Accting., Mgmt. & Info. Tech., Vol. 6, No. 4, pp. 2.55-282, 1996 Copyright 0 1996 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0959-X022/96 $15.00 + 0.00
Pergamon
ORGANIZATIONAL CONTEXT, SOCIAL INTERPRETATION, AND THE IMPLEMENTATION AND CONSEQUENCES OF GEOGRAPHIC INFORMATION SYSTEMS Sundeep Sahay
University of Salford Daniel
Robey
Georgia State University
Abstract-An interpretive approach was applied to understand the social meanings of Geographic Information Systems (GIS) being implemented in two local government agencies in the United States. We assumed that information technologies, like other knowledge systems, are subject to social construction and that resulting social interpretations may shape their implementation and use. Guided by a framework linking social context with organizational processes, the research used inductive methods to describe the social constructed frames of meaning pertaining to GIS. Findings revealed strikingly different patterns of implementation and social consequences from the same technology. These findings demonstrate the role of social interpretations in the implementation of information technology. Social interpretations mediate between organizational context and the process of implementation to produce particular consequences of technology. The results support the notion of emergent causality and the importance of process in understanding the social consequences of technology. Copyright 0 1996 Elsevier Science Ltd
INTRODUCTION
A persistent theme in the literature on information systems is the potential for computer-based information technologies to address a wide variety of social and organizational problems. Information technologies are human creations that embody social purposes, and their potential to improve the social condition is enormous. Yet, the literature is replete with reports of unfulfilled potential. Individual case studies reveal a myriad of problems associated with the implementation of information technologies (e.g. Franz & Robey, 1984; Hirschheim & Newman, 1991; Robey & Newman, 1966; Swanson, 1988; Walsham, 1993). If the social objectives of new technologies are to be achieved, it seems imperative that we understand the nature of the social context and its interaction with human agency and various processes of implementation.
This paper is based on the first author’s doctoral dissertation, completed at Florida direction of the second author. We thank the other members of the dissertation Nielsen, and Carol Saunders-for their contributions. We are also indebted Ruth Chapman, Line Dube, Sarah Maxwell, Wanda Orlikowski, Mahatapa Wishart for their support and guidance throughout this research. 255
International University under the committee-Joyce Elam, Christine to Ana Azevedo, Michelle Brown, Palit, Geoff Walsham, and Nicole
S. SAHAY and D. ROBEY
256
Recent studies of technology in organizations have demonstrated the value of treating information technologies as human artifacts with social constructed meanings (Fulk, 1993; Orlikowski & Gash, 1994; Prasad, 1993). The basic assumption underlying this approach is drawn from current research on the sociology of knowledge and from empirical studies on science and technology. A social constructivist approach assumes that information technology does not directly “impact” upon the social system in which it is developed and used. Rather, information technologies are subject to social interpretation by actors implementing and using them, and the social meanings of technology affect the manner in which they are implemented and used. Accordingly, in this paper we adopt a research framework drawn from recent studies that have linked social context and implementation processes (Orlikowski, 1993; Walsham, 1993). Context represents the social structure that both constrains and facilitates human agency directed toward the implementation and use of information systems. Simultaneously, human agency has the potential to revise both the information systems and the social context within which the technology is used. By tracing reciprocal relationships between context and process, we can better appreciate the implementation and social consequences of information technology. In this paper, we report an interpretive analysis of the implementation of geographic information systems (GIS) in two county government organizations in the United States. The study uses inductive methods within a comparative research design to describe the socially constructed frames of meaning pertaining to GIS. An understanding of these frames of meaning helps us to draw specific conclusions about the links between the process and context of implementation. Understanding these links more clearly should enable the more effective use of GIS and other information systems in organizations. We begin by explaining the technology known as GIS, focusing on the plurality of meanings and uses of GIS in different disciplinary contexts. We then offer a theoretical framework for organizing our empirical inquiry into the implementation and consequences of GIS. This framework is supported by related research on the social interpretation of information technology. Our research method is then explained, followed by the results of the study. Our discussion and conclusion focus upon three specific inferences drawn from the comparisons between the two research sites.
GEOGRAPHIC
INFORMATION
SYSTEMS
In general, the term GIS describes integrated systems of geographically oriented computer technologies-computer aided design, computer cartography, database management, and remote sensing-that are applied to problems involving spatial analysis (Maguire, Goodchild & Rhind, 1991). Like many emerging technologies, however, GIS are not easy to define as discrete entities (Kling, 1987). GIS refers to many different configurations of computer applications rather than a specific system or technology. Many academic and professional disciplines such as geography, engineering, forestry, computer science, medicine, business, and social anthropology have participated in the development and use of GIS (Goodchild, 1995). Consequently, definitions of GIS vary across its various domains of applications. Moreover, the boundaries of GIS technology are more fuzzy than precise. For example, remote sensing, computer assisted design, and computer cartography involve vastly different ensembles of equipment and software, yet all have been defined as GIS (Pickles, 1995). As the commercial application of GIS increases, vendors and developers will probably blur distinctions among GIS applications even further.
GEOGRAPHY
INFORMA~ON
SYSTEMS
257
Despite these sources of confusion, the central technical characteristics of GIS are the digital electronic database and the system of production of electronic spatial representations of those data (Pickles, 1995). In this regard, GIS is a special case of relational information system that represents and processes spatial info~ation’. Although computer hardware and specialized peripheral equipment are essential for GIS, the distinguishing component is database software, which enables the functions of data input, data analysis, and the generation of output reports (Goodchild, 1995). In a GIS database, every object is related to a precise geographical location, and the database contains information on relationships between objects. Goodchild (1995) emphasizes that the most effective means for differentiating one GIS product from another is to examine their data models, which represent the rules, in the form of entities and the relationships among them, for creating geographical representations. There are two main types of data models used in GIS: layer-based and object-based. According to Goodchild, the layer-based model originated with land-use applications in Canada, where land was inventoried using descriptive variables, such as its capacity to support agriculture, recreation and other uses, In such a database, each variable is determined uniquely at every geographical coordinate, so maps can be conceptualized in mathematical terms as a succession of layers. The database captures this information in a “layer cake” that describes each successive field of data over the entire geographic region. By contrast, the object-based data model represents geographic space as populated by various kinds of discrete objects. Any place may be empty or occupied by one or more objects, whereas in a layer-based data model every location has exactly one value on every layer. Such distinctions are relevant to the study of GIS implementation because human interpretations of newer technologies depend upon knowledge about them. In practice, much of the information about GIS is likely to come from other individuals with experience in one or more specific GIS applications. Reliable general info~ation, covering both conceptual and oprational issues, is not likely to be readily available from any single source. To further complicate matters, GIS applications continue to proliferate. Within the last decade, policymakers have acknowledged the potential of GIS to address complex environmental concerns, respond to growing regulatory pressures, manage scarce resources, and allow greater public involvement in environmental decisions. The number of system installations was reported to be doubling every two or three years. with the growth of the GIS market estimated to be increasing at 30-35% annually (Frank, Egenhofer & Kuhn, 1991; Huxhold, 1991; Thompson, 1989). Fletcher, Bretschneider and Marchand (1992) have described GIS as the technology with the biggest impact on the thinking of managers in county departments in the United States. GIS technologies continue to advance with improvements in computer graphics, database management technologies, and the incorporation of satellite images. A 1991 directory listed 371 available software products, representing an enormous diversity of capabilities and approaches (GIS World Inc., 1991). Such rapid evolution and deployment of GIS places additional burdens of understanding on the social systems in which they are implemented and used. Amid the excitement about GIS, concerns have been voiced about the effectiveness of GIS implementation and use (Crosswell, 1991). Both economic and social factors have been identified as potential impediments to the establishment of more effective GIS applications. Investments in GIS present economic risks because they are typically large systems that require major financial outlays, especially during the initial stages of implementation where long-term, ‘Despite their widespread use, GIS have not received much attention in the information systems literature. Ein-Dor and Segev (1993) make no mention of GIS in their classification of 17 basic types of information systems.
258
S. SAHAY
and D. ROBEY
irrevocable commitments must be made (Sieber, 1991). For example, expenses incurred in developing base maps may reach $1.25 million annually and may recur for as many as five years (Thompson, 1989). Social factors often exacerbate these economic risks, especially in governmental settings. The involvement of multiple stakeholders, each with their individual missions and data needs, may lead to the development of incompatible GIS that duplicate data possessed by other agencies. Conflicts may also appear over the specifications for accuracy and precision of data that multiple agencies must share. Along with GIS’s technical scope and complexity, these economic and social aspects of GIS implementation render GIS comparable to many other large information systems used in both public and private organizations. Considerable effort must be exerted by many parties over a long period of time to produce a system with the potential to yield significant benefits. Yet, the realization of those benefits depends upon the acceptance, understanding, and use of the system by its users. From previous research and practical experience, we know that the potential of even much simpler information systems often fail to be realized. For this reason, the implementation and consequences of GIS deserve careful study.
