Structural Aspects

1

Travica * Chapter 3

Chapter 3

Structural Aspects

Organizational structure is a key segment in organizational design. In older organization theory, Organizational structure is organizational design. Although still maintained by some researchers, this equation appears to be untenable in light of advancements in organizational theory that have been made since the time of the “structural school” mentioned in Chapter 2 and later in this chapter (cf. Galbraith, 1973; Galbraith & Lawler, 1993). For example, organizational thought has been enriched with the “cultural school” (Schein, 1985); political view of organizations -- as important as ever – have been expanded through work of several important researchers (e.g, Mintzberg, 1983; Pfeffer, 1992); and it is hard to deny that information systems (e.g., a database or an email system) has nothing to do with organizational structure, or organizational values, or politics of an organization.

The present study defines organizational design in broader terms, which are inspired by these advancements in organization theory. Specifically, design refers to an arrangement of or a fit among structural, cultural, political and information aspects of an organization. So, design can be understood as the shape, or the form of an organization, that is determined by dimensions depicting various facets of the complex puzzle we call organization. What makes a difference between the concept of design used here and in some other studies, is the explicit inclusion of information dimensions in organizational designs. Another point to be noted is that a broad dichotomy of old versus new designs underpins the study. The latter category differentiates between, say, the virtual and the non-traditional organization. Such a differentiation has not been made within the category of old design, which here simply means “bureaucratic organization.” This is not to say that there is just one bureaucratic design; rather, differentiating between bureaucratic deigns is not of interest to this study.

The present study followed Fry's (1982) convincing argument that hierarchy, formalization, centralization and spatial dispersion sufficiently describe organizational structure. These dimensions originated in investigations of Joan Woodward and then were used extensively by the ASTON group that inaugurated the comparative organizational research of organizational structure (see Hall, 1987; Pugh & Hickson, 1989). Stronger hierarchy, formalization and centralization, as well as the positive correlations between these variables, are usually considered to be indications of the bureaucratic organization (cf. Price & Mueller, 1986; Weber, 1947). However, some industry-specific exceptions may occur with regard to centralization: the Aston studies, for example, suggest that centralization may not correlate significantly with other structural variables (see Pugh and Hinings, 1976). When placed in the perspective of the present study, this knowledge gives rise to an expectation that the non-traditional organization (a) needs to have hierarchy, formalization and centralization developed to a smaller extent, and (b) should demonstrate a pattern of the correlations of these three variables that is somewhat different from the one which typifies the bureaucratic organization. Finally, the dimension of spatial dispersion has been suggested both by Fry (1982) and the literature of computer mediated communication. The latter suggests that a higher spatial dispersion may be enabled by information-communication technology (see Hiltz, 1984; Hiltz & Turoff, 1978; Kerr & Hiltz, 1982; Sproul & Kiesler, 1991). The usual concept of spatial dispersion has been somewhat modified in this study for the purpose of investigating the potential for spatial dispersion of local offices, because local offices were usually not dispersed in actuality. No particular expectation was linked to the relationships of the variable with other structural dimensions. Concepts and measures for each structural variable will be discussed in the following discussion.

Hierarchy has been conceptualized as the extent of authority levels (Weber, 1947). A typical way of assessing hierarchy in past research was to count/average the managerial levels in organizational charts (see Blau & Shoenherr, 1971; Hall, 1987; Pugh & Hickson, 1989). This kind of measurement required gaining access to organizational documentation. However, it was learned from the pilot study, which was run prior to the main study, that such organizational documentation was not always accessible. Even more hindering, some organizations investigated did not have organizational charts or at best had inaccurate ones. These facts motivated the author to construct a new measure that would simply ask individual respondents about the number of management levels in a particular office. Also, there were other reasons for using this estimate of hierarchy. Specifically, the accounting literature pertaining to the modern local accounting office (“organization” in this study) typically has four management levels - senior associate, manager, partner, managing partner (Montagna, 1974; Stevens, 1981, 1985). If this is so, then hierarchy estimated from organizational charts may not vary from one office to another, resulting in measurement problems. One such problem would be the possible failure to capture real, “unofficial” ranks, such as “acting manager” or “acting partner” (see Montagna, 1974). Another problem would be ecological: an office level-measure could be attributed to the individual level. As a consequence, at the individual level, hierarchy would no longer act as a variable but rather as a constant, creating problems in correlational analysis. For all these reasons, the new measure of hierarchy mentioned above was preferable to the older, hierarchy chart-based measures.

Formalization was conceptualized in terms of the extent to which rules, procedures instructions and communications were written (Oldham & Hackman, 1981; Pugh et al., 1968). Formalization has been measured as, for example, the availability of particular documents (Inkson et al., 1976), the number of words in a random sample of documentation (Blau & Schoenherr, 1971), and a self-reported assessment of the extent of written regulations - rules, policies, availability of manuals, job descriptions, individual performance records (Oldham & Hackman, 1981). Difficulties anticipated concerning the ability to obtain access to accounting office documents made the Oldham and Hackman (1981) scale favorable to other measures. Their measure also appeared to be exhaustive and did fit the overall reliance on self-reported measures the study employed. On the down-side of the scale is a modest Cronbach Alpha of 0.52 (see Price & Mueller, 1986: 148). The scale used in the present study employed only four of five original items, because one item - as (Price & Mueller, 1986) rightly noted - made no reference to written documents.

