The use of scientific evidence and knowledge by managers
Transcription
The use of scientific evidence and knowledge by managers
Groupe de recherche interdisciplinaire en santé Secteur santé publique Faculté de médecine The use of scientific evidence and knowledge by managers Étude comparée de la collaboration interorganisationnelle et de ses effets : François Champagne N99-01 THE USE OF SCIENTIFIC EVIDENCE AND KNOWLEDGE BY MANAGERS Paper presented at «Closing the Loop : 3rd Conference on the Scientific Basis of Health Care» Toronto, October 1-3, 1999 François Champagne, Ph.D. Département d'administration de la santé et Groupe de recherche interdisciplinaire en santé Université de Montréal and Health Evidence Application Network (HEALNet) 1 Preamble What use do managers make of evidence in their decision-making? To address this question, I would like first to review how and why evidence is supposed to be used. I will propose that clearly managers cannot use evidence this way; that there are good reasons for this; and that it is unrealistic to expect that they ever would. I will then suggest, however, that managers do use evidence in all kinds of other ways, and sometimes very beneficially. I will conclude with some ideas about how to go about to realistically optimize the use of evidence by managers. Models of Knowledge Use in Decision-making The basic premise of what we may now legitimally call this movement for evidence-based decision-making at all levels of health care (clinical, organizational, systemic) is at the same time a simple but heroic assumption: that using evidence is a good thing. More specifically, this basic premise can be broken down as illustrated in Figure 1. The use of scientific evidence and knowledge, that is knowledge derived from systematic investigation conducted in such a way as being recognized as rigorous by a community of researchers should lead to higher quality decisions which should lead to the implementation of higher quality actions and consequently if all goes well, to better outcomes. The causal reasoning that links quality of decisions, quality of action and outcomes seem pretty robust. Unfortunately, the relationship between use of evidence and quality of decision is clearly less well understood. But let's not challenge it for a moment. How is evidence supposed to come into use? Through which process and why would decision-makers and managers use evidence? Several processes or models of knowledge use in decision-making have been suggested in the literature. Indeed, use of scientific evidence and knowledge in decision-making has been a concern of the knowledge utilization field which has been studying the use of social research in policy since the 1930's, of policy analysis, of 2 Figure 1 The basic premise of evidence-based decision-making Use of scientific evidence and knowledge Quality of decisions Quality of implemented actions Outcomes 3 organizational decision-making, of individual/cognitive decision sciences, of the innovation diffusion field, of program evaluation, of operations research and much more recently of evidence-based medicine. Much of the discussion that follows is based on the seminal work of Carol Weiss (1977, 1979) of Harvard University on the use of social science knowledge in social policy. The most traditional and venerable model of knowledge use may be called the knowledge-driven model (Figure 2). This model is derived from the physical sciences. It assumes that the mere fact that knowledge exists naturally and inevitably presses it toward development and use (Weiss, 1977, 1979). According to this model, basic research discloses some opportunity whose practicality is defined and tested by applied research. implemented. If all goes well, appropriate technologies are developed and Although one can find several examples of this model in physics, chemistry and physiology (the development of therapeutic drugs being a prime examplar), very few examples can be found in the social sciences in general and in management in particular at least for 2 central reasons (Figure 3) (Weiss, 1977, 1979). First, management knowledge is not apt to be so compelling or authoritative as to drive inevitably toward implementation. Management is inherently contextual. It is management of context. It cannot be dissociated from it. Consequently, generalizations are hazardous and although knowledge may provide thorough understanding of management situations, its prescriptive value will always be limited. As Whitley (1988) wrote very persuasively: "The production of scientific knowledge which could form the foundation for managerial skills standardized across different hierarchical arrangements and circumstances seems, then, fraught with difficulties. The variety of managerial practices, their organizationally embedded nature and their susceptibility to change and control by enterprise top managers, render attempts at standardizing managerial problems for solution by standardized skills 4 Figure 2 The knowledge-driven model of scientific evidence use