Untitled-14 Information As a Strategic Contingency 145 145 Information As a Strategic Contingency: Applying the Strategic Contingencies Theory of Intraorganizational Power to Academic Libraries Gregory A. Crawford is Head of Public Services in the Heindel Library at Penn State Harrisburg; e-mail: gac@psulias.psu.edu. Gregory A. Crawford This research examines the changes that electronic information tech- nologies have caused on power within organizations. Based on the stra- tegic contingencies theory of intraorganizational power, a model of or- ganizational power is developed and tested. Major features of the model include a technology index, subunit power variables, environmental vari- ables, and bases of power variables (structure, coping with uncertainty, centrality, and substitutability). According to canonical correlation analy- ses, changes in library automation and changes in the environment are related to changes in both the bases of power variables and power itself. The bases of power, in turn, are related to changes in the power of the library as measured by the percent of the institutional budget allocated to the library, the number of library positions, and the percep- tion of power. zational issues, and issues involving the role of the library.2 Finding the money to pay for infor- mation technology (IT) is a great chal- lenge for most libraries. The costs of pur- chasing, maintaining, and replacing com- puter equipment; hiring computer ex- perts or consultants; training librarians and other library staff to use the technolo- gies; and acquiring machine-readable records have stretched the budgets of li- braries greatly. Computing and electronic informa- tion technologies may cause changes in echnology is having a major impact on academic libraries, causing librarians to rethink their positions on collections (e.g., access versus ownership), budgets (e.g., paying for computer hardware and software), buildings (e.g., space needs and wiring for telephone and data connections), staff (e.g., training for new technologies and changes in roles), and services (e.g., educating users about new technologies). 1 As a result, libraries and librarians must face financial issues, social and organi- 146 College & Research Libraries March 1997 certain power payoffs or �power shifts� within organizations.3 Often these shifts alter the political atmosphere of an orga- nization, increasing the power of those who control the decision-making appa- ratus that distributes needed resources within the organization. This study sought to understand the role that electronic information technolo- gies have played in the distribution of power within organizations, specifically, the power of the library within the lib- eral arts college. The strategic contingen- cies theory of intraorganizational power provided the theoretical base for this re- search. Within this theory, power is con- sidered a dependent variable that changes in response to a subunit�s bases of power. The bases of power include the subunit�s ability to cope with uncertainty, its substitutability, and its centrality to the organization.4 A strategic contingency is defined as �a requirement of the activi- ties of one subunit which is affected by the activities of another subunit.�5 Infor- mation is one such strategic contingency. As developed by D. J. Hickson et al., the strategic contingencies theory states that the control of contingencies needed by other subunits within the organization is related to the power of the controlling subunit. The more necessary these con- tingencies are for the work of other sub- units, the more power accrues to the con- trolling subunit. Within this study, the exercise of power is specifically concerned with marshaling resources for use within a subunit (the library) of an organization (the liberal arts college). Resources are defined as both monetary (as indicated by subunit budget allocations) and per- sonnel (as indicated by the number of subunit staff ). In addition, the perception of power on the part of the library direc- tor is included. In addition to power variables, the proposed integrated model includes vari- ables considered to be the bases of power. These include both structural variables and variables associated with the strate- gic contingencies theory of organiza- tional power, namely uncertainty, cen- trality, and substitutability. Because power, according to Jeffrey Pfeffer, is first and foremost a structural phenomenon, various structural variables are measured to determine their changes in response to changes in technology.6 Many previ- ous studies confirm the importance of the existing organizational structure as a base of power.7 Uncertainty can be defined as �a lack of information about future events, so that alternatives and their outcomes are unpredictable.