Integrating New Tools into Information Work: Technology Transfer as a Framework for Understanding Success
T. K. BIKSON AND J. D. EVELAND
A major thesis guiding current research on social aspects of computerization is that it can be understood as an instance of technological innovation in organizations. If so, much of what has been learned about the successful transfer and use of other new technologies can be applied to understanding how best to introduce new computer-based tools into information-intensive work. In this chapter, we propose a technology transfer framework as an appropriate model for understanding this process. This framework incorporates three key sources of effect: features of the new technology; characteristics of the organization; and properties of the implementation process.
The chief conclusion from research based on this conceptual framework is that properties of the implementation process—the sequence of events that starts with the selection of a new tool and ends with its incorporation into ongoing work—are strong predictors of subsequent organizational outcomes. They predict what individuals and work groups will do with or to new tools, and what will happen to the organization as a result.
In what follows we briefly explain why characteristics of the change process are critical determinants of the social impacts of computerization. Then we describe the technology transfer framework in more detail, reviewing research results that have helped to corroborate and refine it, and identify implementation process variables predictive of successful outcomes; namely, people effectively using new computer tools in the office setting.
FOCUS ON THE IMPLEMENTATION PROCESS
The study of technological innovation has a long history of its own and has been the subject of a number of critical reviews (Berman and McLaughlin, 1978; Bikson and Eveland, 1986; Bikson et al., 1981; Rogers, 1983; Tornatzky et al., 1983; Tornatzky and Fleischer, 1990; Yin et al., 1977). Traditionally innovation has been seen as a process that occurs in the stages depicted in Figure 1.
Under the usual assumptions of this framework, the organization is supposed to be in a state of equilibrium before the decision to adopt a new technology. By contrast, the period that immediately follows is expected to involve major changes as the innovation is being implemented. If all goes well, however, this stage should give way to a new stasis as the technology becomes fully routinized in day-to-day work.
In applying this theory to research on computerization, researchers have expected to find that the point of adoption (e.g.,
early or late) and correlates of the adoption decision (e.g., reasons for adopting) were critical predictors of successful innovation. In fact, research generally fails to find any real ''point'' of adoption, and "the adoption decision" is not a single event at a specifiable time (see Bikson et al., 1987). Rather, adoption of computer-based tools is better understood as a more extended process that involves agenda setting, negotiating, experimenting, and the like as well as decision making (see Eveland, 1979). It is difficult to know, either conceptually or empirically, where adoption leaves off and implementation begins. Both the LA Times case and the U.S. Forest Service case exemplify this lengthy process when complex systems are undertaken.
A second hypothesis generated by the traditional view is that the mark of successful implementation is stasis; innovation ends with improved organizational routines. In contrast, research suggests that there is no end to implementation processes while the technological state of the art is rapidly advancing (see Bikson, 1987; Bikson et al., 1985). Successful transfer of flexible interactive tools is associated with continued and reciprocal changes in tasks and their supporting technologies, as demonstrated by the four case studies in this volume, while unchanging routines are more likely to signal failure.
The traditional view of information technology utilization, then, has been characterized by an emphasis on the dimensions of time (when something gets used, particularly relative to other users, and how long it has been used); fidelity (degree of adaptation, usually with an emphasis on the dangers of changing a prototype that works); and institutionalization (the degree of stability achieved by the change over time). While these variables serve well to describe a certain range of technological changes, they generally fail to tap the key element of information systems use—namely, its transactional character. The dynamics of innovation in computerization require an analytical framework that emphasizes interaction over structure. Accordingly, we propose here a framework drawing explicitly on technology transfer as a way of overcoming some of these key problems with more traditional modes of analysis.
TECHNOLOGY TRANSFER FRAMEWORK
Across varied types of innovations and a diverse range of organizational settings, common factors consistently emerge in research addressed to technology transfer and utilization. As Figure 2 shows, an adequate conceptual framework for understanding the
impacts of new technologies in organizations must include three major components: the technology itself, the organizational context into which it is being introduced, and the process of embedding the technology in the context—the implementation process. Further, research supports the inherent interrelatedness of the three components (Bikson and Eveland, 1986; Stasz et al., 1986).
The conceptual nature of this model must be emphasized. It does not claim to be a complete "checklist" that will inevitably guide correct choices when accurate information for specific situations is plugged in. Rather, it is a sort of map to help understand how different components of technology transfer processes are likely to affect the resulting information systems and how they are used by the people involved. These components are reviewed in detail below.
PROPERTIES OF THE TECHNOLOGY
Tools are modifiable, but not indefinitely so. Some aspects of technology that need to be understood are its range of potential applicability (i.e., how much its use is constrained by its very nature), its degree of hardware dependence, its potential for adaptability, and aspects of its "packaging" (i.e., how the different parts are put together and sold). Unlike computer technologies that are heavily bound up with specific pieces of equipment (for example, industrial robots or chips embedded in automobile controls), information technology is generally characterized by a high degree of flexibility, modularity, and adaptability. These very properties, however, make it difficult to test and predict successful imple-
mentation. From this standpoint, it is worthwhile to consider briefly how information technology is and is not like other technologies.
