The Process of Adopting Innovations in Organizations: Three Cases of Hospital Innovations
ANDREW H. VAN DE VEN
It is increasingly being recognized that the process of adopting innovations in and by organizations is far more complex than it is by individuals. The former entails all the social, political, and bureaucratic complexities of large organizations, while the latter is largely a marketing effort that informs and persuades individual consumers of a desirable new product or procedure. Yet, the basic model that most management scholars and practitioners appear to use in organizing their thoughts and actions about adopting innovations in organizations is based on a model of individual innovation adoption. It is not surprising, therefore, that most innovation adoption efforts by organizations fail.
Fortunately, some succeed, as exemplified in this book in the three case studies on the adoption of hospital innovations: the evolution of new technologies and administrative arrangements in the ambulatory care unit of Pennsylvania Hospital described by Robert Cathcart; the introduction of a new technology (ECMO—extracorporeal membrane oxygenation) in a neonatal intensive care unit of Columbia Presbyterian Hospital described by John Driscoll; and experimentation and implementation of alternative working schedules for nurses in Rochester Methodist Hospital of the Mayo Clinic described by Thomas Choi, Helen Jameson, and Milo Brekke. The three cases represent highly selective examples of successful innovation adoption. They are stimulating to read, and provide rich descriptions of the process of innovation adoption in hospitals. But if our knowledge of innovation adoption is to go beyond
description and on to explanation, we need to ask, ''Why were these hospitals successful in adopting their innovations?'' Moreover, we need to ask, What can we learn from these cases to improve our understanding of the innovation adoption process in organizations?
Valid answers to these questions are not possible from these cases alone, because generalizations from three cases are difficult to substantiate, particularly when they include no instances of failure. To address these questions we will expand our sample by leaning on other published studies as well as research currently under way by the Minnesota Innovation Research Program, which since 1983 has been tracking a wide variety of technological, product, and process innovations as they develop from concept to reality in their natural field settings (see Van de Ven, Angle, and Poole, 1989). The three cases presented here will be treated as examples to illustrate key research findings that are relevant to our questions.
Our point of departure will be a brief review of Everett Rogers's basic model, which is perhaps the most widely shared view of the process of innovation adoption and diffusion. Although this model is robust in explaining innovation adoption by individuals, it does not adequately incorporate many complexities often observed in other studies and exemplified in our three cases when the organization is the locus of adoption. Innovation researchers are currently examining these complexities, which is leading to revisions in Rogers's basic model that explains the process of innovation adoption by organizations. Such a revised model, when empirically verified, can make a major contribution by providing practical suggestions for maneuvering the innovation adoption journey in organizations.
ROGERS'S BASIC INNOVATION ADOPTION AND DIFFUSION MODEL
Everett Rogers (in Rogers, 1962, 1983; Rogers and Schoemaker, 1971) has set forth perhaps the most widely accepted view of the innovation process as a result of his own research (encompassing more than a quarter of a century) and a synthesis of more than 3,100 publications on innovation diffusion. Innovation scholars and practitioners would do well to study this model carefully, for it captures much of what we know implicitly or take for granted about innovation adoption. This model, shown in Figure 1, views the process of innovation as a simple linear sequence of three
basic stages: (1) the invention of an idea (which comes from recognition of needs or problems and basic or applied research); (2) its development, production, and testing as a concrete device or program; and (3) its diffusion to and adoption by users. Depending on the innovations being examined, various authors have expanded or modified activities in these three basic stages.
Indeed, specialized fields of study and research have emerged over the years to focus on each stage. In the idea invention stage, psychologists have developed an extensive literature on individual and group creativity (e.g., Amabile, 1983; Angle, 1989; Cummings, 1965); and economists, on "technology push" versus "demand pull" (e.g., Rosenberg, 1982; Thirtle and Ruttan, 1987). Although less extensively studied than the other stages, the process of innovation development is gaining more research attention by management scholars (e.g., Kanter, 1983; Tushman and Romanelli, 1985; Van de Ven, Angle, and Poole, 1989). Finally, Rogers (1983) notes that perhaps no other topic in the social sciences has received as much study as innovation diffusion and adoption.
Whereas most of this research has focused on diffusion, which is largely concerned with the marketing, dissemination, and transfer of an innovation to individual end users, far less has dealt with adoption, or the process by which recipient users select and implement an innovation. Of this smaller subset of adoption studies, most have focused on statistically examining relationships between various "input" factors (characteristic of users, organizations, and the innovation) and "output" (rates of innovation adoption)—leaving the adoption process itself least understood. Yet it is well established that functionally similar organizations respond and perform differently when adopting similar innovations (Barley, 1986; Kimberly and Evanisko, 1981). In other words, the process by which organizations adopt innovations makes a difference on subsequent performance.
As Figure 1 shows, Rogers's basic model focuses on and elaborates five substages in the process of innovation diffusion and adoption. First, the diffusion agency starts by marketing and creating awareness of its innovation through a variety of communication channels, such as journals, advertisements, and leaflets, often followed by personal contacts and informal influence of opinion leaders. Once there is an awareness of an alternative, the next subphase is the arousal of interest by a potential user of the innovation. This arousal of interest is influenced by various preconditions, such as felt need and organizational innovativeness, norms, resources, and communication behavior. The model assumes that
the potential adopter engages in a mental evaluation of the innovation and that likelihood of adoption increases when the innovation: (1) promises to have a strong advantage over alternatives, (2) is highly compatible with existing practices, (3) is not too complex, and (4) when it can be tried out and the results can be observed. An adoption decision typically leads to an actual trial implementation of the innovation. Positive outcomes from the trial will lead to continued use and institutionalization of the innovation by the adopting organization; negative outcomes will lead to rejection.
Although extensive empirical support for this adoption process model has been established for individual adopters (such as farmers adopting best practices promoted by the Extension Service of the U.S. Department of Agriculture), mixed results have been obtained when the organization is the locus of innovation adoption. Organizations (particularly hospitals) are complex political systems consisting of many functional specialties and administrative hierarchies (e.g., different medical and administrative staffs) that
often compete for influence and resources in the adoption and implementation of project priorities. As a consequence, innovation adoption decisions tend to be used for partisan purposes—usually heralded by some, attacked and sabotaged by some, and apathetically ignored by the majority of others who are preoccupied with other organizational priorities (Dahl and Lindblom, 1976).