�8 Such uncertainties can involve the sources and composition of inputs, the work flow or production pro- cess, and the market for products. Hickson et al. defined centrality of a subunit as �the degree to which its activi- ties are interlinked into the system.�9 In her research on colleges and universities, Judith D. Hackman defined centrality as how closely a unit matches the central mission of its parent institution and found that a unit�s centrality crucially affects the internal resources allocated to it by the institution.10 Similarly, Richard H. Hall states that from his observations at a number of universities, �the central- ity of the operation and the scarcity of personnel are major determinants of the power of a particular organizational unit.�11 The final factor included in the con- tingency theory of power is that of sub- stitutability. Hickson et al. defined this term as �the ability of the organization to obtain alternative performance for the activities of a subunit.�12 In other words, the functions of one subunit can be taken over by another subunit. Robert Dubin notes that power within a formal orga- nization is based on the importance of the functions performed by the subunit and the exclusiveness with which it per- forms them.13 The less the activities of one subunit can be taken over by another, the greater that subunit�s power. Information As a Strategic Contingency 147 Integrated Model of Intraorganizational Power The model of intraorganizational power proposed and tested in this study seeks to integrate the strategic contingencies model of intraorganizational power with concepts drawn from the literature of the effects of technology on organizations. Figure 1 presents the integrated model. The dependent variables are three aspects of subunit power; the major independent variables include a technology index and environmental variables; and the inter- vening variables are the bases of power. For this research, technology�specifi- cally, information technology�is the major independent variable. According to the model developed by Carol S. Saunders and Richard W. Scamell in their studies of management information sys- tems (MIS), as MIS usage increases, the power of the subunit that controls the MIS also increases due to the increase in that unit�s centrality, its ability to cope with un- certainty, and its nonsubstitutability.14 Similarly, Andrew M. Pettigrew and Pfeffer and Huseyin Leblebici argue that information access and the control of in- formation technologies are power re- sources.15 Various authors within the li- brary literature have also shown that au- tomation is a change agent within librar- ies.16 The current integrated model also in- cludes variables considered to be the bases of power. These include both struc- tural variables and variables drawn from the strategic contingencies theory of or- FIGURE 1 Integrated Model of Intraorganizational Power Time 1. Time 1 2. Time 2 Information technology Total library automation index Bases of power: A. Structural variables Power: 1. personnel classed as prof. (Dependent variables) 2. personnel classed as other 1. percentage of inst. budget B. Coping with uncertainty 2. number of subunit positions 3. workload/service measure 3. perceived power 4. reference transactions C. Centrality 5. perceived centrality D. Substitutability 6. functions 7. personnel 8. collection development Organizational Environment 1. Potential demand for service (students who could use the library—i.e., size of institution) 2. Collection size 1) titles 2) subscriptions 148 College & Research Libraries March 1997 ganizational power.17 In keeping with the strategic contingencies theory, power is considered to be the main dependent variable. Finally, the model posits that time is an important variable in its own right. The study is longitudinal so that the ef- fects of technology may appear within both the bases of power and the power measures themselves. The environmental variables provide a way to account for influences on the power of the library that lie outside the theories used to construct the integrated model. The size of the user population and the size of the collection are impor- tant environmental variables that must be considered. In brief, the model states that as tech- nology within the subunit changes over time, the bases of power react to those changes, and that as the bases of power change, the power of the subunit changes. As a result, the manifestations of power as revealed by the power vari- ables also change. Research Question One overarching research question guides this research: How has the control of ITs affected power within organizations? In particular, the research question can be made more focused by asking: How has the control of library-related ITs affected the library�s power within the liberal