The institutionalization of computing in complex organizations

Informatization and the Public Sector 2 (1992) 47–73 Elsevier

The institutionalization of computing in complex organizations James L . Perry Indiana University, Bloomington, IN, USA

Kenneth L. Kraemer, John Leslie King and Deborah Dunkle University of California, Irvine, CA, USA Received March 1992

Perry, J .L ., Kraemer, K.L., King, J .L . and Dunkle, D ., 1992, The institutionalization of computing in complex organizations . Informatization and the Public Sector 2 : 47–73. Abstract. Complex organizations spend significant resources acquiring new processes and technologies . However, there is no assurance that organizational members will embrace these technologies once they are adopted . Organizational leaders have a strong incentive to ensure that new practices persist, and that technological changes become so routine as to be taken for granted : a process called institutionalization . The research reported here examines the determinants of institutionalization of computing in a class of complex organizations—municipal governments in the United States—using longitudinal data from 130 organizations collected in 1975 and 1985 . The findings indicate that institutionalization is associated with both internal and external factors, but that the strength of internal factors is less prominent than suggested by earlier, cross-sectional studies.

Introduction Organizations are continuously engaged in replacing existing routines with new routines for performing organizational tasks (Feldman, 1988) . Among important routines are those concerned with use of "core technologies" (Thompson, 1967). In recent decades the opportunities created by information technologies, including computers, office automation and data communications, have compelled Correspondence to : K .L . Kraemer, Center for Research on Information Technology and Organizations (CRITO), University of California, Irvine, CA 92717, USA. 0925-5052/92/$05 .00 © 1992 – Elsevier Science Publishers B .V . All rights reserved

47

J.L . Perry et al. / Computing in complex organisations

many organizations to make large-scale investments in technology. The process of implementing these new technologies often has forced organizations to reevaluate existing ways of performing tasks (Beniger, 1986 ; Kraemer et al., 1989 ; Huber, 1990) . Still, organizational leaders confront considerable uncertainty about how to merge work behavior and technological requirements, and cannot be sure whether their members will conform to new prescriptions for their work behavior when the technology and work prescriptions are implemented. The failure of the organization's members to conform to and embrace the values embedded in the technology raises the interesting research question of why even carefully planned technological reforms often do not meet the critical expectation of institutionalization, wherein a technology becomes a routine and background part of everyday organizational life . More practically, technological innovations that do not become institutionalized cannot be productive assets of an organization, and failure to achieve institutionalization of innovations can impose enormous direct, indirect, and opportunity costs on the organization. Thus, there are both academic and practical reasons to investigate the factors that influence institutionalization of technology use in organizations. This study explores the factors associated with successful institutionalization of computer technology using longitudinal data collected from 130 us municipal governments between 1975 and 1985 . The paper begins with conceptual foundations drawn from innovation research and institutional theory . These are used to construct a set of hypotheses regarding relationships between institutionalization of technology use as the dependent variable, and a set of environmental and intra-organizational factors considered as independent variables . The results show that both environmental and intra-organizational factors are associated with institutionalization, though not all factors produced significant results. The results are interpreted in light of the conceptual foundation and qualitative assessments of information drawn from other sources.

Theoretical framework The first source for conceptual foundations for the research is the field of innovation adoption and diffusion . This field of research is among the broadest in behavioral science, engaging individual, group, organizational and social levels of behavior (Rogers, 1983) . Early innovation research focused primarily on how innovations came to be adopted by individuals and groups, and thus became diffused across populations. In recent years, attention has focused increasingly on the processes by which innovations come to be abandoned or permanently established in use . These studies have used different terms to describe their objectives : Yin (1981) concentrated on "routinization" of innovations, Glaser (1981) investigated "durability," while Gasser (1984) studied "persistence ." Most recently, Goodman and Steckler (1989) used the term "institu48

Informatization and the Public Sector, Vol. 2, No . 1

J.L. Perry et at. / Computing in complex organisations

tionalization" to refer to the process by which innovations become accepted as part of the organization . Although no comprehensive statement has been written to unite these diverse usages of the term, the theme common to all these dependent variables is that of survival over time. The convergence of innovation research on the question of survival, persistence, and acceptance as part of innovative practices parallels the development of the field of institutional theory in organizational sociology (Scott, 1987) . The concept behind institutional theory was clearly expressed by Hughes half a century ago : an institution is a persistent feature of social life that outlasts social participants and survives upheavals in the social order (Hughes, 1939), and institutionalization is the process by which a particular feature becomes an institution . A feature can be any focus of social inquiry, including practices, rules, and belief systems . There is at present no precise definition of what features are appropriate for study under the institutional theory rubric . In fact, the area has not gelled intellectually to the point where it serves as a uniform theoretical base for inquiry . Scott (1987) identified four separate views within institutional theory, and called for a more coherent development of the foundations for this interesting and emerging field. A comprehensive comparison and assessment of the various uses of the institutionalization concept is needed, but such a work is beyond the scope of this paper . Here we discuss a specific aspect of institutionalization arising from the innovation research tradition . We build on the contributions of institutional theory, however, and our conclusions are constructed in light of that perspective. Institutionalization of innovative practices This paper deals with the institutionalization of an innovative practice—the use of administrative computing systems—within individual organizations over time. Our use of the term extends that of Yin's (1981) concept of innovation routinization, meaning incorporation into regular use . Our use is also close to the concept used by Kraemer et al . (1987) to describe the routine use of computing innovations in federal agencies . Institutionalization in this paper describes the process by which the practice of using an innovation takes its place among the established values, norms, and beliefs that have been internalized among members of an organization (Kimberly, 1979) . It corresponds directly to the Goodman et al . (1980) notion that an institutionalized innovation has diffused widely, has been subject to socialization efforts to encourage use, and has been accepted as part of everyday organizational practice. Our use of the term benefits from three useful notions taken from institutional theory (Scott, 1987) . First, the practice of computing is considered institutionalized not when routine use arises due to imposition or directives, nor when use is the product of specific innovation efforts (although both may play a Informatization and the Public Sector, Vol . 2, No . 1

