Due to the heritage and history of operations management, its research ... approach to business education and research (Laidlaw (1988)). ...... embodied in software for expert and decision support systems and in mathematical models of.
JOURNAL OF OPERATIONS
MANAGEMENT
Vol. 8, No. 4, October 1989
Alternative Research Paradigms in Operations JACK R. MEREDITH* AMITABH RATURI* KWASI AMOAKO-GYAMPAH* BONNIE KAPLAN**
EXECUTIVESUMMARY Due to the heritage and history of operations management, its research methodologies have been confined mainly to that of quantitative modeling and, on occasion, statistical analysis. The field has been changing dramatically in recent years. Firms now face numerous worldwide competitive challenges. many of which require major improvements in the operations function. Yet, the research methodologies in operations have largely remained stagnant. The paradigm on which these methodologies are based, while useful, limits the kinds of questions researchers can address. This paper presents a review and critique of the research in operations, itemizing the shortcomings identified by researchers in the field. These researchers suggest a new research agenda with an integrative view of operations’ role in organizations, a wider application of alternative research methodologies, greater emphasis on benefit to the operations manager, cross-disciplinary research with other functional areas, a heavier emphasis on sociotechnical analysis over the entire production system, and empirical field studies. Some of the alternative research methodologies mentioned include longitudinal studies, field experiments, action research, and field studies. Following a description of the nature of research, three stages in the research cycle are identified: description, explanation, and testing. Although research can deal with any stage in this cycle, the majority of attention currently seems to focus on the explanation stage. The paper then discusses historical trends in the philosophy of science, starting with positivism, expanding into empiricism, and then leading to post-positivism. The impacts of each of these trends on research in operations (which remains largely in the positivist mode) are described. Discussion of the importance of a plurality of research methods concludes the section. A framework for research paradigms is then developed based on two key dimensions of research methodologies: the rational versus existential structure of the research process and the natural versus artificial basis for the information used in the research. These dimensions are then further explored in terms of thirteen characteristic measures. Next, research methodologies commonly used in other fields as well as operations are described in reference to this framework. Methodologies include those traditional to operations such as normative and descriptive modeling, simulation, surveys, case and field studies as well as those more common to other fields such as action research, historical analysis, expert panels, scenarios, interviewing, introspection, and hermeneutics. Examples from operations or allied fields are given to illustrate the methodologies. Past research publications in operations are plotted on the framework to see the limitations of our current paradigms relative to the richness of other fields. We tind that operations methodologies tend to Manuscript
received November
*University
of Cincinnati,
**American
Journal
University,
15, 1988; accepted
Cincinnati, Washington,
of Operations
December
22, 1989, after two revisions.
OH 45221-0130 D.C. 20016
Management
297
lie on the more rational end of the framework while spanning the natural/artificial dimension, though the majority of research is at the artificial pole. Last, recommendations are made for applying the framework and paradigms to research issues in operations management. The topics of quality management and technology implementation are used as examples to illustrate how a wide variety of methodologies might be employed to research a much broader range of issues than has currently been researched.
INTRODUCTION The field of operations, or operations management (OM)t, faces multiple new research challenges in the areas of service operations, productivity, quality, technology and many other areas. Never before has the need for pragmatic research, directly useful to the operations manager, been so important to the field, and to industry and society. One perspective on this is offered by Galliers and Land (1987) in reference to information systems, but just as applicable to operations: “[It is] an applied discipline, not a pure science. It follows, therefore, that if the fruits of our research fail to be applicable in the real world, then our endeavors are relegated to Yet, OM researchers still tend to employ a limited range of the point of being irrelevant.” research paradigms to address the new challenges in operations. A part of the reason is historical. Originally, the research orientation of operations was entirely pragmatic: What procedures should be used in what situations? The presentation in and early textbooks on production (e.g., Mitchell (1939)) focused on the organization transformation process through a combination of descriptive and prescriptive discussion (Andrew and Johnson (1982)). The unit of analysis was the production manager and the definition of production management centered around what the production manager did. For example, in explaining why plant location was left out of his book, Mayer (1962, p. v) stated: “ . . there is a reason to believe that the production manager will play a relatively minor role in these areas of decision making.” Then in the 1950s the Ford (Gorden and Howell (1959)) and Carnegie (Pierson (1959)) Foundations’ reports severely criticized business colleges for their lack of rigor or a scientific approach to business education and research (Laidlaw (1988)). Operations research (OR) was moving from war applications into the business and industrial arena and with it, the opportunity for business schools to gain academic respectability. OR quickly developed into a favorite tool (e.g., Holt, Modigliani, Muth, and Simon (1960)) for conducting research in operations (Buffa (1965, p. v.), Nistal (1979-80); Andrew and Johnson (1982)). It also allowed, for the first time, the development of a systematic body of knowledge in operations based on a consistent and rigorous framework. Although useful to numerous areas of business, the predominant application of OR was to the area of operations (Buffa (1968, p. 4)). Marketing, finance, and organizational behavior also used the new tool but found it somewhat limited in its applicability to their problems (Hudson and Ozanne (1988)). Ackoff (1979, p. 94), for example, remarked that the OR approach “came to be identified with the use of mathematical models and algorithms rather than the ability to formulate management problems, solve them, and implement and maintain their solutions in turbulent environments.” Marketing and organizational behavior, in particular, developed a number of other paradigms drawn heavily from the fields of psychology and sociology to address their research problems. But OM researchers were having great success with the new algorithmic modeling tools and found no need to explore other paradigms (Andrew and Johnson (1982)). The new area of operations research/management science (OR/MS) was steadily replacing the function of
298
Vol. 8, No. 4