For example, Clark (1987) reports on a study of a hospital-design team initially consisting of the senior administrative staff and a small group of professional nurses working with the British Department of Health and Social Security. They decided to adopt a standard design package that had been devised and developed in the department for producing cost-effective decisions by placing all treatment activities in a central location. During the early stages of design, the medical staff of the design team made no serious interventions or contributions, even though the proposed designs had considerable implications for the medical hierarchy and for the allocation of space and resources. Then at a very late stage, when it became apparent to them that the new hospital, if built as designed, would involve considerable change in their working conditions and in their professional control of activities, they became much more active, thereby necessitating structural alterations during the commissioning of the new hospital.
REVISIONS NEEDED IN MODEL FOR INNOVATION ADOPTION BY ORGANIZATIONS
As this example suggests, innovation adoption by organizations is far more complex than by individuals. In particular, we will focus on six specific process complexities that are evident in the three cases of hospital innovation adoption in this volume and which are not adequately explained by the basic model in Figure 1.
The three cases exist in organizational contexts that motivated and enabled successful adoption of innovation. Moreover, the stage for adopting the innovations was set over a period of several years and involving many organizational participants.
In each of the three examples of successful hospital innovation adoption, the innovators experienced "shocks" (not merely persuasion) as a result of direct personal confrontations with needs or problems. These shocks were sufficient to trigger their attention and action for innovation. When people become dissatisfied enough with existing conditions, they initiate action to resolve their dissatisfaction.
Once adoption activities begin, the process does not unfold in a simple linear sequence of stages and substages; instead, it proliferates into complex bundles of innovation ideas and divergent paths of activities by different organizational units. In the three cases of successful adoption, the innovation process was kept relatively simple in the face of these inexorable pressures for proliferation.
Setbacks and mistakes are frequently encountered during the innovation process, either because plans go awry or because unanticipated environmental events significantly alter the assumptions of the innovation. These setbacks signal either rejection of the innovation or opportunities for learning through reinvention. Learning fails when events are caused, and consequences are felt, by different people. Through reinvention, participants in the three cases of successful innovation adoption learned by reconnecting the causes and consequences of innovation invention, development, and adoption activities.
In the three successful cases, reinvention of the innovation developed elsewhere was facilitated by modifying the innovations to fit the local organizational situation, having top management extensively involved and committed to the innovation, and using various techniques to maintain task completion and momentum throughout the adoption process.
The adoption processes varied to fit the specific contingencies of the innovations being adopted by the three hospitals.
Many of these six process complexities, which occurred in our three cases of successful adoption, would be viewed as characteristics leading to failure by the adoption model shown in Figure 1. These empirical observations are inconsistent with our conceptual model. We will now discuss how revisions in the basic model might be made to deal with each of the six observed processes. Doing so may not only explain why the three hospitals were successful in adopting their innovations but also propose a revised conceptual model to understand the process of innovation adoption by organizations.
Temporal and Contextual Preconditions for Innovation Adoption
Innovations are not initiated on the spur of the moment, nor are they initiated by a single dramatic incident or a single entrepreneur. As observed in all the innovations considered in a recent
study (Van de Ven, Angle, and Poole, 1989), each of the three hospital innovation adoption cases in this volume shows that there was an extended gestation period, often lasting three or more years, in which medical, professional, or nursing staffs in the hospital engaged in a variety of activities that set the stage for innovation. Many initial events during this period were not intentionally directed toward adopting an innovation. As we will discuss in the next section, some events triggered recognition of the need for innovation (e.g., infant deaths or a shortage of nursing staff), while others generated awareness of the technological feasibility of an innovation (e.g., microscopic surgical procedures, extracorporeal membrane oxygenation, or alternative work schedules). Technology-push and demand-pull events such as these often launched intrepreneurs (Pinchot, 1985) on courses of action that, by chance, intersected with independent actions of others. These intersections provided occasions for people to recognize and access new opportunities and potential resources. Where these occasions were exploited, people modified and adapted their independent courses of action into interdependent collective actions to undertake concerted efforts to initiate an innovation.
Although the basic model in Figure 1 posits that an innovation adoption decision is a relatively straightforward result of knowledge and persuasion, these observations emphasize that chance plays a significant role in affecting the decision and subsequent course of innovation adoption. The sheer volume of initiatives undertaken by a large number of interacting people increases the probability of stimulating innovation. The findings also reinforce the bias-for-action principle of Peters and Waterman (1982). Perhaps Louis Pasteur's adage that Chance favors the prepared mind best captures the process that sets the stage for innovation.
The important practical question then becomes, "What can organizations do to increase their preparedness to capitalize on the chance of innovation?" Angle (1989) provides a core suggestion for dealing with this question: Design and develop the organization's conditions to enable and motivate innovative behavior. These conditions include the legitimacy, resources, structure, and culture of the encompassing organization that innovation groups draw upon to enable and constrain their innovative behaviors. Amabile (1988), Angle (1989), and Kanter (1988) summarize an extensive body of research indicating that innovation is facilitated in organizations that provide both enabling and motivating conditions for innovation; it does not occur where either enabling or motivating conditions are absent.
The three cases of successful adoptions of innovations in hospitals exemplify many of the organizational conditions that enable and motivate innovative behavior. Each of the hospitals housing the innovations is a highly respected, long-established, and a very successful institution located at the hub of its industry and community networks. During their respective periods of innovation adoption, the hospitals were reported to have moderately low personnel turnover rates, long-run strategic time horizons that connected diverse organizational activities to core institutional missions (providing quality care to meet changing patient needs), and a high degree of commitment of top management and medical staffs to their respective innovations; in addition, the hospitals were reported to be making significant investments both in new technologies and in their professional staffs. Although the relative influence of any one of these conditions on innovation is difficult to assess, when combined they exemplify the ingredients of an organizational setting that enables innovation. Moreover, in each case, recognition of the need for innovation was triggered by many (not one or a few) events over an extended period of time (often several years) and involved many different people both within and outside the hospitals.