49

J.L . Perry et al . / Computing in complex organisations

role in institutionalization), but rather when routine use is accompanied by a general sentiment among organizational actors that use of computer automation is taken-for-granted in organizational life (Zucker, 1983) . Innovations are institutionalized when they are more in the "background" than in the "foreground ." Second, organizations can arrive at institutionalization of computing use through different processes, and individual patterns of institutionalization processes can be assumed to conform to individual organizations' particular needs for legitimacy, resources, and survival potential at any given time (Meyer and Rowan, 1977 ; DiMaggio and Powell, 1983) . Third, the process of institutionalization frequently involves conflict within and across organizations, due to differences in views about the appropriateness of the innovation, the particular configuration of the innovative practices, and so on (Friedland and Alford, 1987). These notions from institutional theory are helpful because they lend strength to an emerging view of the practice of computing use as organizational assimilation of complex "packages" of interdependent components (Ellul, 1964 ; Illich, 1973 ; Kraemer et al., 1981, 1989 ; Kling and Scacchi, 1982) . The package includes technology (hardware, software, peripherals), technique (procedures, protocols, practices, organizational arrangements), and knowledgeable people (computer specialists, functional users, and managers) . Moreover, the package is neither stable nor static; it constantly changes with new technological developments, attrition of key organizational actors, and changes in organizational objectives and needs. The "package" view emerged to replace simpler evolutionist concepts in which computing innovations are seen as tools with attributes that "lead" organizations to undergo a process of adoption, implementation, and use . These earlier concepts are best represented by the "stage" models of Nolan (1973) and Glaser et al. (1983), which obtained considerable currency among academics and practitioners of organizational computing from the 1970s to mid-1980s . However, widespread failure and abandonment of technically correct systems (Lucas, 1975 ; Cerveny and Clark, 1981), as well as research into the underlying processes by which computing practices become established (King and Kraemer, 1985 ; Kraemer et al ., 1989), have cast doubt on these models . The emerging view sees technology as an obviously necessary antecedent of technological institutionalization, but by no means a sufficient condition for institutionalization . Also required are changes in organizational values, norms, and beliefs . Put simply, institutionalization has occurred when an innovative practice has become an organizational routine in the sense described by Feldman (1988), that not only aids the organization in dealing with rationalizable and predictable tasks, but also allows organizations " . . . to act under conditions that mitigate against deliberation and conscious coordination" (Feldman, 1988, p . 4). Our question is, how does such institutionalization take place in the case of the innovative practice of computing use in the complex environment of us municipal governments?

50

Informatization and the Public Sector, Vol . 2, No . 1


Hypotheses An organization's decision to adopt an innovation has been found to be strongly influenced by opinion leaders—individuals in interorganizational and organizational networks who are influential in convincing others to be supportive of or antagonistic to stated needs for innovation (Rogers, 1983 ; Leonard-Barton, 1985) . Opinion leaders are similarly likely to be significant in the institutionalization process, since a key factor in both adoption and institutionalization is the "mobilization of bias" (Kraemer et al ., 1981) for or against the reforms implied by the innovation; a conjecture supported by Yin (1981) and others . Chaiken (1978) and Lawless (1982, 1987) found that failures of innovation institutionalization could be traced to loss of staffing and budgetary support from key managers . Yin's (1981) study of the routinization of innovations in local government agencies also demonstrated the enabling role of support by top executives, especially through allocation of staffing and budgetary resources for the innovation effort . Yin's study further found support for the views of Downs (1967) and Niskanen (1971) that successfully routinized innovations often received support from managers because the innovation provided justification for expanded agency budgets . These influences of top managers on successful routinization, and by extension, institutionalization of innovations, suggest the following hypothesis: Hypothesis 1 . The greater the commitment of top managers to innovative practices, the greater their institutionalization. In order for innovations to become fully incorporated into organizational life, they must not only receive persistent support from organizational elites, but they must be accepted by organizational members . Full institutionalization requires acceptance of and incorporation by organizational members in daily activity (Feldman, 1988) . Goodman et al . (1980) measure institutionalization by the extent of innovative practice within an organization, acceptance of the practice by organizational members, and the persistence of the practice over time. Widespread performance of new behaviors and acceptance of them as social facts are likely to be facilitated by efforts to indoctrinate organizational members about the new practices . Such indoctrination can include training groups, support groups, and similar methods to draw organizational members to the values represented by the technology . The importance of active efforts to develop grass roots support for institutionalization imply the following hypothesis: Hypothesis 2 . The greater the indoctrination of organizational members to innovative practices, the greater the institutionalization of those practices. Although organizations frequently use persuasion to socialize members to new practices, the alignment of organizational rewards to encourage acceptance and use of new practices may also enhance institutionalization . Reward systems may Informatization and the Public Sector, Vol . 2, No . 1

51

J.L. Perry et al. / Computing in complex organisations

affect institutionalization in several different ways (Goodman et al ., 1980; Goodman and Dean, 1982) . It is likely that the provision of different types of rewards will produce greater institutionalization than reliance on a single type of reward . Another important factor is discrepancies between expected and actual rewards . Innovations that live up to the expectations of adopters are likely to be self-reinforcing and continue to be used while those that fail to meet expectations are likely to fall into disuse . These arguments suggest the following hypothesis: Hypothesis 3 . The greater the organizational reinforcement of innovative practices through rewards, the greater the institutionalization of those practices. Two alternative strategies confront top managers when implementing an innovative practice . One is to make the change incrementally, by beginning in selected subunits and expanding from there . This strategy has the advantage of allowing learning and adjustment to the change, and the making of improvements in the implementation effort . However, it has the disadvantage of providing an opportunity for opponents of the innovative practice to form coalitions that can block progress (Keen, 1981) . Another strategy is to make the change comprehensively, throughout the organization simultaneously . This is expensive and sometimes difficult, but it has the advantage of constraining the formation of counter-innovation coalitions (Goodman and Dean, 1982) . Differences in the relative risks of either strategy will depend on the nature of the organization studied . Local governments are complex, multi-functional organizations with competing internal interest groups that can mobilize to block initiatives . Thus, we agree with Goodman and Dean (1982) and hypothesize: Hypothesis 4. The greater the initial diffusion of innovative practices throughout the organization, the greater the subsequent institutionalization of those practices. Institutionalization requires incorporation of new norms, values, and structures within the framework of existing patterns of norms, values, and beliefs (Kimberly, 1979) . Institutionalization will therefore be affected by the ability of organization members to control the adaptation process . Member control will hasten the process through which the innovation becomes identified with the values of members, and subsequently "owned" by those members (Leonard-Barton, 1985) . Member input to the process of innovation redesign is also likely to enhance member identification and ownership by reducing conflicts between existing and emerging norms, values and structures . Thus: Hypothesis 5 . The greater the member control over innovative practices, the greater the subsequent institutionalization of those practices. Although prior empirical research (Yin, 1981 ; Laudon, 1985) indicates that internal processes are especially critical for institutionalization, organization environments are also likely to be influential . The environment's effects on institutionalization are likely to occur as a result of both coercive and normative 52