operations in academia and operations was becoming seen as an applied portion of the OR/MS field. Simultaneously, the field of operations lost considerable interest as its sister functions grew in size and importance both in industry and in academia. As Galbraith (1958) put it some years earlier, the United States had “solved the problem of production. ” The attention and resources of firms were thus directed to marketing and finance instead of “toward improving manufacturing capabilities” (Hayes and Wheelwright (1984, p. 20)). In response, operations academics began to feel an “identity crisis” (Andrew and Johnson (1982)) reflecting the nature of those teaching in the field as well as the content of the courses. Raiszadeh and Ettkin (1988), in their survey of operations curricula, found a preponderance of “diversity” in the faculty teaching operations and, therefore, also in the course content. There are two major milestones in the discipline of operations. The first, the virtually overnight reorientation of the field to an analytical approach based on quantitative modeling, was the most revolutionary. Another reorientation response to the Ford and Carnegie reports was the systems analysis conception of the field, though this approach never grew the way the quantitative approach did. Two textbooks offering this latter perspective of the field were Starr (1964) and Greene (1965). The second major milestone occurred in the late 1970s and early 1980s as the Operations Management Association (OMA) and the OMA-United Kingdom were formed and their journals began publication. At this point, researchers in OM took stock of the field and found its name being inappropriately applied, its original faculty largely retired and replaced with quantitatively-oriented faculty, and the business world seriously in need of its attention (Grayson (1973)). Moreover, OM researchers addressing the problems of production and productivity through the now-standard quantitative modeling paradigm were more and more simply talking among themselves (Buffa (1968, p. 5)). Managers looked at this “research” and found that they could neither understand the solutions being proposed nor the problems OM researchers thought they were addressing (Andrew and Johnson (1982)). Along similar lines, Anderson, Chervany, and Narasimhan (1979) noted “. . . managers felt that the existing implementation research had little practical value for them in their day-to-day responsibilities . . . .” This attitude is confirmed by Buxey (1984, p. 529): “. . . it is debatable whether the practice of production management is much influenced by what appears in leading production research periodicals . . . it does not capture the essential flavour of what the manager has to do and is therefore unlikely to be of use to him . . .” And as McKay, Safayeni, and Buzacott (1988, p. 87) concluded about job-shop scheduling: “The problem definition is so far removed from job-shop reality that perhaps a different name for the research should be considered.” Our point is not that OR/MS methodology is inappropriate for research in operations (e.g., see Orden (1988)), but that it should not be the only methodology. We would reiterate the sentiments of Buffa (1980) who noted that “. MS/OR methodology does not define the OM field nor point the way of the future.” Yet, research in operations has still not changed significantly (Amoako-Gyampah and Meredith (1989)). What is needed at this point is a broader understanding by OM researchers, as well as journal editors and referees, of the variety and acceptability of alternative research paradigms that other fields use and have accepted as rigorous. This paper is intended as a first step in this direction.
Journal
of Operations
Management
299
A REVIEW
OF RESEARCH
IN OPERATIONS
AS described by the following critics, past research in operations a variety of shortcomings:
has too frequently
exhibited
1. Narrow instead of broad scope Focused on problems with a narrow scope (Buffa (1980)) Largely micro-oriented (Chase (1980)) Concerned a subsystem rather than a whole system (Buffa (1980)) Used only a single-criterion quantitative model (Buffa (1980)) 2. Technique instead of knowledge orientation Dominated by the application of techniques (Chase (1980)) Assumed to be simply applied operations research (Voss (1984), p. 29) 3. Abstract instead of reality perspective Used approaches largely confined to the laboratory and based on model formulation and manipulation (Chase (1980)) Emphasized equipment rather than people (Chase (1980)) Rarely involved field studies (Chase (1980)) Even in the few studies using real-world settings, the research approaches were characterized by one-day visits, interviews, and the use of questionnaires (T. Hill (1987)). l
l l
l
l
l
l
l
l
l
In sum, it appears that OM research has failed to be integrative, is less sophisticated in its research methodologies than the other functional fields of business, and is, by and large, not very useful to operations managers and practitioners. Operations is an applied field and its research should be usable, in some fashion, in practice. It is not, like management science or organizational behavior, a tool area but a functional discipline. As Voss (1984, pp. 29, 30) notes: “. . the production/operations management person is concerned with procedure and process . as well as . linking operating decisions and policies with company policies and the decisions, technologies, and procedures they should adopt to maximize company competitiveness.” It is worth noting that this extends far beyond the normative perspective of management science; it includes exploration and interpretation of procedures and processes which may not be embedded in a rational, single-objective, or value-free context. The decision environment of the real-world operations manager is not usually driven just by quantitative elements susceptible to mathematical modeling-where does politics, law, ethics, or the environment fit in? From the very first issue of the Journal of Operations Management, authors have called for research that examines the unstructured real-world problems of operations practitioners, of considers multiple evaluation criteria, recognizes both the inter- and intrarelationships organizational units, and incorporates both parametric and nonparametric statistical tests (Buffa (1980), Chase (1980)). But such prescriptions are rarely accompanied by an overview of the systematic and methodological changes in research techniques that will support these forays into new research settings. We hope to provide this overview here. Several researchers have addressed the content problems in their proposals for a new OM research agenda. For example: l
l
Miller and Graham (198 1) called for an under the broad categories of operations turing strategy, the role of the customer technologies on operations policies were As an extension to Miller’s agenda, Groff
300
integrative view of operations’ role in organizations policy, control, productivity, and services. Manufacin service delivery systems, and the effects of new mentioned as key issues in the agenda. and Clark (198 1) called for a wider array of research
Vol. 8, No. 4
l
l
l
l
l
l
l
methodologies to broaden the theoretical foundations of operations. Hax (198 1) suggested a redirection of OM research so as to be of more benefit to the operations practitioner. Focusing on services, Mabert (1982) suggested studying the interrelationships between formal planning systems and the databases to support these systems. Sullivan (1982) advocated cross-disciplinary research between operations, marketing, and organizational behavior for a more holistic approach to real-world problems. More recently, Chase and Prentis (1987) recommended more interdisciplinary research and the application of sociotechnical analysis to the total production system. In a review of recent operations dissertations, Hill et al. (1987) identified the need for new research methodologies to address more managerial and macro-level issues. In an update of the progress on the original Miller et al. (198 1) agenda, Amoako-Gyampah and Meredith (1989) reviewed journal and proceedings publications between 1982-1987. They concluded that the dominant strategies used for research in the operations discipline continue to be model building and laboratory simulation. Finally, Swamidass (1988a) reviewed the difficulties operations has been having with its existing research paradigms and pointed out the crucial need in operations for empirical field studies, which he identifies as a “new frontier.”