THEORETICAL
FOUNDATION:
SOCIAL INTERPRETATIONS TECHNOLOGY
OF INFORMATION
Orlikowski and Baroudi (1991) suggest that interpretive research, which emphasizes the subjective meanings that human actors ascribe to technology in its context of implementation and use, is appropriate for investigating the implementation of information systems. Three basic assumptions guide an interpretive approach. First, neither human actions nor technologies are assumed to exert direct causal impacts. Rather, consequences are assumed to result from the interplay of computing infrastructures, conflicting objectives of different social groups, and the operation of chance (Markus & Robey, 1988). The consequences of a technology like GIS, therefore, are assumed to be indeterminate because of the inherently unpredictable nature of social processes. Second, actors are assumed to endow technology with social meanings as they engage in processes to propose, design, develop, implement, and use the technology. These meanings can shape the implementation process and subsequent use of the technology, independent of technology’s material properties (Fulk, 1993; Hirschheim & Newman, 1991; Orlikowski & Gash, 1994; Prasad, 1993; Robey & Azevedo, 1994; Walsham, 1993). Third, the aim of interpretive research is to increase understanding of phenomena within a specific social context. Inferences developed from a particular case analysis cannot be generalized to other settings because of the inherently contextual nature of this knowledge. Thus, the level of generality at which knowledge claims are made needs to be defined very carefully in such studies. However, this assumption does not remove the opportunity to build theory inductively by drawing inferences from the data of specific cases. Taken together, these basic assumptions reduce the temptation to regard information technology, including GIS, as capable of producing social results directly. Rather, the social meanings that emerge from the processes of designing and using applications of information technology lead to technology’s social consequences. The results of a growing number of studies support the validity of these basic assumptions. Especially revealing are studies of similar or identical technologies in comparable organizational settings. For example, Barley (1986) reported on the divergent social consequences of the same medical technology (computerized tomography) in two neighboring hospitals. Although the material features of the technology were virtually identical, divergent social processes were triggered by its implementation in each organization. Barley argued that technology was an
GEOGRAPHIC
INFORMATION
SYSTEMS
259
“occasion for”, not a determinant of, social change. Orlikowski’s (1993) comparison between two organizations adopting computer aided software engineering (CASE) technology also reveals a diversity of outcomes not explained by the characteristics of CASE alone. Elsewhere, Robey and Rodriguez-Diaz (1989) reported the divergent experiences encountered by an implementation team installing the same system in different offices of a multinational corporation. They concluded that the cultural context of implementation influenced the formation of different social interpretations in each office, which in turn helped to shape the patterns of adoption and use. Divergent interpretations of identical technologies were also reported by Orlikowski and Gash (1994) in their study of Lotus Notes, a software product designed to facilitate collaborative work. In this case, “technologists” and “managers” in the same organization held different interpretations of Notes, which affected its implementation and use. Recent interpretive research on implementation also helps to explain why information technologies are prone to different adaptations of use, or “re-inventions” (Fulk, 1993; Johnson & Rice, 1987; Kraut, Dumais & Koch, 1989; Orlikowski, 1996; Tyre & Orlikowski, 1994). Interpretive research on information systems is related to more fundamental work in the sociology of knowledge, which subjects the content and use of all knowledge systems, including technology, to sociological analysis. Berger and Luckmann’s (1967) seminal contribution regards all knowledge systems as products of human activity, that is, as socially constructed cultural artifacts. The epistemological assumption underlying the social constructivist view is that knowledge is subject to multiple interpretations which are related to the social processes surrounding the generation and use of knowledge. Knowledge is “produced” through specific practices employed within highly organized social systems, and scientific “facts” reflect the assumptions and biases embedded in the social order governing their origin (Hicks & Porter, 1991). The foundation of social construction provides a useful platform for interpretive studies of information technology. Three related concepts drawn from studies on the social construction of technology are especially significant for interpretive studies of information systems. Below, we define and discuss the concepts of relevant social groups, interpretive flexibility, and technological frames of meaning. Relevant social groups In traditional usage, the concept of a social group refers to a set of people who share a common geographical space, or occupy the same functional boundaries, such as those that separate departments in an organization. In the literature on the social construction of technology, however, relevant social groups are delineated based on similarities among interpretations of technology. The term may be “used to denote institutions and organizations (such as the military or some specific company), as well as organized and unorganized groups of individuals” (Pinch & Bijker, 1987, p. 30). The concept of relevant social group is similar to Boland andTenkasi’s (1995) notion of “communities of knowing”, in which groups develop unique social and cognitive repertoires that guide their interpretations of the world. Fleck’s (1979) concept of “thought collective”, defined as “a community of persons mutually exchanging ideas or maintaining intellectual interaction” (1979, p. 39) is also similar to the idea of relevant social group. For Fleck, cognition is considered a socially conditioned activity in which groups or communities play an essential role. Studies of social interpretation, therefore, must investigate the community within which collective understandings are constructed and used. In the case of GIS implementation, our interest was to identify groups that were capable of influencing the implementation and use of technology. Since GIS development and use normally occurs within the context of formal organizations, the structure of the organization becomes an
260
S. SAHAY and D. ROBEY
important influence on social interpretations of information systems. This necessarily involves the participation of multiple, interdependent social groups, typically representing specialized perspectives related to departmental or professional affiliations. For example, different groups are typically responsible for the component technologies of a GIS. Digitizers, analysts, programmers, remote sensing technologists, system administrators, GIS technicians, users, and policy makers all participate in one form or another in GIS implementation. These groups differ in their motivations for use, political and social interests, educational backgrounds, occupational culture, and power in the organization. These differences lead groups to develop different interpretations about GIS, and these interpretations impinge on each group’s behavior during implementation. As these groups interact, they shape the use and consequences of GIS. Because the nature and composition of social groups is likely to differ among organizations, the implementation and consequences of GIS are also likely to differ between organizations. The subjective meanings ascribed to technologies by relevant social groups may differ, and the differences in meanings may account for differences in implementation and use (Kling & Gerson, 1978; Orlikowski & Gash, 1994). Identifying relevant social groups, therefore, becomes an important element of any research on the social construction of technology. Several recent studies have revealed the importance of social groups and their social constructions of technology (Anderson, 1988; Bijker, 1987; Fulk, 1993; Mackenzie & Wajcman, 1985; Noble, 1984; Orlikowski & Gash, 1994; Pinch & Bijker, 1987). Interpretive fZexibility Interpretive flexibility refers to the capacity of a specific technology (or other knowledge system) to sustain the divergent interpretations of multiple relevant groups. In general, information technologies may be more interpretively flexible than, say, production technologies that are configured within specific locations. Information technology (especially its software component) presents fewer material constraints and is easily portable over different locations. GIS, in particular, consists of various component technologies for retrieving, storing, manipulating and presenting data. Data input is provided by digital maps, satellite imagery, and aerial photographs. Data are converted from paper maps and digital storage to GIS formats using different data conversion technologies. Data is stored in relational databases and interfaced with existing corporate databases like Oracle and Ingres through relational database integrators. Data are processed in different modules within the GIS system for editing, plotting, manipulating and representing data. In the front end, the system may be connected with decision support and expert systems. Because these many components are subject to multiple interpretations by many different groups, a GIS is exposed to a wide variety of social interpretations. Technological frames of meaning Orlikowski and Gash (1994) described the technological frame as a component of the larger frames of meaning by which groups understand organizational realities. Technological frames include the basic assumptions, beliefs, and expectations that people hold about a specific application of technology (Bijker, 1987), including not only the nature and role of the technology itself, but the specific conditions, applications, and consequences of that technology in particular contexts (Orlikowski & Gash, 1994, p. 178). Technological frames are likely to be shared within a relevant social group because members share common relationships and experiences with a particular technical application (Burkhardt, 1994; Fulk, 1993). Yet, each group’s technological frame may differ from those of other groups. For example, developers and users of information systems typically construct different frames of meaning because their interactions with a particular application differ and they come from different social positions,
GEOGRAPHIC
INFORMATION
SYSTEMS
261
educational backgrounds, historical circumstances, and interests (Beath & Orlikowski, 1994; Newman & Robey, 1992). Orlikowski and Gash’s (1994) notion of “congruence” of technological frames refers to the alignment of frames of different groups on key elements or categories. Congruence suggests a similarity in structure (common categories of frames) and in content (similar values on common categories) between groups, and implies that these groups have similar expectations for the role of information technology. Incongruence of frames suggests different expectations for different groups. The capacity for information systems to support multiple, and potentially incongruent, frames of meaning is greater for applications with more interpretive flexibility. A framework
for inductive research
The social interpretation of GIS was examined within the research framework illustrated in Figure 1. Depicted are several elements of the social context within which the implementation of information technology occurs. The social context is shown as influencing and being specifically those activities that comprise the influenced by organizational processes, implementation and use of information technology. It is assumed that the context is able to constrain implementation processes, but that implementation outcomes may be instrumental in revising or reinforcing that context. These mutual influences occur not once, but are ongoing, as the vertical arrows in Figure 1 denote. Also, the organizational processes are sequentially related, with their time-ordering represented by left-to-right movement in Figure 1. The proposed framework guides the inductive research process without indicating precisely how specific contextual conditions or organizational processes are related. In other words, no hypotheses are offered. Inductive research is most appropriate where existing theories are judged inadequate to explain the phenomenon of interest. Currently, our knowledge about the implementation of information systems in general, and GIS in particular, has not been synthesized into a coherent theoretical statement capable of supporting many testable propositions. It more closely resembles a jigsaw puzzle (Swanson, 1988). Although many of the important variables that contribute to successful implementation can be identified, their assembly into a comprehensive theoretical model is problematic and probably premature. Therefore, we adopted an inductive approach that focuses broadly upon the role of human agency in enacting the processes of implementation within a social context.