In the short term there is little that managers can do directly to change organizational culture, legitimacy, and prestige, because they are the historical by-products of all previous activities and interactions of an organization with its environment. Thus, it is erroneous to expect that these innovation-enabling characteristics can be changed quickly. However, long-term macroconsequences are produced by the accumulation of many microactions that preoccupy the short-term attention of organizational participants.
The immediate setting for most innovations is the organization itself, and much can be done to modify the immediate operating conditions of an organization. Organizations are complex social systems that provide templates for playing out many distinctive roles important to an innovation. Organizational attributes, such as structure, systems, and practices influence the likelihood that innovation ideas will be surfaced, and once surfaced that they will be developed and nurtured toward realization. Furthermore, the organization is the most direct source of material, financial, and other resources needed to support innovation efforts.
With respect to structure, there are several features that will affect the gestation of innovative activities. The more complex and differentiated the organization, and the easier it is to cross boundaries, the greater the potential number of sources from which
innovative ideas can spring. However, with increasing organizational size and complexity comes segmentation (Kanter, 1983) and bureaucratic procedures; these often constrain innovation unless special systems are put in place to motivate and enable innovative behavior. Key motivating factors include providing a balance of intrinsic and extrinsic rewards for innovative behaviors (Amabile, 1983). Pay, in itself, seems to be a relatively weak motivator for innovation; it more often serves as a proxy for recognition. Individualized rewards tend to increase idea generation and radical innovations, whereas group rewards tend to increase incremental innovations and their implementation (Angle, 1989).
However, the presence of motivating factors, by themselves, will not ensure innovative behavior. Enabling conditions are equally necessary. Examples of such enabling conditions include the following:
Resources for innovation
Frequent communication across departmental lines, among people with dissimilar viewpoints
Moderate environmental uncertainty and mechanisms for focusing attention on changing conditions
Cohesive work groups with open conflict resolution mechanisms that integrate creative personalities into the mainstream
Structures that provide access to innovation role models and mentors
Moderately low personnel turnover
Psychological contracts that legitimate and solicit spontaneous innovative behavior
In short, normal people have the capability and potential to be creative and innovative. The actualization of this potential turns on whether management can create an organizational context that not only motivates but also enables individuals to innovate.
"Shocks" That Trigger Innovation
Although a conducive organizational climate appears to set the stage for innovation, concrete actions to undertake specific innovations are triggered by "shocks" from sources either internal or external to the organization. In the three hospital cases, these shocks included the introduction of the Diagnotic Related Group payment reimbursement system, increasing competitiveness of the hospital industry, infant deaths in the neonatal intensive care unit, as well as results of an employee survey.
The reason why shocks are needed to trigger innovation is based on a simple model of decision making that is embedded in Rogers's basic model: when people reach a threshold of dissatisfaction with existing conditions, they initiate action to resolve their dissatisfaction. Thus, we have no problem with Rogers's basic model since it recognizes that necessity, opportunity, or threat is the mother of invention. Instead, the problem lies in not appreciating the physiological limitations of human beings and the conditions that trigger their thresholds for action.
Human beings are unconsciously highly adaptable. Small and gradual changes over time provide insufficient stimulation to reach people's recognition thresholds for action (Helson, 1964). People adapt to gradually changing conditions and often fail to notice that conditions have signaled the appropriateness (through opportunity or threat) of a change. As a consequence, unless the stimulus exceeds their action thresholds, i.e., a shock, people do not move into action to correct their situation, which over time may become deplorable. Opportunities for innovation are either not recognized or not accepted as important enough to motivate innovative action.
With regard to perception, organizations establish structures and standard operating procedures to run efficiently and reliably. But these structures also have the effect of programming people into cognitive routines or habits, which desensitize them to novel events. Ironically, this habit-bound perception is particularly prevalent where the people are most competent. As argued elsewhere (Van de Ven, 1986, p. 595), "what people do most is often what they think about least."
People's reluctance to accept change is equally relevant. Change is often threatening, because it offers the possibility that coping mechanisms that were at least adequate under the old situation may no longer suffice. People who were "winners" may now be lucky to break even. Adding to the relatively passive blinders that people wear, because of habit or inattention, are the active blinders related to such defense mechanisms as denial. So for a variety of reasons, it may be difficult for people to notice change of the sort that should by all rights stimulate them to innovate. The management of attention is a central problem in managing the innovation journey (Van de Ven, 1986).
What can organizations do to solve the problem? Although there is no panacea, Angle (1989) suggests that mechanisms can be put into place for redirecting and jostling the attention of organizational members, so that subtle changes and needs will be
noticed. For example, Normann (1985) observed that well-managed companies are not only close to their customers, as Peters and Waterman (1982) suggest, they also search out and focus on their most demanding customers. Empirically, von Hippel (1981) has shown that ideas for most new product innovations in an industry come from customers. Utterback (1971) also found that about 75 percent of the ideas used to develop product innovations came from outside of the organization.
In each of the three hospital innovation adoption cases in this volume, nurses and physicians often came face to face with demanding patient needs, people at the cutting edge of technology, hospital nursing staffs, and consultants. These personal exposures increased the likelihood of triggering actions to pay attention to changing technological, patient, and community needs. In general, direct personal confrontation with sources of problems and opportunities is needed to reach the threshold of concern and appreciation required to motivate most people to act (Van de Ven, 1980b).
Adoption Activities Become Complex, Divergent Progressions
As Schroeder et al. (1986) found across a wide variety of innovations, the three hospital cases in this volume show that shortly after the adoption decision was made, the process became increasingly complex to manage, as the initially simple innovation process proliferated into diverse pathways. More specifically, after the onset of a simple stream of activity to develop or adopt an innovative idea, the process quickly diverged into multiple and parallel paths of activities. Some of the proliferation is produced by dividing the labor among functions and organizational units (e.g., nurses, doctors, administrators) necessary in developing different conceptions of the innovation. Some of the proliferation is produced because a given innovation typically entails a bundle of related innovations (e.g., a new technological procedure requires adopting new administrative procedures, new occupational roles, and new conceptions of patient care). Each path represents a different development and adoption process. Finally, other organizational activities may appear unrelated to the innovation but often compete for scarce resources and thwart the innovation adoption process (e.g., the introduction of new computerized reporting procedures during the nursing schedule experiment). As a consequence, after a short initial period of simple unitary activities, the management of innovation soon proliferates into an effort to direct controlled chaos (Quinn, 1980).