Basic research Applied research Technological development Use (adoption of technology) Quality of implemented actions Outcomes 5 Figure 3 Limits to the applicability of the knowledge-driven model in management Basic research Applied research Management knowledge is not apt to be so compelling or authoritative as to drive inevitably toward implementation Technological development Management knowledge does not readily lend itself to conversion into technology Use (adoption of technology) Quality of implemented actions Outcomes 6 across firms and industries of limited value. (...) In this view, attempts to establish a general «science of managing» which would generate knowledge of highly general relations between limited and standard properties of separate, standard objects are doomed to failure since «managing» is not a standardized activity." Secondly, management knowledge does not readily lend itself to conversion into replicable technology. Although managers may and do use manage technology, the technology of management itself is fluid, imprecise, one could say prehistorical. A second model, maybe the more common and indeed the one that lies behind both evidence-based medine and much of the current efforts by funders of research to increase the relevance of research to practice, is the problem-driven (or problem-solving or decision-driven) model of scientific evidence use (Figure 4) (Weiss, 1977, 1979). This model, like the preceding one is an instrumental model, in the sense that evidence is seen as directly useful and applicable to a decision. According to this model, the path to scientific evidence use starts with the definition of a problem situation. Information or understanding is lacking either to generate a solution to the problem or to select among alternative solutions. Research will provide the answer (Weiss, 1977). In variant A of this model, the one underlying evidence-based medicine, research predates the recognition of the problem. Decision-makers faced with a decision go out and search for information from preexisting research. Or the information is brought to their attention by staff, analysts, colleagues, consultants or they see it somewhere in a journal, newsletter or at conferences (Weiss, 1977, 1979). In variant B, the one underlying efforts to establish partnerships between researchers and practitioners, research is commissioned to fill the knowledge gap. In both variants, newly acquired knowledge is used to better understand the problem situation and to choose among alternatives courses of action. 7 Figure 4 The problem-driven model of scientific evidence use Variant A acquisition of knowledge from pre-existing knowledge reservoir Definition of problem Identification of missing knowledge Interpretation for the problem situation Variant B acquisition of knowledge through commissioned research Use (adoption of technology) Quality of decision Quality of implemented actions Outcome 8 In variant A, where research predates the problem, the usual prescription for increasing the use of knowledge is to make findings more easily accessible through improved communication and transfer of research to decision-makers. In variant B, improved use will result from more control by funders or decision-makers on the specification and conduct of the requested research. As with the previous knowledgedriven model, there are several limits to the applicability of this problem-driven model to managerial decision-making (Figure 5). First in many managerial situations, the parameters of the problem may not be clearly defined or even more likely, there might not be consensus of these parameters. Secondly, even when there are clear-cut problems to be solved and clear-cut decisions to be made, decision-makers do not always know what kind of information they need to know. Thirdly, in many cases, indeed in all complex cases, research findings may be ambiguous, less than clear-cut, with weak support and low power. Research is rarely comprehensive or convincing enough to supply the correct answer. In addition, time constraints may preclude thorough knowledge search (in variant A) or patient waiting for results (in variant B). Even more importantly, acquisition of knowledge may be impeded by the widely acknowledged cultural gap between the research and users communities. It may indeed be the most persistent observation in the literature on knowledge utilization that researchers and decisionmakers belong to separate communities with different values and ideologies and that these differences impede utilization (Beyer and Trice, 1982). Finally, the implication of knowledge for action may not be clear-cut either. Understanding may not directly lead to recommendation for action. Implications of findings for action may be far from clear (Scriven, 1991, 1995). So there appears to be serious limits in the applicability to management of the theoretical models that underlie the current calls for greater instrumental use of scientific evidence in decision-making. Clearly, management knowledge will not