Organizational Processes
Fig. 1. Suggested relationships amongst the social context of implementation process.
and organizational
262
S. SAHAY and D. ROBEY
The reciprocal link between context and process in Figure 1 is drawn from structuration theory (Giddens, 1984) which has recently been used as an organizing frame for research on information technology (Barley, 1986; Orlikowski, 1992; Orlikowski & Robey, 1991; Orlikowski & Yates, 1994; Walsham & Han, 1993). Structuration theory does not provide a priori hypotheses about the degree or direction of influence of context or process in any given situation, so its use as an inductive framework is appropriate. The value of the framework is to guide research efforts by sensitizing data collection and data analysis. Research findings may also be organized within the framework to give them coherence and form (e.g. Orlikowski, 1993, 1996; Walsham, 1993).
METHOD Research design and data collection The study employed a research design that enabled a comparative analysis between two different county government organizations that were in the process of implementing GIS. Given the pseudonyms of “North County” and “South County”, these two neighboring counties used the same GIS software (called Arc/Info). To allow for both intra- and inter-site comparisons, the research design was deliberately chosen to include one technical support group and two user departments from each organization. In North County, the technical support group was the Technical Services Department and the user groups were the Planning and Land Management Departments. In addition, two persons outside of these three departments (one each from the Legal and the Regulation Departments) were identified as relevant to the implementation of GIS and were also interviewed. In South County, the technical support group was the Office of Computer Services, and the two user groups were the Planning Department and the Environment Resource Management Department. Within each organization, the sampling plan was flexible and evolved with the research needs. Key respondents were initially identified from the selected departments based on the criterion of potential relevance to GIS implementation. Discussions with these key people provided a deeper understanding of the social networks within and across departments, and additional respondents were selected from these networks. The field work lasted 6 months, during which a total of 60 interviews were conducted. In North County, 32 interviews were conducted with 31 different people. (One respondent was interviewed twice, the second interview serving as a point of cross-verification of certain issues). Twenty-nine of the 3 1 respondents belonged to one of the pre-selected departments, and two belonged to departments not previously identified (Legal and Regulation). In South County, 28 interviews were conducted with 29 people (in one case two people were interviewed together). Prior to the interviews in these two sites, a pilot study including 15 interviews was conducted in an agency of a third county that was using GIS. The objective of the pilot study was to refine the research methods, especially the sampling and data gathering procedures. The primary method of data collection was through personal interviews conducted by one of the authors. Interviews were typically 1 hour long and semi-structured, so as to elicit responses about specific aspects of the technology and its implementation and consequences. Some initial questions about the features and limitations of GIS guided respondents toward the topics of research interest. Demographic details related to each respondent’s age, education, training and experience with technology were also obtained in the interviews. Questions were also asked about the process of implementing GIS. For example, respondents were asked how the organizational structure was established to oversee GIS implementation, which department originated the
GEOGRAPHIC
INFORMATION
263
SYSTEMS Least detail Most abstraction t
/*\ Technological Frames of Meaning
/
Aligning Themes with Relevant Social Groups
\
Grouping Split Data into Themes Splitting Data into Coded Categories
Raw Data: Transcribed Interviews \I
Most detail Least abstraction
Fig. 2. Inductive process followed in data analysis.
technology, and whether the GIS was deployed as a distributed or centralized system. Interview reports for this information were supplemented by archival sources. For example, in North County, relatively more objective information about the deployment of GIS was obtained through the study of a document that described the organization’s business plan. Data analysis The method for analyzing the interview data was conceived as a process of continuing refinement, moving from the raw transcribed interview text toward more general theoretical inferences. Figure 2 portrays this process as a pyramid, with the apex representing the general inferences and the base representing the transcribed interviews. The intermediate steps represent the process of theoretical induction through data reduction. As the apex is approached, the level of abstraction increases while the level of detail decreases. The specific actions taken on the data included coding and splitting, integrating the split data to form themes, aligning themes with relevant social groups, producing context/process models for each research site, and drawing theoretical inferences from comparisons between research sites. Each of these steps is described below in greater detail. Coding and splitting. Data preparation required the development of a coding scheme to provide a means for identifying and later combining statements with similar meanings (Miles & Huberman, 1984). The authors began with a simple coding scheme that included broad categories describing the initiation, implementation and impacts of technology. Within these broad categories, we then tried to identify more detailed coding schemes that included the more specific interpretations expressed by the respondents. For example, text coded initially because it referenced the period of GIS initiation, was further coded to capture statements about the “need” that different groups of people felt for GIS, and the “capability” perceived at the individual and organization levels to use the technology and adapt to the changes that its use created. The initial coding scheme was refined through an iterative process whereby the authors independently read sample transcripts, coded text into preliminary categories, discussed
264
S. SAHAY
and D. ROBEY
Table 1. Coding schemes broad coding categories Broad coding categories
North County
South County
1. Initiation: What kind of interpretations people expressed about the organizational dynamics that surrounded the introduction of Arc/Info in the organization.
1.1 Need: Interpretations about the existing organizational inefficiencies in map production and spatial analysis and the need for acquiring the state of art in GIS technology Arc/Info. 1.2 Capability: How individuals felt their past experience in the organization with mapping technology would support the use of Arc/Info; the existing “digital culture” in the organization would enable a smooth transition to Arc/Info. 1.3 Software Selection: How the nature and content of the software selection process helped to initially position Arc/ Info in the organization.
1.1 Capability: How individuals felt their past experience in the organization with data processing and their own computer science kind of backgrounds would impact their use of Arc/ Info. 1.2 Software Selection: Interpretations about the lack of involvement of the various agencies in the selection process; the criteria for selection being primarily financial rather than the functional capability of the software.
2. Implementation: What people felt about the manner in which the implementation of Arc/Info was organized and managed.
2.1 Organization: What kind of organizational structures were set up to oversee the implementation process; the manner in which the knowledge transfer process from the developers to the users was managed and its perceived effectiveness. 2.2 Technology: The manner in which the technology was deployed in the organization (centralized versus distributed) and the nature of technical and organizational problems this configuration created.
2.1 Intra-organization: The organizational structure set up within the individual county agencies to facilitate GIS implementation. 2.2 Inter-organization: Relations between different agencies, for example, communication and knowledge transfer with respect to GIS. 2.3 Technology, intra-organization: Existing technical capabilities within individual agencies to absorb new technology. 2.4 Technology, inter-organization: The nature of access provided by the technology providers to the users, and the kind of issues it created for both groups.
3. Impacts: What people felt were the impacts that Arc/Info had had and was going to have on their work lives and the way in which the organization did business.
3.1 Current: Interpretations about the consequences that Arc/Info had had till date since its implementation in the organization. 3.2 Future: What consequences people felt that Arc/Info would have in the future in terms of jobs, nature of analysis and organizational culture.
3.1 Current: Interpretations about the consequences that Arc/Info had had till date since its implementation in the organization. 3.2 Future: What consequences people felt that Arc/Info would have in the future in terms of jobs, nature of analysis and organizational culture.
GEOGRAPHIC
INFORMATION
SYSTEMS
265
problems in the preliminary schemes, and modified the scheme. The process continued until the coding scheme shown in Table 1 was developed. Using this scheme, all interviews were split into coded segments. Formation of themes. We defined a theme to be a unifying idea representing the interpretations found in multiple coded segments. Themes were developed by integrating the split data on the basis of similarity in the meaning of concepts. The process began with the assignment of conceptual labels to the split data. Similarly labelled segments were then combined and assigned new labels that reflected the common theme of the combined statements. Aligning themes with relevant social groups. The social groups subscribing to the themes were identified by detecting commonalities among the respondents connected with the coded segments that made up each theme. Relevant groups formed around departments (e.g. the Planning Department in North County), organizational levels (e.g. managers), disciplines (e.g. geographers), and functions (e.g. initiators of GIS in North County). The different themes to which each group subscribed were then aggregated to describe the group’s technological frame of meaning. Producing context/process models for each research site. Using Figure 1 as a model, the relevant social groups and their frames of meaning were positioned between the components of social context and organizational processes. This resulted in a descriptive model specific for each research site that connected context with process. Derivation of theoretical inferences. The final step in data analysis was the derivation of more abstract theoretical inferences. The frames of meaning for the different groups were compared across organizations to reveal differences in interpretations of GIS. These interpretations were examined in relation to the organizational context and process in order to identify possible reasons for the differences. This analysis led to the generation of theoretical statements describing the relationships between social interpretations of GIS, the social context, and processes of implementation. Because these inferences were drawn from the contrasts between two sites, they represent statements of more general theoretical interest than the more limited descriptions of individual cases.