Except for Swamidass (1988a), none of the previous work addressed the causes of the industry-academia gap. Individually, these researchers recognized the need to bridge the gap and offered a wide range of prescriptions. These papers clearly drew attention to the problem and provided incentive to work on real problems-yet it did not happen. Since the prescriptions were not embraced widely, it would seem that the problem lay with the inadequate fit between the problems being addressed and the research paradigm used by OM researchers. The current paradigm is typically prescriptive, deterministic, non-contextual, and exhibits a preponderance of “rational” constructs. A major contributing factor to this narrowness of current OM research is a lack of knowledge about appropriate alternative research paradigms. Reisman (1988) has defined a number of generic research strategies for the management and social sciences. Also, some of the authors mentioned above outlined a few alternative methodologies for the issues they recommended investigating. These alternative methodologies, as well as a number of others, will be discussed in more detail in a later section. Next, we explore the concept of research as a foundation for the paradigms we describe later. We start with some definitions of research and then examine how knowledge is accumulated through a repetitive cycle of research. A background section, briefly describing the history of research and post-positivist thought with examples from the field of operations, is appended following the references. This section is for those seeking a broader foundation in the philosophical development of research methodology. THE NATURE OF RESEARCH The Research Cycle Arguments frequently arise about whether the best research is that which proposes knowledge or that which validates knowledge. Borrowing from Emory’s (1985) three research tasks of description, prediction, and explanation, we suggest that all research investigations involve a continuous, repetitive cycle of description, explanation, and testing (through prediction), as
Journal
of Operations
Management
301
illustrated in Figure 1. Thus, proposing knowledge (explanation) and validating knowledge (testing) simply are two stages in the ongoing cycle of research. An individual research study may involve only one of the stages in the cycle at a time. We consider each of the three stages in detail. It should be noted that, in practice, the stages rarely are as clear and distinct as we portray them here. As with many stage models, the boundaries between the stages are purposely drawn sharply here for analytical reasons. In actuality, a researcher may intertwine activities from different stages, step through the stages in a different order, or backtrack as the research progresses. We initiate our discussion with “description” since this must precede explanation and testing.
THE ONGOING
FIGURE 1 CYCLE OF RESEARCH
STAGES
Description. Descriptive research seeks to report and chronicle elements of situations and events. As noted previously, the predominant activity of early research in operations was descriptive. The approaches and techniques available for capturing this information depend on the field and on the nature of the situations and events of interest. The result is a welldocumented characterization of the subject of interest. This characterization then may be used for generating or testing theories, frameworks, and concepts regarding the situation. For example, Meredith (1984) describes the complications that arose in the simple process of attempting to purchase a copying machine for a university department and Heller (1951) describes investment decisions in general. A finer, more detailed level of description about a particular facet of the subject may require what is sometimes known as exploratory research. Here, a particular aspect is investigated more fully, based on the understanding that the preliminary descriptive research gave. This
Vol. 8, No. 4
understanding may have illuminated areas of confusion, unearthed contradictions in previous concepts or “facts” about the situation, or given further meaning to areas of interest or to existing knowledge. The result of the exploratory research is more detailed description that may lead to further insight and understanding. Miles (1983) argues that such qualitative data are attractive for many reasons: “They lend themselves to the production of serendipitous findings and the adumbration of unforeseen theoretical leaps.” A number of areas in operations need realistic descriptions. Examples include MRP and shop floor control systems, new manufacturing technologies, operational problems relating to new technologies and systems, what operations managers’ jobs consist of, and even the organizational decision-making process concerning the adoption of new operational imperatives such as JIT, TQC, FMS, CIM. Explanation. On the basis of, or in the process of producing, a description, some initial concepts about the situation may be postulated. Perhaps some action-reaction or cause-effect relationships may be inferred. Or possibly a more complex set of reactions or relationships may be constructed to explain the observed behavior or events. If a complex, relatively closed set of relationships appears to be operating, a “framework” may be constructed to explain the dynamics of the situation. A framework offers a conceptual frame of reference to help researchers design specific research studies, interpret existing research, and generate testable hypotheses. One example in operations is Saladin’s (1984) conceptual model of the scope of the field. At a more integrative level and with further testing, the framework or sets of frameworks may be developed into a theory describing the principles operating in the situation. There are many definitions of “theory” such as “a set of general principles that explain observed facts,” but more importantly, Dubin (1969) has identified a number of characteristics typical of all theories: a theory must include the interrelationships between its variables and/or attributes as well as some criteria that define its boundaries. The theory must also improve our understanding of the non-unique phenomenon or help us make predictions about it. Finally, the theory must be interesting (Davis (1971)), that is, non-trivial. Note that, as Striven (1962) argues, a prediction is not the same as an explanation; the former can be inferred from correlation but the latter has to address the underlying causal structure of the theory. How can research activity in a field be conducted without the researcher having an understanding of what it means to “explain”? Studies complaining about the lack of “usable” research in operations lead to a fundamental issue: not what research topics need investigation, but what research perspective we should hold. Striven concludes that explanation is “a topically unified communication, the content of which imparts understanding of some scientific phenomenon” (1962, p. 224). A description that does not explain, although research, is incomplete. A prediction based on constructs that cannot be explained, such as a crystal ball, is magic. Hospers (1956) presents three common interpretations of the explanation of a phenomenon. These may be summarized as: (1) Stating the phenomenon’s goal or purpose. Research in operations strategy often gets trapped here. Apart from the limited number of case studies, few researchers have devoted time and effort to disseminate any knowledge about the resultant overall impact when a firm develops and implements an operations strategy. Most arguments here are purposive: A firm should develop an operations strategy because its purpose is sacrosanct. Similar arguments are used for a number of other initiatives like inventory reduction.