Validity and Efficacy
In considering the validity and efficacy of technology, the basic question is whether it works—and if so, with what degree of consistency. Another question is how we know—against what criteria is efficacy determined? It would seem essential that for any technology to be successfully transferable from developers to users, it must actually do what it says it will do, with reasonable predictability and few possibilities for catastrophic failure. Yet these conditions are not easy to achieve, especially with information technologies.
Some of the difficulty has to do with the different environments experienced by the developers and the users. Unfortunately, performance in an R&D lab or vendor demonstration is seldom a predictor of performance anywhere else. An information tool that performs reliably in one setting may not perform equally well in another setting because of greater work load in the real world, less predictable user demands, a more heterogeneous base of installed hardware and software into which the new tool must be integrated, and other context-specific factors. A good deal of the empirical analysis of technologies in terms of transferability has to do with determining performance limits—the conditions that bound effective functioning.
The environment for new computer-based information tools has proved to be too rich, variable, and uncontrolled for the technology to be deployed uniformly. This has posed difficult challenges for behavioral scientists attempting to predict what individuals, work groups, and organizations will do. This should not be interpreted as a criticism of the effort to create a behavioral science base for interactive information technologies; rather, it is to say that as yet such a science has not progressed to the point where it can make reliable predictions about the effects of new computer-based tools.
This element of variability has proved both a blessing and a curse. As is clear, information technology can be applied to a great many parts of human activity, and there is little inherent in the tools themselves that conditions how they are to be used. By the same token, however, there are relatively few cues provided by the tools themselves as to their use, and the burden falls on
their adopters (perhaps with the aid of developers or implementers) to create patterns of effective use. Thus the effective use of information tools typically requires more creativity—and more introspection into the user context—than many other types of technology.
Scope, Testing, and Scale
The scope, testability, and scale of a new technology are interrelated features that have to do with how broadly a technology will be applied, how much of the user organization it could potentially affect, over what time frame, and how one might go about finding out. It is noteworthy that all judgments about scope, or how substantially an innovation will affect a particular kind of user, are subject to a very high margin of error. It is one of the aspects of new technologies that is the hardest for either developers or users to assess. It is not hard to see why—it requires looking at a very complex, working whole and visualizing something else. Because new elements in the way work is done reposition existing elements, there is no way to anticipate with confidence how matters are likely to evolve.
Differences between anticipated and realized scope may operate in either direction. A technology that is anticipated to have major repercussions may turn out to be useful only in a limited context (e.g., teleconferencing); or one originally thought to be of relatively limited effect may turn out to be revolutionary (e.g., personal computers). Moreover, these differences—in either direction—may be evaluated rather differently before and after the fact.
To some degree, the unpredictability of scope can be dealt with through trial and experimentation. On the other hand, technologies differ in the degree to which they can be put to meaningful field trials before full deployment. For example, it would be feasible to buy one microcomputer, put it in a room with a lot of conventional office machines such as typewriters and calculators, and evaluate its performance against specified criteria. It is much harder to experiment with, say, integrated local area networks, where the emphasis is on effects that stem from an entire system (see Markus, 1984).
In general, field trials of technologies are progressively less efficient the larger the behavioral or human component of the technology, a fact particularly salient in the implementation of information tools (Johnson and Rice, 1986). Behavior is notoriously context-sensitive, and the context of a trial inevitably differs from the usual world of work. While there is much to be said for
careful experimentation with advanced technologies before implementation, most such experiments are difficult at best and misleading at worst; they should be undertaken only with the guidance of extensive prior investigation, as illustrated by the year-long pilot trial of news-editing technology undertaken by the Los Angeles Times (in this volume).
A factor that invariably complicates questions of scope and testing is organizational politics, which includes the local standards for handling and distributing costs and benefits among the users of the new system. These accounting mechanisms are essentially opaque to the developers of technologies, who concentrate their attention on context-free criteria of technical performance. On the other hand, they are central to an organization's rationale for installing a new technology and play a key role in the way users absorb the tools. 1 Thus they can play a peculiar role in trials ostensibly aimed at assessing the technical effectiveness of a new information tool. Those responsible for pilot projects would do well to bear such conditions in mind.
Finally, the idea of scale is somewhat similar to scope and testing but more straightforward. Scale refers to how broadly the technology must be applied, how much effort will have to go into using it, and how much it will cost to get started. Some technologies can be applied in limited circumstances with high returns for the organization; others require a broad base and a relatively long time frame for their benefits to be felt. Likewise, economies of scale can be expected to differ significantly. In general, innovations that are limited in scale have a better chance of moving rapidly and incrementally through the organization; by contrast, they may not get the degree of publicity or enthusiasm of more broad-scale innovations.
The degree to which an innovation can be modified or tailored to its anticipated or actual organizational slot is an extremely critical feature. Any new technology must eventually be mapped into an existing organization with established personnel and procedures. Implementation (see below) is inherently a process of mutual adaptation of the technology to its environment. To the degree that the burden of all changes must be borne by the organization (e.g., by changing tasks to fit the tools or by training or replacing former employees to obtain new skills) rather than being absorbed in part by the technology, the implementation effort
is likely to be much more difficult and susceptible to abortion or ultimate failure.