Centralization has been conceptualized as the extent to which control is concentrated (Pugh et al., 1968; Tanenbaum, 1968). Centralization has been measured by a number of self-reported measures, such as assessments of the level of participation in decision making, autonomy at work, authority for making specific past decisions, and control graph (see Price & Mueller, 1986, 50-67). The control graph measure, whose author is Tannenbaum (Tannenbaum, 1961), asks respondents about the amount of "say or influence" in an organizational setting. Respondents are required to assess this amount of say or influence for organizational ranks that are specified in advance (e.g., from top executive through "me and my co-workers"). This scale was chosen for the present study because it contains a smaller number of items than any other scale, which saves time on the respondent’s side. The scale also provides rich information -- the distribution of influence across organizational ranks as well as the total amount of control (see Tannenbaum & Cooke, 1979). The reliability of the scale, however, seems to be on the trade-off side. Markham and colleagues (1984) found split-half reliability coefficients, ranging from 0.18 (for the slope) through 0.39 (for the "me and my co-workers" item). Still, this study did not completely falsify the measure and, therefore, further utilization of it seems to be warranted.

Spatial dispersion refers to the distribution of an organization’s units or operations in the geographical space (cf. Hall, 1987). In traditional organization theory, spatial dispersion is assessed by counting units of a particular organization that are somehow defined as operating sites (cf. Blau & Schoenherr, 1971). In the study of computer mediated communication, this dimension has been revamped so that dispersed individuals or teams are being investigated in relation with communication computers can; unit of analysis was, therefore, changed (cf. Hiltz, 1984). There are general problems in measurement of this dimension, concerning the definition of operational site (see Price & Mueller, 1986). Put another way, what spatial distance between two units of an organization qualifies them to be considered dispersed can be subject of the researcher’s arbitrary decision. The same problem applies to IT-related research to the extent that it focuses on individuals/teams that are bound to spatially separated facilities. A more complicated problem, however, occurs if the linkage between the person/team and the facility is broken, that is, the people are mobile.

The present study did not study actual spatial dispersion, but latent one, which poses in the attitudes of organization members. The thinking behind this approach was that actual differentiation of local accounting offices could be insignificant, while a potential for differentiation can be high. This potential was assumed to be associated with the capability of overcoming spatial limitations owing to IT (for example, portable computers) and was deemed important for understanding intricacies of design of these organizations. The study developed a scale for measuring this latent spatial dimension.

Predictions

Leavitt and Whisler (1958) speculated about the impact of computers on management in the 1980's. At the time they wrote the article, even the term “information technology” was new. The authors used this term to refer to computers that processed large amounts of data, applications of statistical and mathematical methods, and simulations of “higher-order thinking through computer programs” (Leavitt & Whisler, 1958, 41). In addition, they took into consideration middle and large size organizations, predicting three main organizational changes associated with computers: centralization, increased formalization of middle management jobs, and the shrinking of the middle management level. The first effect meant that top managers would gain more power in decision making due to the computer-supported centralization of organizational information. The formalization of middle manager jobs would occur as part of the formalization trend that started with Taylorism, which formalized the jobs of bottom-level managers. As a consequence, middle manager jobs could be encoded in computer programs, and this would in turn lead to a shrinking of these management levels. In his later work, Whisler subsequently developed the thesis that middle management shrinks with the employment of IT (Whisler, 1970). This thesis is supported by Dawson and McLaughlin's (1986) study (see above), and by evidence on the reduction of middle managers which started in the mid-1980's.

Another of Leavitt and Whisler's (1958) predictions is that innovators, and information specialists (data processing, statisticians, etc.), would work in amorphous groups, have significant freedom in designing their work, and their performance would be subject to peer evaluation rather than formal procedures. Whisler tested some of these propositions in his later study (Whisler, 1970; see the discussion below).

Olson and Lucas (1982) speculate about the impacts of office automation on individuals, groups and organizational units. They define office automation as the application of integrated computing and communication systems which could support administrative procedures in an office environment (p. 838). These systems include: text processing, electronic mail, teleconferencing, fax, electronic filing, electronic calendars, etc. The authors suggest a number of consequences that office automation might bring about, such as:

• an increase in managers’ span of control;
• an increase in informal communication and social links;

• expansion of communication beyond departmental boundaries;

• increased chances for conflict between organizational units due to expansion of communication spaces; and

• an increase in individuals' isolation, which can cause a greater degree of stress and fewer chances for promotion due to a lowered visibility of individuals. Olson has tested some of these propositions in her later empirical study (e.g., Olson & Bly, 1991), which is discussed later in this chapter.