Journal
of Operations
Management
303
(2) Showing the phenomenon to be an instance of a familiar phenomenon. This interpretation of explanation shows the subject phenomenon to be an instance of a more familiar phenomenon. Essentially: “ the essence of an explanation consists in reducing a situation to elements with which we are so familiar that we accept them as a matter of course, so that our curiosity rests.” (Bridgeman (1968, P. 37))
An example here is the common reinterpretation of the just-in-time philosophy as simply “reducing waste” (Schonberger (1987)). Eliminating waste through increased flexibility (crosstraining), responsiveness (quick setups and changeovers), and responsibility (zero defect quality) are the cornerstones of this approach. However, as Klein (1989) points out, this reductionist spiral ignores our conventional notions about workers. The entire philosophy hinges primarily on “more and more strictures on workers’ time and action” (Klein (1989, p. 60)). In the process of reducing the phenomenon to a more familiar form (waste reduction), we wind up making certain assumptions about autonomy and worker cooperation that may not conform to reality. (3) Bringing the phenomenon under a law. This last interpretation is that an explanation makes a phenomenon more familiar or removes mystery from it by bringing it under an existing law or making it a new law. The law itself may be unfamiliar, but its consistency with other laws gives us comfort and removes our concern about it. Hence, to explain an event is to “simply bring it under a law, and to explain a law is to bring it under another law.” (Hospers (1956), p. 98). This interpretation reflects the current axiomatic, prescriptive conceptualization of research activity in operations where managerial problems are addressed primarily with mathematical models. Thus, a complex phenomenon is simplified and then “solved” (or explained) by treating it with an algorithmic model. Given the nature of the real-world questions asked of operations, this paradigm severely limits the scope of OM research activity, the usability of the results, and the ability of researchers to understand operations phenomena better. For example, many operations studies are concerned with cost minimization or output maximizationproduction scheduling, capacity planning, process design, and so on. But in an organizational setting, cutting costs may negatively affect operations managers by reducing their budgets, personnel, flexibility, or power. Testing. The final stage in the cyclic process of research (Figure 1) is testing the concepts to determine which are correct, which are false, and how to modify or expand them. The process commonly involves a prediction based on the explanation constructed in the previous stage, and then observation to determine if the prediction was correct. Alternatively, a prediction may be postulated and then checked against observations already made or included in the description. This testing stage often is claimed to be “true research,” perhaps because of the rigor the commonly used tools of statistics or experimentation seemingly lend to this activity. The dominant mode of testing in operations is simulation. While simulation provides for uncertainty in outcomes, very few simulations are actually grounded in reality. Following testing of the concepts, more description follows concerning other or more detailed aspects of the situation, more exploratory investigation is conducted, and new concepts (or modifications) are developed to be tested in turn. Thus, the cycle of learning and research continues.
304
Vol. 8, No. 4
A FRAMEWORK
FOR RESEARCH
PARADIGMS
A research paradigm is a set of methods that all exhibit the same pattern or element in common. However, there are a number of dimensions on which research activity may be classified. For example, it may be classified according to the technique used to gather the data (model, literature, survey, observation, interview, experiment, laboratory, etc.), the methods used to analyze the data (statistics, protocol analysis, taxonomy), the immediate purpose of the research (exploration, description, evaluation, hypothesis generation, hypothesis testing), the nature of the units of analysis (individuals, groups, processes), the duration/time points of data collection, and so forth. Though limited, there have been other frameworks offered for classifying research paradigms. Beged-Dov and Klein (1970), for example, classify research in management science in terms of formalism or empiricism, Reisman (1988) categorizes the range of management and social science research strategies in terms of a Venn diagram-type of framework, and Chase (1980) offers a matrix framework for classifying the research conducted in operations. A more generic and comprehensive framework, similar in that sense to the framework constructed by Mitroff and Mason (1984) for business policy, is presented by Paulin, Coffey, and Spaulding (1982) for the field of entrepreneurship. Here we present a generic framework for a classification of paradigms based on the framework generated by Mitroff and Mason (1984). In discussing the underlying metaphysical assumptions inherent in business policy research, Mitroff and Mason specify two key dimensions that shape the philosophical basis for research activity. We have redefined their two dimensions, illustrated in Figure 2, to better fit operations. The first is the “rutionallexistential dimension” which concerns the nature of truth and whether it is purely logical and independent of man or whether it can only be defined relative to and concerns the source individual experience. The second dimension is “naturallartificial” and kind of information used in the research. The Rational/Existential
Dimension
This dimension relates to the epistemological structure of the research process itself. It involves the benefits and limitations of the philosophical approach taken to generating knowledge; that is, the viewpoint of the researcher. At one extreme is rationalism, which uses a formal structure and pure logic as the ultimate measure of truth. At the other extreme is existentialism, the stance that knowledge is acquired through the human process of interacting with the environment. Thus, in existentialism an individual’s unique capabilities, in concert with the environment, are regarded as the basis of knowledge. The former conforms to the traditional deductive approach to research; the latter to an inductive approach. Our view of the rational/existential dimension includes four generic perspectives that structure the research by different degrees of formalism. These four perspectivesin order of degree of formal structure, are axiomatic, logical positivist/empiricist, interpretive, and critical theory. We explain these briefly, using examples from operations. The axiomatic perspective represents the theorem-proof world of research. A high degree of knowledge is assumed, a priori, about the goals and the socio-technical structure of the organization. The key organizing concepts are the presence’of formal procedures (e.g., lot sizing), consensus, consistency of goals (such as cost minimization), and a work place ideology characterized by scientific management principles. Operations research (OR) studies tend to fall in this category, such as the many variations of the economic order quantity model. Additionally,
Journal
of Operations
Management
305
A GENERIC
FIGURE 2 RESEARCH FRAMEWORK
RATIONAL r
gp; DIRECT , , ARTIFICAL OBSERVATION’ PERCEFTIONS ‘RECONSTR’CTN .9
EXISTENTIAL
Hounshell(1988, p. 61) provides a historical perspective of some basic axioms of manufacturing management. The logical positivist/empiricist perspective assumes that the phenomenon under study can be isolated from the context in which it occurs and that facts or observations are independent of the laws and theories used to explain them. This is the basis for most survey research. For example, Anderson, Schroeder, White, and Tupy (1980) use this perspective to derive conclusions about critical MRP implementation factors. Isolated from the context, one concludes that top management commitment is essential for implementation success. The question that naturally follows is what leads to lack of commitment. Are there competing demands for management commitment? If so, then the phenomenon is much more complex than we have assumed. If not, the results are tautological. “Good” management is essential for implementation success, but “good” is defined by a successful implementation.