As far as characteristics of the technology are concerned, the key question is the degree to which the tools can in principle be modified by users themselves without damage to their output. Computer-based tools vary in the extent to which their design or implementation imposes a single required method of use. There are differences between product and process innovations on this dimension, although the issue applies to both. That is, the developer of a new microprocessor or a new data base manager undoubtedly takes it for granted that those who buy the product will use it in widely differing ways and for widely differing purposes, by users with strong stylistic preferences for how a given information task should be carried out. Thus, however efficient it may be, a software application that can be used only in one particular way with rigidly prescribed inputs and outputs will be quite limited in the range of sites within which it can be deployed.2 That is to say, the task of transferring and broadly deploying a non-adaptable technology is likely to be more difficult and require other kinds of mechanisms and resources than are needed with a technology where the user has greater power over how it operates. Much of today's information technology comprises generic process innovations that are designed in almost context-free ways and then deployed into highly variable user contexts (see, for instance, the Los Angeles Times and U.S. Forest Service cases in this volume).
The importance of fidelity to the developers' concept of the technology versus the ability of users to adapt and "reinvent" the innovation has engaged lively debate—particularly in relation to process innovations in education (e.g., Berman and McLaughlin, 1974) and health (Kaluzny et al., 1974; Calsyn et al., 1977), and to the use of information technology in a variety of contexts (Eveland et al., 1977; Bikson et al., 1981; Johnson and Rice, 1986). In general, what is important to our analysis at this point is that, other things being equal, adaptability of an innovation both in how it can be used and under what circumstances simplifies the burden on the technology transfer system and increases the chance of success.
The ability of a new technology to be packaged as a salable product reflects the degree to which the subcomponents of the
technology—all its different physical elements and associated behaviors—form "tight" or "loose" bundles, and can be aggregated or disaggregated in the transfer process. Does the user organization have to acquire an entire system all at once, or can it be bought in stages? Does it have to be purchased up front, or can it be leased without making a final commitment? How extensive a range of associated goods and services—e.g., peripheral equipment, consulting services—will be required to make initial use of the innovation?
Wrapping up innovations in a complete package for "turnkey" deliverability is often a desirable strategy from the point of view of developers. It simplifies their transfer responsibilities, puts the major burden on the marketplace as a technology transfer mechanism, and cuts down on the degree of messy involvement with after-sale processes.3
For some innovations—primarily those of limited scale and scope, high reliability, low behavioral content, and low adaptability—this approach works well. Users can scan the marketplace and make their decisions to go with one package or another. This strategy does not work nearly so well with more complex technologies—including most information tools—where significant amounts of organizational and social change are likely to accompany implementation. The larger the bundle of technical and social components an organization will have to accept, the more difficult the process of making that decision is likely to be. The best illustrations of the problems of trying to deal with complex technologies as single products are to be found in the experiences of large organizations trying to install single, large-scale all-purpose information systems. While such efforts can have effective outcomes, the amounts of time and energy expended in trying to keep such complex innovations within the bounds of the package are frequently staggering (Stasz et al., 1986, 1991).
PROPERTIES OF INVOLVED ORGANIZATIONS
The preceding discussion suggests that in fact there are two aspects of organizational context that influence technology transfer and use. First, there is that of the developer or vendor—the set of interrelated goals, rules, expectations, criteria, assumptions, and technical know-how that have shaped the technology. Second, there is that of the user organization, a similarly complicated set of individual and social characteristics that strongly affect the ability of work groups to assimilate and use the technology.
Although the two contexts are likely to have a number of features in common, this review gives greatest attention to properties of user organizations that have been shown to influence an organization's capability for technological innovation. While we will single out these properties below, it should be acknowledged that their importance and their consequences depend partly on purely situational factors—how they interact with other characteristics of the user organization and the technology under consideration.
The great variety of properties potentially affecting technology transfer operate at one or another of two major levels.
Firm level: Organizational structure and process characteristics include the pattern or arrangement of jobs, authority, work flow, and communications over an organization. They include such variables as size, resources, levels of hierarchy, centralization, and formalization, and need to be assessed as firm-wide properties.
Work group level: Work design variables have to do with the combinations of tasks, skill requirements, responsibility, authority, and other attributes that make up particular jobs. Characteristics such as variety, challenge, autonomy, and rewards fall within this category. They may be assessed at the individual level but are most reliably measured at the work group level.
Previous research literature has given greatest attention to properties of a user organization's structure and processes as influences on how it goes about acquiring and introducing new technologies. Four factors appear to be especially salient in shaping an organization's technology transfer stance. These include the organization's size; the resources available (with attention both to resources already committed ["sunk costs"] and those uncommitted ["slack"]); the conceptual rigidity of the organization (e.g., is the "not-invented-here'' syndrome a major concern?); and the dependence on technology of the organization specifically and its business sector generally.