Informatization and the Public Sector, Vol . 2, No . I


isomorphic processes (DiMaggio and Powell, 1983 ; Friedland and Alford, 1987). Two institutional features of public organization environments are likely to have significant influence on organizational decisionmaking, and, in turn, institutionalization of innovative practices . The first is pluralism, wherein many interest groups are given access to the policy making process . Pluralism is a mechanism of coercive isomorphism that injects multiple values into organizational decisionmaking processes, and stimulates increased competition for resources . Pluralism also can disrupt the continuity of organizational leadership, reducing stability and continuity of organizational goals . Pluralism may make it difficult to achieve goal congruence, and by extension, agreement on means for accomplishing goals (Selznick, 1948) . Thus: Hypothesis 6. The greater the level of community pluralism, the lower the eventual institutionalization of any given innovative practice. Another feature of some local governments is extensive use of "good government" structural reforms, flowing from the movement that began in the early 20th century to reduce graft, corruption, and patronage abuse in cities (Stone et al., 1941). Such reforms were seen as a check on several negative consequences of unchecked politicization of governmental processes, and were accomplished by developing a cadre of politically-protected, professional managers schooled in the use of modern administrative practices, and reducing partisanship in election practices . Both the desire to use innovations, and the political protection afforded by the reformed government structure, increase the likelihood that innovation will be pursued vigorously . Thus: Hypothesis 7. The greater the degree of reformed government practice, the greater the institutionalization of any given innovative practice. An environmental factor likely to affect institutionalization is the population size of the government jurisdiction . The mechanism of effect is indirect, through the fact that jurisdictions of large population require proportionately large government bureaucracies . Increased size of the bureaucracy brings increased control and coordination costs, as well as weaker goal congruence, both of which make implementation of innovative practices throughout the organization difficult . On the other hand, large bureaucracies are more likely to possess the need, capacity and slack resources necessary for successful application of costly and complex innovative practices such as computing . On balance we believe the presence of available resources, plus the generally centralized decision structures of us municipalities (Kraemer et al ., 1981) suggest that larger bureaucracies will have greater success in innovation . Thus: Hypothesis 8 . The greater the size of the community the greater the institutionalization of any given innovative practice. The structures established for carrying out innovation are likely to influence the degree of institutionalization (Goodman and Dean, 1982) . A most significant Informatization and the Public Sector, Vol . 2, No . 1

53


structural feature involves the assignment of administrative authority for the innovation . The allocation of authority for advocacy, operation, and promotion of the technology's use can take several forms (King, 1983) . Most organizations, including local governments, initially assigned responsibility for the innovation to functional units such as the finance or administrative services department. Other organizations assigned authority for computing innovation to independent computing service departments . The latter strategy is believed to be more conducive to institutionalization of computing by reducing the intra-departmental goal conflict that occurs when one functional subunit (e .g., police) must obtain resources from another functional subunit (e .g., finance) (Danziger et al ., 1982) . It is also likely to empower managers of the computing innovation to mold use of the technology to organization-wide norms and values, thereby generating allegiance to computing reforms among top management (Kraemer et al., 1989). Thus, it is hypothesized: Hypothesis 9. The greater the amount of administrative independence granted to the unit responsible for new practices, the greater the degree of institutionalization of those practices.

Methods Tests of these hypotheses were conducted using data from the URBIS (uRBan information systems) Research Project, an ongoing, longitudinal study of computing evolution in all us local governments over 50,000 in population . Data for this study were collected in 1975 and 1985, providing a ten-year interval for analysis . The ten-year interval between surveys permitted a sufficiently long period during which changes in levels of institutionalization would become apparent . Data used here were taken from three questionnaires completed by each government in each panel . Two of the instruments were mailed to data processing managers and one was sent to the chief executive or primary deputy .The logic of analysis was to use 1975 data about the determinants of institutionalization described above to predict the extent of institutionalization actually achieved in 1985. In order to test the hypotheses, scores for organizational and environmental predictors, based upon 1975 data, were regressed against an institutionalization index based upon 1985 data. Sample The sample consisted of 130 organizations for which complete data were available . There are approximately 400 local governments over 50,000 in population . All were surveyed, and response rates exceeding 70% were obtained in both 1975 and 1985 . To be used in this study, a respondent organization had to complete all necessary questions on all surveys in both 1975 and 1985 . Our 54

Informatization and the Public Sector, Vol . 2, No. 1


sample of 129 is approximately 30% of all us cities over 50,000 in population. The sample is smaller than the response for each survey because some cities responded to one survey but not the other, some cities with automation in 1985 were not automated in 1975, and some cities with 50,000 population in 1985 had less population in 1975 and therefore were not surveyed. Measures Table 1 presents the means, standard deviations, and zero-order correlations for all variables used in the study . The Measurement Appendix presents operational details for the more complex variables. Dependent variable Institutionalization was operationalized as a composite of three indicators from the 1985 survey . The first was organizational members' knowledge about and preferences for computing use . This was created from four measures of perceived problems with computing use (see Measurement Appendix) . The second was the proportion of the organization using computing, measured by an extensiveness index of the proportion of municipal functions using computing. The third was the sophistication of the applications in use, which was a count of applications in use according to their information processing task complexity (see Measurement Appendix) . The composite institutionalization index was a sum of the standardized scores for each of these three indicators. Organizational variables All organizational variables were measured using data from the 1975 study. Indoctrination was measured by two indicators : a simple measure of the percent of the computing budget spent on training, plus a more complex measure of user involvement in computing system design based on information system manager answers to thirteen questions (see Measurement Appendix) . Both indicators were seen as reflecting the extent to which management draws organization members into the practice of computing use to familiarize them with the technology . Management commitment to the practice was measured by two indicators : a dummy variable indicating whether there was long-range planning for computing, and a manager support index based on five Likert-type questions about chief executive dedication to computing innovation (see Measurement Appendix). Reward allocation was measured with two indicators : chief executives' views on the incentives provided to middle management to learn about computing; and the views of information systems managers on the extent of chief executives' involvement on six aspects of decision making related to computing . Increasing scores on both indicators indicated use of management rewards for involvement in computing innovation. Informatization and the Public Sector, Vol . 2, No . 1

55

J.L . Perry et al . / Computing in complex organisations

Diffusion of computing technology use was measured by the total number of computer applications in operation in 1975 . Member control over computing activity was measured by two indicators : an assessment of the means by which users evaluate computing services ; and a dichotomous measure of the extent to which overall evaluation of computing services was centralized in top management, or decentralized in the heads of user departments. Environmental variables Three indicators were used to measure key aspects of each municipal government's environment . Log of total population in '1970 tapped the size of the jurisdiction . An index of reformed structure measured the extent to which the local governance system had incorporated electoral and administrative characteristics associated with the good government movement . A community pluralism scale measured the diversity of interests represented within the community. Structure of the change A dummy variable indicating whether or not computing was an independent organizational department reporting to the organization's CEO was used to indicate structure for the innovation. Controls Three variables, growth in government employees (1970–1985), growth in population (1970–1980), and rate of change in government employment (1910–1985), were initially identified to control for trends in the data over time . The two indicators of employment growth were highly intercorrelated and, therefore, growth in government employees (1970–1985) was dropped from further analysis . Growth in population and rate of change in government employment (1970–1985), which tapped different time trends and were moderately correlated (r = 0.28), were selected for inclusion in the analysis . One additional control variable, prior experience with EDP, measured the number of years the government had been involved with computing.

Results The regression results for the composite institutionalization index are presented in Table 2 . Regressions for each component of the overall index are presented in Tables 2a–2c . The procedure used in calculating the regressions was to enter the control variables first, followed by the structure variable, and finally the environmental and organizational predictors . This assured that variations in institutionalization attributable to time trends and different structural arrangements would be accounted for prior to entering the explanatory variables. Informatization and the Public Sector, Vol. 2, No. 1

57

J.L. Perry et al. / Computing in complex organisations Table 2 Regression analysis of institutionalization Independent variables Environmental variables Community pluralism Reformed government structure Log of total population, 1970 Organizational variables Commitment: Top management support Long-range plan Indoctrination: Percent data processing budget spent on training User involvement in system design Reward allocation: Middle-management incentives to learn Data processing perception of chief executives interest Diffusion: Log of total applications operational, 1975 Member control: User control of data processing Decentralization of service evaluations Structure of the change Independent data processing department Control variables Growth in government employment, 1970-1985 Percent growth in population, 1970-1980 Prior experience with DP

Beta

Standard error

F

(sig .)