arts college? Methodology The unit of analysis for this study was the library within the liberal arts college. The population frame consisted of all those colleges classified as Liberal Arts Colleges I or Liberal Arts Colleges II by the Carnegie Foundation for the Ad- vancement of Teaching.18 Only those col- leges that completed both the 1982 HEGIS and the 1990 IPEDS surveys and that were not part of a multilibrary re- porting group were included in the study. The final total population frame for the study was 487 institutions. Variables in the Model Most of the variables in the tested model used two measures (Time1 = 1981�82 aca- demic year and Time2 = 1989�90 aca- demic year) to examine changes occur- ring over this time span, although a few measured perceptions of change over the time period using only one. The time periods included in the study reflect the data available from the Higher Education General Information Survey (HEGIS) in 1982 and its 1990 replacement, the Inte- grated Post-Secondary Educational Data System (IPEDS). These two dates pro- vided a basis for studying the changes in libraries during the 1980s, a period in which great strides were made in library automation. The dependent variables were three measures of power: (1) percentage change in the institutional budget allo- cated to the library, (2) percentage change in the number of personnel in the library, and (3) perceived change in the power of the library. Independent variables included the technology index and the environmen- tal variables. Intervening variables in- cluded the structural variables, coping with uncertainty variables, centrality variables, and substitutability variables. The technology index was developed recursively using existing measures and various data sources for library automa- tion.19 The library technology index was calculated simply by adding together the number of technologies incorporated into the library over the time span stud- ied. These included automated acquisi- tions, automated serials, automated cata- loging, automated circulation, online In keeping with the strategic contingencies theory, power is considered to be the main dependent variable. Information As a Strategic Contingency 149 public access catalog, network utility membership (OCLC, RLIN, WLN, etc.), network-based interlibrary loan, local area network within library, node on campus network, local/regional consor- tia or networks, telefacsimile, public computing workstations, CD-ROM in- dexes, reference database searching, and librarian/staff workstations. The bases of power consisted of two major groups of variables: structural vari- ables and contingency variables. The con- tingency variables included three sub- groups: coping with uncertainty vari- ables, centrality variables, and substitut- ability variables. These variables acted as dependent variables for analyses involv- ing the technology indexes and environ- mental variables, as independent vari- ables for analyses of the power variables, and as intervening variables in analyses of the entire model. The structural variables examined the personnel of the subunit and consisted of two subvariables: percentage change in the number of personnel classed as professional and percentage change in the number of personnel classed as other. �Other� personnel included those indi- viduals not classed as professional, such as support staff, nonprofessionals, and clerical workers. Coping with uncertainty was defined as involvement within the work flow of the organization. Two coping with un- certainty measures were included: workload of the subunit (percentage change in the circulation per staff mem- ber) and instructional service of the sub- unit (percentage change in the number of reference transactions). Centrality examined the ability of the library to support the primary mission of the college. This research used a single measure of centrality�perceived change in centrality. Substitutability referred to the ability of other campus subunits to perform functions similar to those performed by the library. The three substitutability vari- ables measured changes in the ability of other subunits to perform the library�s information functions, the difficulty in hiring technologically qualified librar- ians, and the collection development re- sponsibility of librarians. The environmental variables exam- ined changes within the subunit�s envi- ronment and included change in poten- tial demand for service (number of stu- dents) and change in collection size as measured by changes in the total title count and the number of periodical sub- scriptions. Sources of Data The sources of data included a mailed questionnaire and archival data. The questionnaire sought data for the follow- ing variables: amount of library automa- tion; perceived change in power; per- ceived