This last is an interesting factor. In today's world, most organizations depend on technology to a greater or lesser degree for their competitive edge. These may be production technologies, process technologies, managerial technologies, or others. Some fields are much more dependent than others on a core technology;
if so, they are likely to be responsive to changes in the underlying science. Microelectronics and biotechnology firms, for example, cannot afford to operate with less than leading-edge computer tools. In contrast, a sizable proportion of white-collar enterprises as well as many manufacturing firms can survive happily with technology that is nearer to the lagging edge. The greater the dependence of the firm on its advanced technological position, the more it must scan for and be open to tools developed elsewhere.
On the other hand, size has been a more frequently studied predictor of technological innovation. Typically, large firms are shown to be more likely to adopt and deploy new tools. However, this finding is subject to a number of caveats. First, size is hard to disentangle from other variables (most notably, from available budgetary and human resources). Second, if large firms are quicker to make new procurements, small firms are faster at accomplishing the changes required for a new technology to be fully implemented and diffused throughout the organization (see Tornatzky and Fleischer, 1990). Finally, large organizations—by virtue of having more levels of hierarchy and more entrenched bureaucratic procedures—may be more prone to the conceptual rigidities that impair change.
Finally, there are a host of variables such as centralization, formalization, and specification that appear to be negatively associated with technological innovation. On the other hand, there seems to be a positive association between ability to innovate and such properties as openness of communication and lateral boundary-spanning interactions. While the contrast has some face validity, we cannot draw consistent conclusions. With respect to the rigidity-based properties (e.g., centralization), it is often unclear whether they represent organizational structures or processes; and the flexibility markers (e.g., open communications) are quite clearly process characteristics. But it is not uncommon for de facto processes to take on shapes and patterns not defined by official organization charts (e.g., a highly centralized structure with a lot of lateral interaction). Our own examination of these and other organization-wide properties in a cross-sectional study of innovation found them to be associated with successful outcomes either weakly or not at all (Bikson et al., 1987).
Job design and work group variables make better predictors of the successful incorporation of new tools for ongoing tasks. For example, Bikson et al. (1987) found that a variety of job demands
and availability of adequate resources to work group members predict successful innovation at that level for information technology. We believe this reflects the fact that in large organizations, technological innovation starts at different times and proceeds at different rates. Some parts of an organization may be using cutting-edge computer tools (e.g., the R&D department) while other parts are just beginning to use microcomputers. Consequently there may be more variability between departments within large organizations than there is between common types of departments (e.g., legal departments, R&D departments) in different organizations.
An important but readily unnoticed part of the framework we have proposed for understanding the transfer and use of new technology is the idea of an organization's boundaries. The notion of transfer implicitly includes boundaries. Most obviously, technology transfer involves moving a new tool out of a developer or vendor organization and into a user organization. But first, boundaries are crossed by information.
Each party going into a technology transfer relationship is presumed to know something about the other participant(s).4 The key question at this point is, "Who knows what about what?" The question involves not only the amount of relevant knowledge available at any given time but also the ability of the organization to generate new knowledge when it needs it. This in turn requires recognition that not all needed or desired knowledge is currently available.
Consequently an important boundary factor is the capacity of the organization to search outside itself—to engage in systematic environmental scanning. From the perspective of technology developers, this usually takes the form of market research—defining who are likely targets for the technology in question, developing ways of approaching them, and creating an understanding of how the needs of markets should be factored back into the organization's planning strategy. For technology users, scanning means being able to find out what opportunities are becoming available and to help shape their development before they are cast in concrete. Where users are able to interact creatively with technology developers in the early stages of development, both sides benefit (von Hippel, 1988).
Generally in technology transfer, the users seek information
about the technology—both its technical and behavioral dimensions. When attained, the information must then be interpreted for the receiving organization. Typically, both parties to the interaction tend to hear what they want to hear. Since developers are generally more comfortable with the technical aspects of their innovations than the behavioral component, they can be expected to hear most questions as technical ones and respond with technical answers. Users, by contrast, are more preoccupied with what they are going to do with the new tool, and have attached social and behavioral meanings to their questions; the mismatch often means they fail to get satisfactory answers. The only solution is frequent communications across the boundaries between developers and users—whether these are boundaries between firms or between departments within large organizations. Frequent iterations are required to get over the information barrier.
Boundary crossing may be regarded as a form of interorganizational interaction in which boundaries become permeable and organizations often in effect ''move into" each other. In technology transfer, the boundary crossing may take a variety of forms; but usually one organization crosses farther than the other. For example, a vendor representative who expects to sell the firm's products to a particular organization must know considerably more about that customer than the customer can be expected to know about the vendor. On the other hand, user organizations who know reasonably well what they want may wind up understanding considerably more about the potential products available and those who sell them than any of the potential suppliers know about the client.
Besides informational interactions, market transactions may also be thought as a form of boundary crossing. Market mechanisms are all those that involve reciprocal transactions of value exchange. These include direct sales, technology licensing or leasing, partnerships and cooperative arrangements of varied types, and other exchange processes.5
From the perspective of this review, it is sufficient to note that the boundary crossing does not necessarily end when the acquisition decision is made. Particularly with large acquisitions, it is not unusual for members of the user organization to work closely with system developers or integrators. On the other hand, the vendor may work closely with the user to facilitate implementation, training, maintenance, technical help support, and system extension or upgrading. The vendor may well set up an office within the user organization for these purposes, or may take spe-
cial steps to ensure rapid and easy interaction by other means (e.g., by providing a guest account on its electronic mail system for the user organization's technical staff or establishing a voice mail link). Finally, the user organization may come to regard environmental scanning as a regular part of the way it should operate.