-0 .15 0 .11 0 .28

0 .09 0 .09 0 .10

3 .01 1 .45 8.35 * *

(0 .085) (0 .213) (0 .005)

- 0 .15 0 .02

0 .09 0 .09

2.99 0.05

(0 .087) (0.827)

0 .14 - 0 .07

0 .08 0 .09

2.92 0 .60

(0.091) (0.440)

0 .12

0 .08

1 .92

(0.168)

0 .14

0 .09

2 .51

(0.116)

0.01

0.09

0 .01

(0 .937)

0.27 - 0 .05

0 .09 0 .09

8 .82 * * 0 .29

(0 .004) (0 .595)

0 .17

0 .09

3 .99 *

(0.048)

-0 .06

0 .09

0 .45

(0.506)

0 .15 0 .04

0 .09 0 .09

3 .19 0 .19

(0.077) (0.664)

Adjusted R 2 = 0 .21 F = 3 .14 n = 130

As Table 2 indicates, the control variables, rate of government growth, percent growth in population, and prior experience with data processing were not significant . This verifies that institutionalization was not simply a linear trend over time, comparable to local population or government size changes, or a product of the period when a municipal government first adopted the technology . The dichotomous variable measuring the structure for change was significant and positive . This indicates that organizations that located administrative responsibility for computing in an independent unit reporting to the CEO 58


J.L . Perry et al. / Computing in complex organisations Table 2a Regression analysis of knowledge and preference for computing Independent variables

Beta

Standard error

F

(sig .)

-0 .22 0 .24 -0 .04

0 .10 0.10 0 .11

5 .15 * 6 .02 * 0 .17

(0 .025) (0 .016) (0 .684)

0 .18 -0 .07

0 .10 0 .09

3 .52 0 .60

(0 .063) (0 .441)

0 .16 0 .01

0 .09 0 .10

3 .28 0 .02

(0 .072) (0 .883)

0 .11

0 .09

1 .52

(0 .221)

0 .06

0.10

0.40

(0 .529)

- 0 .08

0 .10

0.69

(0 .409)

0.03

0 .10

0.07

(0 .792)

- 0.14

0 .10

2.00

(0 .160)

0.17

0 .09

3 .38

(0 .069)

- 0 .00

0 .10

0 .00

(0 .966)

- 0 .10 0 .06

0 .09 0 .10

1 .07 0 .38

(0 .304) (0 .539)

Environmental variables Community pluralism Reformed government structure Log of total population, 1970

Organizational variables Commitment: Top management support Long-range plan Indoctrination: Percent data processing budget spent on training User involvement in system design Reward Allocation: Middle-management incentives to learn Data processing perception of chief executives interests Diffusion: Log of total applications operational, 1975 Member control: User control of data processing Decentralization of service evaluations

Structure of the change Independent data processing department

Control variables Growth in government employment, 1970-1985 Percent growth in population, 1970-1980 Prior experience with DP Adjusted R 2 = 0 .08 F = 1 .65 n = 126

experienced higher levels of institutionalization, while those that located responsibility in functional units such as finance experienced less institutionalization. Conversely, organizations that assigned computing to an existing administrative unit such as finance or administration, or to multiple units, achieved a significantly lower degree of institutionalization. Informatization and the Public Sector, Vol . 2, No. 1

59

J.L . Perry et al. / Computing in complex organisations Table 2b Regression analysis of extensiveness of computing Independent variables Environmental variables Community pluralism Reformed government structure Log of total population, 1970 Organizational variables Commitment: Top management support Long-range plan Indoctrination: Percent data processing budget spent on training User involvement in system design Reward Allocation: Middle-management incentives to learn Data processing perception of chief executives interests Diffusion: Log of total applications operational, 1975 Member Control: User control of data processing Decentralization of service evaluations Structure of the change Independent data processing department Control variables Growth in government employment, 1970-1985 Percent growth in population, 1970-1980 Prior experience with DP

Beta

Standard error

F

- 0 .13 0.08 0.27

0 .08 0.08 0.09

2 .39 0 .80 8 .39 * *

(0.125) (0.374) (0.004)

- 0 .25 0 .02

0 .08 0 .08

8 .99 * 0 .06

(0 .003) (0 .800)

0 .13 - 0 .06

0 .08 0 .09

2 .64 0 .45

(0 .107) (0 .502)

0 .10

0 .08

1 .50

(0 .223)

0 .10

0 .08

1 .42

(0 .236)

0 .06

0 .09

0.47

(0 .495)

0 .30

0 .09

0 .02

0 .08

0.03

(0 .858)

0 .14

0 .08

2.90

(0 .091)

-0 .09

0 .09

1 .17

(0.281)

0 .16 0.04

0 .08 0.08

4.12 * 0.23

(0 .045) (0 .631)

(sig .)

12.33 * *

(0 .001)

Adjusted R 2 = 0 .28 F = 4 .19 n = 130

Two environmental or organizational predictors were significant : the log of total population in 1970, and member control of computing . The other predictors, commitment, indoctrination, reward allocation, and diffusion, were not significant. The regressions for the three components of the institutionalization index, presented in Tables 2a, 2b and 2c, revealed the significance of several variables 60


J.L . Perry et al. / Computing in complex organisations Table 2c Regression analysis of sophistication of computing applications Independent variables Environmental variables Community pluralism Reformed government structure Log of total population, 1970 Organizational variables Commitment: Top management support Long-range plan Indoctrination: Percent data processing budget spent on training User involvement in system design Reward allocation: Middle-management incentives to learn Data processing perception of chief executives interests Diffusion: Log of total applications operational, 1975 Member Control: User control of data processing Decentralization of service evaluations Structure of the change Independent data processing department Control variables Growth in government employment, 1970-1985 Percent growth in population, 1970-1980 Prior experience with DP

Beta

Standard error

F

(sig .)

-0 .04 - 0.05 0.41

0 .08 0 .08 0 .09

0 .29 0 .37 20 .47 * *

(0.594) (0.544) (0.000)

- 0 .23 0 .10

0 .08 0 .08

8 .12 * * 1 .46

(0 .005) (0 .229)

0 .03 -0 .08

0 .08 0 .09

0 .14 0.82

(0 .712) (0 .369)

0 .06

0 .08

0 .56

(0 .457)

0 .14

0 .08

3 .17

(0.078)

- 0 .01

0.09

0 .02

(0 .890)

0 .26

0 .08

9 .68 * *

(0 .002)

0 .02

0 .08

0 .06

(0 .811)

0 .09

0 .08

1 .30

(0 .257)

0 .04

0 .09

0 .20

(0 .654)

0 .25 - 0 .00

0 .08 0 .08

9 .51 * * 0 .00

(0 .003) (0 .958)

Adjusted R 2 = 0 .31 F = 4 .63 n = 130

that did not attain significance for the overall index . Two aspects of the institutional environment, community pluralism and reformed government structure, were significant determinants of knowledge and preference for computing. Three variables, log of total population, 1970, top management support, and percent growth in population, 1970-1980, were significant in the regressions for both extensiveness and sophistication of computing . The significance of the Informatization and the Public Sector, Vol. 2, No. 1

61

J.L . Perry et al. / Computing in complex organisations Table 3 Relative contribution of organizational and environmental predictors of institutionalization R 2 change

F

(sig .)