inexorably lead to development and action as in the knowledge-driven approach and 9 Figure 5 Limits to the applicability of the problem-driven model in management Definition of problem - unclear problem - no consensus Identification of missing knowledge - difficulty in identifying information needs Variant A acquisition of knowledge from pre-existing knowledge reservoir Variant B acquisition of knowledge through commissioned research - ambiguous findings - time constraints - cultural gap Interpretation for the problem situation Use (adoption of technology) Quality of decision Quality of implemented actions - troublesome link between knowledge and action Outcomes 10 only in very rare circumstances would all conditions posited in the problem-driven model be met in the managerial world for research to directly influence decisions. Types of Managerial Knowledge To add more gloom to this picture, let us come back to the basic premise that use of scientific evidence leads to better decisions. It is safe to say that although the use of evidence might contribute, might enhance or at least might be related to the quality of decisions, there is little doubt that many other factors influence the quality of any decision (Figure 6). First, it has been accepted since the early work of Herbert Simon (1945) that decision-making in any complex or unprogrammed situation cannot be perfectly rational as would be a decision uniquely based on evidence. Rationality is limited by the limits in human cognitive processes and by the presence and interaction of multiple values, stakes and stakeholders. Indeed, most decision situations do involve multiple participants with multiple views and interests. Secondly, managers will be informed and influenced by all kind of knowledge other than scientific. First, practical knowledge, that is personal knowledge derived from experience, is almost universally highly valued by managers. In addition, managers lend much credence to anecdotal evidence, that is knowledge they derive from their personal observation either fortuitous or provoked of particular situations, usually in organization very similar to theirs. Indeed, many managerial decisions may be based on imitation of esteemed peers. Second, intuitive knowledge may be the most highly regarded form of knowledge in some management circles. For some, intuition is some type of holistic and configurational thinking process distinct from the usual linear, analytical mode of thinking (Mintzberg, 1976, 1989). For others, intuition is analysis frozen into habit (Simon, 1987). Whatever your view on intuition, you will agree that it tends to be a major factor in a managerial decision-making. A closely 11 Figure 6 Use of scientific evidence in a context of limited rationality decision-making Instrumental use of scientific evidence and knowledge Quality of decisions Quality of implemented actions Characteristics of decisions - stakes - stakeholders - processes Knowledge production paradigm Use of - practical knowledge - anecdotal evidence - intuitive knowledge/wisdom - expert reasoning (pattern recognition, forward reasoning and parallel processing) Outcomes 12 related issue is that in some familiar circumstances, managers may be considered as expert decision-makers, that is they may use expert reasoning involving pattern recognition, parallel processing and forward reasoning (Patel et al., 1996). In other words, they will not decide by carefully evaluating competing alternatives, but by recognizing a problem and its solution as a function of a highly structured knowledge base but with minimal use of explicit knowledge. An additional factor which may influence the instrumental use of evidence is the epistemological foundation according to which this evidence was produced. Traditionally, the positivist paradigm was the dominant management research paradigm. this paradigm posits the existence of a single reality, of a single truth independent of any observer's interest. It also asserts that the observer can and should remain detached and distant from its object of study. Competing paradigms include critical theory, post-modernism and constructivism. All of these share a belief that knowledge should be constructed by using actors' subjectivity rather than trying to exclude it and that the inquirer and the inquired are interlocked in such a way that the findings are constructed through their interaction (Guba and Lincoln, 1989, 1990). When knowledge is produced as it usually is, that is according to the positivist paradigm, with its emphasis on isolating the object of study from its content and eliminating subjectivity, managers will tend to dismiss it as theoretical, which is a word they use for things they do not like and that they find fatally flawed and not grounded in reality. Types of Use of Knowledge by Managers So, should we be discouraged about the possibility and value of using scientific evidence in managerial decision-making? Although managers may not or may 13 with difficulty use evidence instrumentally, that is to