While the preceding discussion has focused on organizations, it should be recalled that ultimately it is individual-to-individual transfer of information and skills that must take place along with a market transaction. The individuals who make the flow of information happen between organizations are (for analytical purposes) generally termed "boundary spanners" in the research literature. This is seldom, of course, an organizational title. Those playing this role regularly may be called "salesmen," "field representatives,'' or "consultants." A number of organizational members have occasion to span boundaries in the course of their ordinary work, usually without thinking much about it. It is a difficult role to play well. To a significant degree, an effective boundary spanner must operate in multiple organizations simultaneously—a position nearly guaranteed to make the participants in each organization believe this individual has been "co-opted" or somehow unduly influenced by the other. Few organizations provide much support for playing the role. On the contrary, many organizations have created barriers to boundary spanning; these may take the form of specific procedural constraints or may be more informal cultural inhibitions on interactions that cross organizational lines. Nevertheless, boundary spanning—both laterally within an organization as well as extramurally in the development community—has long been associated with successful technological innovation.
PROPERTIES OF IMPLEMENTATION PROCESSES
Implementation refers to the transition from a workable technology idea to its manifestation in day-to-day work. The study of organizational change suggests that when adopted, most innovations are not new per se but rather are new to a particular user organization (Tornatzky et al., 1983). Interactive computer tools of the sort represented by the cases in this volume were not new inventions; but many organizations were introducing them as tools for use by white-collar employees who were not computer professionals. Thus, while a great deal was known in general about computer equipment and software, not much information was avail-
able to any work group about how best to make that knowledge relevant to and operational in the group setting—the heart of an implementation effort.
Implementation variables characterize the processes by which participating groups arrived at their current level of incorporation of computer-based work tools. There is growing support from both empirical research and case studies for the thesis that the nature of the implementation process itself significantly influences the outcomes of technological innovation efforts (Bikson and Eveland, 1986; Bikson, Gutek, and Mankin, 1981, 1987; Bikson, Stasz, and Eveland, 1990; Johnson, 1985; Tornatzky et al., 1983). While some consistent findings emerge from this literature, they do not rely on a common set of constructs or terminology. Consequently they are best summarized in terms of the set of themes they embody.
Planning for Change
In earlier research on innovation processes, perhaps the most widely studied variable was the reason for change (see the review in Bikson et al., 1981). In general, successful innovation was thought to develop from reasons internal to the organization (e.g., perceived needs rather than technology opportunities) and from positive rather than negative initiatives (e.g., improving performance rather than reducing cost). With respect to the introduction of computers into office work environments, however, these incentives operate concurrently. In particular, successful transitions to computerization need to take into account properties of the technology becoming available in the marketplace and characteristics of the organizational setting. If technology scanning is not part of an organization's planning process, it will not be able to initiate change; rather, it will by default be in a reactive position (typified by fire-fighting or playing catch-up, or both). Planning initiatives, then, should represent both needs and opportunities.
A second planning variable found to be important in prior innovation research is the thoroughness of the effort. For instance, the existence of organization-wide policy documents directing the move toward electronic information tools would be treated as an index of careful planning; another index might be the comprehensiveness of the transition (measured, for example, by the proportion of departments or work groups affected by the planned change). Such indicators were associated with positive implementation out-
comes in earlier studies; consequently, they were included in large cross-sectional study of the introduction of computers into white-collar work (Bikson et al., 1987).
In this study, slightly more than half the research sites had organization-wide implementation policies, represented in many instances by a formal planning document. Paradoxically, however, the existence of organization-wide change policies in that study was strongly correlated with rigidly detailed, top-down planning and control, and with highly centralized implementation processes; they were not associated with successful outcomes. Flexible approaches to planning are important if an organization is to take advantage of rapid technical developments that threaten to make any long-term blueprint for action obsolete before implementation is completed. Organizations need to be able to make general strategic information technology plans, allowing more detailed decisions to be made at later points in the implementation process.
A final planning theme related to positive outcomes is the extent to which means have been devised in advance to facilitate the assimilation of new work methods by implementing departments. For example, working out procedures for converting manual or batch tasks to interactive systems, and assessing the likely impact of such changes on job design and employee attitudes, are key components of successful planning. However, organizations tend to focus their planning efforts on computer hardware, software, and networks, leaving critical social and behavioral questions unaddressed. On the other hand, a mutual adaptation model of successful implementation suggests that both social and technical factors should be given equal attention in the implementation process; their interlinkage means that if change processes address only half the relevant factors, a number of unplanned, unmanaged, and potentially deleterious consequences may well develop in the other half. Sociotechnical balance, then, characterizes successful planning for new computer-based systems (Pava, 1983; Tappscott, 1982).
