and host-based EDI rigidly define the format, such as the sequence of elements, and allows little room for deviation from the original definition. Based on these observations, we grouped the media as follows:
The degree of explicitness in the definition of protocols
High group: fax OCR (44)・NIFTY-Serve (168)・host-host EDI(9)・customer’s form sent by mail (9): total 230
Low group: telephone (16)・Misumi’s order form sent by facsimile (328)・customer’s order form sent by facsimile (67): total 411
Independent Variables
<Uncertainty>
Uncertainty in the order placement context can be interpreted as the volume of information necessary in placing orders. The volume of information can be measured in terms of order placement frequency, number of items placed per single order, and number of digits of product code (length of protocols) necessary in placing orders. These three were thus used as operationalized indices for uncertainty.
<Equivocality>
Equivocality in the order placement context can be interpreted as the extent to which explanation is necessary beyond information presented in catalogs. We have operationalized this by asking experiences in (1) requesting products not described in catalogs, (2) making inquiries about products, (3) requesting shorter delivery lead time, and (4) negotiating for discounts. The yes-or-no answers were added to create an integrated index. (See Figure 7.)
Model 2
Dependent Variable
With the exception of telephones, which allow high switching flexibility for both sides of the trade partnership, media used for placing orders to Misumi allow high switching for either Misumi alone or the for customer but not for both.
It is at the format level that there is a large variance among media in the degree of switching flexibility. Fax and traditional mail orders employing customers’ formats allows a high degree of switching flexibility for the customers but not for Misumi. Host-based EDI, on the surface, uses EIAJ standard protocol, so we may judge that it adopts widely shared protocol and allows high switching flexibility for both Misumi and the customers. The reality, however, is that the protocol formats are proprietary to the customers, and there is little switching flexibility on Misumi’s side. On the other hand, fax orders using Misumi formats, fax OCR orders that use Misumi format, and NIFTY-Serve EDI that employs Misumi formats all allow high switching flexibility for Misumi but not for the customers. We have thus categorized the media on their switching flexibility as follows:
The degree of switching flexibility of protocols for customers
High group: customer’s order form sent by facsimile (67)・host-based EDI (9)・customer’s form by mail (9): total 85
Low group: Misumi’s order form sent by facsimile (328)・fax OCR (44)・NIFTY-Serve (168): total 540
Dependent Variables
<Balance of resource-dependency>
An index that can be used to represent the power balance, i.e., resource-dependency, is the relative size of companies. Another is the share of a particular partner’s transaction in the total volume of a firm’s total transactions. In the case of this research, since the only seller is Misumi, we have used the customers’ annual sales volume as the index for the relative size of the partners to Misumi. Also the share of purchase from Misumi in the customers’ total purchase was obtained from the questionnaire. Finally, the relative share of each customer for Misumi was categorized into four classes as shown in Figure 8-1.
<Installed base of internal protocols>
The degree of openness of the installed base of communication systems of a focal firm was quantified by the answers to the following six questions: (1) whether or not there are carbon slip systems (i.e., shitei denpyo) specified by the focal firm (yes 1 point, no 0 point), (2) whether or not receipt forms (i.e., kenshusho, which are proprietary format in all cases) are sent to the vendors (yes 1 point, no 0 point), (3) whether or not internal documents of the focal firm are used for order placement (yes 1 point, no 0 point), (4) product code used (focal firm’s 1 point, Misumi’s 0 point), (5) ownership of computers (mainframe 3 points, workstations 2 points, personal computers 1 point, none 0 point), and (6) the use of CAD/CAM (integrated CAD and numerical controlled manufacturing system 2 points, stand-alone use of CAD 1 point, do not use CAD 0 point). (See Figure 8-2.)
5. Results of the Survey and Analyses
Model 1
The Table 2 shows the results of the analysis on Model 1. Chi-square for covariates of -2 log likelihood, which is an indicator of the model fit was significant at p<0.01 level. As for the significance level of the independent variables, among three of the uncertainty indices, (a) transaction frequency and (b) volume of transaction per transaction proved to be significant at p<0.05 level (Wald chi-square), as hypothesized. The equivocality variable was significant at p<0.01 level. Parameter estimates were 0.162 for transaction frequency, -0.198 for volume of transaction per transaction, and 0.297 for equivocality.（Note: Transaction frequency is higher when the index is lower, so the signs for frequency and volume are in opposite directions, as presumed.)
The above analysis leads us to the
following statements:
・The greater the uncertainty, the greater the likelihood that media with a higher degree of explicitness in the definition of protocols will be chosen.
・The greater the equivocality, the greater the likelihood that media with a low degree of explicitness in the definition of protocols will be chosen.
