Introduction

It is time that we in organization sciences develop and implement a new mental model for cause and effect relationships. The dominant model in research dates at least to the 1700s and no longer serves the full purposes of the social science research problems of the 21^st century. I am not the first to make this suggestion, but by focusing an article specifically on this subject, I hope to draw more attention to the question. This article will briefly touch on the foundations of the traditional notion of cause and effect, provide examples from the literature of researchers and theorists who have found this traditional notion ineffective for their study, discuss some alternative framings of the cause/effect relationship, define mutual causality, introduce Maturana’s work on ontology and epistemology, offer an alternative model to replace the traditional cause and effect construct, demonstrate the benefits of this model for two existing research studies, and critique the interactional model.

The Classic Model of Cause and Effect

The notion of causation has always been controversial in philosophy. David Hume, in 1748, was perhaps the first person to lay out clearly the requirements for causation in the classic model (Singleton, Straits, and Straits, 1993). His three requirements of association, direction of influence, and nonspuriousness have been elaborated to the following four conditions to state that X causes Y.

1. Temporal order - changes in X must precede changes in Y.

2. Associative variation - changes in X must be associated with changes in Y.

3. Nonspurious association (absence of other causes) - There must not be a factor Z, which if introduced, explains the association between X and Y.

4. Theoretical support - if X causes Y is consistent with the theory proposed, and the theory has explained other phenomena, then the theory provides support for the assertion that X causes Y.

Item 2 is necessary, but not sufficient for causation. It is more useful for suggesting that causation is not in effect. The absence of association between X and Y under certain conditions is very strong evidence that X and Y are not causally related. Item 3 has a further complication. Cause-effect relationships usually occur in chains, not simple pairings. Given that X causes Y which causes Z, many people try to eliminate Y if they want to avoid Z. However, it is frequently the case that even if Y is eliminated, X will continue to produce Y and Z will be recreated (Ackoff, 1981, p. 192). This situation occurs when someone treats the symptoms rather than the root of a problem. As long ago as 1948, Russell wrote that even though this elemental form of causality still appears in books (as it does in Singleton, Straits, and Straits (1993) among others), it never takes this form in any advanced science (Russell, 1948, p. 315). Russell provides a number of qualifiers, for example, omitting the environment as a factor which must be present, in order for such a direct cause and effect relationship to be asserted.

Concerning Item 1 above, Bunge (1959) reviewed the common formulations of causality and found that none seems to require the temporal precedence of cause prior to effect. Bunge concludes that the principle of antecedence and the causal principle are independent of each other. This independence suggests that cause and effect can occur simultaneously. Russell (1948, p. 309) draws the same conclusion. However, the requirement of the temporal precedence of cause and effect was so deeply ingrained within Carl Jung, for example, that when he noted many simultaneous events that he felt were pivotal to science and therapy, he coined a new term, synchronistic, to avoid disrupting accepted notions of cause and effect (Slife and Williams, 1995, p. 102-103).

Another reason supporting the linear cause and effect assumption has been well-discussed by Karl Weick (1985, 1995). Weick suggests that “when people look back at prior events once they know the outcomes of those events, they ‘see’ an orderliness and inevitability that suggests that the events unfolded in a rational manner” (1985, p. 112). Weick also notes that people know relatively little about how they got things done and that people “misremember the process of accomplishment” (1985, p. 132).   So, within “a complex prior history of tangled, indeterminate events” (Weick, 1995, p. 28), many people will extract a meaning which suggests a number of direct, linear cause and effect relationships.

The literature is replete with examples of phenomena in which the traditional cause and effect construct does not allow for greater understanding and insight into the phenomena (Begun, 1994). Several examples, chosen for the wide timeframe they cover and the wide range of phenomena, are offered here.   Perhaps one of the earliest is Mary Parker Follett’s (1924) assertion that “we shall never catch the stimulus stimulating or the response responding” (p. 60). Other researchers have suggested that personality and environment vary as functions of each other (Dill, 1958); that it is difficult to distinguish empirically between power and influence (Rossi, 1958); that evaluation and empirical analysis are intertwined (Lindblom, 1959); that language and thought are highly dependent on each other (Engel, 1980); that meaning and community are co-constructive (Drath and Palus, 1994); that individual and social activity are not separable (Weick, 1995); that the arrow points both ways between individual and organizational learning (Argyris and Schon, 1996); and, that empowerment is a mutual interaction of leader and follower (Vaill, 1998, p. 128).

Three of the above examples are briefly elaborated to show the richness of these cases. At the time Lindblom (1959) wrote (and perhaps among many policymakers today) the conventional wisdom was that Lindblom’s audience of public administrators first determined the objectives of the public and then selected or created policy to achieve those objectives. Lindblom argues that “one simultaneously chooses a policy to attain certain objectives and chooses the objectives themselves” (p. 82). In other words, “one chooses among values and among policies at one and the same time” (p. 82). Evaluation and empirical analysis are so intertwined that their relationship is not illuminated by the traditional cause and effect construct.