Inevitably implementation is a people-based process. It turns on hundreds of choices made and actions taken on a daily basis by various actors throughout a process that can take months or years. This theme roughly includes the ways in which general implementation plans are carried out and the requisite changes supported. Prior studies of innovation had identified user involve-
ment in implementation decisions as a major influence on the results of the change process (Bikson and Eveland, 1986; Bikson et al., 1983; Mumford, 1983). Current research on computer-based information tools has further reinforced the value of user participation in design and decision processes (Bikson et al., 1987).
Whether using or not using a computer-based tool for a particular task is a voluntary decision is one way of looking at the user participation theme (evidenced, for example, in RAND research as well as research by Lucas, 1978, and others). Since white-collar work is often susceptible to discretionary and variable procedures (Strassman, 1985), leaving the choice of how to accomplish it up to employees is one measure of participatory decision making. More often, however, user participation is represented by the type and extent of influence users have over the particular computer-based tools available to them. At RAND, for instance, we have studied the role of users in decisions about hardware, software applications, and new task procedures (Bikson et al., 1987). In contrast, user participation in system design has been the focus of much Scandinavian research. Finally, providing higher-level means for users to program or otherwise tailor the systems they use is a third avenue for user participation. While results of current research on user involvement in implementation decisions are not entirely consistent and conclusive, the preponderance of evidence to date suggests that it can be a powerful influence on success. It is, however, not trivial to arrange effective user participation and, even where it is well done, it is not a substitute for high-quality technical resources.
Next, supporting innovative change in an organization inevitably requires providing opportunities for learning (Berman and McLaughlin, 1978; Bikson et al., 1981). Training programs and other procedures for enabling employees to learn how to incorporate computer-based tools into their work should be regarded as central components of an implementation strategy (Bikson and Eveland, 1989; Bikson and Gutek, 1983). The importance of good training for effective computer use is often acknowledged but seldom enacted. The discrepancy between anticipated learning needs and common organizational practices led us to coin the term "humanware," designating the knowledge, skills, and technical resources necessary to take advantage of computer-based tools. Underinvestment in humanware, relative to hardware and software, is often responsible for the underuse of the latter.
When organizations provide training, moreover, it is typically for beginning-level use. Very little effort is directed toward users'
continued learning, or even to providing back-up assistance while they try to come up to speed on new tools. Consequently, even high-level professionals may remain amateur users of tools closely implicated in the tasks for which they are responsible. Further, day-to-day technical help is likely to be generic and low in quality; thus users most often turn to colleagues who happen to be proficient users of the tool in question. But the availability of local "gurus" is by no means ensured, and often organizations do not encourage them to act as sources of technical help. Consequently, Kling and others regard the social infrastructure for computing support as a key implementation variable (e.g., see Bikson et al., 1985; Kling and Scacchi, 1982; Stasz et al., 1990).
Commitment to Change
Historically innovation literature has emphasized the importance of attitudinal factors, chiefly addressing their role in adoption rather than implementation of new technology (Bikson, 1980; Bikson et al., 1981). However, as we have explained above, variables associated with adoption of computers for information work have not typically been able to predict successful implementation. On the other hand, attitudinal variables representing the nature of a work group's commitment to the change process itself have turned out to be more closely linked to implementation outcomes.
One attitudinal variable that has figured in recent research on computerization might be termed "diffusion status." It represents a group's conception of its role in spreading a technological innovation. That is, does the implementing group regard itself as a leading-edge model for the organization, as setting a positive example, or as having little or no directional influence? Recent innovation theory is unclear as to the consequences of early adoption of interactive technologies. On the one hand, early adopters tend to enjoy some advantages (the absence of formalized routines, the perception of being state of the art, and perhaps even the special "hand-holding" that an official prototype group enjoys). On the other hand, some features of this status are negatively correlated with implementation success (lack of a critical mass of users or lack of institutionalized structures to ensure the continued existence). Our research suggests that, from an attitudinal perspective, being at the forefront of a wave of diffusion is associated with success (Bikson et al., 1985, 1987). However, the larger the user community for interactive innovations—and espe-
cially for networked systems—the greater the advantage to all; so, objectively speaking, very late adopters may have more to gain from introducing the technology than early adopters did (Markus, 1984; Rogers, 1983).
A second critical implementation attitude is "change orientation": the extent to which participants in an innovation process view the change as a positive, problem-solving, and achievable goal—one that will benefit both management and employees alike. In one cross-sectional study of organizations introducing computerized information tools, this attitude was a highly significant predictor of success (Bikson et al., 1987). It is important to include here because of the widely shared belief that people are naturally resistant to change. More than 10 years of research on the computerization of information work at RAND have failed to substantiate this belief. On the contrary, resistance to change is more often observed in organizations than in employees. That is, organizations are notably resistant to recognizing new task demands and new competencies acquired by employees with commensurate changes in titles, job grades, pay, or career paths.
Employees are often justified in the belief that changes are being introduced primarily to achieve reductions in labor costs. While this has seldom been the actual result, there is typically a significant lag between change in employee performance and its formal organizational acknowledgment. A positive change orientation, then, taps the belief that the benefits of technological change will be fairly shared.