RESULTS The results are presented first for North County and then for South County. For each case, presentation of the results is divided into five parts. The first part provides an overview of GIS implementation in each site. In the second part, the social context is described, which refers to conditions that are capable of influencing the implementation processes. Third, the relevant social groups in the implementation process and their technological frames of meaning are described. Fourth, five organizational processes surrounding the implementation of GIS are presented: the initiation of technology, the transition to the new technology from existing methods of conducting such work, the deployment of technology in the organization, spread of knowledge about GIS, and the consequences of technology. Finally, each site is briefly summarized. The presentation of results for the individual sites is followed by a summary of the comparisons between sites. North County Overview of implementation. The implementation of Arc/Info in North County began in October, 1989. Arc/Info was initiated by means of a business plan, a prospective business
266
S. SAHAY and D. ROBEY
case justifying the need for Arc/Info, and a 72-page document commissioned by a software committee comprising senior officers from different user departments. The software selection process consumed 3 months, and Arc/Info was selected over two competing packages based on its technical and functional superiority. A distributed environment was chosen for implementation, wherein users held primary responsibility for developing applications, maintaining different coverages, and controlling the quality of individual data sets. Prior to the introduction of Arc/Info, the Geographic Sciences Division had been responsible for the analysis and maintenance of spatial data, using the Computer Vision and PC Arc/Info systems. The core group of five technicians in Geographic Sciences was disbanded and redistributed over five user departments when Arc/Info was first introduced on Sun workstations. Implementation of GIS was enabled through a three-tier committee structure. At the top level was the GIS sponsor’s group, which included representatives from top management. The second tier of the structure included GIS co-ordinators who were responsible for ensuring that their departmental practices adhered to organizational procedures and data standards for GIS. The coordinators were closely linked to the GIS sponsor’s group, with some overlapping membership. A subcommittee of the co-ordinators’ group, the GIS Technical Advisory Committee, included GIS experts drawn from each department. This committee was responsible for shaping the technical future of GIS through such tasks as the design of the data dictionary, development of procedures for data archiving, and repair of software bugs. The third tier of this structure was the GIS Users’ Group, which existed to enhance user awareness and to provide feedback from users to top management. Another group of GIS experts, called the GIS Support Group, was responsible for keeping the system running, loading the software, configuring the workstations, maintaining the data sets, managing the archiving of data, assisting the users with backups, solving user problems, and conducting training. Social context. The two facets of social context that we found to be relevant in our analysis were “structure” and “capability”. Structure refers to the configuration of the organization with respect to departmental responsibilities, especially relating to GIS. North County was configured as a unified structure in which the different departments were included under a common umbrella relating to mission, governance mechanisms and budget. This unified structure facilitated the implementation of GIS as a distributed system because the departments shared a common mission. The unified structure also increased the visibility of management’s efforts with GIS. For example, most departments were aware of the rigorous benchmarking process and the involvement of a popular consultant in writing the business case. Arc/Info originated in the Geographic Sciences Division in North County, and the influence of geographers on GIS implementation was very evident. The geographers subscribed to a holistic philosophy, consistent with their academic discipline, in which users were assumed to be mature enough to appreciate a system that required data and programs to be widely shared. The geographers favored a high degree of user training. Capability refers to the organization’s ability to deal with the demands of new technology. The geographers’ extensive experience with mapping software equipped them with the capability to absorb technologies that dealt with spatial data. Although most of the other groups (technicians, surveyors, and managers) were more familiar with digital data systems, their general understanding of spatial data and mapping systems enhanced their capability to deal with Arc/Info. Moreover, geographic databases already existed in the organization, and it was easy for most groups to negotiate contracts for data conversion with external vendors. Thus, North County was very capable of developing and maintaining spatial databases to support GIS applications.
GEOGRAPHIC
INFORMATION
SYSTEMS
261
Relevant social groups. While a number of different groups were identified as relevant to the process of GIS implementation, the influence of the two most significant groups is discussed here. These groups are labelled as “technology initiators” and “geographers”. The technology initiators included senior geographers from various departments and some staff from the Planning Department, which was sponsoring the Arc/Info effort. The members of this group had all been involved in the early initiation activities, such as serving on the Software Selection Committee or in the formulation of the implementation plan. They positioned Arc/Info as a state-of-the-art technical solution to an existing organizational problem, but one which would not radically disturb the status quo. The group of geographers included senior people with a formal education in geography, and with significant experience in mapping software, including GIS. They played a significant role in the initiation of Arc/Info and later in implementation. Geographers showed a strong affinity toward GIS and its power to manipulate and display spatial information. They were also largely responsible for the training and support functions. Organizational processes. As mentioned earlier, five organizational processes are described: initiation of technology, transition to the new technology, deployment of technology, spread of knowledge, and consequences of technology. 1. Initiation. Arc/Info was positioned as a state-of-the-art technical response to the organizational need for new methods to conduct spatial analysis. The inefficiencies of existing methods and systems were described by many respondents, but especially by members of the technology initiators group. One member of the Software Selection Committee described the time delays that had previously existed when getting maps from the Geographic Sciences Division: “Whatever you wanted maps or information, you requested it from them, and they put it in their stack of priorities and gave it back to you whenever they could get to it”. The technology initiators referred to Geographic Sciences as a “closed shop”, a term also documented in the business case and the implementation plan. The metaphor of closed shop conveyed the idea that relevant knowledge was not moving outside the Geographic Sciences Division. Inefficiencies in Geographic Sciences were seen to be the main impediments preventing the spread of knowledge. The initiators described Arc/Info as a new order of spatial analysis that would help to reduce these inefficiencies, and they positioned Arc/Info as an organizational effort, rather than a departmental one. A senior project manager of the Land Management Department described some of the steps taken to promote an organization-wide commitment to Arc/Info: “We had to build a business case. What is the pay-off of converting, establishing a GIS system, getting people to buy into it, especially the upper management?” To accomplish this, initiators sought representation of different departments in the software selection process, prepared a 72-page document detailing objective criteria and benchmarks, selected the criterion of demonstrated functionality, and employed a popular consultant to develop the business case. All of these actions supported the interpretation that Arc/Info was an organizational initiative in North County and produced the shared understanding that, indeed, “Arc/Info was the right choice”. A senior manager in the Land Management Department commented that the selection of Arc/Info “was a credit to the objectivity of the process”. 2. Transition. Prior experience and demonstrated capability in working with spatial technology enabled a smooth transition to Arc/Info in North County. Most people using Arc/Info had prior experience with spatial mapping software like Computer Vision, Autocad, and PC Arc/ Info. While there were mixed opinions about the value of such prior experience, people agreed that it produced familiarity with geographic terms and concepts. Also, a number of respondents had training in disciplines that encouraged spatial thinking, such as geography, graphic arts, urban planning and architecture. For example, one technician commented: “I went into art
268
S. SAHAY
and D. ROBEY
school and got out and went to DuPont and got into technical illustrations and charts and graphs”. Another technician had been exposed to different geographical concepts while working as a secretary in the Geographic Sciences Division. She commented: “It is just that any computer mapping program has been useful because we already know about calibrating and geo-referencing and real world coordinates. X’s and Y’s, lats and longs, and all those things were already familiar to me”. Also contributing to the smooth transition was the relatively long period of employment of many people working with GIS at North County. Many of the respondents interviewed had been employed by the organization between 10 and 15 years, and this long association had endowed them with the experience of shifting through successive generations of paper maps and automated systems. A change to GIS, therefore, was not seen as a radical change. As one technician described it: “Making use of digital mapping equipment was part of the culture of the district. Most people got used to the idea of digital maps instead of paper maps”. 3. Deployment. The geographers were primarily responsible for deploying Arc/Info in North County. Their holistic world view, which emphasized looking at the spatial relationship between different entities, allowed them to see problems against a larger, organizational backdrop rather than taking a parochial, departmental view. One senior geographer described the distributed arrangement for GIS using the analogy of a football game, in which team success depended on the performance of its individual team members and their co-ordination. The distributed deployment of the GIS was compatible with this team-based approach in which the challenge was to provide enough capability to the user departments to make them self-sustaining. A comment by a senior geographer, who held advanced degrees in both geography and computer science, symbolizes the geographers’ preference for distributed deployment over central control: “Maybe it is the geography part in me which says you need to distribute certain things. And then there is the computer science part in me that says that certain things need to be centralized”. As stated in the business case, the objectives for the distributed deployment of Arc/Info were to ensure user participation in development and to spread the responsibilities for database maintenance among those with the greatest interest in individual datasets. Management’s decision to disband the Geographic Sciences Division and to deploy its technicians to user departments was regarded very favorably. As a senior geographer saw it: “The smartest move we did was getting GSD [Geographic Sciences Division] kind of disbanded. It is not an IS or GSD thing. This is really an unselfish posture”. 4. Spread of knowledge. North County’s geographers emphasized the development and spread of conceptual knowledge. Training programs were designed to encourage spatial thinking and geography based concepts rather than learning by rote procedures for the use of menudriven software. The geographer in charge of the training programs explained: “I spend a few hours on making a map. The objective is to get information and not to make maps”. Training was focused on teaching users to visualize problems and to translate the need for geographical products into workable GIS solutions. This involved an understanding of relatively advanced concepts relating to spatial relationships such as topology, geo-referencing, and geo-coding. The philosophy underlying training was to make the users become self-reliant with the software because a distributed environment could be effective only when a mature user community understood how to meet its own needs. Other means for spreading knowledge were the efforts of the GIS Technical Advisory Committee, a group dominated by geographers, and the GIS Users’ Group, which included at least 50 active users. The Users’ Group helped to provide greater visibility to GIS technology, which further enabled the spread of knowledge. A common language emerged among users as they grew comfortable with speaking in terms of latitude, longitude, geo-coding, and other
GEOGRAPHIC
INFORMATION
SYSTEMS
269
geographical concepts. This further reinforced the informal spread of knowledge throughout the organization. 5. Consequences. Even though GIS was only in the second year of its 3-year implementation program, it was almost universally felt in the organization that Arc/Info had produced changes in departmental structure and the conduct of business. Arc/Info was seen as the cause of organizational changes, such as the manner in which intra-organizational communication took place. Individual departments communicated with each other more because they had become jointly responsible for maintaining high standards of integrity for the data that they all shared. Such high levels of interaction among departments was unthinkable prior to the implementation of GIS. A senior technician described this process as follows: “I think it has helped a lot with the environment here where everybody had put up walls around themselves. These walls have come down”. Arc/Info was also seen to have changed the conduct of spatial analysis in the organization, specifically its diversity, efficiency and accuracy. A greater range of applications, from monitoring permit violations over time to examining socio-economic shifts in the population, had occurred because of Arc/Info. Individual analyses also were more detailed than before. A geographer in the Planning Department explained that users asked for more detailed breakdowns of land use into the various sub-classifications compared to the basic land-use map previously required: “Now they want maps broken down in every possible category you can think of”. Most respondents felt that there was no danger that their skills would become obsolete. To the contrary, Arc/Info was viewed as creating both the need for more geographical information and the capacity to meet this need. A technician voiced the widely held belief that jobs in the organization were not threatened: “I see jobs as very secure because the demand for information is going to be 100 times more than what it was before”. Summary. In general, respondents in North County interpreted Arc/Info as a technology that was the state-of-the-art in spatial analysis, and one which could bring about improvements in existing methods of working with maps, without radically disturbing the existing organizational status quo. The distributed configuration in which Arc/Info was deployed was seen as a means to empower the user departments with the capability of maintaining and supporting their respective databases, and developing applications that were relevant to their individual needs. The users, while accepting the opportunity of self-empowerment, were quite sensitive to the need that their databases and applications satisfy organizational standards established and monitored by Geographic Sciences. The links between the social context, relevant groups, and the processes of Arc/Info implementation in North County are summarized in Figure 3, which adopts the form of the more general framework in Figure 1. The unified organizational structure made it easier to develop shared interpretations about the implementation effort, thus facilitating an organization-wide commitment to Arc/Info. The initiation of Arc/Info in the Geographic Sciences Division placed geographers in a prominent role during the implementation process. Their holistic conception of technology influenced the deployment of GIS and the spread of knowledge in the organization. The prior capability of users in dealing with spatial concepts developed through working with different forms of mapping systems, and this experience enabled a smooth transition to Arc/Info. These favorable elements provided momentum for GIS implementation, and by the second year more than 50 workstations had been installed compared to the 12 originally budgeted. The consequence of GIS implementation was the universal acknowledgement that GIS had transformed not only the conduct of spatial analysis in North County but also its organizational structure.