change in centrality; and per- ceived change in substitutability of func- tions, personnel, and collection development responsibility. Seven-point Likert scales were used for perceived change in power, centrality, and substi- tutability (three measures). A coin flip decided whether each question stressed an increase or a decrease in the variable under consideration. For the technology index, a list of technologies was pre- sented and respondents indicated the year when each technology was made available in their libraries. After pretesting, the questionnaire was sent to the directors of the 487 identified libraries, except those directors who par- ticipated in the pretest. If it was not re- turned within eight weeks of mailing, another copy was sent by mail. Usable responses were received from a total of 416 institutions (85.4%). The HEGIS and the IPEDS surveys This research used a single measure of centrality�perceived change in centrality. 150 College & Research Libraries March 1997 provided data for the following variables for both Time1 and Time2: library bud- get as a percentage of the institutional general and educational budget, number of positions, potential demand for ser- vice (size), collection size (title and sub- scription counts), classification of person- nel (professional and other), and workload/service of the library (circula- tion and reference transactions). Methods of Analysis Full and partial canonical correlation analyses were used to test the model which consisted of independent (auto- mation indexes and environmental vari- ables), intervening (bases of power), and dependent (power) variables. The signifi- cance level was set at a = .05. In canonical correlation, each set of variables (i.e., the independent, interven- ing, and dependent variables) may be represented by at least one linear combi- nation of variables, called a canonical vari- ate. The correlation between these canoni- cal variates is called the canonical corre- lation. Thus, canonical correlation analy- sis examines the relationship between two sets of variables and also can be used to partial out a third set of variables. Table 1 provides descriptive statistics for the variables included in the study including the mean, standard deviation, maximum, minimum, and number of cases for which data were available. Test of the Proposed Model The proposed model was tested using a set of three canonical correlations. Table 2 provides the detailed statistics for all three steps of the analysis. In the first step, the canonical correla- tion included the automation index and the environmental variables as indepen- dent variables and the bases of power variables as the dependent variables. The canonical correlation analysis yielded one significant canonical correlation be- tween the automation index and the en- vironmental variables and the bases of power (Wilks� Lambda = .77, F = 2.03, df = 32/901.4, p < .001). The amount of vari- ance of the bases of power variables ex- plained by automation and the environ- TABLE 1 Summary Descriptive Statistics for Variables Variable Mean Std. Dev. Min. Max. N Automation Index Total automation 5.06 2.91 0 14 416 Bases of Power Variables Change in professionals .23 .68 -1.00 9.00 487 Change in other staff .34 1.31 -1.00 14.00 457 Change in circulation 5.78 129.20 -1.00 2830.40 480 Change in reference transactions 1.61 8.41 -1.00 141.86 444 Perceived centrality 4.74 1.47 1 7 375 Substitutability of functions 2.96 1.87 1 7 382 Substitutability of personnel 3.42 1.65 1 7 378 Substitutability of collection 2.32 1.50 1 7 390 development Environmental Variables Change in number of students .29 .80 -.42 10.40 476 Change in number of titles .31 2.09 -.85 41.24 405 Change in number of subscriptions 1.59 16.62 -1.00 306.69 463 Information As a Strategic Contingency 151 TABLE 2 Canonical Correlation Analyses of Integrated Model: Standardized Structure Coefficients (Variate Loadings) Step One Canonical Variate: 1 Independent Variables (Automation Index & Environmental Variables) Total automation .30 Change in students .64 Change in titles .67 Change in subscriptions -.06 Dependent Variables (Bases of Power Variables) Change in professionals .73 Change in other .57 Change in circulation .20 Change in ref. trans. .35 Perceived centrality .16 Subs. of functions -.21 Subs. of personnel -.16 Subs. of col. dev. .47 Squared Canonical Correlation: .12 Wilks’ Lambda: F 32,901.4 = 2.03* *p<.001 Step Two Canonical Variates: 1 2 3 Independent Variables (Bases of Power Variables) Change in professionals .60 .16 .70 Change in other .92 .01 -.45 Change in circulation -.01 .02 -.25 Change in ref. trans. .01 -.04 -.13 Perceived centrality -.07 .95 -.22 Subs. of functions .06 .02 .05 Subs. of personnel .01 .07 -.01 Subs. of col. dev. .06 -.01 -.27 Dependent Variables (Power