directly, specifically and punctually influence decisions, they do use evidence in all kind of other ways (Figure 7). First, they may come to be influenced and informed by evidence indirectly through disorderly interaction with researchers. Researchers are one of the actors within the policy and decision-making arena and as such may propagate evidencebased opinions, ideas, understanding and solutions. Evidence is thus used as part of a public deliberation involving multiple stakeholders Weiss (1977, 1979) labeled this model the «interactive model». Evidence may also be used by managers to support their a priori position on an issue involving multiple actors and multiple interests. Research is used politically «as ammunition for the side that finds its conclusions congenial and supportive» (Weiss, 1979). There is also another way that managers use evidence strategically. This time, managers can use the research process itself as a tactic either to delay decision and action, to avoid taking responsibility for a decision, or as a public relation exercise. In all of these situations, it is the sheer fact that research is being done and will be used that is invoked not the content of the findings (Weiss, 1979). Finally, the most common use of scientific evidence by decision-makers including managers may be conceptual, that is as background, integrated knowledge for understanding. In this process which has been called an enlightenment process (Janowitz, 1972, Crawford and Biderman, 1969, Rich and Caplan, 1976; Pelz, 1978; Weiss, 1979), the concepts and theoretical perspectives of social and management sciences permute the decision-making process. Knowledge used in decision-making is constituted by accumulated and integrated evidence and generalizations rather than findings from a specific study. 14 Figure 7 Models of non-instrumental use of scientific evidence • The interactive model : researchers interact with all others stakeholders. They all engage in a disorderly fashion in mutual consultations that progressively move closer to potential decision options ⇒ deliberative use of evidence • The political model : scientific evidence is used as ammunition in a political debate ⇒ political use of evidence • The tactical model : the research process is used to delay decision and action, to avoid taking responsibility for a decision or as a public relations exercise ⇒ tactical use of evidence • The enlightenment model : managers use background, integrated knowledge for understanding ⇒ conceptual use of evidence 15 So managers do not often use evidence instrumentally but do use it deliberately, politically, tactically and conceptually. What can we make of this to inform and advance the debate on evidence-based decision-making? (Figure 8). Is Evidence-based Managerial Decision-making Possible? First, we should recognize and accept the fact and value of noninstrumental use of scientific evidence. The conceptual use of research in particular may well represent use in the truest sense of the word. Knowledge internalized by managers gets used constantly with much impact. In addition, the influence of conceptual use of evidence does not rest on untenable assumptions of rational and consensual decision processes. Similarly, the deliberative use of evidence in the public arena acknowledges a proper use of scientists' expertise and capacity to synthesize knowledge so as to meaningfully contribute to the solution of salient problems. The political use of evidence should not be distrusted or undervalued either. Given the inherently political nature of managerial decision-making, the political use of evidence is inevitable. Contrarily to population opinion, there is nothing wrong with using research to support a predetermined position. This is neither unimportant nor improper. Only distortion and misinterpretation of findings are illegitimate (Weiss, 1977, 1979). Secondly, we should, in particular funders should, accept the fact that some research will never be used. As with any human enterprise, some research will be flawed and should not be trusted nor used. More importantly, research is a risky business. Some will not lead to usable results or indeed to anything. We usually value risk taking We should do so in research too. 16 Figure 8 Some points for discussion • Recognize and accept the prevalence and value of non-instrumental use of scientific evidence • Accept the fact that some research will never be used • Beware of the “error of jumping to recommendation” • Critically review current attempts at partnerships between decision-makers and researchers • unrealistic expectations • conflicts of interests/managerialism • epistemologically unsound ? • Adopt a comprehensive theoretically informed framework for the analysis and optimization of the use of scientific evidence and knowledge in decision-making 17 Thirdly, it should be acknowledged that the link between research findings are evidence on the one hand and recommended action on the other may be tortuous and ambiguous. Pressure for the use of evidence