270
S. SAHAY and D. ROBEY
+
f
TechnologyInitietDE
/ Various
Groups
Geographers
Various
Groups
positionedAtvlnfo as total orgsnizatlonsl response to
sharedneeds
prior capabIlities effectedsmooth bansfuon 01 AICllllfCl
distributed
environment
provided conceptual knowledge abouiG/S
perce~tfonof GfS sffecthg organizatlonslsmlcture
\ I’
Initiation
Transition
Deployment
Organizational Processes
Fig. 3. Intra-site model of context and process: North County.
South County Overview of implementation. South County was organized so that discrete, legally independent agencies had their own mission statements, governance mechanisms and budgets. The Office of Computer Services was the central unit which provided data-processing services to all other agencies in the country. With the introduction of Arc/Info, GIS was added to this range of services. The other two agencies that were studied (Environmental Resource Management and Planning) were users, depending upon Computer Services to create their GIS applications. GIS was initiated in South County in 1987 through the efforts of a small group of people at the Office of Computer Services, including the Director and the Technical Support Supervisor. The primary initiation efforts were aimed at getting financial support from potential user agencies for GIS development. The Technical Support Supervisor succeeded in getting county agencies such as Police, Zoning, Waste, and Elections to contribute to a pool of funds in the first year. In future years the different agencies were expected to contribute based on need and capability. South County’s Office of Computer Services also contracted with a private utility company to develop a common geographic database for the county on a cost-sharing basis. Unlike most of its data-processing functions, which were designed to run on an IBM mainframe computer, the Office of Computer Services implemented Arc/Info on a VAX computer system. A centralized configuration was adopted, with Computer Services maintaining the central databases, providing application development and support services to other county agencies, and conducting training programs for users. The Office of Computer Services’ centralized Application Development and Technical Support groups were made responsible for developing GIS applications, which were provided for users on a contractual basis. Users could
GEOGRAPHIC
INFORMATION
SYSTEMS
271
meet their GIS needs only by hiring the Computer Services analysts, who then assumed responsibility for a project. GIS was introduced to the Environmental Resource Management and Planning Departments under different circumstances. Prior to the development of actual GIS applications, Environmental Resource Management decided to develop a data management system. Further impetus was provided by federal programs such as the Storm Water Utility and Federal Emergency Management Agency which mandated that governments produce maps for storm water outfall and flood protection layers using a GIS. The Planning Department’s interest in GIS was initiated through an informal chance encounter with some members of North County’s GIS group. This led to an exchange of ideas on the use of GIS to support South County’s planning functions. Social context. As in our description of North County’s social context, both structure and capability proved to be relevant aspects. South County’s organizational structure consisting of independent agencies was seen to be the most relevant aspect of the social context. The arrangement whereby the Office of Computer Services provided GIS services to county agencies produced fragmented development of discrete applications. Because the funding for GIS projects was initially drawn from the pool of funds contributed by user departments, the Office of Computer Services tended to support the “richer” departments that had contributed earlier. As a result, the remaining departments experienced difficulty in initiating and sustaining GIS projects. The Office of Computer Services’s capability was primarily as a traditional “data-processing shop” running most of its applications on its IBM mainframe. The traditional training and educational experiences of Computer Service’s staff in data processing favored the centralization of technical resources rather than their distribution into the user community. Controls and standards were regarded as essential to all system development efforts, including GIS. Moreover, Computer Services personnel generally lacked experience in dealing with spatial information, which made the transition to Arc/Info relatively abrupt for most individuals. Relevant social groups. Relevant groups in South County were distinguished mainly on the criterion of whether the group was providing or receiving technology. The providers were from Computer Services and were involved in the initiation of Arc/Info in the county. Only after Arc/ Info was procured and deployed as a centralized resource, did the users become relevant to the process of implementation. Computer Services believed that their client agencies should focus on activities specific to their respective mission statements. Users, however, regarded such arguments as self serving and not respectful of users’ needs. From the users’ perspective, the Office of Computer Services purposefully kept its clients dependent on overpriced services, which were undertaken according to priorities assigned by Computer Services. While personnel in the Office of Computer Services felt that Arc/Info had some impacts in South County, the users generally disagreed. Besides their organizational affiliations, the educational and work backgrounds of the two relevant social groups differed. People with data processing backgrounds emphasized technical standards, controls, formats, and product appearance in terms of menus and screen layouts. Users came with backgrounds specific to different application domains, such as environmental science, marine biology, and planning. Users were more interested in the capabilities of the system to support their needs, e.g. whether the scale and resolution of a map appropriately reflected the real world phenomenon in which they were interested. Organizational processes. As in North County, five organizational processes were associated with implementation of GIS. 1. Initiation. Initiation was a relatively isolated effort by a small group of staff from the Office of Computer Services, headed by the Technical Support Supervisor who had no prior experience
272
S. SAHAY
and D. ROBEY
with mapping systems. The participation of users was limited to attendance at fund raising seminars. Arc/Info was positioned as a product that the Office of Computer Services had added to its existing range of data-processing services, which would then be marketed to the rest of the county. Arc/Info was initiated in South County with little explicit attention to the need for spatial analyses or the methods for selecting software. The project manager referred to economic criteria as the primary basis for choosing Arc/Info over two competing systems: “The deciding factor was mainly the cost, and the other proposals were double that of Arc/Info”. A central pool of funds was established to support GIS development in South County. Some of the county’s departments, like Police and Elections, made substantial contributions to this fund while others waited to see how GIS succeeded before they made significant financial commitments. The Director in the Planning Department noted that his department was limiting its contribution to the country-wide fund: “Right now we are hoping that the land use [contribution in the form of maps] will allow us to participate without having to make an out-ofpocket contribution”. A user from Environmental Resource Management acknowledged that disparities produced by the funding formula led to a mismatch between the efforts by Computer Services to develop GIS applications and the actual need for GIS products in user organizations: “We have great disparities in terms of which departments can afford them. The rich departments move ahead like gang-busters”. 2. Transition. The transition to Arc/Info was relatively abrupt because of Computer Services’ lack of experience in working with spatial systems and geographical databases. For 20 years it had functioned as an IBM data-processing shop without having worked with mapping programs. Most of the respondents involved with GIS came from traditional data-processing backgrounds and had little or no experience or training in GIS. For example, the Director of GIS had 16 years of data-processing experience, and the Technical Support Supervisor had worked in this area for more than 10 years. A technician in the Office of Computer Services felt that this prior experience made the transition to map-based systems abrupt and problematic: “I have been in IBM in all my career. I have never dealt with graphics before. I was strictly a number cruncher. So there was that other stumbling block I had to overcome”. 3. Deployment. In the centralized computing environment of South County, Computer Services played the lead role in development, support and training functions. Both the users and data-processing analysts perceived technical advantages arising out of centralization-the reduction of data redundancy, improved quality control and gave speedier access. However, opinions varied about the organizational benefits of centralization. The analysts believed that centralization was essential to the rapid growth of GIS capabilities among users. A technician in the Office of Computer Services commented: “In GIS systems, control was necessary because there was a need to set standards and really enforce them”. A senior manager in Computer Services felt that centralization prevented user departments from hiring technical personnel to develop their own systems: “[Users] hire people and they don’t know what to do with them. The problems arise when departments try to do things which are not part of their mission”. The users disagreed with the argument that centralization enabled the growth and spread of GIS applications. A user from Environmental Resource Management observed: “If anything, the centralized GIS has inhibited growth because of the problems in product development with people seeing it as more trouble than it is worth’. Users suspected the Office of Computer Services of keeping them in a dependent relationship, restricting training, controlling purchases of all computer equipment in the county, and limiting users’ access to the system. One user in Environmental Resource Management described Computer Services’ practice of limited access: “They have put on a hardware lock. Office of Computer Services does not give command level access to run the system”. Another user felt that analysts in Computer Services wasted time on
GEOGRAPHIC
INFORMATION
SYSTEMS
213