Variables) Change in budget -.09 .04 1.05 Change in staff 1.02 .06 -.19 Perceived power -.02 .99 -.21 Squared Canonical Correlation: .76 .58 .04 Wilks’ Lambda: F 24,850.4 = 44.7* F 14,588 = 24.6* F 6,295 = 2.2** *p<.001, **p<.05 (continued) 152 College & Research Libraries March 1997 ment was small (canonical correlation = .28, variance explained = 12%). The canonical variate for the indepen- dent variables was characterized by the following significant loadings: total li- brary automation index, .30; change in the number of students, .64; and change in the number of titles, .67. For the ca- nonical variate of the dependent vari- ables, the highest loadings were change in the number of professionals (.73), change in the number of other staff mem- bers (.57), change in reference transac- tions (.35), and change in collection de- velopment responsibility (.47). In step two of the analysis, the bases of power were considered the indepen- dent variables and the power variables were the dependent variables. The analy- sis yielded three significant canonical correlations. The first canonical correla- tion (canonical correlation = .87, Wilks� Lambda = .09, F = 44.7, df = 24/850.4, p <. 001) explained the most variance (76%); the second explained 58 percent of the variance (canonical correlation = .76, Wilks� Lambda = .40, F = 24.6, df = 14/588, p <. 001); and the third explained 4 percent (canonical correlation = .21, Wilks� Lambda = .96, F = 2.2, df = 6/295, p <. 05). The standardized canonical coeffi- cients for the dependent variables (i.e., the power variables) showed that each variate had only one high loading: vari- ate 1, change in the number of staff (1.02); variate 2, perceived power (.99); and vari- ate 3, change in the budget allocation (-1.05). For the independent variables (i.e., the bases of power variables), the highest loadings for the first canonical variate were change in the number of professionals (.60) and change in the number of other staff members (.92), showing a strong relationship between the change in the number of profession- als and the change in the number of other staff members variables with the overall change in staff power variable. For the second variate only perceived centrality loaded highly (.95), exhibiting a strong relationship with perceived power. Fi- nally, for the third variate both change in the number of professionals (.70) and TABLE 2, cont. Canonical Correlation Analyses of Integrated Model: Standardized Structure Coefficients (Variate Loadings) Step Three (Bases of Power Partialed Out) Canonical Variate: 1 2 Independent Variables (Automation Indexes and Environmental Variables) Total automation .19 .99 Change in students -.70 .27 Change in titles .60 -.16 Change in subscriptions -.23 .20 Dependent Variables (Power Variables) Change in budget .46 .28 Change in staff -.69 .75 Perceived power .63 .48 Squared Canonical Correlation: .10 .05 Wilks’ Lambda: F 12,627.3 = 3.5* F 6,476 = 2.5** *p<.001, **p<.05 Information As a Strategic Contingency 153 change in the number of other staff mem- bers (-.45) exhibited high loadings and were significantly related to the change in budget variable. The third step of the analysis used the automation indexes and the environmen- tal variables as the independent variables and the power variables as the depen- dent variables with the bases of power variables being partialed out of the analy- sis. Two canonical correlations were sig- nificant: canonical correlation = .32, Wilks� Lambda = .84, F = 3.5, df = 12/ 627.3, p < .001, 10 percent variance ex- plained; and canonical correlation = .23, Wilks� Lambda = .93, F = 2.5, df = 6/476, p < .05, 5 percent variance explained. The variates representing the independent variables were characterized by high loadings of change in the number of stu- dents (-.70) and change in the number of titles (.60) for the first variate and total library automation (.99) for the second. All three power variables loaded highly on the first variate of the dependent vari- able: change in budget (.46), change in the number of staff (-.69), and perceived power (.63). For the second dependent variate, two of the power variables loaded highly: change in the number of staff (.75) and change in perceived power (.48). This step of the analysis showed that, although there is a significant rela- tionship between automation and the environment and the power of the library, the effects are small. Limitations This research has several limitations. First, the study was conducted using data from only one broad classification group of academic institutions. As a result, the findings may not be generalizable to other types of academic libraries or to other types of libraries. Second, the time span used as the ba- sis of the study may not have provided a long enough time for changes caused by automation