This proliferation of activities over time appears to be a pervasive but little-understood characteristic of the innovation adoption process. The basic model in Figure 1 assumes that the innovation in concept and scope remains intact as it is adopted. These observations suggest a need to revise the model to address the continuous redefinitions or enactments (Weick, 1979) of the innovation that are made by organizational participants into terms that they can understand and are compatible with their cultures. Moreover, they show that it is not correct to depict the user as adopting only a single innovation; many users are simultaneously choosing from diverse sets of innovations in many different areas, such as equipment, organizational structures, and new work practices (Clark, 1987).
So what might be done to cope with this proliferating complexity? Like Peters and Waterman (1982), we recommend Kiss (Keep It Simple, Stupid!). We believe that much of this complexity is the result of striving to achieve too much too soon and thereby becoming embroiled in many activities that are not necessary or essential to adopting an innovation. For example, in the ECMO case, one way of decreasing complexity of the innovation was to focus on adopting it in the neonatal intensive care unit before efforts were made to introduce it for regional use with other hospitals in the metropolitan area.
In our studies of innovation, we observed organizational adopters to exhibit an impatient quest to leapfrog into a large program or technology without evaluating the merits of the core innovation (Van de Ven, Venkataraman, Polley, and Garud, 1989). Ironically, this tendency often delays adoption of core innovations because organizational participants become preoccupied with efforts peripheral to the immediate tasks needed to develop the basic idea. Instead, much effort is devoted to preengineering systems that may be needed to adopt families or generations of the innovation. In the process of doing so, basic problems inherent in the core idea are often masked and go unquestioned until setbacks arise (as discussed below).
Administrative reviews are periodically conducted to evaluate innovation adoption progress. However, administrative reviews tend to be poor substitutes for the acid test of the market. Restricting and simplifying adoption activities to the core innovation idea decreases implementation cost and time. Moreover, it tends to decrease the costly mistakes of investments in innovations that do not meet the market test. Small mistakes are more tolerable and correctable than large costly mistakes.
Setbacks Provide Learning Opportunities
All three hospital cases show that mistakes and setbacks were frequently encountered during the process, either because initial plans go awry or because unanticipated environmental events significantly alter the ground assumptions and context of the innovation. Although detailed analyses of these setbacks was not reported in the three hospital cases, other studies can provide insights into the anatomy of setbacks and the opportunities they provide for learning and reinvention during the innovation adoption process.
In recent innovation studies in Minnesota (Van de Ven, Angle, and Poole, 1989) the typical initial response to setbacks was to adjust resources and schedules, which provided a grace period for innovation development. But with time, many of the problems snowballed because additional resources or slack time only masked more fundamental problems; namely, difficulties in detecting, correcting, and learning from mistakes. Many of these setbacks and errors went uncorrected for the following reasons: (1) it was difficult to discriminate substantive issues from ''noise'' in systems overloaded with mixed signals about performance; (2) innovation champions escalated their commitments to a course of action by ignoring naysayers and proceeding full speed ahead; (3) some innovation participants became hypervigilant, calling prematurely for changes in a course of action when minor or correctable problems were encountered; and (4) in-process criteria of innovation success often shifted over time as the initial euphoria with an innovation waned and new, more exciting alternatives became apparent to organizational participants. Thus, although extensive errors were detected in the Minnesota innovation studies, very few were corrected, and they snowballed to crisis proportions before they were addressed. Many learning experiences could not be acted on because of the time lag required in changing a course of action.
In their classic study, Pressman and Wildavsky (1973) examined the efforts of the federal Economic Development Administration to implement a new program that appeared to be destined for success in providing jobs for Blacks in Oakland through public works grants and loans to local enterprises. But more than six years later, public works construction had not been finished and there were few new jobs for unemployed Blacks. Initial agreements among federal, regional, and local officials slowly dissolved into a host of disagreements over the details of implementation. These details included changing actors, diverse perspectives, and
multiple negotiations and clearances among decision makers, all leading to a geometric growth of interdependencies and delays. Delays came from (1) unplanned accidental occurrences, (2) blocking efforts by participants who wanted to stop the program, (3) alternative time priorities, and (4) delays caused by delay itself—with every delay momentum declines in the commitment to seeing the program through to completion.
An immediate recommendation that emerges from these findings is to structure opportunities and resources to detect and correct mistakes when they occur and before they become vicious cycles, as suggested in the chapters by Driscoll and by Choi et al. However, it is not clear that additional slack time, alone, will result in adaptive learning. We have found no empirical evidence in the literature that additional resources and slack time will facilitate trial-and-error learning in highly ambiguous situations.
Trial-and-error learning may be far more difficult than has been assumed. With highly novel undertakings, one can seldom rely on past routines or plans to guide behavior. Moreover, the highly ambiguous information that participants typically receive and the idiosyncratic experiences they encounter greatly circumscribe rational learning processes (March and Olsen, 1976). Instead, these conditions spawn "superstitious learning" (Levitt and March, 1988). Because information is often unreliable, ambiguous, or late, and because most innovation participants have not experienced other innovations from which to develop inferences, it becomes difficult to identify cause-and-effect relationships. As a consequence, there is a blurring between "success" and "failure" as results are interpreted in relation to differing personal perspectives and frames of reference (Dornblaser et al., 1989).
Perhaps the root problem for why little learning is observed as setbacks arise exists in the basic adoption process model itself, which many organizations have come to use to guide their action. The linear sequence of invention, development, and adoption stages in Figure 1 minimizes opportunities for learning. As Pressman and Wildavsky (1973) emphasize, "Learning fails because events are caused and consequences are felt by different organizations. Just as "planners should not be separated from doers" (p. 135), and "implementation should not be divorced from policy" (p. 143), neither should innovation adoption be separated from innovation invention and development. In other words, as Rogers recognizes (Rice and Rogers, 1980), reinvention facilitates adoption.