superficial details during the systems development process rather than concentrating on the analytic contents of the system: “They spend 18 to 20 weeks on things like making the black line that goes around the border thicker”. The analysts felt that the users were always in a rush, wanting products within unreasonably short times and constantly modifying product specifications. While the users felt the analysts should use newer systems development methods that consumed less time and expense, an analyst expressed the belief that a return to traditional system development practices would clear up some existing problems: “There is a need to step back and get more into traditional data processing methodologies than we have in the past”. 4. Spread of knowledge. The GIS training conducted by the Office of Computer Services for county agencies was largely procedural, encouraging the spread of operational, rather than conceptual knowledge. For example, training showed users how to digitize and how to enter information about water sources into the database. The users called this stance a “closed-shop” attitude because of Computer Services’ unwillingness to transfer much conceptual knowledge about GIS. The users felt restricted to operating menu-driven systems and entering data. One user in Environmental Resource Management felt that GIS training had given her limited exposure: “I have not worked with GIS before this project, and I have not learned a whole lot about it, other than how to enter data into the system”. A staff member from Office of Computer Services confirmed that procedures were emphasized in training but felt that the programs were adequate: “Well, in most cases if we design a menu-driven system for a department which will maintain a map, we provide a level of training such that the user can pull up the menu”. 5. Consequences. Arc/Info had limited consequences in South County. In the Office of Computer Services, some effects on the size of the department were apparent with more supervisors and programmers being hired. This trend was expected to continue as more departments initiated GIS applications. In the user departments, increases in staff were not apparent because the applications were developed and maintained by the Computer Services analysts. As a user from Environmental Resource Management bluntly stated: “I don’t see any organizational change because we have GIS”. A biologist in the same department noted that the only changes in work associated with GIS involved data entry, and that too was accomplished by using temporary interns. The Planning Department had yet to make a firm commitment to GIS and had not experienced any significant consequences. While some surveyors had apprehensions about losing their jobs, the consensus was that GIS technology would not affect employment. Changes in skill requirements had been insignificant, focused narrowly on skills needed for data entry in geographic databases. Overall, consequences in South County were limited to GISs use as a device for data entry and to produce attractive maps, while supporting a small amount of spatial analysis. Some managers in user agencies feared that people might waste time by indulging in geographical analysis just because the tools were available, not because a project required it. These managers were guardedly enthusiastic about the possible consequences of GIS, and they realized that changes would be slow, given budget and manpower restrictions and the customarily sluggish operation of the county’s bureaucratic structure. Summary. Broadly, control and power can be viewed as the central interpretations surrounding Arc/Info in South County. The group of providers from the Office of Computer Services drew upon their technical norms of standards and formats, which are emphasized in the computer science education and data-processing work experience, to legitimize the users’ dependence upon them. The users viewed this form of control untenable, and attributed it to Computer Services’ attempts to profit at their expense. These differing perspectives on control and power led to numerous operational problems related to communication, training and
274
S. SAHAY and D. ROBEY Social Context
I
I
4
Is
4 ,
ocs
be marketed to county agencies
regarded GIS as transition to a new unhw-se
disagreed on centralized deplojment
provided procedual knowledge about G/S
exlsflng
.stNctUreS reinforced, with no major consequences
Initiation
Organizational Processes
Fig. 4. Intra-site model of context and process: South County.
support, and budgets, which impeded the process of implementation and limited GIS’s organizational impact. Figure 4 summarizes the links between social context, relevant social groups, and the processes of implementation for South County. The sharp differences between the technological frames of the users and the Office of Computer Services staff helped to shape the organizational processes surrounding Arc/Info implementation. The providers initially positioned Arc/Info as a revenue producing service to be added to the inventory of services rendered to the county. The lack of prior experience with spatial systems made the transition to Arc/Info a relatively abrupt process. The providers deployed Arc/Info as a centrally controlled resource, both to protect traditional data processing concerns about efficiency and security and to fulfill their commercial objectives. The providers emphasized the spread of procedural knowledge in their training, and this inhibited the development of GIS capabilities among the users. Consequently, the dependence of users on the Office of Computer Services was maintained. Summary
of intersite
comparisons
Table 2 summarizes the main contrasts between North and South Counties. With regard to social context, two main differences were observed. First, North County’s more unified organizational configuration accommodated the decentralized management of information resources whereas South County’s differentiated structure placed the management of information technology within a distinct, centralized agency. Second, the sites differed in their capacity to absorb Arc/Info. North County had greater capability for dealing with spatial data and mapping systems, whereas South County had negligible experience with such systems.
GEOGRAPHIC
INFORMATION
215
SYSTEMS
Table 2. Summary of comparisons North County
South County
Organizational structure was unified, with decentralized management of information technology. High capability to absorb GIS due to continuous experience with earlier versions of mapping systems. GIS was positioned as an organization-wide response to shared needs. Initiating group was geographers with holistic view of organization and application systems. Transition to GIS was smooth and continuous due to existing capability.
Organizational structure was differentiated with centralized management of information technology. Low capability to absorb GIS due to Office of Computer Services’ traditional orientation as data processing department. GIS was positioned as revenueadding service provided by Office of Computer Services. Initiating group was focused on marketing service to user departments. Transition to GIS was abrupt and discontinuous due to lack of prior capability with mapping systems. GIS was deployed in a centralized technical configuration, controlled by the Office of Computer Services. Data processing personnel had responsibility for developing applications for users. Procedural knowledge was conveyed to users through training conducted by the Office of Computer Services.
Basis for comparison Configuration
Capability
Initiation
Transition
between sites
Deployment
GIS was deployed in a distributed technical configuration supported by a central group. Users had primary responsibility for developing and maintaining applications.
Spread of knowledge
Conceptual knowledge about GIS was spread among users through training conducted by the initiating group of geographers. GIS was widely acknowledged as responsible for transforming the conduct of work and the structure of North County’s organization.
Consequences
The interpretations
surrounding
Arc/Info
were remarkably
GIS was slowly implemented with little consequence for users. Minor increases in staffing in Office of Computer Services were attributed to GIS.
different
in the two organizations.
In North County, Arc/Info was seen as a state-of-the-art tool in spatial analysis, one that helps to empower the user departments with the capability to work independently with geographical
datasets while contributing to overall organizational needs. By contrast, in South County Arc/ Info was seen as an instrument of control and power. The providers of technology were seen to use Arc/Info as a medium by which they could reinforce the dependence of the users upon themselves. The users resented this form of control, and the opposing frames of meaning led to a number of operational problems that impeded the process of implementation. Differences in the social context, and the nature of interpretations of Arc/Info, contribute to the differences in the organizational processes in the two sites. Initiation of Arc/Info took place very differently in the two sites. In North County, geographers positioned Arc/Info as a functionally superior solution to organization-wide needs, whereas South County’s Office
276
S. SAHAY and D. ROBEY
of Computer Services positioned Arc/Info as a revenue-adding product. The transition to Arc/ Info was markedly different in the two sites. North County’s tradition of working with spatial systems enabled a smooth transition, while in South County transition was more abrupt. Intersite comparisons also showed contrasts in deployment. North County adopted a distributed computing environment where users were expected to assume responsibility for developing and maintaining applications. South County adopted a centralized computing environment in which the Office of Computer Services retained responsibility for building applications and maintaining data. The spread of knowledge about Arc/Info in North County was focused on giving users a conceptual understanding of GIS, while in South County procedural knowledge was emphasized. Finally, North County respondents saw Arc/Info as a causal agent that redefined their ways of conducting business, patterns of intraorganizational communication, and the distribution of responsibility. In South County, changes were limited to some hiring in the Office of Computer Services, with little changes in jobs or structure in the user groups.
DISCUSSION The comparisons between research sites allow the formulation of specific inferences in three general areas: (1) the relationship between structure and initiation, deployment and spread of knowledge; (2) the relationship between capability and transition, deployment, and spread of knowledge; and (3) organizational consequences of GIS. Each group of inferences is based on specific comparisons between two sites and used to build general theoretical arguments about the implementation of information systems (Lee, 1989; Yin, 1984). These inferences are illustrated in Figure 5.
Organizational Context
Organizational Processes Fig. 5. Inferences
supported by the contrasts between sites.