should not lead to errors of jumping to unwarranted recommendations (Scriven, 1995). Fourth, current attempts to promote greater use of evidence in decisionmaking usually consists in establishing close partnerships between decision-makers and researchers involving various linkage and exchanges throughout the research process. These should be critically reviewed on several grounds. First, they create unrealistic expectations of usable results. It is foolish for a scientist to guarantee to discover explanations or provide usable recommendations as a result of investigation of a particular situation (Scriven, 1991, p. 304)). Science is not plumbing. Secondly, overemphasis on utilization may create strong conflicts of interest situation for researchers because it places pressure on them to adjust findings to what decisionmakers are willing to do rather than to what they should do (Scriven, 1991, p. 370). According to Guba and Lincoln (1989), this tendency toward managerialism is detrimental since it precludes critical analysis of managerial action and is almost inevitably disempowering and unfair with the ultimate power resting with decisionmakers. An even more serious criticism of current partnerships attempts is that most of them involve problem-driven commissioned research projects, that is they are partnerships around specific research projects aimed at providing understanding or solution to specific problems. This approach is potentially troublesome on epistemological grounds. Knowledge is not simply data or information. Knowledge is not a simple statement of acts emerging from a local situation (Kerr, 1984). Knowledge is constituted by the accumulation of evidence. As a scientist, I am concerned about the possibility of any decision being too overly influenced by any single research project. There are few critical experiments in research and none in social science and management research. Evidence-based medicine with its emphasis on research 18 synthesis does avoid this flaw of variant B of problem-driven research. Research partnerships in health policy and management research could likewise be built around research synthesis rather than around research projects. Finally, the determinants and impact of the use of scientific evidence in organizational decision-making should be studied using a comprehensive and theoretically informed framework. Various literature streams mentioned earlier could indeed inform such a framework which could lead to optimizing, by which I purposefully mean finding the proper role and place, not maximizing, the use of scientific evidence in managerial decision-making. Figure 9 presents an interpretation of the type of factors that could be considered1. Although this framework would obviously require much discussion, my purpose here is simply to highlight major factors. You are already familiar with the right hand side of this figure which depicts the relationship between the instrumental use of scientific evidence and knowledge, quality of decision and action, and outcomes in a limited rationality context involving multiple stakeholders and where managers also rely on other types of knowledge. This framework also recognizes that scientific evidence and knowledge will be used in other non-instrumental ways and will accordingly impact on the quality of decision. All forms of use depend first on evidence and knowledge characteristics, that is its availability, accessibility and validity. Data availability can be influenced by research funding and targeting practices. Data accessibility can be influenced by efforts to improve exposition of decision-makers to evidence and knowledge, most notably research transfer and brokering practices. Validity of scientific evidence and knowledge rests on the strength of the knowledge base and on the methodological and epistemological approaches that drive the production of knowledge. Indeed, my earlier 1 This model is an elaboration of a model proposed by Oh and Rich (1996) Figure 9 A framework for the analysis and optimization of the use of scientific evidence and knowledge in decision-making Communication and brokering Funding and commissioning Characteristics of scientific evidence availability accessibility validity Characteristics of data Characteristics of decision-makers -values and beliefs -attitudes/motivation -skills -scientific literacy Non-instrumental use of scientific evidence and knowledge deliberative political tactical t l Instrumental use of scientific evidence and knowledge Characteristics of organizational and systemic context specialization formalization managerial attention culture interorganizational collaboration Quality of decisions Quality of implemented actions Outcomes Characteristics of decisions - stakes - stakeholders - processes Knowledge production paradigm Organizational action and performance Health system action and performance Societal context Use of - practical knowledge - anecdotal evidence - intuitive knowledge/wisdom - expert reasoning (pattern recognition, forward reasoning and