306
Vol. 8, No. 4
The interpretive perspective includes the context of the phenomenon as part of the object of study. Interpretive researchers study people rather than objects, with a focus on meanings and interpretations rather than behavior. The purpose is to understand how others construe, conceptualize, and understand events and concepts. In contrast to the implicit absolutism of positivism, interpretivism is relativistic because facts are not considered independent of the theory or the observer. Interpretive researchers explain by placing behaviors in a broader context in which the behaviors make sense. An excellent example of an interpretive study is provided in the context of just-in-time (NT) manufacturing by Klein (1989, p. 66). In a review of the JIT movement she concludes that greater employee responsibility does not mean greater discretion over time and work. She exemplifies it with the attitude of the typical operations manager: “. . . it seemed obvious to him that increased participation was precisely what the workers wanted . . .” Observing that with JIT, tasks are more tightly coupled than ever before, her conclusion is hard hitting: “They ought not to promise workers autonomy when they mean them to deliver an unprecedented degree of cooperation. ” Critical theory is a recent influential contribution to post-positivist thought, primarily through the work of Jurgen Habermas (1979a, 1979b). The critical theory perspective is an attempt to synthesize the positivist and interpretive perspectives and get past their dichotomy by placing knowledge in a broader context of its contribution to social evolution. The positivist and interpretivist perspectives are considered dialectically interrelated. Critical theorists transcend the contradiction between the way people behave in practice and the way they understand themselves to be acting. An example here is the evolution of quality practice in organizations where a symbiotic thrust has emerged between the positivist traditions related to cost of quality and the highly interpretive findings related to quality of work life (Alexander (1988) and Raturi and McCutcheon (1989)). Measures of the dimension. A number of measures, as illustrated in Figure 2, can be placed on this dimension that help clarify the continuum. At the rational pole, the research process: l
l l
l l
l
l
tends to be deductive, is more formally structured, entails a high degree of objectivity, is methodologically prescribed, restricts environmental interaction lest the findings be biased by the researcher’s orientation, requires a priori assumptions concerning primary constructs, establishes the truth of its findings by coherence with the truth of other statements or “laws.”
This contrasts with the existential pole where the process is more inductive, less structured, typically subjective, and requires more interaction with the environment. The process of knowledge creation requires “detective work” and then a “creative leap” (Mintzberg (1983)). Further, researchers at this pole are concerned more about the correspondence of their findings to the real world than their coherence with existing theories or laws. The Natural/Artificial
Dimension
This second dimension concerns the source and kind of information used in the research. At the natural end of the continuum is empiricism (deriving explanation from concrete, objective data), while at the artificial end is subjectivism (deriving explanation from interpretation and artificial reconstruction of reality). The progression from natural to artificial on this dimension
Journal
of Operations
Management
307
parallels the historical periods presented in the Appendix. The researcher’s perception of reality is molded by the mechanisms used to study the phenomenon. In a very broad sense, these mechanisms may be classified into three categories: object reality, people’s perceptions of object reality, and artijcial reconstruction of object reality. Object reality refers to direct observation by the researcher of the phenomenon. It assumes that there is an objective reality and that human senses can detect it. It corresponds to the pure empiricism extremum exemplified by Locke. As with the other categories, the observation may be subjected to formal structured analysis (or axiomatization, as in econometric studies) or to interpretation using critical theory. People’s perceptions of object reality relate to research conducted “through somebody else’s eyes, ’ ’ as in surveys, interviews, or many laboratory experiments. Thus, the primary concern is with the perception or abstract representation of the reality of individuals exposed to the phenomenon. These are second source methods, but may be the only efficient or effective way to obtain information about the phenomenon of interest. A number of constructs in operations, like the effect of layout on the productivity of an operation or the success of a new piece of equipment, are difficult to study through direct observation. The opportunity may not be there, or the results may be clouded by the Hawthorne effect. In such situations, an assessment of people’s perceptions may yield significant insights into the underlying explanation of the phenomenon. Descriptive information about the phenomenon, as well as people’s constructs/models about what relationships are operative, can be ascertained through these second source methods. An arti$cial reconstruction of object reality is attempted in almost all the modeling and systems analytic efforts in operations. These approaches recast the object reality, as originally determined from one of the above two categories (usually the researcher’s own belief concerning the object reality), into another form that is more appropriate for testing and experimentation, such as analytical models, computer simulations, or information constructs. Measures of the dimension. As with the rational/existential dimension, there are a number of measures that describe this dimension, as shown in Figure 2. At the artificial pole, the research: l
l
l
l
l
l
uses highly abstracted and simplified models such as linear representations; tends to yield conclusions with high reliability and internal consistency; is often characterized by a significant separation of the phenomenon from the researcher, as with an abstract representation; is highly controlled since the researcher uses a priori constructs or models to specify the information to be collected; process is highly efficient since aberrations (classified as “noise”) do not have any causal source; is dated, since the specification of the constructs or models takes most of the researcher’s time and pushes the natural phenomenon further into the past.