GEOGRAPHIC
INFORMATION
SYSTEMS
217
The relationship between organizational structure and initiation, deployment and spread of knowledge Organizational structure was identified in this research as a major point of contrast between the two sites. We can associate differences in organizational structure with three aspects of the implementation process: initiation, deployment, and spread of knowledge. Specifically, a centralized and unified organizational structure is associated with initiation of GIS as an organization-wide technology (la), deployment of GIS as a distributed technology (lb), and more rapid spread of GIS knowledge (lc) compared with a decentralized and differentiated organizational configuration. These effects are realized because a unified organizational structure enables greater commonality among the social interpretations of new technology and the creation of a single vision. Information is more widely shared in a unified organization than in a differentiated structure, allowing congruent technological frames of meaning to emerge. The uniformity of governance mechanisms and a common mission statement allow management to convey its commitment to new technology, which further strengthens shared interpretations. In a decentralized, differentiated configuration, deployment of the technology is more likely to be restricted to one organizational unit, thus keeping knowledge from spreading very rapidly. Differences in unit goals and their spatial separation may impede the formation and spread of shared interpretations and produce multiple, incongruent frames of meaning. Overall, configuration is likely to produce important implications for the initiation, deployment, and spread of knowledge about new technology. These observations are consistent with institutional theory (DiMaggio, 1988; Oliver, 1991) which argues that the proximity of sub-units to a central unit affects the formation of lasting social interpretations. In a differentiated structure, units tend to be more distant from each other and from central co-ordinating units. The differences in mission statements, funding capabilities, and physical locations lead to contrasting frames among relevant social groups. In a unified organization these barriers do not exist, making it easier for common technological frames to develop. Support may also be obtained from contingency theories of organizational structure. Decentralized, differentiated structures are more likely to produce conflict among sub-goals, and differentiated structures must devise integrating mechanisms in order to be effective (Galbraith, 1977; Lawrence & Lorsch, 1967; Thompson, 1967). This first group of inferences is supportive of King’s (1983) analysis of centralized and decentralized computing. King argued that the organization of computing resources is anchored in more fundamental questions of organizational power and control, and the concentration of computing resources in a single unit of an organization is likely to preserve the unit’s power over the users of the technology. By contrast, a distributed deployment of computing resources encourages the spread of knowledge, thereby empowering the users. Technological capabilities are more likely to expand and spread in an organization where new technology is configured in a distributed rather than a centralized manner. The distributed computing environment assumes the presence of a self-reliant and mature user community, one that possesses both conceptual and procedural knowledge of the technology. A centralized technological environment prevents the spread of knowledge by placing control in the hands of a single group. Centralized-computing environments, in effect, fulfill their own prophesy that users lack the discipline and skills to develop and use information technology efficiently. The technological frames of meaning developed by relevant groups provide the explanation of the effects summarized in the first set of inferences. The more closely shared frames of meaning in a unified organizational configuration produce initial agreement on the prospective value of technology and affect the technical configuration adopted for the deployment of the technology. Key elements of the initiators’ technological frame are the
278
S. SAHAY
and D. ROBEY
group’s assumptions about control and power, the degree of faith and trust they have in others, and their more global orientation to organizational needs. These frames of meaning also explain the interest in spreading knowledge about new technology widely rather than restricting it. The relationship between capability and trunsition, deployment,
and spread of knowledge
Our second set of inferences concerns the links between organizational capacity and the organizational processes of transition, deployment, and spread of knowledge. We propose that greater prior capability to absorb new technology produces a smoother transition to new systems (2a). Where new technology reinforces existing technological capabilities, a transition may be interpreted as an extension or continuation of prior learning. Where new technology is interpreted as a departure from existing technological capabilities, the transition becomes more abrupt and difficult. Continuities of technology build upon an established base of knowledge whereas discontinuities require the creation of new knowledge before the transition can occur. Capability, in turn, is affected by deployment and spread of knowledge. Specifically, distributed deployment of information technology increases the capability of the user community, whereas centralized deployment restricts user capabilities (2b). Furthermore, capability bears a reciprocal relationship with spread of knowledge, indicated by the two-headed arrow (2~): spread of knowledge about technology increases capability while capability is affected by spread of knowledge. In other words, the capacity to understand technological innovations is enhanced when the knowledge about them is spread among the user community, and knowledge spreads faster when user capability is high. The relative continuity of transitions to new technology has been the focus of prior research undertaken at the industry level of analysis (Tushman & Anderson, 1986). Tushman and Anderson suggested that technologies are “competence enhancing” when existing capabilities are shifted incrementally and “competence destroying” when entirely new competencies have to be developed. Their logic can be extended to the organizational level of analysis, where it becomes consistent with explanations of organizational change that involve learning (e.g. Attewell, 1992; Cohen & Levinthal, 1990; March, 1991). Organizational learning requires that knowledge barriers be overcome when new technologies are adopted. Gradual technological innovation is more likely to be accepted than discontinuous, quantum innovation because the latter creates significant knowledge barriers that must be overcome before new technologies can be effectively used. The capability to absorb new technology is enhanced by members’ prior experience, as affected by the deployment of similar (often earlier generations of) technologies. Capability and spread of knowledge are thus interwoven in a process of continuous learning, enabling smooth transitions of related innovations. Organizational
consequences
of GIS
Our findings do not support any simple conclusions regarding the consequences of GIS technology, despite the striking differences in the consequences realized in the two research sites. We conclude, however, that the organizational consequences of information technology are conditioned by the interplay between social interpretations, context and the organizational processes that occur during implementation (3). This inference is offered as a means of capturing the influence of the processes of implementation, as affected by the contextual conditions of configuration and capability. Simply stated, there is no single contrast between sites that logically accounts for Arc/Info’s different consequences in the two organizations studied. The consequences can certainly not be attributed to the features of the technology, which were essentially the same. Rather, we conclude that technology’s organizational
GEOGRAPHIC
INFORMATION
SYSTEMS
219
consequences arise out of a complex interplay of the technological frames of different relevant social groups within the context and processes of implementation. This conclusion is consistent with prior efforts to explain the consequences of information technology by regarding technology as an occasion for revising organizational structure. Empirical findings by Barley (1986) Burkhardt and Brass (1990) Orlikowski (1993, 1996) Orlikowski and Yates (1994), and Walsham (1993) are supportive of this conclusion. These studies avoid the deterministic logic that still governs many predictions of information technology’s consequences, choosing rather to employ the idea of emergent causality to explain technology’s consequences for organizations (Markus & Robey, 1988). Thus, similar technologies may be introduced to support similar kinds of work, but the process of implementation may be so different that extreme contrasts in their social consequences may be realized. Our inference provides for diversity of consequences from the same technology because outcomes emerge from different social processes.
CONCLUSIONS The inferences drawn from this interpretive study are abstractions based on the differences in the process of GIS implementation in two comparable organizations. The striking contrasts between the sites suggest that new technology, in these cases GIS using Arc/Info software, is subject to differing interpretations depending upon the social context and the process of implementation. The differences in social context appear to condition the processes of initiation and deployment of the technology. As the GIS technology was introduced and implemented, differences in social meaning and their importance to the implementation and consequences of GIS emerged. Our research framework emphasizes the links between social interpretations and the context and processes of implementation. Examining how different groups interpret GIS technology within their social frames of reference yields valuable insights for understanding actual patterns of implementation. This approach to analysis is drawn from the stream of work on the social construction of technologies. As artifactual systems of knowledge, information technologies acquire social meanings that affect the consequences of the technology. The research method adopted for this study is sensitive to the assumptions underlying social construction and seeks to identify the relevant social groups and their technological frames. The value of this research lies not in the ability to generalize results directly to other contexts. Each context is different, and it is expected that different contextual and processual elements will interact to produce different consequences. What is true for GIS in local county governments may be untrue for GIS or other technologies in, say, private enterprises. Even settings that appear to be comparable to the two organizations in this study may manifest different dynamics of implementation and different consequences. The value of this research, rather, lies in its demonstration that similar technologies may occasion different social processes and consequences. Technology itself should not be regarded as a determinant of organizational impacts; the consequences of nearly identical technologies can (and do) vary considerably. Our research focuses upon the manner in which technology is introduced and how it acquires social meaning within the context of implementation. Information technologies can transform organizations, but only when participants in the process share such an understanding. Where shared technological frames do not emerge out of the process of implementation, even the most radical technologies may be implemented without significant consequence.
280
S. SAHAY and D. ROBEY
While these conclusions respect the tenets of emergent, non-determinate causality, our framework and findings help to locate the elements of context and process that potentially influence social interpretations of technology. Any research seeking to connect the introduction of information technology with changes in organizational functioning and effectiveness must include some consideration of the implementation process that unfolds within the existing social context. The framework that guided this investigation can provide this more general guidance for studying other technologies in other organizational settings. The framework may also be an effective sensitizing device for managers planning the implementation of a new technology. Managers should avoid the simple assumption that technology will be understood the same way by different groups of people. Variations in social meaning arise because of different connections that groups have with the technology and their own historical backgrounds. As we have seen, these meanings can influence the implementation process and its outcomes, including the evaluations of ultimate success or failure. Managers should be sensitive to their own technological frames, especially their definition of the overall “benefit” of technology for the organization. If managers can identify other relevant groups and influence their interpretations of new technology, they may reduce the chance that technology initiatives will fail. However, the interpretations of other relevant groups are not subject to simple managerial manipulation, so this recommendation cannot be taken as a quick fix to the problem of implementing new systems. The results of this study are especially relevant to those interested in the implementation of GIS. The primary characteristic of the GIS literature in this area to date has been the dominance of a “factors approach”, wherein the conditions under which implementation is impeded or enabled, are studied (e.g. Crosswell, 1991). This approach has been criticized for omitting detail and being too static and simplistic, given that GIS implementation is inherently dynamic and complex in nature. In this vein, Campbell and Masser (1992) noted the “urgent need for in-depth case studies that evaluate the experience of users in relation to the vital processes of implementation”. Similar criticisms have been expressed in the more general literature on information systems (e.g. Franz & Robey, 1987). Our research directly responds to this demand and demonstrates the importance of social interpretation to understanding the implementation and consequences of information systems.