Allen and Hauptman (1987) developed a conceptual model to represent the capability of IT to support the structure of project-driven research organizations. These authors used the term “communication-related information technologies” to refer to electronic mail, computer conferencing, electronic bulletin boards, and document retrieval systems. In explaining the purpose of the model, the authors point out that project-driven organizations employ a dual structure: one is functional, and the other is task-related. A functional organization groups people according to their specialties, while a task structure requires grouping according to the project at hand. People can stay in touch with information within their specialties, while in functional units. However, once they are assigned to project teams, people tend to lose touch with this information because of project requirements.

Allen and Hauptman suggest that these structural limitations and specialist information needs can be reconciled by the employment of IT. The needs for specialist information on the part of project people can be satisfied by using document retrieval systems, an expert-based, selective dissemination of information and bulletin boards or electronic discussion forums. At the same time, project members can get project-related information quickly by means of hierarchically organized bulletin boards for project status reporting and configuration control.

Wigand (1988) analyzes evidence on the effects of office automation and telecommunication, and predicts that organizations will become flatter in the future because the mid-level management will shrink. In addition, IT and satellite communications will reduce the importance of location, which in turn will enable organizations to become more spatially dispersed. The overall trend, contends Wigand, is that more flexible, less centralized, and more fluid organizational structures will emerge. All these structural changes are enabled by IT. Even more extensive results of the employment of IT are to be expected, concludes Wigand, when managers cease relating IT to traditional efficiency and begin thinking about employing IT capabilities in order to reshape organizations.

Drucker (1988) speculates about “information-based organization,” which depends upon the employment of IT and the new decision-making, management and work organization methods. Drucker calls the organization that prevails today the “command-and-control organization,” and argues that the information-based organization differs from it in possessing

• fewer management levels;
• smaller self-governing units (e.g., task forces) for the basis of work organization; and

• professionals that are capable of being self-disciplined in carrying out their responsibilities for jobs, relationships, communication and information.

Drucker finds examples of this organization both in the past and present. According to him, the British administration in India exemplifies a flat organization of professionals, while a symphonic orchestra and hospitals are modern examples of the information-based organization. In addition, Drucker points to potential problems of the information-based organization. One of these structural problems is that of devising a management structure for an organization of task forces; a cultural problem may be how to create a unified vision in an organization of specialists.

Huber (1990) developed a normative theory of the effects of advanced IT on organizational design, intelligence, and decision making. The part of the theory concerning organizational design is relevant for this review. By “advanced IT” Huber refers to devices that transmit, manipulate, analyze or exploit information, and in which a digital computer processes information integral to the user's communication or decision task (e.g., electronic mail, electronic bulletin boards, and decision-support systems). According to Huber's theory, the following effects are to be expected in connection with these technologies:

• a more uniform distribution of decision making across organizational levels (this is a centralization variable);

• a decrease in the number of authority levels (hierarchy); and
• a canceling effect on IT-related formalization.

The proposition about centralization is linked to two expectations in connection with the employment of IT: in more centralized organizations, centralization can decrease, while in highly decentralized ones centralization can increase. Huber points to evidence indicating that lower-level managers can use IT to stay better informed about the organization's overall situation, problems, policies, etc. But IT can also work towards the already existing trend, be it centralization or decentralization. Consequently, the proposition is expected to show correct “on balance” when the results of several studies are combined.

The rationale behind the hierarchy proposition is that IT can enable greater commonality of information and knowledge across hierarchical levels. Consequently, a given hierarchical level can become more qualified to apply more criteria and decision rules than it could without IT. In this way, fewer authority levels could be needed in authorizing a decision. Huber, however, warns that evidence of this reduction does not warrant separating the contribution of IT from that of other possible forces (e.g., attempts to reduce direct labor costs).

Regarding formalization, Huber (1990) has hypothesized that the long-term effects of IT on formalization associated with it could be nil: when IT is viewed from the perspective of a life cycle, the amount of rules regulating the use of a system can be high in the early stages, and can decrease as the system becomes familiar, approaching the degree of formalization associated with the old, replaced system. This is a correction of Huber’s (1984) earlier theorizing in which he posited that IT can trigger requirements for formalizing the handling of non-routine information, the scope and amount of environmental information, and decision making. The neutral role of IT Huber (1990) proposed later is in agreement with the finding of Pfeffer and Leblebicci (1977) discussed in this chapter.

Inconclusive Evidence

Empirical studies on effects of IT are agreeable with the prediction discussed above, if one takes these in a summative fashion. At the level of single study, however, deviations from predictions are extant, making evidence inconclusive.

With regard to hierarchy, a number of researchers have found that the telephone can make hierarchical relations more democratic (see Pool, 1983). For example, a foreman can talk to a president of Steel Trust if the job demands it (p. 42). It has also been observed that the use of the telephone can facilitate the centralization of decision making at the upper management levels. The former occurred at the turn of the century when the introduction of the telephone contributed to the field manager's losing his autonomy in ongoing operations because his remote supervisor used the telephone for issuing orders.