This contrasts with the natural pole where the research process is more directly concerned with the real phenomenon, less concerned with reliability and more with externally generalizable validity, closer to reality, less controllable, less efficient, and more current. The critical issue here is the balance between reliability and external validity. Like IQ tests, survey instruments provide very reliable data but their validity in actually measuring constructs is suspect. Clearly, the most valid information is that obtained by direct involvement with the phenomenon. This section has proposed a new research framework and illustrated a broad range of
308
Vol. 8, No. 4
paradigms available to researchers. Current research in operations has tended to lie in the rational-artificial quadrant and thereby has limited not only the phenomena that can be researched effectively but also the utility of the findings. In the next section, we describe a number of research methodologies that fall across the quadrants of the framework and discuss their potential application to research in operations. PARADIGMS
OF RESEARCH
METHODS
Figure 3 presents the two dimensions we established in the last section, with the methodologies available to researchers placed in their appropriate cell(s). Note that some methodologies logically could fall into a number of cells, or relate to only one of the dimensions. Also some methodologies can fairly easily be used for any of the three stages of researchdescription, explanation, or testing-and these are occasionally pointed out in passing. For example, case studies can describe, explainor disprove a hypothesis. The methods listed in this figure are described briefly below. Methodological references are FIGURE 3 FOR RESEARCH
A FRAMEWORK NATURAL
RATIONAL A
METHODS > ARTIFICIAL
ARTIFICIAL
< PEOPLE’S PERCEPTIONS OF OBJECT REALITY
DIRECT OBSERVATION OF OBJECT REALITY
RATIONAL
publications
ARTIFICIAL RECONSTRUCTION OF OBJECT REALITY
A MS DS JOM
AXIOMATIC
1977
1987
60% 82% -
70% 58% 33%
TOTAL
MS DS JOM
LOGICAL POSITIVIST/ EMPIRICIST
1977
1987
-
4% -
TOTAL
MS DS JOM
INTERPRETIVE
1977
1987
8%
7% 7%
TOTAL
CRITICAL
1977 MS DS JOM
5%
1977
1987
28% 18% -
15% 42% 47%
TOTAL
1% 1987
4% TOTAL
MS DS JOM
I977 MS DS JOM
-
28% 1987 4%
13% TOTAL
1%
62%
3%
THEORY
V
EXISTENTIAL SAMPLE SIZE BY JOURNAL MANAGEMENT SCIENCE DECISION SCIENCES JOURNAL OF OPERATIONS
MANAGEMENT TOTAL
MS DS JOM
1977
1987
25 17
26 12 15
95
The inescapable conclusion is that our research in operations is still overwhelmingly artificial in nature, though breaking the methodological tie with the field of management science has allowed us to begin moving toward more existential (primarily interpretive) paradigms and to move away from the more rationalistic, “scientific” paradigms (both axiomatic and logical positivist/empiricist). We believe that a much stronger movement toward naturalistic paradigms (especially direct observation via case, action, and field studies) and existential (primarily
Journal
of Operations
Management
317
interpretive) paradigms is called for. The methods are accessible, their legitimacy is proven, and the need is great. In general, the newer, more interrelated, more situation- or people-dependent topics in operations require the additional perspective afforded through the natural and existential methodologies. As an example, we describe the possible application of these methodologies to two major topics in operations: quality management and technology implementation. In so doing, we only sketch the major outlines of the potential application, leaving the methodological details to the creativity of the research readers. Potential Research in Quality
Management
One rich area within operations for discussing the application of the different paradigmatic research stances is quality management. There are two primary reasons. The first is that quality management recently has become a primary concern and even strategic thrust of many business organizations. Secondly, the concept of the quality of a product or a service is multi-dimensional and, to some extent, nebulous in definition. The accepted definition concerns “fitness for use”; yet evolving social values and perceptions continually redefine the interpretation of this phrase. Much difficulty in conducting research in quality is due to how products and, especially, services are perceived. Abstract, subjective, and commercial characteristics (e.g., a restaurant’s “atmosphere” or their staffs “courtesy”) can only be measured through people’s perceptions. Moreover, each person may define these characteristics differently. Because the basic definition of quality involves the degree of consumer satisfaction with a product or service, people’s perceptions of the major components of quality provide a basis for quality assessment. The notion of quality is created by a combination of such diverse attributes and thus, diverse research methods may also be required (Alexander (1988)). We first consider the application of the natural-artificial dimension of our framework for research in quality management. The three major paradigms include direct observation, determining people’s perceptions, and artificially reconstructing object reality. Quality characteristics presumed inherent in the product or service (defects, durability) are, by and large, directly observable. Field experiments and field studies are appropriate for analyzing these characteristics of a product because the variables are clear and potentially subject to control. For the more contextually-defined or situation-dependent quality characteristics (such as aesthetic appeal), case studies and action research are required because of their deeper analysis of context. When assessing perceptions, methods such as surveys and structured interviews provide an empirical approach to studying quality. But significant progress in this area may require more interpretive research. Here, intensive interviewing, expert panels, scenario analysis, and even introspective reflection could be valuable for uncovering basic attitudes, perhaps even subconscious feelings, about what constitutes quality. Reflection offers the possibility of merging the more rational/empirical quality concepts with the interpretive findings so as to move to a higher level of understanding of the concept of quality. Most of the existing research pertaining to quality has been conducted at the artificial end of the scale through the methods of modeling, laboratory experimentation, and simulation. It might be noted here that other “direct, observable” characteristics such as reliability and maintainability are, in practice, typically measured through artificial reconstructions of the object reality. For example, a system reliability of 0.99 either is a statistical estimate or a guess based on the hypothetical reconstruction of past experience. However, conceptual modeling for the purpose of interpretation has been rare.