to have become appar- ent. The decade of the 1980s, however, was one of great changes in comput- ers and library automation. Some dif- ferences due to automation may not have had time to manifest themselves over this time period. Third, the underlying economic con- ditions of the institutions included in the study may have affected its results, es- pecially the changes in budget percent- age and staff power variables. Fourth, the model proposes causality when the relationships between variables may merely be correlational. Fifth, much of the research upon which the tested model was built was drawn from the administration and MIS literatures. The conclusions of such re- search that ITs have acted as change agents within organizations, especially businesses, may not be applicable to the academic library setting. Finally, other exogenous variables not included in the model may have caused any changes that were identified. Such variables may include those that exam- ine the economic condition of the colleges and the leadership styles of the college administrators and head librarians, both of which could affect the funding of the library significantly. Conclusion The driving question behind this research was: How has library automation af- fected the academic library and its posi- tion on college campuses, echoing the statement of Kenneth L. Kraemer and John L. King that �The fundamental question about computing and organi- zational politics is who gains and who loses from computing.�20 The present The main idea expressed in the proposed model was that as auto- mation within a library changed, the bases of power of that library also changed. 154 College & Research Libraries March 1997 study developed and tested a model of intraorganizational power based largely upon the strategic contingencies theory of intraorganizational power. The strate- gic contingencies theory states that power within an organization can be viewed as a function of a subunit�s abil- ity to cope with uncertainty, its central- ity to the organization, and the substi- tutability of its functions and personnel within the organization. The tested model also added automation, structural, and environmental variables to provide an in- tegrated model of intraorganizational power. The main idea expressed in the pro- posed model was that as automation within a library changed, the bases of power of that library also changed. Such changes in the bases of power would then cause changes in the power of the library on campus. The analyses showed that automation and the environment affected specific bases of power variables posi- tively. In accordance with the strategic contingencies theory, the bases of power variables showed significant relation- ships with the power variables. The analyses also revealed a significant posi- tive relationship between automation and the environment and power. These results provide limited support for the major theory presented by this research, namely, that automation is a change agent within libraries. In the analyses presented in this study, the ef- fect of automation, though significant, was weak. However, the core of the stra- tegic contingencies theory was upheld with the bases of power accounting for 76 percent of the variance of the power variables. Understanding the effects of library automation and obtaining a better view of the nature of power should prove to be beneficial to the directors of liberal arts college libraries. In various comments given by library directors in the course of this research, the directors believed that automation had been a major help in providing the library with increased power. Although automation was shown by the analyses to influence the bases of power, it had only a weak direct relation- ship to power itself. This research provides one way to study the relationship between IT and power. The data presented and the results generated by them can help enlighten li- brary directors as to the state of the field and provide insight into the effects that ITs have had on subunit power within the college. As Pfeffer says, �Power and politics are often part of organizations, and need to be understood as fundamen- tal and important processes.�21 This re- search is one step toward such an un- derstanding within liberal arts college libraries. Notes 1. Barbara B. Moran, �The Unintended Revolution in Academic Libraries: 1939 to 1989 and Beyond,� College & Research Libraries 50 (Jan. 1989): 25�41. 2. William G. Potter, �Insurmountable Opportunities: Advanced Technology and the Aca- demic Library,� in Academic Libraries: Research Perspectives, eds. Mary Jo Lynch and Arthur Young (Chicago: ALA, 1990), 165�91. 