Reinvention is fundamentally a learning process that is triggered by the inevitable setbacks and mistakes people encounter as
they attempt to implement innovation. Learning is a trial-and-error process, and its essential steps include idea invention, development, implementation, evaluation, and adjustment. When invention and development activities are divorced from implementation and adoption, the learning process is short-circuited because different people experience these activities. As Pressman and Wildavsky (1973) conclude, implementation and adoption must not be conceived of as processes that take place after innovation invention and development. Learning requires that these activities become fused in the innovation process, and that interaction occurs among the people principally concerned with invention, development, and implementation. This happens more easily when innovations are "home grown" than when they are developed elsewhere.
In the organizations where innovations are "home grown," Schroeder et al. (1986) found that implementation and adoption activities often occur throughout the development period, by linking and integrating the new with the old, as opposed to substituting, transforming, or replacing the existing organizational arrangements with the new innovation. The implication of this observation is that, because of limited organizational resources, innovations cannot often be mere additions to existing organizational programs. Neither is it practical for many innovations simply to replace existing organizational programs because of the history of investments and commitments made to make these programs (yesterday's innovations) work. However, such a possibility is often perceived by organizational participants not involved in the development of the innovation. Therefore, new innovations often represent a threat to the established order. Instead, if they are to be implemented and become institutionalized, the new innovations must overlap with and become integrated into existing organizational arrangements.
Concentrated efforts to link the new with the old throughout the development period provide not only more time but also more opportunities to address problems and modify the developing innovations to applied situations than is possible for organizations that adopt innovations developed elsewhere. Therefore, we expect "home-grown" innovations normally to require less time to implement and institutionalize than externally induced innovations (Van de Ven, 1980a).
Reinventing Innovations Developed Elsewhere
As was done in the three hospital cases, organizations can take a number of steps to facilitate reinvention and learning as they
adopt innovations initially developed elsewhere. These steps include (1) modifying and adapting the innovation to the organization's local situation, (2) active involvement of top management, and (3) applying techniques that facilitate coordination among diverse and distracted groups of people to meet key deadlines and maintain momentum for innovation adoption.
First, conventional wisdom suggests that successfully implemented innovations start small and spread incrementally with success (Greiner, 1970). Although this approach provides one way to deal with proliferating complexity (others will be suggested below), this may often not be a wise strategy to deal with organizational political life. Lindquist and Mauriel (1989) compared two common alternative strategies for adopting and implementing innovations: a breadth strategy in which the innovation is implemented across all organizational units simultaneously, and a depth strategy in which the innovation is implemented and debugged in a demonstration site before it is generalized to other organizational units. They found that the breadth strategy was more successful than the depth strategy in adopting and institutionalizing site based management in two public school districts. Several explanations were provided for this surprising finding:
Once the depth strategy is introduced and heralded by top management, the demonstration project loses visible attention, as their agendas become preoccupied with other pressing management problems.
With a breadth strategy, top management stays in control of the innovation implementation process—thereby increasing (rather than decreasing) its power. Moreover, slack resources within the control of top management can ensure success better than limited budgets for innovation to a demonstration site.
There is a trade-off between implementing a few components of an innovation in breadth versus implementing all components in depth in a particular demonstration site. Less resistance to change is encountered when a few (and presumably the easy) components of an innovation are implemented across the board to a few (and presumably supportive) stakeholders than when all (both easy and hard) components of a program are implemented in depth with all partisan stakeholders involved.
With a depth strategy, it is easier for opposing forces in other parts of the organization to mobilize efforts to sabotage a "favored" demonstration site than it is to produce evidence of the merits and generalizability of an innovation.
Experimenting with alternative forms of an innovation introduces several variations of the depth strategy in an organization. For example, Choi, Jameson, and Brekke describe their efforts to conduct an experiment of three different hospitals in Rochester Methodist Hospital. Rigorous methods of scientific experimentation (such as making random assignment, minimizing contamination between groups, standardizing observations, and replicating procedures) were used to determine which schedule was most appropriate. Although it might be argued that these scientific methods enhanced the validity of findings from the experiment, they had the trade-off effects of creating animosity among the staff because experimental procedures permitted no choice or participation in their design or in the work schedules to which nurses were randomly assigned. Furthermore, as is common to any depth innovation adoption strategy, nurses in each experimental schedule were isolated from those working on other schedules. Such isolation restricts communication between groups and may explain the decrease in perceived teamwork among nursing staff. When compounded with other setbacks or "glitches" that inevitably occur during the process, the experiment significantly jeopardized the successful adoption of the innovation. Indeed, only the staunch support of the nursing department managers prevented the staff from rejecting the innovation.
As it turned out, the experiment found that nurse staffing, not scheduling, influenced the cost of care. Further, the experiment findings had little affect in determining the work schedules that the nurses actually adopted. Most nursing units chose the select-a-plan, which permits each unit the greatest flexibility to design and adopt its own work schedule. This is predictable and again underscores a consistent research finding: performance, satisfaction, and motivation of individuals are higher when they implement their own plans than when they carry out someone else's plan (Bass, 1971; Bennis et al., 1962; Delbecq et al., 1975).
From this experience, Choi, Jameson, and Brekke conclude that "good management does not necessarily allow for good science." Over the years, Argyris has argued a different conclusion: "Good science calls for good management." Precisely because of the kinds of unintended consequences of rigorous research illustrated in this case, Argyris (1968, p. 194) has called for adopting good management principles to improve the quality of scientific studies:
In our experience the more subjects are involved directly (or through representatives) in planning and designing the research,
the more we learn about the best ways to ask questions, the critical questions from the employees' views, the kinds of resistance each research method would generate, and the best way to gain genuine and long-range commitment to the research.
In another study, Bryson and Roering (1989) examined the introduction of an administrative innovation (the adoption of new planning systems) in six local governmental agencies. They found that each attempt to adopt the innovation was prone to disintegration because:
external events and crises frequently occurred, distracting participants' attention and taking away any slack resources available to adopt the innovation;
the adoption process itself was partially cumulative—what occurred before was remembered and had to be accounted for; and
participants became bogged down with information overload, conflicting priorities, and divergent issues that were outside their jurisdictions.