Drath and Palus (1994) offer a new way of viewing leadership, as meaning-making in a community of practice. Their leadership definition rests upon the assumption that meaning and community are co-constructive (p. 13). This suggests that leadership is something people do together, with individual leadership only a special case of one individual helpfully framing what has been unknown or not articulated among the relevant community. Drath and Palus (1994) point out that the work of Escher often evokes mutual causality. In one well-known work, a right hand is drawing the left hand at the same time the left hand is drawing the right hand. These authors suggest that in the same manner, “meaning constructs community which constructs meaning” (p. 11). It is impossible to isolate one or the other as cause or effect.

A final, most recent example comes from Argyris and Schon (1996) who write about organizational learning and how it relates to individual learning. Their own words are clear and concise and nicely capture the mutually causal relationship they see between the two forms of learning.

We recognize the complex interactions that occur between individuals and organizational learning. We see the causal arrow pointing in both directions: The learning of individuals who interact with one another is essential to organizational learning, which feeds back to influence learning at the individual level (p. xxii).

Research inadequacies and policy outcomes such as these, together with significant earlier critiques of the linear causal model (Brand, 1976; Giddens, 1976; Rakover, 1990; Caws, 1993 and many others), have led to improvement efforts in the understanding of cause and effect relationships. In this section I will introduce alternative framings suggested by Ackoff (1981), Argyris and Schon (1996), and Schwartz and Ogilvy (1979). Von Bertalanffy (1968) notes that there are a wide field of relationships in which unidirectional causality and models with few variables seem appropriate. However, such appropriate settings are not commonly found in biological, behavioral, and social sciences (Begun, 1994). Yet, the unidirectional model is predominant in the behavioral sciences (Slife and Williams, 1995, p. 101).

Producer-product

Ackoff (1981, p. 20) writes about a distinction between producer-product and cause-effect. Producer-product is terminology introduced by the American philosopher E. A. Singer, Jr. A producer-product relationship exists when X is necessary, but not sufficient to cause Y. Singer uses the example of an acorn and an oak tree. An acorn is necessary to cause an oak tree, but if it is not placed in a suitable environment, the acorn will not grow into an oak. In producer-product relationships, the producer alone cannot be the cause of the product. There are always other necessary conditions. In the case of the acorn, the necessary conditions are the sunlight, soil, and other environmental conditions.

From the view of producer-product, the environment becomes central to understanding and explanation. Ackoff (1981) notes that "the use of the producer-product relationship requires the environment to explain everything whereas use of cause-effect requires the environment to explain nothing. Science based on the producer-product relationship is environment-full, not environment-free" (p. 21). Consequently, by definition, any principle offered about producer-product relationships must stipulate the conditions under which the principle applies. If the principle were to apply in all conditions, then the environmental conditions are not co-producers of the effect.

For many years, Chris Argyris has been identifying several phenomena which are related to each other in ways not best described by traditional cause and effect (Argyris, 1957, 1958). Argyris and Schon (1996) make arguments similar to others presented here, with some additional nuances. The authors begin by noting that the purpose of the traditional conception of cause and effect (which they refer to as the normal-social-science model of causality) is to establish “covering laws” for relating variables X and Y, so that given the values of X and the knowledge that X has occurred, Y will result, independent of any other features of the contexts in which X and Y occur. This model centers on the definition of a variable, “a named attribute extracted from the complexity of observed phenomena which is treated as essentially the same in whatever local context it occurs” (p. 38). This conception stands on the assumption of the constancy of the definition of a variable. Several researchers dating back to Follett (1924) have found this assumption not to hold with many organizational phenomena. Follett (1924), for example, suggests that in a presumed causal interaction between two people, that the nature of the “variable” changes (not just the value of the variable) with every interaction between the people.

Argyris and Schon (1996) continue by differentiating between design causality and efficient causality. Design causality is defined as “the causal relation that connects an actor’s intention to the action he or she designs in order to realize that intention” (p. 39). Efficient causality is “the causal connection between an act and its consequences, intended or unintended” (p. 39). Argyris and Schon (1996) intimate that a greater understanding of organizational phenomena will be achieved by focusing on design causality rather than efficient causality. The latter has relatively little utility in practice because of the nature of covering laws which tend to be “precise, quantitative, probabilistic, abstract, and complex” (p. 41). Design causality is environment-ful in Ackoff’s sense of the term, including all of the subjective, idiosyncratic, qualitative and other factors of the content and context.

Four Classes of Causal Models

Another view of causation is suggested by Schwartz and Ogilvy (1979) who view causality in terms of four classes of causal models, only one of which closely matches the classical definition. Their first class is direct, linear causality. Probabilities are introduced as a factor in the second class. Cybernetics launched the third class of models which "permit feedback from effects to causes; however, the primary focus is on negative [compensating] feedback" (p. 56).