318
Vol. 8, No. 4
On the rational-existential dimension, the axiomatic perspective historically has been most common in quality control research. Statistical laws are used to make judgments about the quality of products/processes/services and quality planning efforts are directed toward reducing the overall cost of quality. But is cost the real driver? The trade-offs implicit in this analysis are becoming suspect, as the Japanese approach to quality has indicated (Hayes and Abernathy ( 1980)). Logical positivist/empiricist notions are implicit in research using surveys that relate quality to product/firm success, and also in experimental work related to the efficacy of small groups (quality circles) in quality improvement programs. In the more artificial vein, simulation of processes and systems have been conducted to identify improved quality management strategies. Similarly, Fine and Bridge (1984) address issues concerning the explicit and implicit cost and direct measures of quality. Interpretive efforts in the area of quality might address the interactions of organizational functions in the provision of quality products and services, including the place of organizational politics, leadership, consensus-building, and so on. A major issue here is the strategic role of quality in the firm’s competitive strategy and how quality is included in the formulation of the firm’s policies. Critical theory research could address issues such as the firm’s reward and motivation systems’ effects on quality. Issues of safety, health, ethics, and social benefit as related to quality and how it is delivered also could be researched here. Given the highly individualistic nature of interpretations of quality, other research topics might include the differing perceptions of quality within the firm, the relative importance of its different facets, and the organizational policies that influence them.
Potential Research in Technology
Implementation
The same kind of arguments might be made for the topic of technology implementation. We will only briefly sketch some of these points here. Research on the implementation of new technologies is an area that lends itself particularly well to the use of the interpretive paradigm. Previous operations studies on implementation have been mainly in the positivist/empiricist mode and employed surveys and field studies (e.g., see Ettlie (1988) Nutt (1986), White et al. (1982), Leonard-Barton and Kraus (1985)). Rogers (1983) and Voss (1988) argue that the study of implementation must spread over the life span of the implementation process. Action research, where the researcher is involved with other parties in the analysis, planning, and execution of the implementation process, could help significantly in improving OM researchers’ understanding of the implementation process (Warmington (1980)). And case studies, particularly longitudinal, would allow detailed exploration of unfolding processes in real time and avoid the frequently distorted responses of surveys and structured interviews. And since the implementation of new technology usually spans some years and involves many participants, historical analysis could be used to capture the context of the social, political, economic, and cultural issues that invariably come into play during such an extended process. This might even allow the development of a framework of recurring patterns in implementation processes. And also because of the extended time span of the implementation, Delphi and scenario analyses would be feasible and possibly desirable to anticipate normally unexpected developments.
Journal
of Operations
Management
319
CONCLUSIONS Research in operations has employed a limited set of paradigms for too long. Heavy on the artificial and rational ends of the scales and light on the object reality and existential/interpretive ends, research in operations has typically exhibited high reliability and internal validity but almost no external validity. We emphasize objectivity in research and thus stress predictive power, but with little understanding of the phenomenon. It is time to expand our limited set of worn-out paradigms and consider new research methods from paradigms used in our sister fields. This broadening of our perspective would contribute substantially to addressing the research problems we face. If we do not expand our approaches to research, managers will continue to perceive us as irrelevant academics who address fictitious problems and are not interested in the real world. To make true contributions to both research and practice, we must enlarge our repertoire of methodologies and apply those that are most appropriate, efficient, and effective for the situations at hand.
ENDNOTES ?Throughout this paper we follow the example of the other functional resources and drop the unnecessary “management.”
business fields of finance, marketing,
and human
REFERENCES
1. Abernathy, 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.
320
W.J., and J.E. Corcoran. “Relearning from the Old Masters: Lessons of the American System of Manufacturing.” Journal of Operations Management. vol. 3, no. 4, August 1983, 155-167. Ackoff, R. “The Future of Operational Research is Past.” Journal of the Operational Research Society, vol. 30, no. 2, 1979, 93-104. Agar, M.H. Speaking of Ethnography. Sage University Paper Series on Qualitative Research Methods, vol. 2. Beverly Hills, CA: Sage Publications, 1986. Alexander, C.P “Quality’s Third Dimension.” Quality Progress. July 1988, 21-23. Amoako-Gyampah, K., and JR. Meredith. “The Operations Management Research Agenda: An Update.” Journal of Operations Management, vol. 8, no. 3, 1989, 250-262. Anderson, J.C., R.G. Schroeder, E.M. White, and SE. Tupy. “MRP: The State of the Art.” APICS Monograph. Falls Church, VA: American Production and Inventory Control Society, Inc., Fall 1980. Anderson, J.C., N.L. Chervanay, and R. Narasimhan. “1s Implementation Research Relevant for the OR/MS Practitioner?” Inferfaces, vol. 9, no. 3, May 1979, 52-56. Andrew, C.G., and GA. Johnson. “The Crucial Importance of Production/Operations Management.” Academy of Managemenr Review, vol. 7, no. 1, 1982, 43-147. Antill, L. “Selection of a Research Method.” In Research Methods for Information Systems, E. Mumford and R. Hirschheim (eds.) Amsterdam: North-Holland, 1985. Argyris, C., R. Putnam, and D.M. Smith. Action Science: Concepts, Methods, and Skills for Research and Intervention. San Francisco: Jossey-Bass, 1985. Babbie, E.R. Survey Research Methods. Belmont, CA: Wadsworth, 1973. Basu, A., and R.G. Schroeder. “Incorporating Judgments in Sales Forecasts: Application of the Delphi Method at American Hoist & Derrick.” Interfaces, vol. 7, no. 3, May 1977, 8-27. Beged-dov, A.G., and T.A. Klein. “Research Methodology in the Management Sciences: Formalism or Empiricism.” Operational Research Quarterly, vol. 21, no. 3, 1970, 31 l-326. Benbasat, I., D.K. Goldstein, and M. Mead. “The Case Research Strategy in Studies of Information Systems.” MIS Quarterly, vol. 11, no. 3, Sept., 1987, 369-386. Systems of the Future-A Delphi Study.” Benson, PG., A.V! Hill, and T.R. Hoffmann. “Manufacturing Production and fnventoy Management, vol. 23, no. 3, Third Quarter 1982, 87-106. Bonoma, T.V. “Case Research in Marketing: Opportunities, Problems, and a Process.” Journal qf Marketing Research, vol. 22, no. 2, May 1985, 99-208.