Bryson and Roering developed several useful recommendations for enhancing innovation adoption efforts. First, not only have a powerful innovation sponsor but also an effective process facilitator who is committed to continuing with the adoption process, particularly when difficulties and setbacks occur. Second, since disruptions and setbacks cause delays, and interest wanes with time, structure the process into key junctures—deadlines, conferences, and peak events. These structured junctures in the adoption process establish key deadlines to perform planned intermediate tasks, force things to come together, and facilitate unplanned intersections of key ideas, people, transactions, and outcomes. Finally, adopt a willingness to be flexible not only about what constitutes acceptable innovation adoption, but also in constructing arguments geared to many different evaluation criteria. In short, innovation-adoption success more often represents a socially constructed reality than an objective reality (Dornblaser et al., 1989).
Finally, because organizations are complex hierarchical systems, contradictory part-whole relations are often produced when system-wide innovations are introduced. The invention and development of an innovation at one organizational unit or level often represents an externally imposed mandate that other (often lower-level) organizational units also adopt the innovation. Thus, we have often observed a top management or policy unit express euphoria about the innovation it developed for the entire organi-
zation, while frustrations and opposition to that same innovation are expressed by affected organizational units. Such a situation often presents itself when a systemwide innovation is attempted by mandating that all organizational units adopt the innovation.
Such a situation was reported by Marcus and Weber (1989) who studied actions taken by 28 American nuclear power companies in response to a new set of nuclear safety procedures mandated by the U.S. Nuclear Regulatory Commission and the implications of those actions for organizational effectiveness. They found that the nuclear power plants with relatively poor safety records tended to respond in a rule-bound manner that perpetuated their poor safety performance. On the other hand, those plants whose safety records were relatively strong tended to retain their autonomy by adapting the standards to their local situations, a response that reinforced their strong safety performance. Ironically, those least ready or willing to adopt the innovation may be those that need it the most.
The Marcus and Weber study provides an important and generalizable inference for the management of externally imposed innovations: Be forewarned of the possible consequences of passive acceptance of external dictates by those who strictly follow the letter of the law; they may do so in "bad faith" that may not achieve the results intended. Some autonomy is needed for an adopting unit to identify with and internalize an innovation; mere formal compliance is insufficient. The disposition of innovation adopters is likely to be negatively affected if they are not granted a sufficient level of autonomy; and it is their disposition that is often critical in assuring successful adoption. The "not invented here" syndrome is well known in all sorts of organizations. Adopting agencies or organizations that have not developed any sense of commitment to those innovations may well behave bureaucratically and simply do what the letter of the law requires.
Contingencies in the Innovation Adoption Process
We may never find one best way to innovate. As the three cases of hospital innovation adoption in this volume suggest, a sophisticated manager of innovation will try instead to identify those contingent factors that influence what works and what does not. In particular, we believe that many of the key innovation processes described in the previous section are more pronounced for innovations of greater novelty, size, and duration.
Radical versus Incremental Innovations
Some innovations change the entire order of things, obsoleting the old ways and perhaps sending entire businesses the way of the slide rule or the buggy whip. Others simply build on what is already there, requiring only modest modification of one's old world view. We expect that innovations of different levels of novelty need to be managed differently. Indeed, some organizations may be well suited to one type of innovation but not another. For example, an organization that values and rewards individuals may have the advantage in radical innovation, while a more collectivist system may do better at incremental innovation (Angle, 1989). Empirically, we found that statistical relationships between perceived effectiveness and various measures of innovation ideas, people, transactions, and context were weaker for highly novel than for less radical innovations (Van de Ven and Chu, 1989). Pelz (1985) found that the stages of the innovating process were more disorderly for technically complex innovations than for technically simple innovations.
Innovation Stage and Temporal Duration
Transitions from innovation invention to development to adoption activities often entail shifts from radical to incremental and from divergent to convergent thinking. As innovations approach the culminating institutionalization step, they become more highly structured and stabilized in their patterns and less differentiated from other organizational arrangements (Zaltman et al., 1973; Lindquist and Mauriel, 1989).
The developmental pattern and eventual success of an innovation are also influenced by its temporal duration. The initial commitments and investments in launching an innovation represent an initial stock of assets that provide an innovation unit a "honeymoon" period to perform its work (Fichman and Levinthal, 1988). These assets reduce the risk of terminating the innovation during its honeymoon period in case setbacks arise or if initial outcomes are judged unfavorable. The likelihood of replenishing these assets depends on how long it takes to complete the adoption process. Interest and commitment wane with time. Thus, after the honeymoon, innovations terminate at disproportionately higher rates in proportion to the time needed for their adoption. As Pressman and Wildavsky (1973, p. 130) state, "The advantages of being new are exactly that: being new. They dissipate quickly
over time. The organization ages rapidly. Little by little the regulations that apply to everyone else also apply to it."
Size and Scope of the Innovation
It may be that small organizations have the advantage in starting up an innovation but that larger organizations with more slack resources have the advantage in keeping an innovation alive until it is completed. We found that venture capital was more risky and more difficult to obtain than was internal corporate venture funding (Van de Ven et al., 1989). Larger organizations offer a more fertile ground for sustaining and nurturing spin-off innovations. Also, there may be more places to "hide" something in a larger organization, until an innovation can stand on its own. Yet, large organizations seem to need bureaucratic systems in order to survive, and this is not particularly conducive to innovation. The message to managers is to keep finding ways to remain flexible, to permit sufficient power to concentrate on innovation, to build access to technical competence, and to listen attentively to the views of those directly responsible for implementation—factors that Nord and Tucker (1987) found were critical to successful adoption of innovations in large organizations.
This chapter has attempted to explain why the three hospital cases presented in this book were successful in adopting their innovations and what can be learned from these cases to understand organizational innovation adoption. To address these issues, we relied on an extensive and growing body of research literature on innovation adoption, and used the cases to illustrate some of the key findings in this literature. Our starting point was a review of Rogers's basic model, which is perhaps the most widely shared view of the process of innovation adoption and diffusion. Although this model is robust in explaining the adoption of innovations by individuals, it must be revised to incorporate the complexities exemplified in our three hospital cases and often observed when the organization is the locus of adoption.