Organizational predictors: Diffusion Commitment Reward allocation Indoctrination Member control

0 .000 0 .018 0 .025 0 .021 0 .055

0 .01 1 .50 2 .07 1 .69 4 .46 * *

(0 .937) (0 .227) (0 .131) (0.189) (0.014)

Environmental predictors

0 .059

3 .22 *

(0 .025)

control variable, percent growth in population, indicates a linear trend for these indicators of institutionalization. The explanatory power of the environmental, diffusion, commitment, reward allocation, indoctrination, and member control variables was tested by comparing their relative contributions to the regression . These comparisons are provided in Table 3 . The block of environmental variables explain a significant part

Table 3a Relative contribution of organizational and environmental predators of knowledge and preference for computing R 2 change

F

(sig .)


0.005 0.027 0.013 0.024 0.015

0 .69 1 .84 0 .89 1 .64 1 .01

(0 .409) (0 .164) (0 .412) (0 .200) (0 .369)

Environmental Predictors

0 .103

4 .63 * *

(0 .004)

Table 3b Relative contribution of organizational and environmental predictors of extensiveness of computing R 2 change

F

(sig .)


0 .003 0 .051 0 .015 0 .017 0 .069

0 .47 4 .56 ** 1 .36 1 .49 6 .25 **

(0 .495) (0 .013) (0 .260) (0 .229) (0 .003)


0 .050

3 .01 *

(0 .033)

62


J.L . Perry et al. / Computing in complex organisations Table 3c Relative contribution of organizational and environmental predictors of sophistication of applications of computing R 2 change

F

(sig .)


0 .00 0 .046 0 .019 0 .005 0 .053

0 .02 4 .28 ** 1 .78 0 .46 4 .94 * *

(0 .890) (0 .016) (0 .174) (0 .632) (0 .009)


0.127

7 .93 *

(0 .000)

of the change in R 2 in each of the four regression equations . Among the organizational predictors, only the member control block is significant for the composite index of institutionalization.

Discussion The analysis provided only moderate support for the hypotheses . The environmental predictors and structure of the change were significantly related to the overall institutionalization index, but among the five organizational predictors, only member control was significant . Commitment, indoctrination, reward allocation, and diffusion were not significant in the regression for the composite institutionalization index. In general, we can conclude that institutionalization of computing innovations is a function of both environmental and internal factors . While the environmental factors show up strongly in this analysis, the managerially controllable internal factors of member control and structure of the change were significant as well . The surprise was that the other organizational factors were not significant. Earlier, historically-oriented case study research of computing innovation over time by Laudon (1985) and Kraemer et al . (1989) suggested that environmental stimuli had a significant effect on organizational decisions to adopt computing practices, but that organizational factors under control of managers played by far the greatest role in establishing computing use as a routine, embedded, institutionalized organizational practice . We explore the likely reconciliation of the current findings with those studies below. First, we discuss the implications of the significant predictors . The analyses suggest that the ability of the organization to focus its institutionalization efforts is important to success . In keeping with Hypothesis 6, community pluralism appears to influence institutionalization of innovative practices by altering member preferences for the technology . Increasing community interest group activity appears to reduce the ability to inculcate new habits, norms, and values Informatization and the Public Sector, Vol . 2, No . 1

63

J. L . Perry et al.

/

Computing in complex organisations

within the organization . This is also in keeping with Meyer's (1979) finding that ambiguity and indeterminateness in decision making in public bureaucracies increase with greater societal (as opposed to internal) determination of organizational goals and practices. Pluralism is a broad concept, applicable both with respect to an organization's relations with its larger community, and with respect to intra-organizational units and the larger organizational ecology . Thus we also found, in keeping with Hypothesis 9, that increased administrative independence was associated with increased institutionalization . This suggests provision of a means for focusing top-level, independent attention on the needs of the innovation and institutionalization process reduced the potentially crippling effects of pluralism within the organization . Creation of an independent computing service provider not only removes the institutionalization process from parochial views of one or a few functional departments, but it creates a locus of dedicated and focused advocacy, possibly extending to missionary zeal, for use of the innovation . For example, Danziger et al . (1982) found that such organizational units engaged in active promotion of the technology, rapid expansion of computer applications, and moralization about the potential of the technology to bring about organizational change. We raise a note of caution here, however . Under some circumstances, the creation of the independent computing services department produces "empire building" behavior in which widespread organizational use of computing is directly tied to the bureaucratic welfare of the computing services department . These "skill bureaucracies," to use Danziger et al .'s (1982) expression, seek to secure and sustain their independence from both top management and the user community by maintaining a monopoly on expertise even while expanding by encouraging the use of computing throughout the organization. Regardless of whether the overall welfare of the organization is best served by establishment of computing skill bureaucracies, it seems likely that such self-interested focus can play a role in successful institutionalization of innovations. Our findings also suggest that focused efforts to promote computing use and avoid the divisive aspects of internal pluralism are not sufficient to produce institutionalization of computing use . True institutionalization arises when organizational members incorporate computing use as routine, "background" activity. This requires, at least to some extent, diffusion of real control over computing use throughout the organization . In keeping with Hypothesis 5, member control of computing innovation was a significant factor in institutionalization. Such control facilitates member acceptance of the innovation as an integral part of the organization's activities, and creates a distributed cadre of believers who will help to bring more reluctant organizational participants along. Member control provides feedback from the user community to the computing service providers (and their superiors) about the character of systems and the consequences of their use . In this way, the creation of an independent computing department and maintenance of member control may be consonant, in that an organization-wide independent support unit is likely to be more willing than 64