Older frameworks for understanding innovation emphasized stasis as the mark of a successful transition. Technological innovation in general and the transfer of new computer-based tools in particular, however, are better understood as a dynamic and on-going process. The technology transfer framework provides a model.
First, the attributes of a technology itself have a distinct effect on how it will be transferred and incorporated into an organization (just as they reflect a number of properties of the developing organization). They also influence the kinds of decision-making processes and implementation strategies that may be invoked. The same attributes are not necessarily applicable to all computer-based tools. They are subject to widely differing perceptions among the participants involved, and they are susceptible to reinterpretation or reinvention. But they are a useful starting point for look-
ing at what is and is not manipulable in technological innovation processes.
Second, several important conclusions can be drawn about relationships between organizational characteristics and successful innovation. In attempting to understand how new technologies are transferred and deployed in user organizations, it is important to distinguish between firm-wide and group-level characteristics. Moreover, it is wise to take into account the work groups and the design of jobs within them in planning for as well as predicting successful innovation. At that level organizational characteristics will be most closely linked to properties of the tools in relation to tasks to be performed.
Another point to bear in mind about organizational characteristics—regardless of level—is that they are not fixed or absolute, but rather are susceptible to shifts reflecting the state of equilibrium or disequilibrium in the system generally. Research models that treat such characteristics as "independent" variables for "predicting" technology transfer results risk missing this point. Technology transfer is an iterative and interactive process; throughout that period the organization is itself changed. Although features of the organization that characterize it before participation in the innovation process may initially bias the effort one way or another, it is much more interesting to think of them as ''dependent" variables to be shaped and modified in the course of technology transfer.
Similarly organizational boundaries, originally thought to be rigid, are local and permeable. That is, they exist over limited parts of the system and for limited periods of time. While some boundaries are more permanent than others, the opportunity to achieve a competitive advantage may blur even the most long-standing of them. Any given part of an organization may be required to open a window to the outside in order to achieve an objective of interest, and it should structure ways of doing so that put the least possible pressure on those inside (March and Simon, 1958; Thompson, 1967).
Finally, the technology transfer framework includes the implementation process. Few of the properties found useful for thinking about the stages of innovation in previous literature are helpful for understanding the implementation of innovative computer-based tools in organizations. Perhaps it is better to regard an organization as being at a "mature" stage of computerization if its work groups take for granted that their information tasks will be done using interactive tools, even though the technology is undergoing
continued change. Even relatively late stages in the implementation process cannot properly be understood in terms of approximation to stasis or completion of initial plans—especially when the base science is continuing to advance.
Rather, there is increasing evidence that the more important factors are properties of the implementation process itself: thorough planning that focuses on people as well as tools, user participation and training, and commitment to change on the part of managers and employees. A key element identified in implementation analyses is the role of "mutual adaptation" of tools and use settings. That is, new tools can and do change in the course of being integrated into particular settings, even as work groups must and do change in assimilating new technical capabilities (Bikson et al., 1981). The concepts of sociotechnical systems analysis (Trist, 1981) that emphasize the integration of the technical system of the organization (tools and procedures) with the social system (roles and relationships among participants) are particularly relevant. The important part is to recognize that such internal processes are both natural and necessary to the incorporation of new technology into organizations, and are shortcircuited or overruled only with serious risk to the success of the venture.
Technological innovation is frequently said to be hampered by employees' resistance to change. Research provides little support for that view. There are countless examples of individuals doing old tasks in new ways and doing new tasks they would not have anticipated when they entered their current jobs. By self-report as well as by management assessment (Bikson et al., 1987), they tend also to be doing more work and meeting higher performance standards.
We suggest that resistance to change is observed more often in the organization than in its employees. For instance, organizations seldom acknowledge changes in employee skills, tasks or standards with changes in job titles, job descriptions or grades, or pay. They are reluctant to invest in training and learning resources that would better support employees' use of advanced tools. And they remain ambivalent about the need to facilitate technological innovation—only a minority see a continuing role for organizations in this area (Mankin et al., 1988). Given that there is no foreseeable end to the advance of computer-based tools, organizations will be involved in implementation processes for a long time to come. They would do well to aim at managing change successfully rather than at minimizing it.
Berman P., and M. McLaughlin. 1974. Federal Programs Supporting Educational Change: A Model of Educational Change. (R-1589/1-HEW). Santa Monica, Calif.: The RAND Corporation.
Berman P., and M. McLaughlin. 1978. Federal Programs Supporting Educational Change: Implementing and Sustaining Innovations. (R-1589/8-HEW). Santa Monica, Calif.: The RAND Corporation.
Bikson, T. K. 1980. Getting it Together: Gerontological Research and the Real World. Santa Monica, Calif.: The RAND Corporation, Report P-6447.
Bikson, T. K. 1987. Cognitive press in computer mediated work. Pp. 353–364 in Social, Ergonomic and Stress Aspects of Work with Computers, G. Salvendy, S. L. Sauter, and J. J. Hurrell, Jr., eds. Amsterdam: Elsevier.
Bikson, T. K., and J. D. Eveland. 1986. New Office Technology: Planning for People. New York: Pergamon Press.