We conclude that our data supports hypotheses 1.1 and 1.2.
One variable that we did NOT find significantly related was the average number of product code digits used in order placements. We think that this was the case because the average length of product codes, the question we asked in the questionnaire, was too vague and failed to capture actual uncertainty reduction necessity.
Model 2
The results for model 2 are shown in Table 3. -2 log likelihood, which is an index for model fit, was significant at p<0.01 (chi-square for covariates). As for the significance level of the three independent variables on the balance of resource-dependence, firm size was found to be significant at p<0.01 (Wald chi-square), and purchase volume (share of the firm for Misumi) was significant at p<0.05. The index on the size of the installed base of systems employing internal protocols was found to be significant at p<0.05 level. Parameter estimate for the firm size was -4.13, -0.316 for purchase volume, and -0.177 for installed base. Signs for the parameter estimates all agreed with the expectations.
Thus, hypotheses 2.1 and 2.2 are generally supported as follows.
・The greater the imbalance is in resource dependence, the greater the tendency is for the stronger firm to choose a medium that allows large switching flexibility for itself and small switching flexibility for its partners.
・The larger the installed base of systems that employs internal protocols, the greater the tendency is for the firm to choose a medium that allows large switching flexibility for itself and small switching flexibility for its partners.
The share of Misumi in the customers’ purchasing volume, an index we used for resource dependence (Misumi’s powers over the focal customer), was found not to be significantly related to the media choice. We think this is because concerning punch and die/automated machinery components, customers have a wide enough rage of alternative vendors to choose from to let a vendor (however large it may be) exercise any power over them.
In this research we have attempted to identify when media with a high level of protocol definition and a high level of switching flexibility, two characteristics we feel are important, might be employed in inter-organizational communications. We have confirmed that the level of uncertainty and equivocality (factors often used for intra-organizational communications analysis) significantly affect the use of (or aversion of) highly defined media. Likewise, we found that resource dependency and the largeness of the installed base of non-switching flexible media are significant in the adoption of switching flexible media.
There are several important questions that we have fallen short of answering in this research. First, because we did not have sufficient data on the use of media that use socially shared standard protocols, we could not analyze how “mutually switching flexible media” are adopted.
Nevertheless, we can hypothesize how socially shared standard protocols may be adopted by the data we obtained on how a firm decides to adopt other firms’ protocols. That is, we have confirmed that resource dependence and the largeness of sunk costs in the installed base of systems that use proprietary protocols are two factors that give strong incentives for a firm to force its proprietary protocol on others.
This leads us to assume that it is when the resource dependence relation is balanced between trade partners, and when the installed base of proprietary systems is small, that socially shared standards are accepted. This generally supports the authors’ observation that in Japan, because firms were cash rich and could afford to invest heavily in proprietary systems, they are now in a rather difficult position of having to throw away large systems in order to meet the needs for open computer communications with other firms.
In the future, we think it will be useful to incorporate the strategic intent of firms into our analysis. That is we can study how the following factors might affect the choice of media:
・The intent to increase the number of trade partners.
・The intent to have flexibility in switching trade partners.
・To avoid having vendor depend excessively on you.
Second, this study, which looks at the order placement of standardized products, primarily deals with inter-organizational communications that require a low level of context sharing. This is reflected in our choice of relatively mechanistic models for the explanation of media choice. We are aware that this is only a small part of the larger picture. In many cases, firms have to communicate with a high level of context sharing. Cooperative development of new products and cooperative marketing would be examples of cases when such a high-level context sharing becomes necessary. In these cases, our analytical model should be expanded to incorporate social constructionish factors, such as how each medium is perceived by the users.
Third, we used “the most frequently used medium” as our dependent variable. In reality, firms use media in certain combinations of media instead of relying on a single medium. The structure of such media mix tend to become more complex as the need for high context building becomes stronger. Thus, in situations when equivocality is more critical, we would need to approach the question of “how a particular media mix is chosen” instead of how a particular medium is chosen.
Clemons, E. K. (1991), “Sustaining IT Advantage: The Role of Structural Differences,” MIS Quarterly Vol. 15, No. 3, Sept., pp. 275-292.
Clemons, E. K., S. P. Reddi, and M. C. Row (1993), "The Impact of Information Technology on the Organization of Economic Activity: The ‘Move to the Middle’ Hypothesis," Journal of Management Information Systems, Vol. 10, No. 2, pp. 9-35.
Clemons, E. K. and M. C. Row (1991), “Information Technology at Rosenbluth Travel: Competitive Advantage in a Rapidly Growing Global Service Company,” Journal of Management Information Systems, Vol. 8, No. 2, pp. 53-79.