a narrowly-focused functional unit to extend opportunities for widespread member control. Summarizing these findings, institutionalization of computing use is influenced by the organization's ability to focus its efforts in the absence of pressure from multiple external and internal interest groups, while maintaining a coherent program of innovation promotion . Establishment of an internal, independent service provision unit helps with both objectives . However, to obtain widespread acceptance of the innovation among organizational members, it is necessary to make the executive authority of the independent service provision unit accountable by extending member control over evaluation of the service unit's performance . The issue is not as simple as executive vs . pluralistic approaches, but rather the maintenance of a balance of executive authority for some activities and pluralistic processes for others. We return now to discussion of the failure of the other organizational predictors to reach significance in the regressions. The failure of diffusion to appear as a significant predictor is surprising because earlier cross-sectional analyses of the 1975 IRais data showed those organizations with the most diffused use of the technology to be the most sophisticated in use as well (Kraemer et al., 1981) . It stands to reason that a combination of extensive and sophisticated use would bode well for downstream institutionalization. However, two factors that cannot be accounted for in this analysis might intervene . First, diffusion is significantly dependent on the character of the technology itself, and improving price-performance characteristics of computing during the decade between 1975 and 1985 could well have altered organizational predispositions to invest further in computing innovations . Second, diffusion itself is a component of the larger question of institutionalization, and is subject to the influence of factors such as indoctrination, commitment, and so on . Recent case studies in a selected set of the cities included in the 1975 study have shown that some of the "early adopters" of computing in 1975 fell into the middle of the population by 1985, and some "late adopters" surged ahead during this period (Kraemer et al ., 1989) . These findings suggest that comprehensive change strategies may have been less efficacious than incremental change strategies, but we were unable to detect this distinction with the way diffusion was measured . Whether the lack of significance of the diffusion variable is due to problems in measurement or in the precision of our theoretical constructs, it is apparent that both require additional attention in future research. The lack of significance of the other three variables, commitment, indoctrination, and reward allocation is more puzzling . For purposes of assessment, we can combine these variables by noting that they are all "top–down" factors involving the top management . Top management commits to the innovation, dedicates the organization's resources to indoctrination mechanisms, and extends suitable rewards to middle managers for promoting the innovation . There is little doubt that these factors do matter at some level . For example, it seems hard to imagine how computing innovations could ever be institutionalized if top Informatization and the Public Sector, Vol. 2, No . 1

65


managers withheld their support or resources, or if they punished middle managers for use of the technology . At minimum, top management must tolerate the innovative practices. From another perspective, however, it is not clear that active top management support is essential or even conducive to institutionalization . Dutton and Kraemer (1979) and Perry and Kraemer (1977) found the role of municipal chief executives not to be uniformly positive with respect to implementing computing innovations . Executives often set unrealistic expectations and priorities for computing . They also tended to allocate resources for computing use in ways that produced political advantage even if those ways were suboptimal to the overall organization . Further, it was not uncommon for chief executives to support questionable projects simply because of faith in technology rather than knowledge of the issues . Thus, active chief executive attention to computing may play a mixed role in institutionalization. The lack of significance for the top management support variable may also involve limitations of the measure used in the present study . The role of top management support is likely to be more complex and subtle than the measures from the 1975 data can address . Careful study of the histories of computing in seven municipal-governments by Kraemer et al . (1989) revealed critical roles for managers in the character and success of computing initiatives . How can we reconcile these differences? First, the term "top management" is imprecise . The 1975 survey measured only chief executive attitudes, but top management consists of more than the CEO . In fact, while CEO support was occasionally critical in the histories of the seven case study cities, the most influential senior managers were usually one or two steps down at the assistant city manager and department head level . The 1975 survey probably aimed too high in attempting to identify the role of managerial influence on institutionalization . It would be ideal to check this conjecture using the 1975 data set, but data for the key variables are only available from the chief executive level . One cannot go back and ask the questions that have subsequently proven to be important in later analysis . This is an obvious but seldom articulated limitation with longitudinal, empirical research in the social sciences. Second, the study of seven organizational histories revealed that managerial actions that influenced institutionalization of computing did not manifest themselves as clear-cut behaviors waiting to be turned into measurable variables. Certainly, some factors such as managerial knowledge of the issues and sensitivity to the organizational and political realities of the situation proved important in every case . But it is very difficult to measure such managerial attributes remotely via self-administered instruments, or by any other means, for that matter (Feldman, 1988). Thus, we find it somewhat surprising that the organizational predictors of commitment, indoctrination and reward allocation were not significant . However, we believe our results reflect more on the character of our 1975 data 66



collection instruments than on the sensibility of the ideas we were testing. Indeed, despite extensive efforts to make the 1975 instruments as sound as possible (cf Kraemer et al ., 1976, 1981), so little was known about computing innovations in vivo that shortcomings in some of the measures derived from the instruments were inevitable . Discussion of these shortcomings can be found in several places, including Kraemer and King (1981), King and Kraemer (1985), and Kraemer et al . (1989) . Nevertheless, as Medawar (1982) suggests, the way of science includes both learning about the world and learning about how to learn. The obvious call in following this research is to learn from our results, improve and expand our measures of the organizational factors that might play a role in institutionalization, and to apply them in future longitudinal research.

Measurement appendix Institutionalization index This index was calculated by summing the Z-scores of three indexes : (1) knowledge about and preference for computing, (2) extensiveness of computing, and (3) sophistication of computing . Descriptions of the three indexes are provided below . The intercorrelations of the three indexes:

knowledge/preference extensiveness sophistication

knowledge/ preference

extensiveness

sophistication

1 .00

0.21 1 .00

0 .13 0 .87 1 .00

Coefficient alpha = 0 .67.

Knowledge about and preference for computing index (coefficient alpha = 0 .80) was computed by obtaining the mean response to four Likert-type items in the Data Processing Manager Survey, 1985 . Four response categories were provided: not a problem, at times a problem, often a problem, very often a problem. The four items are : (1) EDP lacks acceptance from top government officials, (2) EDP lacks acceptance from major department heads, (3) EDP lacks acceptance from staff of user departments, and (4) User departments are not knowledgeable about EDP . Scores ranged from a low of 1 .50 to a high of 4 .00 (mean = 3 .14, s.d. = 0.57). The average score was then converted into a Z-score. Extensiveness of computing index was computed by calculating the proportion of 34 specified functional areas that were automated in the city . Scores ranged from a low of 0 .06 to a high of 0 .88 (mean = 0 .44, s .d . = 0.15). A functional area was considered automated if there were at least two applications automated in that area . The proportion was then converted into a Z-score. Informatization and the Public Sector, Vol . 2, No . 1

67

J.L. Perry et al. / Computing in complex organisations

Sophistication of computing index is the sum of all applications automated in the city with each application first weighted by degree of information task complexity of the applications . All applications listed in the Data Processing Manager's Survey, 1985 were scored with respect to the level of information task complexity involved in the application . Information task complexity weights ranged from a low of 1 to high of 5 ; 1 = recordkeeping (activities which primarily involved the entry, updating, and storage of data), 2 = calculating/ printing (activities which primarily involve sorting, calculating, and printing of stored data to produced specific operational outputs), 3 = a record-restructuring (activities which involve reorganization, reaggregation, and/or analysis of data), 4 = sophisticated analytic (activities which utilize sophisticated visual, mathematical, simulation or other analytical methods to examine data), and 5 = process control (activities which approximate a cybernetic system ; data about the state of the system are continually monitored and fed back to a human or automatic controller which steers the system toward a performance standard) . Scores ranged from a low of 37 to a high of 510 (mean = 177 .70, s .d. = 94.73). The total weighted score was then converted to a Z-score. Pluralism of community influence scale Scale constructed by computing the average degree of influence (1 = not influential ; 2 = somewhat influential, 3 = quite influential, 4 = extremely influential) over 16 community groups that were rated by chief executives in 1975 (coefficient alpha = 0.80). Scores ranged from a low of 1 .25 to a high of 3 .13. The groups included: Democratic Party, Republican Party, newspapers, bar association, local medical groups, labor unions, minority groups such as Blacks, Chicanos, Puerto Ricans, other ethnic groups, neighborhood groups, church leaders, chamber of commerce, industrial leaders, building and real estate people, bankers and executives of other financial institutions, good government organizations such as the League of Women voters and citizens's leagues, and environmental/ ecology groups. Reformed government structure index The measure was constructed by summing three indicators of progressive reforms in local governments : type of government structure, electoral system and election laws . Government structure was scored as 2 = council-manager system, 1 .5 = mayor-council with chief administrative officer and 1 = all others; electoral system was scored as 2 = at-large elections, 1 .5 = mixed and 1 = ward; non-partisan election laws was scored as 2 = nonpartisan ; 1.2 = local parties only and 1 = partisan . Scores on this index range from a low of 1 to a high of 2. Intercorrelation matrix for these items is:

government structure electoral system election law 68

government structure

electoral system

election law

1 .00

0 .29 1 .00

0.26 0 .14 1 .00

Informatization and the Public Sector, Vol . 2, No . 1


Log of total population, 1970 Measure is the log of the total population as recorded in the 1970 census. Values on this variable range from a low of 4 .7 to a high of 6 .29. Prior experience with DP Prior experience with DP is operationalized as the total number of years that a city has had computing . Scores range from a low of less than one year (0) to a high of 20 years. Top management support scale The scale is constructed by computing the mean of the chief executive responses in 1975 to five Likert-type items (coefficient alpha = 0 .70) . Responses to the items were scored as 0 = strongly disagree, 25 = disagree, 50 = undecided, 75 = agree, 100 = strongly agree . The items in this scale include: (1) the computer is an essential tool in the day-to-day operations of this government ; (2) in the future, the computer will become much more essential in the day-to-day operations of this government ; (3) for the most part, computers have clearly increased the speed and ease of performance of government operations where they have been applied ; (4) the elected legislative body here is generally favorable to expanding the use of computers and data processing ; and (5) the department heads are generally favorable to expanding the use of computers and data processing . Scores for this scale range from a low of 45 to a high of 100.

Long-range plan This dichotomous variable (1 = no, 2 = yes) is obtained from the responses of the data processing managers in 1975 in each city to the question : does this installation have a long range (two years or more) EDP plan . In those cities in which there was more than one data processing installation, the city-level measure was obtained by calculating the mean response across all data processing managers in the city who responded to the Data Processing Managers Survey, 1975 . Scores range from a low of 1 to a high of 2. Percent data processing budget spent on training This variable was computed using the responses to two items in the Data Processing Managers Questionnaire, 1975 : total expenses associated with training and education of programmers and analysts for the current year divided by the total expenditures budgeted for the current year . In those cities in which there was more than one data processing installation, the city-level measure was obtained by first summing all budget expenditures by installation and then computing the percent spent on training . Scores ranged from a low of 0% to a high of 5 .38%. Informatization and the Public Sector, Vol. 2, No. 1

69


User involvement in system design scale This scale was computed by calculating the mean response of data processing managers to four items in the Data Processing Managers Questionnaire, 1975. Responses to the items were scored as : 1 = never, 2 = seldom, 3 = often, and 4 = always . The four items were designed to tap four dimensions of user involvement : adoption, design, development, and evaluation . The following items were used to represent these four dimensions : (1) adoption : the frequency with which users initiate major changes of EDP applications (such as changing the flow of information, the input or output) ; (2) design : the frequency with which users work as a member of a technical group in designing an application; (3) development : the frequency with which users provide test data for an application ; and (4) evaluation : the frequency with which users formally evaluate applications they use . Average scores ranged from a low of 1 (never) to a high of 3 .75. Coefficient alpha = 0 .75. Middle-management incentives to learn This measure is a single item taken from the chief executive's response to the following question on the Chief Executive Survey, 1975 : Middle management should be provided incentives either by pay, promotion ; or free instruction to encourage them to learn more about computers and data processing . Response categories include : 1 = strongly disagree, 2 = disagree, 3 = undecided, 4 = agree, and 5 = strongly agree . Scores on this variable range from a low of 2 to a high of 5. Data processing perception of chief executives interest scale This scale was computed by calculating the mean response of data processing managers to six items in the Data Processing Managers Questionnaire, 1975. Dichotomous response categories in which 0 = no and 1 = yes were used for each of the items . Data processing mangers were asked to respond to whether the chief executive was involved in : (1) providing a major input into whether or not a new set of EDP applications will be adopted ; (2) has the authority for setting application priorities ; (3) approves budget requests for new computer mainframes and systems ; (4) approves requests for new peripheral equipment in user departments; (5) is primarily responsible for evaluating the services provided by the installation ; and (6) must approve major reorganizations such as changing the departmental status or location of EDP, or consolidating several independent EDP units . Scale scores ranged from a low of 0 to a high of 1. Coefficient alpha = 0 .74. Log of total applications operational, 1975 This measure is the log of the total count of applications that were operational in the city in 1975 . Data Processing Managers completed the Local Government Computer Applications Questionnaire, 1975 which provided an inventory of 254 70



computer applications and files divided into 33 functional areas of local government . They were asked to indicate which of the 254 computer applications and files were operational in their installation . Scores ranged from a low of 0 to a high of 4 .78. User control of data processing scale This scale was computed by calculating the mean responses of data processing managers to two items in the Data Processing Installation Questionnaire, 1975. Response categories include : 1 = never, 2 = seldom, 3 = often, 4 = always . Exact wording of the items are : What is the frequency with which users of your data processing unit (1) sit on a policy board overseeing the . computer unit, and (2) complete questionnaires or evaluation forms on their satisfaction with data processing services? Scores range from a low of 1 to a high of 4 . Coefficient alpha = 0 .62. Decentralization of service evaluation This dichotomous single item scale is based on the responses of data processing managers to the following question : Who is primarily responsible for evaluating the services provided by the installation? The following response items were provided : data processing manager, department head over data processing, user department heads, chief executive official, local legislative body, inter-departmental board or steering committee, inter-governmental board or steering committee, or other . If "user department heads" or "inter-governmental Board or steering committee" was selected, the city was scored as "decentralized", otherwise, the city was scored as "centralized" . Under conditions of multiple installations within a city, the individual installation scores were mean averaged to obtain the city-level score . Scores range from 1 (centralized) to 2 (decentralized). Independent data processing department Cities were scored a "1" if there was a single installation within the city and the installation was an independent data processing department under the chief executive . All other organizational arrangements, as well as all cities with more than one installation within the city were scored a "0". Growth in government employment, 1970—1985 The variable was computed as the rate of change in the number of employees per 1000 population in 1980 with the number of employees per 1000 population in 1985 . Scores range from a low of — 0.67 to a high of 1 .21. Percent growth in population, 1970—1980 The variable was computed as a percent rate of change in the total population between 1970 and 1980. Informatization and the Public Sector, Vol . 2, No . 1