Bikson, T. K., and J. D. Eveland. 1989. Technology transfer as a framework for understanding social impacts of computerization. Pp. 28–37 in Work with Computers: Organizational, Management, Stress and Health Aspects, M. J. Smith and G. Salvendy, eds. Amsterdam: Elsevier.
Bikson, T. K., and B. G. Gutek. 1983. Advanced Office Systems: An Empirical Look at Utilization and Satisfaction. Santa Monica, Calif.: The RAND Corporation, Report N-1970-NSF.
Bikson, T. K., B. G. Gutek, and D. Mankin. 1981. Implementation of Information Technology in Office Settings: Review of Relevant Literature. Santa Monica, Calif.: The RAND Corporation, Report P-6697.
Bikson, T. K., B. Quinn, and L. Johnson. 1983. Scientific and Technical Information Transfer: Issues and Options. Santa Monica, Calif.: The RAND Corporation, N-2131-NSF.
Bikson, T. K., C. Stasz, and D. Mankin. 1985. Computer-Mediated Work: Individual and Organizational Impact in One Corporate Headquarters. Santa Monica, Calif.: The RAND Corporation, R-3308-OTA.
Bikson, T. K., B. G. Gutek, and D. Mankin. 1987. Implementing Computerized Procedures in Office Settings: Influences and Outcomes. Santa Monica, Calif.: The RAND Corporation, R-3077-NSF.
Bikson, T. K., C. Stasz, and J. D. Eveland. 1990. Plus Ca Change, Plus Ca Change: A Long-term Look at One Technological Innovation. Santa Monica, Calif.: The RAND Corporation, WD-5032-USDAFS.
Calsyn, R., L. G. Tornatzky, and S. Dittmar. 1977. Incomplete adoption of and innovation: The case of goal attainment scaling. Evaluation 4:128–130.
Eveland, J. D. 1979. Issues in using the concept of 'adoption' of innovations. Journal of Technology Transfer 4(1):1–14.
Eveland, J. D., E. M. Rogers, and C. Klepper. 1977. The Innovation Process in Public Organizations. Ann Arbor, Mich.: University of Michigan.
Johnson, B. M. 1985. Innovation in Office Systems Implementation. Department of Communication, University of Oklahoma.
Johnson, B. M., and R. Rice. 1986. Managing Organizational Innovation. New York: Columbia University Press.
Kaluzny, A., J. T. Gentry, and J. Veney, eds. 1974. Innovation in Health Care Organizations. Chapel Hill, N.C.: University of North Carolina Press.
Kling, R., and W. Scacchi. 1982. The web of computing: Computer technology as social organization. Advances in Computers 21:2–60.
Lucas, H. C., Jr. 1978. Empirical evidence for a descriptive model of implementation. MIS Quarterly 2(2):27–42.
Mankin, D., T. K. Bikson, B. A. Gutek, and C. Stasz. 1988. Managing technological change: The process is key. Datamation 34(18):68–80.
March, J. G., and H. A. Simon. 1958. Organizations. New York: John Wiley and Sons.
Markus, L. M. 1984. Systems in Organizations: Bugs and Features. Boston: Pitman.
Mumford, E. 1983. Designing Human Systems for New Technology: The ETHICS Method. Manchester, England: Manchester Business School.
Pava, C. 1983. Managing New Office Technology: An Organizational Strategy. New York: Free Press.
Rogers, E. M. 1983. The Diffusion of Innovations. New York: Free Press.
Stasz, C., T. K. Bikson, and N. Z. Shapiro. 1986. Assessing the Forest Service's Implementation of an Agency-Wide Information System: An Exploratory Study. Santa Monica, Calif.: The RAND Corporation, N-2463-USFS.
Stasz, C., T. K. Bikson, J. D. Eveland, and B. Mittman. 1990. Information Technology in the U.S. Forest Service: An Assessment of Late Stage Implementation. Santa Monica, Calif.: The RAND Corporation, R-3908-USFAFS.
Stasz, C., T. K. Bikson, and J. D. Eveland. 1991. Assessing Benefits of the U.S. Forest Service's Geographic Information System: Research Design. Santa Monica, Calif.: The RAND Corporation, WD-4590-USDAFS.
Strassman, P. 1985. Information Payoff. New York: Free Press.
Tappscott, D. 1982. Office Automation: A User-driven Method. New York: Plenum Press.
Thompson, J. D. 1967. Organizations in Action. New York: McGraw-Hill.
Tornatzky, L. G., and M. Fleischer. 1990. The Processes of Technological Innovation. Lexington, Mass.: Lexington Books.
Tornatzky, L. G., and others. 1983. The Process of Technological Innovation: Reviewing the Literature. Washington, D.C.: National Science Foundation.
Trist, E. W. 1981. The Evolution of Sociotechnical Systems. Toronto, Canada: Ontario Quality of Work Life Centre.
Von Hippel, E. 1988. The Sources of Innovation. New York: Oxford University Press.
Yin, R. K., K. A. Heald, and M. E. Vogel. 1977. Tinkering with the System. Lexington, Mass.: Lexington Books.