The key points raised by the case studies in this volume are as follows:
The cases of hospital innovation adoption took place in organizational contexts that motivated and enabled successful adop-
tion. Moreover, the stage for adopting the innovations was set over a period of several years and involved many organizational participants.
In each of the three examples of successful hospital adoption of innovation, the innovators experienced "shocks" (not mere persuasion) as a result of direct personal confrontation with needs or problems. These shocks were sufficient to get their attention and trigger action for innovation. When people experience sufficient dissatisfaction with existing conditions, they initiate corrective action.
Once adoption activities begin, the process does not unfold in a simple linear sequence of stages and substages; instead, it proliferates into complex bundles of innovation ideas and divergent paths of activities by organizational units. The three cases of successful adoption demonstrated ways to keep the innovation process relatively simple in the face of these inexorable pressures for proliferation.
Setbacks and mistakes are frequently encountered during the innovation process, either because plans go awry or because unanticipated environmental events significantly alter the assumptions of the innovation. These setbacks signal either rejection of the innovation or opportunities for learning through reinvention. Participants in the three cases of successful innovation adoption learned through reinvention by reconnecting the causes and consequences of innovation invention, development, and adoption activities.
In the three successful cases, reinvention of the innovations developed elsewhere was facilitated by modifying the innovations to fit the local organizational situations. Moreover, top management was extensively involved and committed to the innovation, and various techniques were used to maintain momentum throughout the adoption process.
The adoption processes varied to fit the specific contingencies of the innovation being adopted by the three hospitals.
No doubt, other explanations can be offered to answer why the three hospital cases were successful in adopting their innovations. However, this chapter focused on six explanations because they suggest the revisions needed in Rogers's basic model to explain more accurately the process of innovation adoption by organizations. These revisions have important implications for the practice of organizational innovation adoption. Some of these implications were presented as normative guidelines for maneuvering the innovation adoption journey in organizations. When verified
in subsequent research, we believe they can substantially improve the odds of successful innovation adoption.
It should be emphasized that the applied principles offered here will not ensure success in adoption of an innovation. The reason for this caveat rests with a concluding lesson that we believe underlies all the key processes along the innovation journey: management cannot control innovation success; only its odds (Angle and Van de Ven, 1989). This lesson implies that a fundamental change is needed in the control philosophy of conventional management practice.
Professor William McKelvey at the University of California, Los Angeles, tells a story of the 1976 Winter Olympics, where Franz Klammer won the men's downhill skiing competition. When interviewed after the event and asked how he managed to turn in such an incredible performance, Klammer said that he had chosen to ''ski out of control.'' He knew that there were many other toplevel skiers entered against him, several of whom might outpace him on any given day, if he were to ski his normal speed—that is, under control. He chose, instead, to ski so fast that he abandoned any sense of control over the course. Although this was obviously not a sufficient condition for victory, he saw it as a necessary condition. Staying in control would virtually ensure a loss, whereas skiing out of control would make it at least possible to win.
Innovation managers may have an important lesson to learn from this vignette. By definition, an innovation is a leap into the unknown. In order that innovations have a chance to succeed, traditional notions of managerial control may need to be relaxed.
A number of practical consequences follow if innovation success is recognized to be a probabilistic process. First, innovation success or failure would more often be attributed to factors beyond the control of innovators. This, in turn, will decrease the likelihood that the careers of innovation managers will be stigmatized if their innovation fails, and increase the likelihood that they will be given another chance to manage future innovations. After all, one cannot become a master or professional at anything if only one trial is permitted. As we reported, relatively little trial-and-error learning occurred once the journey was begun for a given innovation. Repeated trials over many innovations are essential for learning to occur, and for applying these learning experiences to subsequent innovations. It is largely through repeated trials and the accumulation of learning experiences across these trials that an organization can build an inventory of competence, and thereby progressively increase its odds of innovation success.
I appreciate comments from Bright Dornblaser (University of Minnesota) and Rosabeth Kanter (Harvard University) on an earlier draft of this chapter.
Amabile, T. M. 1983. The Social Psychology of Creativity. New York: Springer-Verlag.
Amabile, T. M. 1988. A model of creativity and innovation in organizations. In Research in Organizational Behavior, Vol. 10, B. M. Staw and L. L. Cummings, eds. Greenwich, Conn.: JAI Press.
Angle, H. A. 1989. Psychology and organizational innovation. Chapter 5 in Research on the Management of Innovation: The Minnesota Studies, A. Van de Ven, H. Angle, and M. S. Poole, eds. New York: Ballinger/Harper & Row.
Angle, H. A., and A. H. Van de Ven. 1989. Suggestions for managing the innovation journey. Chapter 21 in Research on the Management of Innovation: The Minnesota Studies, A. Van de Ven, H. Angle, and M. S. Poole, eds. New York: Ballinger/Harper & Row.
Argyris, C. 1968. Some unintended consequences of rigorous research. Psychological Bulletin 70(3):185–197.
Barley, S. 1986. Technology as an occasion for structuring: Evidence from observations of CT scanners and the social order of radiology departments. Administrative Science Quarterly 31:78–108.
Bass, B. 1971. When planning for others. Journal of Applied Behavior Science 6:151–172.
Bennis, W. G., K. D. Benne, and R. Chin. 1962. The Planning of Change. New York: Holt, Rhinehart and Winston.
Bryson, J., and W. Roering. 1989. Mobilizing innovation efforts: The case of government strategic planning. Chapter 18 in Research on the Management of Innovation: The Minnesota Studies, A. Van de Ven, H. Angle, and M. S. Poole, eds. New York: Ballinger/Harper & Row.
Clark, P. 1987. Anglo-American Innovation. Berlin: Walter de Gruyter.
Cummings, L. L. 1965. Organizational climates for creativity. Academy of Management Journal 8:220–227.
Dahl, R. A., and C. E. Lindblom. 1976. Politics, Economics, and Welfare. Chicago: University of Chicago Press.
Delbecq, A. L., A. H. Van de Ven, and D. H. Gustafson. 1975. Group Techniques for Program Planning. Glenview, Ill.: Scott-Foresman.