REFERENCES Anderson, H.W. (1988). Technological trajectories, cultural values and the labour process: The development of NC machinery in the Norwegian shipbuilding industry. Social Srudies of Science, 18, 465-482. Attewell, P. (1992). Technology diffusion and organizational learning: The case of business computing. Organization science, 3, 1-19. Barley, S. (1986). Technology as an occasion for structuring: Evidence from observations of CT scanners and the social order of radiology departments. Administrative Science Quarrerly, 31, 78-108. Beath, C.M. & Orlikowski, W.J. (1994). The contradictory structure of systems development methodologies: Deconstmcting the IS-user relationship in information engineering. Information .Systems Research, 4, 350-377. Berger, P.L. & Luckmann, L. (1967). The social construction of reality: A treatise in the sociology of knowledge. New York Doubleday. Bijker, W.E. (1987). The social construction of bakelite: Towards a theory of invention. In Bijker, W.E., Hughes, T.P & Finch, T. (Eds), The social consfruction of technological systems, Cambridge, MA: MIT Press, pp. 159-190. Boland, R.J. & Tenkasi, R.V. (1995). Perspective making and perspective taking in communities of knowing. Organization
Burkhardt, Academy
Science,
6, 350-372.
M.E. (1994). Social interaction of Management
Journal,
effects following
37, 869-898.
a technological
change:
A longitudinal
investigation.
GEOGRAPHIC
INFORMATION
281
SYSTEMS
Burkhardt, M.E. & Brass, D.J. (1990). Changing patterns or patterns of change: The effects of a change in technology on social network structure and power. Administrarive Science Quarterly, 35, 104-127 Campbell, H. & Masser, I. (1992). GIS in local government: Some findings from Great Britain. Internarional Journal of Geographic
Cohen,
Information
W.M. & Levinthal,
Administrative
Science
Systems, 6, 529-546.
D.A. (1990).
Absorptive
Crosswell, P.L. (1991). Obstacles to GIS implementation Journal,
capacity:
A new perspective
on learning
and innovation.
Quarterly, 35, 128-152.
and guidelines to increase the opportunities
for success. URISA
3, 43-56.
P.J. (1988). Interest and agency in institutional theory In Zucker, L.G. (Ed.), Insti~tional Patterns and Cambridge, MA: Ballinger. Ein-Dor, P. & Segev, E. (1993). A classification of information systems: Analysis and interpretation. Information Sysrems Research, 4, 166-204. Fleck, L. (1979). Genesis and development of a scientific fact. In Trenn, T.J. & Merton, R.K. (Eds), Chicago, IL: University of Chicago Press, (first published in German 1935). Fletcher, P.T., Bretschneider, S.I. & Marchand, D.A. (1992). Managing information technology: Transforming country governments in the 1990s. Syracuse, New York: Center for Science and Technology, School of Information Studies, Syracuse University. Frank, A.U., Egenhofer, M.I. & Kuhn, W. (1991). A perspective on GIS technology in the nineties. Photogrammetric DiMaggio,
Organizarion.
Engineering
and Remote Sensing,
57, 1431-1436.
Franz, C.R. and Robey, D. (1984). An investigation of user-led system design: Rational and political perspectives. Communications of the ACM, 27, 1202-1209. Franz, C.R. & Robey, D. (1987). Strategies for research on information systems in organizations: A critical analysis of research purpose and time frame. In Boland, R. & Hirschheim, R. (Eds), Critical issues in information systems research, New York: Wiley, pp.205-225. Fulk, J. (1993). Social construction of communication technology. Academy of Management Journal, 36, 921-950. Galbraith, J.R. (1977). Organization Design, Reading MA: Addiwon-Wesley. Giddens, A. (1984). The constitution of society: Outline of the theory of structure. Berkeley, CA: University of California Press. GIS World Inc. (1991). 1991/92 international GIS source book, Fort Collins, CO: Author. Goodchild, M.F. (1995). GIS and geographic research. In Pickles, J. (Ed.), Ground truth: The social implications of geographical information systems. New York: Guilford, pp.3 l-50. Hicks, D. & Porter, J. (1991). Sociology of scientific knowledge: A reflexive citation analysis, or science disciplines and disciplining science. Social Studies of Science, 21, 459-501. Hirschheim, R. & Newman, M. (1991). Symbolism and information systems development: Myth, metaphor and magic. Information
Systems Research,
2, 29-62.
Huxhold, W.E. (1991). The GIS profession: Title, pay, qualifications. Gee Info Systems, 12, (March 22). Johnson, B.M. & Rice, R.E. (1987). Managing the organizational innovation: The evolution from word processing to ofice information systems. New York: Columbia University Press. King, J.L. (1983). Centralized versus decentralized computing: Organizational considerations and management options. Computing
Surveys, 15, 319-349.
Kling, R. (1987). Defining the boundaries of computing across complex organizations. In Boland, R.J. & Hirschheim, R.A. (Eds), Critical issues in information systems research, Chichester, UK: Wiley, pp.307-323. Kling, R. & Herson, E. (1978). Patterns of segmentation and intersection in the computing world. Symbolic Interaction, 1, 24-43.
Kraut, R., Dumais, S. & Koch, S. (1989). Computerization,
productivity,
and quality of work-life. Communications
of
the ACM, 32, 220-238.
Lawrence,
P. & Lorsch, J.W. (1967). Differentiation
and integration
in complex organizations.
Administrative
Science
Quarterly, 12. l-47.
Lee, A.S. (1989). A scientific methodology for MIS case studies. MIS Quarterly, 13, 33-50. Mackenzie, D. & Wajcman, J. (1985). The social shaping of technology, Milton Keynes, UK: Open University Press. Maguire, D.J., Goodchild, M.F. & Rhind, D.W. (1991). Geographical information systems, Volume I, London: London Scientific and Technical. March, J.G. (1991). Exploration and exploitation in organizational learning. Organization Science, 2, 71-87. Markus, M.L. & Robey, D. (1988). Information technology and organizational change: causal structure in theory and research. Management Science, 34, 583-598. Miles, M.B. & Huberman, A.M. (1984). Qualirarive data analysis: A source book of new methods, Beverly Hills, CA:
282
S. SAHAY and D. ROBEY
Sage Publications, Newman, M. & Robey, D. (1992). A social process model of user-analyst relationships. MIS Quarterly, 16, 249-266. Noble, D.F. (1984). Forces of production: A social histor); of industrial production. New York: Oxford University Press. Oliver, C. (1991). The antecedents of deinstitutionalization. Organization Studies, 13, 563-588. Orlikowski, W.J. (1992). The duality of technology: Rethinking the concept of technology in organizations. Organization Science,
3, 398-427.
Orlikowski, W.J. (1993). CASE tools as organizational change: Investigating incremental and radical changes in systems development. MIS Quarterly, 17, 309-340. Orlikowski, W.J. (1996). Improvising organization transformation over time: A situated change perspective. Information Systems Research, 7 (in press). Orlikowski, W.J. & Baroudi, J. (1991). Studying information technology in organizations: Research approaches and assumptions. fnformation Systems Research, 2, l-28. Orlikowski, W.J. & Gash, D.C. (1994). Technological frames: Making sense of information technology in organizations. ACM Transactions
on Information
Systems, 12, 174-207.
Orlikowski, W.J. & Robey, D. (1991). Information technology and the structuring of organizations. Information Systems Research, 2, 143-169. Orlikowski, W.J. & Yates, J. (1994). Genre repertoire: The structuring of communicative patterns in organizations. Administrative
Science
Quarterly, 39, 541-574.
In Pickles, J. (Ed.), Ground Press, pp. l-30. Pinch, T.J. & Bijker, W.E. (1987). The social construction of facts and artifacts: Or how the sociology of science and the sociology of technology might benefit each other. In Bijker, W.E., Hughes, T.P. & Pinch, T. (Eds), The social construction of technological systems, Cambridge, MA: MIT Press, pp. 17-50. Prasad, P. (1993). Symbolic processes in the implementation of technological change: A symbolic interactionist study of work computerization. Academy of Management Journal, 36, 1400-1429. Robey, D. & Azevedo, A. (1994). Cultural analysis of the organizational consequences of information technology. Pickles, J. (1995). Representations
in an electronic
truth: The social implications of geographical
Accounting,
Management
and Information
age: Geography,
GIS and democracy.
information systems, New York: Guilford
Technologies,
4, 23-37.
Robey, D. & Newman, M. (1996). Sequential patterns in information systems development: An application of a social process model. ACM Transactions on Information Systems, 14, 30-63. Robey, D. & Rodriguez-Diaz, A. (1989). The organizational and cultural context of systems implementation: Case experience from Latin America. Information and Management, 17, 229-239. Sieber, R.E. (1991). Assess employee GIS perceptions early-A simple, radical, and cheap approach to addressing management issues. Proceedings of the 1991 Urvan & Regional Information Systems Association, Vol.11, 66-77. Swanson, E.B. (1988). Information System Implementation, Homewood, IL: Irwin. Thompson, B. (1989). The dazzling benefits (and hidden costs) of computerized mapping. Governing, December, 40-45. Thompson, J.D. (1967). Organizations in Action, New York: McGraw-Hill. Tushman, M.L. & Anderson, P. (1986). Technological discontinuities and organizational environments. Administrative Science
Quarterly, 31, 439-465.
Tyre, M.J. & Orlikowski, W.J. (1994). Windows of opportunity: Temporal patterns of technological adaptation in organizations. Organization Science, 5, 98-l 18. Walsham, G. (1993). Interpreting information systems in organizations. Chichester, UK: Wiley. Walsham, G. & Han, C.K. (1993). Information systems strategy formation and implementation: The case of a central government agency. Accounting, Management and Information Technologies, 3, 191-209. Yin, R.K. (1984). Case study research: Design and method. Beverly Hills, CA: Sage Publications.