Vol. 8, No. 4
17. Bourgeois, L.J., III, and K.M. Eisenhardt. “Strategic Decision Processes in High Velocity Environments: Four Cases in the Microcomputer Industry.” Management Science, vol. 34, no. 7, July 1988. 816-835. 18. Bridgeman, PW. The Logic of Modern Physics. New York: Macmillan, 1968. 19. Buffa, E.S. “Research in Operations Management.” Journal qfUperations Managemenr. vol. I, no. I, 1980. l-8. 20. Buffa, E.S. Operations Management: Problems and Models. 2nd ed. New York: Wiley, 1968. 21. Buffa, E.S. Modern Production Management. New York: Wiley, 1965. In Research in Production/Operations Management, 22. Buxey, G. “Research Needs in Production Management.” C.A. Voss. Brookfield. VT: Gower, 1984. 23. Chase, R.B. “A Classification and Evaluation of Research in Operations Management.” Journal of Operations Management, vol. 1, no. 1, 1980, 9-14. 24. Chase, R.B., G.B. Northcraft, and G. Wolf. “Designing High-Contact Service Systems: Application to Branches of a Savings and Loan.” Decision Sciences. vol. 15, no. 4, Fall 1984, 542-555. Journal qf Management, vol. 25. Chase, R.B., and E.L. Prentis. “Operations Management: A Field Rediscovered.” 13, no. 2, 1987, 351-366. 26. Choi, R.H., and E.M. Malstrom. “Evaluation of Traditional Work Scheduling Rules in a FMS Using a Physical Simulator.” Journal of Manufacturing Systems, vol. 7, no. I, 1988, 33-45. 27. Davis. M. “That’s Interesting!” Phil. Sot. Sci.. vol. 1, 1971, 309-344. 28. Denzin, N.K. The Research Act. 2nd ed. New York: McGraw-Hill, 1978. 29. Dexter, L.A. Elite and Specialized Interviewing. Evanston, IL: Northwestern University Press, 1970. 30. Donham, W.B. “Essential Groundwork for a Broad Executive Theory.” Harvard Business Review, vol. 1, no. 1, Oct. 1922, I-10. 3 I. Dubin, R. Theor?/ Building. New York: The Free Press, 1969. 32. Emory, W.C. Business Research Methods. 3rd ed. Homewood, IL: R.D. Irwin. 1985. 33. Ettlie, J. Taking Charge of Manufacturing. San Francisco: Jossey-Bass, 1988. 34. Ferdows, K., J.G. Miller, J. Nakane, and T.E. Vollmann. “Evolving Global Manufacturing Strategies: Projections into the 1990s.” International Journal of Operations and Production Management, vol. 7, no. 4, 1987, 6-16. M.I.T. Working Paper #1607-84. Boston: 35. Fine, C.H., and D.H. Bridge. “Managing Quality Improvement.” Massachusetts Institute of Technology, 1984. 36. Freiman, J.M., and B.O. Saxberg. “Impact of Quality Circles on Productivity and Quality: Research Limitation of a Field Experiment.” IEEE Transactions on Engineering Management, ~01.36, no. 2, May 1989, 14-118. 37. Friedman, M. “Explanation and Scientific Understanding.” Journal of Philosophy, vol. LXXI, no. I, Jan. 17, 1974, 5-19. 38. Galbraith, J.K. The Afluent Society. Boston: Houghton Mifflin, 1958. 39. Galliers, R.D. “In Search of a Paradigm for Information Systems Research.” In Research Methodsfor Information Systems, E. Mumford and R. Hirschheim (eds.) Amsterdam: North-Holland. 1985. 40. Galliers, R.D., and FE Land. “Choosing Appropriate Information Systems Research Methodologies.” Communications of the ACM, vol. 30, no. Il. Nov. 1987, 900-902. 41. Gerwin, D. “A Theory of Innovation Processes for Computer-Aided Manufacturing Technology.” IEEE Transactions on Engineering Management, vol. 35, no. 2, May 1988, 90-100. 42. Goodman, PS., and S. Garber. “Absenteeism and Accidents in a Dangerous Environment: Empirical Analysis of Underground Coal Mines.” Journal of Applied Psychol0g.x vol. 73, no. I, 1988. 81-86. 43. Gordon, R.A., and J.E. Howell. Higher Eduction for Business. New York: Columbia University Press, 1959. 44. Graham, M.B.W., and S.R. Rosenthal. Institutional Aspects of Process Procurement ,for Flexible Machining Systems. Research report. Boston: Boston University, September 1986. 45. Grayson. C.J. “Management Science and Business Practices.” Harvard Business Revif)