Dornblaser, B. M., T. Lin, and A. H. Van de Ven. 1989. Innovation outcomes, learning, and action loops. Chapter 7 in Research on the Management of Innovation: The Minnesota Studies, A. Van de Ven, H. Angle, and M. S. Poole, eds. New York: Ballinger/Harper & Row.
Fichman, M., and D. A. Levinthal. 1988. Honeymoons and the Liability of Adolescence: A New Perspective on Duration Dependence in Social
and Organizational Relationships. Pittsburgh, Pa.: Graduate School of Industrial Administration, Carnegie Mellon University. Unpublished paper.
Greiner, L. E. 1970. Patterns of organizational change. In Organizational Change and Development, G. Dalton, P. R. Lawrence, and L. E. Greiner, eds. Homewood, Ill.: Irwin-Dorsey Press.
Helson, H. 1964. Current trends and issues in adaptation-level theory. American Psychologist 19:23–68.
Kanter, R. M. 1983. The Change Masters. New York: Simon and Schuster.
Kanter, R. M. 1988. When a thousand flowers bloom: Structural, collective and social conditions for innovation in organizations. In Research in Organizational Behavior, Vol. 10, B. M. Staw and L. L. Cummings, eds. Greenwich, Conn.: JAI Press.
Kimberly, J., and M. Evanisko. 1981. Organizational innovation: The influence of individual, organizational, and contextual factors on hospital adoption of technology and administrative innovations. Academy of Management Journal 24:689–713.
Levitt, B., and J. G. March. 1988. Organizational Learning. Annual Review of Sociology, 14. Greenwich, Conn.: JAI Press.
Lindquist, K., and J. Mauriel. 1989. Depth and breadth in innovation implementation: The case of school-based management. Chapter 17 in Research on the Management of Innovation: The Minesota Studies, A. Van de Ven, H. Angle, and M. S. Poole, eds. New York: Ballinger/Harper & Row.
March, J. G., and J. P. Olsen. 1976. Ambiguity and Choice in Organizations. Bergen: Universitetsforiaget.
Marcus, A., and M. Weber. 1989. Externally induced innovation. Chapter 16 in Research on the Management of Innovation: The Minnesota Studies, A. Van de Ven, H. Angle, and M. S. Poole, eds. New York: Ballinger/Harper & Row.
Nord, W. R., and S. Tucker. 1987. Implementing Routine and Radical Innovations. Lexington, Mass.: D. C. Heath.
Normann, R. 1985. Towards an action theory of strategic management. In Strategic Decision Making in Complex Organizations, J. Pennings, ed. San Francisco: Jossey-Bass.
Pelz, D. C. 1985. Innovation complexity and the sequence of innovating stages. Knowledge Creation, Diffusion, Utilization 6(3):261–291.
Peters, T. J., and R. H. Waterman, Jr. 1982. In Search of Excellence: Lessons from America's Best Known Companies. New York: Harper & Row.
Pinchot, G. III. 1985. Intrapreneuring. New York: Harper & Row.
Pressman, J. L., and A. B. Wildavsky. 1973. Implementation. Berkeley: University of California Press.
Quinn, J. B. 1980. Strategies for Change: Logical Incrementalism. Homewood, Ill.: Irwin.
Rice, R., and E. Rogers. 1980. Reinvention in the innovation process. Knowledge: Creation, Diffusion, Utilization 1(4):499–514.
Rogers, E. 1962. Diffusion of Innovation. First Edition. New York: Free Press.
Rogers, E. 1983. Diffusion of Innovations. Third Edition. New York: Free Press.
Rogers, E., and F. Schoemaker. 1971. Communication of Innovation: A Cross-Cultural Perspective. New York: Free Press.
Rosenberg, N. 1982. Inside the Black Box: Technology and Economics. Cambridge: Cambridge University Press.
Schroeder, R. G., A. H. Van de Ven, G. D. Scudder, and D. Polley. 1986. Managing innovation and change processes: Findings from the Minnesota innovation research program. Agribusiness Management 2(4): 501–523.
Thirtle, C. G., and V. W. Ruttan. 1987. The Role of Demand and Supply in the Generation and Diffusion of Technical Change. New York: Harwood Academic Publishers.
Tushman, M., and E. Romanelli. 1985. Organizational evolution: A metamorphosis model of convergence and reorientation. Pp. 171–222 in Research in Organization Behavior, Vol. 7, B. M. Staw and L. L. Cummings, eds. Greenwich, Conn.: JAI Press.
Utterback, J. 1971. The process of technological innovation within the firm. Academy of Management Journal 14:75–88.
Van de Ven, A. H. 1980a. Problem solving, planning, and innovation. Part I. Test of the program planning model. Human Relations 33(10): 711–740.
Van de Ven, A. H. 1980b. Problem solving, planning, and innovation. Part II. Speculations for theory and practice. Human Relations 33(11): 757–779.
Van de Ven, A. H. 1986. Central problems in the management of innovations. Management Science 32(5):590–607.
Van de Ven, A. H., and Y. Chu. 1989. A psychometric assessment of the Minnesota innovation survey . Chapter 3 in Research on the Management of Innovation: The Minnesota Studies, A. Van de Ven, H. Angle, and M. S. Poole, eds. New York: Ballinger/Harper & Row.
Van de Ven, A. H., H. Angle, and M. S. Poole, eds. 1989. Research on the Management of Innovation: The Minnesota Studies. New York: Ballinger/Harper & Row.
Van de Ven, A., S. Venkataraman, D. Polley, and R. Garud. 1989. Processes of new business creation in different organizational settings. Chapter 8 in Research on the Management of Innovation: The Minnesota Studies, A. Van de Ven, H. Angle, and M. S. Poole, eds. New York: Ballinger/Harper & Row.
von Hippel, E. 1981. Users as innovators. In Corporate Strategy and Product Innovation, R. R. Rothberg, ed. New York: Free Press.
Weick, K. 1979. The Social Psychology of Organizing. Reading: Addison-Wesley.
Zaltman, G., R. Duncan, and J. Holbek. 1973. Innovations and Organizations. New York: Wiley.