Daft, R. L., and R. H. Lengel, (1986), “Organizational Information Requirements, Media Richness and Structural Design,” Management Science, Vol. 32, No. 5, pp. 554-571.
Fulk, J., J. A. Schmitz and C. W. Steinfield (1990), ‘A Social Influence Model of Technology Use,’ in Fulk, J. and C. W. Steinfield (ed.), “Organizations and Communication Technology,” Sage, pp. 117-139.
Fulk, J., C. W. Steinfield, J. A. Schmitz and J. G. Power (1987), “A Social Information Processing Model of Media Use in Organizations,” Communication Research, Vol. 14, pp. 542-543.
Galbraith, J. R. (1973), "Designing Complex Organizations," Addison-Wesley.
Galbraith, J. R. (1977), "Organization Design," Addison-Wesley.
Granovetter, M. (1978), “Threshold Models of Collective Behavior,” American Journal of Sociology, Vol. 83, No. 6, pp. 1420-1443.
Huber, G. P. and R. L. Daft (1987), ‘The Information Environments of Organizations,’ in Jablin, F. M. et al. (ed.), “Handbook of Organizational Communication,” Sage, pp. 130-164.
Kokuryo J. (1992), “The Impact of the Retailing Industry’s EDI-Based Quick Response Systems on Vender Logistics Operations,” Unpublished Doctoral Dissertation, Harvard University.
Kokuryo, J. (1995), “Open-network Management,” Nihon Keizai Shinbunsha.
Krone, K. J., F. M. Jablin and L. L. Putnam (1987), ‘Communication Theory and Organizational Communication: Multiple Perspectives,’ in Jablin, F. M. et al. (ed.), “Handbook of Organizational Communication,” Sage.
Lea, M., T. O’Shea and P. Fung (1995), “Constructing the Networked Organization: Content and Context in The Development of Electronic Communications,” Organization Science, Vol. 6, No. 4, pp. 462-478.
Malone, T. W., J. Yates, and R. I. Benjamin (1987), "Electronic Markets and Electronic Hierarchies," Communications of the ACM, Vol. 30, No. 6, pp. 484-497
Markus, M. L. (1987), "Toward a 'Critical Mass' Theory of Interactive Media," Communication Research, Vol. 14, No. 5, Oct., pp. 491-511.
Markus, M. L. (1994), "Electronic Mail as the Medium of Managerial Choice," Organization Science, Vol. 5, No. 4, pp. 502-527.
Mata, F. J., W. L. Fuerst and J. B. Barney (1995), “Information Technology and Sustained Competitive Advantage: A Resource-Based Analysis,“ MIS Quarterly, Vol. 19, No. 4 , pp. 487-505.
McFarlan, F. W. (1984), "Information Technology Changes the Way You Compete," Harvard Business Review, Vol. 62, No. 3, May-June
Mead, G. H. (1934), “Mind, Self, and Society,” The University of Chicago Press.
Mintzberg, H. (1973), “The Nature of Managerial Work,” Harper Collins.
Oliver, P., G. Marwell and R. Teixeira (1985), “A Theory of the Critical Mass. I. Interdependence, Group Heterogeneity and the Production of Collective Action,” American Journal of Sociology, Vol. 91, No. 3, pp. 522-556.
Rice, R. E. and D. E. Shook (1990), “Relationships of Job Categories and Organizational Levels to Use of Communication Channels, Including Electronic Mail: A Meta-Analysis and Extension,” Journal of Management Studies, Vol. 27, No. 2, pp. 195-229.
Takeda, Y. (1996 spring), “Standardization of Processes and Media in Inter-organizational Transactions: A Case Study of Distribution of Industrial Goods,” Proceedings of the Conference of the Japan Society for Management Information.
Takeda, Y. and J. Kokuryo (1995), “Media Choice in Interorganizational Transactions: An Exploratory Study of the Distribution of Dry Groceries in Japan,” Proceedings of the 16th International Conference on Information Systems, 1995.
Trevino, L. K., R. H. Lengel, and R. L. Daft (1987), “Media Symbolism, Media Richness, and Media Choice in Organizations: A Symbolic Interactionist Perspective,” Communication Research, Vol. 14, No. 5, pp. 553-574.
Tushman, M. L. and D. A. Nadler (1978), “Information Processing as an Integrating Concept in Organizational Design,” Academy of Management Review, Vol. 3, pp. 613-624.
Williamson, O. E. (1975), "Markets and Hierarchies," Free Press.
Williamson, O. E. (1985), "The Economic Institutions of Capitalism," Free Press
Copyright (c) 1996 Yoko Takeda and Jiro Kokuryo