parallel processing) 20 criticism of partnership between researchers and decision-makers should not in any way be taken to be a criticism of the partnership between researchers and inquired participants called for in non-positivist paradigms of research. Proponents of these and most notably constructivists, strongly feel that those interactive, intersubjective, naturalistic and participative research processes will enhance usefulness and use of research findings. Individual decision-makers characteristics should also be taken into account. Indeed, their values and beliefs, attitudes and motivation, skills and scientific literacy will influence the propensity of individuals to use scientific evidence and knowledge. Organizational characteristics will also facilitate or impede the use of scientific evidence and knowledge. Less formalization, that is a more flexible and organic structure, more specialization, that is a more complex division of labor, more managerial attention to use and more participation in interorganizational collaboration and networking should all be associated with greater use of evidence and knowledge by managers. This last factor may indeed be one of the most important given what we know about the prevalence of imitation behaviors among managers. Indeed, this observation of a central influence on knowledge use of esteemed peers and fellow network members is supported by many different lines of inquiry including innovation diffusion theory, social learning theory, communities of practice and communities of knowing approaches to organizational learning as well as academic detailing and opinion leaders approaches to the implementation of practice guidelines. 21 Conclusion Managers can and do use scientific evidence and the quest for increased evidence-based decision making in health care management is clearly legitimate. However, this quest should be pursued with realistic expectations given the nature of management knowledge and of managerial decision-making, with proper acknowledgement of the value and legitimacy of indirect, non-instrumental use of evidence and with a broad and thorough understanding of the factors involved in use and consequently in the optimization of the use of scientific evidence in decisionmaking. 22 References Beyer JM, Trice HM (1982). The Utilization Process : A Conceptual Framework and Synthesis of Empirical Findings. Administrative Science Quarterly 27:591-622. Crawford ET, Biderman AD (1969). The Functions of Policy-Oriented Social Science. In Crawford and Biderman (eds) Social Scientists and International Affairs. New York : Wiley. Guba EG, Lincoln YS (1989). Fourth Generation Evaluation. Newbury Park : Sage. Guba EG, Lincoln YS (1990). Competing Paradigms in Qualitative Research. In Denzin N, Lincoln YS (eds) Handbook of Qualitative Research. Newbury Park : Sage. Janowitz M (1972). Professionalization of Sociology. American Journal of Sociology. 78:105-135. Kerr DH (1984). Barriers to Integrity : Modern Modes of Knowledge Utilization. Badder: Westview Press. Mintzberg H (1976). Planning on the Left Side, Managing on the Right. Business Review. July-August. Harvard Mintzberg H (1989). Mintzberg on Management. New York : The Free Press. Oh CH, Rich RF (1996). Explaining Use of Information in Public Policy-Making. Knowledge and Policy : The International Journal of Knowledge Transfer and Utilization 9(1):3-35. Patel V, Kaufman DR, Magder SA (1996). The Acquisition of Medical Expertise in Complex Dynamic Environments. In Erickson KA (ed) The Road to Excellence: The Acquisition of Expert Performance in the Arts and Sciences, Sports and Games. Hillsdale, NJ : Lawrence Erlbaum. Pelz DC (1978). Some Expanded Perspectives on Use of Social Science in Public Policy. In Yinger JM, Cutler SJ. Major Social Issues : A Multidisciplinary View. New York : The Free Press. Rich RF, Caplan N (1976). As quoted in Pelz (1978). Scriven M (1991). Evaluation Thesaurus. Newbury Park : Sage. 23 Scriven M (1995). The Logic of Evaluation and Evaluation Practice. New Directions for Program Evaluation. Winter, no 68. Simon H (1945). Administrative Behavior. New York : The Free Press. Simon H (1987). Making Management Decisions : The Role of Intention and Emotion. The Academy of Management Executive 1(11):57-64. Weiss CH (1977). Introduction. In CH Weiss (Ed.). Using Social Research in Public Policy Making. Lexington, MA : Lexington Books. Weiss CH (1979). The Many Meanings of Research Utilization. Public Administration Review sept.-oct., 425-431. Whitley R (1988). The Management Sciences and Managerial Skills. Studies 9(1):47-68. Organization Adresse de correspondance Prière d'adresser toute correspondance concernant le contenu de cette publication ou autres rapports déjà publiés à : Groupe de recherche interdisciplinaire en santé Secteur santé publique Faculté de médecine Université de Montréal C.P. 6128, Succ. Centre-Ville Montréal (Québec) H3C 3J7, Canada Téléphone : (514) 343-6185 Télécopieur : (514) 343-2207 Adresse de notre site Web http://www.gris.umontreal.ca/