71


References Beniger, J ., 1986 . The Control Revolution : Technological and Economic Origins of the Information Society . Harvard University Press, Cambridge, MA. Cerveny, R .P . and Clark, T.D ., Jr ., 1981 . Conversations on "why information systems fail—and what can be done about it . " Systems, Objectives, Solutions 1 : 149-154. Chaiken, J ., 1978 . Transfer of emergency service deployment models to operating agencies. Management Science 24 : 719-731. Danziger, J .N ., Dutton, W .H ., Kling, R . and Kraemer, K.L ., 1982 . Computers and Politics. Columbia University Press, New York. DiMaggio, P .J . and Powell, W .W ., 1983 . The iron cage revisited : institutional isomorphism and collective rationality in organizational fields . American Sociological Review 48 : 147-160. Downs, A ., 1967 . Inside Bureaucracy . Little, Brown, Boston, MA. Dutton, W .H . and Kraemer, K.L., 1979 . Urban technology, executive support, and computing. The Urban Interest 1 : 35-42. Ellul, J ., 1964. The Technological Society . Knopf (Vintage ed .), New York.Feldman, M ., 1988. Understanding organizational routines : stability and change . Working paper, University of Michigan Institute of Public Policy Studies. Feller, I . and Menzel, D .C ., 1975 . Diffusion milieux as a focus of research on innovation in the public sector. Paper presented at the annual meeting of the American Political Science Association. Friedland, R . and Alford, R .R ., 1987 . Bringing society back in : symbols, structures and institutional contradiction . Paper presented at the Conference on Institutional Change, Center for Advanced Study in the Behavioral Sciences, Stanford, CA, 15–16 May. Gasser, L ., 1984 . The social dynamics of routine computer use in complex organizations. Information and Computer Science Department, Irvine, CA. Glaser, E .M ., 1981 . Durability of innovations in human service organizations : a case-study analysis . Knowledge: Creation, Diffusion, Utilization 3 : 167-185. Glaser, G ., Torrance, A .L . and Schwartz, M .H ., 1983 . Administrative applications. In : A . Ralston and E .D . Reilly, eds . Encyclopedia of Computer Science and Engineering . Van Nostrand-Reinhold, New York : 23–40. Goodman, P .S ., Bazerman, M . and Conlon, E ., 1980 . Institutionalization of planned organizational change . Research in Organizational Behavior 2 : 215-246. Goodman, P .S . and Dean, J .W ., Jr ., 1982 . Creating long-term organizational change . In : P .S. Goodman and Associates, Change in Organizations . Jossey-Bass, San Francisco : 226-279. Goodman, R .M . and Steckler, A ., 1989 . A framework for assessing program institutionalization. Knowledge in Society: the International Journal of Knowledge Transfer 2: 57-71. Huber, G ., 1990 . A theory of the effects of advanced information technologies on organizational design, intelligence and decision making . Academy of Management Review 15 : 1–47. Hughes, E ., 1939 . Institutions . In : R. Park, ed . An Outline of the Principles of Sociology. Barnes & Noble, New York : 283–346. Ilich, I ., 1973 . Tools for Conviviality . Harper & Row, New York. Keen, P .G .W ., 1981 . Information systems and organizational change . Communications of the ACM 24 : 24–33. Kimberly, J .R ., 1979 . Issues in the creation of organizations : initiation, innovation, and institutionalization . Academy of Management Journal 22: 437-457. King, J .L ., 1983 . Centralized vs . Decentralized computing : organizational considerations and management options . ACM Computing Survey, December : 319–345. King, J .L . and Kraemer, K .L ., 1985 . The Dynamics of Computing . Columbia University Press, New York. Kling, R . and Scacchi, W., 1982 . The Web of computing : computing technology of social organization . In : M . Yovits, ed . Advances in Computers, Vol . 21 . Academic, New York: 245–294. 72

Informatization and the Public Sector, Vol. 2, No . I

J.L. Perry et al. / Computing in complex organisations Kraemer, K . and King, J . 1981 . Computing policies and problems : toward a stage theory. Telecommunications Policy, 5 : 198-215. Kraemer, K., Dutton, W. and Northrop, A., 1981 . The Management of Information Systems. Columbia University Press, New York. Kraemer, K .L ., Dickhoven, S ., Fallows Tierney, S . and King, J .L ., 1987 . Datawars: the Politics of Modeling in Federal Policymaking . Columbia University Press, New York. Kraemer, K .L ., King, J .L ., Dunkle, D . and Lane, J .P ., 1989 . Managing Information Systems. Jossey-Bass, San Francisco, CA. Laudon, K ., 1985 . Environmental and institutional models of system development : a national criminal history system . Communications of the ACM 28 : 729-740. Lawless, M ., 1982 . A policy and process analysis of computer model implementation in criminal justice agencies . Applied Management Science 2: 217-231. Lawless, M .W ., 1987 . Institutionalization of a management science innovation in police departments . Management Sciences 33 : 244-252. Leonard-Barton, D ., 1985 . Experts as negative opinion leaders in the diffusion of a technological innovation . Journal of Consumer Research 11 : 914-926. Lucas, H ., 1975 . Why Information Systems Fail . Columbia University Press, New York. Medawar, P ., 1982 . Phato's Republic . Oxford University Press, Oxford. Meyer, M ., 1979 . Change in Public Bureaucracies . Harvard University Press, Cambridge, MA. Meyer, J . and Rowan, B ., 1977 . Institutionalized organizations : formal structure as myth and ceremony . American Journal of Sociology 83 : 340-363. Miller, C .L ., 1983 . How to successfully resist a computer system and avoid its benefits : a victory for the bureaucracy? Systems, Objectives, Solutions 3 : 3-12. Niskanen, W ., 1971 . Bureaucracy and Representative Government . Aldine & Atherton, Chicago, IL. Nolan, R .L ., 1973 . Managing the computer resource ; a stage hypothesis . Communications of the ACM 16 : 339-405. Perry, J .L . and Kraemer, K.L., 1977 . The chief executive in local government information systems: catalyst or barrier to innovation . Urban Systems 2 : 121-131. Robey, D ., 1983 . Information systems and organizational change : a comparative case study. Systems, Objectives, Solutions 3 : 143-154. Rogers, E ., 1983 . Diffusion of Innovations, 3rd ed . Free Press, New York. Scott, W .R ., 1987 . The adolescence of institutional theory . Administrative Science Quarterly 32: 493-511. Selznick, P ., 1948 . TVA and the Grass Roots . University of California Press, Berkeley, CA. Stone, H ., Price, D . and Stone, K., 1940 . City Manager Government in the United States . Public Administration Service, Chicago, IL. Thompson, J .D . 1967 . Organizations in Action . McGraw-Hill, New York. Tolbert, P .S . and Zucker, L .G ., 1983 . Institutional sources of change in the formal structure of organizations : the diffusion of civil service reforms, 1880-1935 . Administrative Science Quarterly 23 : 22-39. Yin, R ., 1981 . Life histories of innovations : how new practices become routinized . Public Administration Review 41 : 21-28. Zucker, L., 1983 . Organizations as institutions . Research in the Sociology of Organizations 2 : 1-47.

Informatization and the Public Sector, Vol. 2, No. 1

73