3. Kenneth L. Kraemer and William H. Dutton, �The Automation of Bias,� in Computers and Politics: High Technology in American Local Governments, eds. James N. Danziger, William H. Dutton, Rob Kling, and Kenneth L. Kraemer (New York: Columbia Univ. Pr., 1982), 170�93; Alvin Toffler, Powershift: Knowledge, Wealth, and Violence at the Edge of the 21st Century (New York: Bantam Bks., 1990). 4. D. J. Hickson, C. R. Hinings, C. A. Lee, R. E. Schneck, and J. M. Pennings, �A Strategic Contingencies� Theory of Intraorganizational Power,� Administrative Science Quarterly 16 (June 1971): 216�29; Ran Lachman, �Power from What? A Reexamination of Its Relationships with Struc- tural Conditions,� Administrative Science Quarterly 34 (June 1989): 231�51; Gerald R. Salancik and Information As a Strategic Contingency 155 Jeffrey Pfeffer, �Who Gets Power�and How They Hold on to It: A Strategic-Contingency Model of Power,� Organizational Dynamics 5 (winter 1977): 3�21. 5. Hickson et al, �A Strategic Contingencies� Theory,� 222. 6. Jeffrey Pfeffer, Power in Organizations (Cambridge, Mass.: Ballinger, 1981). 7. W. Graham Astley and Paramjit S. Sachdeva, �Structural Sources of Intraorganizational Power: A Theoretical Synthesis,� Academy of Management Review 9 (Jan. 1984): 104�13; Clyde C. Caufield, �An Integrative Research Review of the Relationship between Technology and Struc- ture: A Meta-analytic Synthesis� (doctoral diss., Univ. of Iowa), Dissertation Abstracts International 51 (1989): 553A; C. R. Hinings, D. J. Hickson, J. M. Pennings, and R. E. Schneck, �Structural Condi- tions of Intraorganizational Power,� Administrative Science Quarterly 19 (Mar. 1974): 22�44; Henry Mintzberg, The Structuring of Organizations: A Synthesis of Research (Englewood Cliffs, N.J.: Prentice-Hall, 1979); Max Weber, The Theory of Social and Economic Organizations, trans. by A. M. Henderson and T. Parsons (New York: Oxford Univ. Pr., 1947); Joan Woodward, Industrial Organi- zation: Theory and Practice (New York: Oxford Univ. Pr., 1965). 8. Hickson et al, �A Strategic Contingencies� Theory,� 219. 9. Ibid, 221. 10. Judith D. Hackman, �Power and Centrality in the Allocation of Resources in Colleges and Universities,� Administrative Science Quarterly 30 (Mar. 1985): 61�77. 11. Richard H. Hall, �Power in an Academic Setting,� in Power in Organizations, ed. Mayer N. Zald (Nashville, Tenn.: Vanderbilt Univ. Pr., 1970), 50. 12. Hickson et al, �A Strategic Contingencies� Theory,� 221. 13. Robert Dubin, �Power, Function, and Organization,� Pacific Sociological Review 6 (spring 1963): 16--24. 14. Carol S. Saunders, �Management Information Systems, Communications, and Department Power: An Integrative Model,� Academy of Management Review 6 (July 1981): 431�42; Carol S. Saunders and Richard W. Scamell, �Organizational Power and the Information Services Depart- ment: A Reexamination,� Communications of the ACM 29 (Feb. 1986): 142�47. 15. Andrew M. Pettigrew, �Information Control As a Power Resource,� Sociology 6 (May 1972): 187--204; Jeffrey Pfeffer and Huseyin Leblebici, �Information Technology and Organizational Struc- ture,� Pacific Sociological Review 20 (Apr. 1977): 241�61. 16. Moran, �The Unintended Revolution�; Lewis. D. Cartee Jr., �Is Library Automation Pro- ducing a New Kind of Manager?� Journal of Library Administration 13 3/4 (1990): 99�115; Michael Gorman, �The Organization of Academic Libraries in the Light of Automation,� Advances in Li- brary Automation and Networking 1 (1987): 151�68; Peggy Johnson, Automation and Organizational Change in Libraries (Boston: G. K. Hall, 1991); Verna L. Pungitore, �Development and Evaluation of a Measure of Library Automation,� Library and Information Science Research 8 (Jan.�Mar. 1986): 67�83. 17. Hickson et al., �A Strategic Contingencies� Theory�; Salancik and Pfeffer, �Who Gets Power�; Pfeffer, Power in Organizations; Pfeffer, �Power and Resource Allocation in Organizations,� in New Directions in Organizational Behavior, eds. Barry M. Staw and Gerald R. Salancik (Chicago: St. Clair Pr., 1977), 235�65; Hackman, �Power and Centrality�; Saunders, �Management Information Sys- tems, Communications, and Department Power�; Carol S. Saunders, �The Strategic Contingen- cies Theory of Power: Multiple Perspectives,� Journal of Management Studies 27 (Jan. 1990): 1�18. 18. Carnegie Foundation for the Advancement of Teaching, A Classification of Institutions of Higher Education (Princeton, N.J.: Carnegie Foundation for the Advancement of Teaching, 1987). 19. Pungitore, �Development and Evaluation of a Measure of Library Automation�; Maxine K. Sitts, ed., Automation Inventory of Research Libraries (Washington, D.C.: ARL, 1987); Beverly K. Duval and Linda Main, Automated Library Systems: A Librarian�s Guide and Teaching Manual (Westport, Conn.: Meckler, 1992). 20. Kenneth L. Kraemer and John L. King, �Computing and Public Organizations,� Public Ad- ministration Review 46 (Nov. 1986): 492. 21. Pfeffer, Power in Organizations, 370.