The icon of the third class is homeostasis, "a dynamic process of regulation in which a set of essential variables regulate each other via the mechanism of a complex set of interlocking feedback loops" (Clemson, 1984, p. 237). The third class is also represented by the second circular causality principle: "given negative [compensating] feedback (i.e., a two-part system in which each part tends to offset any change in the other), the equilibrial state is invariant over a wide range of initial conditions" (Clemson, 1984, p. 201).

The fourth class is called complex, mutually causal models and is defined as those relationships which "evolve and change together in such a way (with feedback and feedforward) as to make the distinction between cause and effect meaningless" (Guba, 1985, p. 88). This class incorporates and often focuses on amplifying feedback which is discussed in circular causality principle one:

"given positive feedback (i.e., a two-part system in which each stimulates any initial change in the other), radically different end states are possible from the same initial conditions" (Clemson, 1984, p. 201).

Mutual Causality

Certainly there are phenomena which are adequately explained in each of the categories of Schwartz and Ogilvy’s (1979) classification. However, increasingly, researchers of today are discovering that the problems of interest are not as easily explained by the first three classes. Part of the argument here uses the same rationale as Lindblom (1959) who decried that researchers were focusing on and perfecting policy formulation in situations with clear objectives, explicit evaluation, a high degree of comprehensiveness of overview, and a quantification of values for mathematical analysis. Perfection was being achieved, but only in theoretical situations which had no real world counterparts. Lindblom noted that researchers had a common tendency to describe complex problems as though these theoretical conditions held. Today, the same is true of direct cause and effect relationships. Very few issues being addressed by organizational scientists meet the conditions of direct cause and effect. Yet, organizational scientists are imposing that formulation upon phenomena as if the conditions did prevail.

The definition of mutual causality above is carefully chosen because the term has different meanings in the literature. In fact, perhaps the first paper to use these terms in its title (Maruyama (1968: 1963)) isn’t really about mutual causality as defined here. The primary feature of that article is amplifying feedback loops. Maruyama’s implicit definition of mutual causality does allow for simultaneous “cause” and “effect” relationships, but nearly all of the article’s examples are of the alternating variety. Maruyama emphasizes the importance of the “initial kick,” a term which implies discrete steps. In this paper, mutually causal relationships are those in which it is impossible to clearly identify discrete steps in a causal chain. This condition, however, violates the traditional requirements for a cause and effect relationship, that “events are bounded into causal chains by two relations: spatio-temporal continuity and statistical relevance. Explanation requires the exhibition of such chains” (Kitcher, 1991, p. 321).

Perhaps another useful way to think about mutual causality is that it is the special case of circular causality when the time lag is effectively reduced to zero. When temporal precedence can no longer be determined, cause and effect become jumbled in such a way that traditional research techniques fail to offer insights into the nature of the relationship between two or more phenomena.

Maturana’s Ontology and Epistemology

One of the reasons for the great limitations of the traditional cause and effect model is that it is predicated on an epistemology which only narrowly illuminates organizational phenomena of today. What is needed is a shift in worldview (Dent, 1997) away from the exclusive dominance of philosophical assumptions such as objectivity, reductionism, and rationality. Maturana’s work on a new ontology and epistemology holds great promise for a quantum leap in understanding of organizational phenomena. The reader is referred to Maturana (1998) and Maturana and Varela (1987) for a more complete explanation of his ontology and epistemology.

Maturana begins by strongly suggesting that the assumption of an objective world is not viable. He offers a simple example to demonstrate the inadequacy of objective reality. People cannot distinguish among an illusion, a hallucination, or a perception. All three are experienced identically by human beings. The distinction among these three is socially made “through the use of a different experience as a metaexperiential authoritative criterion of distinction, either of the same observer or of somebody else subject to similar restrictions” (Maturana, 1998, 5:1).

Rather than an objective world, Maturana and Varela (1987) advocate that for humans, reality is subjective because perception is interrelated with language to such a degree that “languaging [is] the act of knowing, in the behavioral coordination which is language, [which] brings forth a world” (p. 234). “Objects are in the process of languaging, consensual coordinations of actions that operate as tokens for the consensual coordinations of actions that they coordinate. Objects do not pre-exist language” (Maturana, 1998, p. 8:4). Dell (1985) has suggested that a major contribution of Maturana’s work is the realization that his two major research questions, (1) what takes place in the phenomenon of perception, and (2) what is the organization of the living, are one and the same question. Cognition and the process of living are the same biological phenomenon.

Maturana’s assertion poses a tremendous challenge to decades of organizational research. If reality is not objective and cognition is a biological phenomenon, then control is ontologically impossible (Dell, 1985, p. 11) and so is the traditional notion of cause and effect. What then do we make of our everyday experience in which many cause-effect situations seem obvious and vivid? An answer lies in distinguishing between traditional cause and effect - what Maturana and Varela (1987) refer to as “instructive interaction” - and structural coupling. In situations of instructive interaction, for example, a manager can completely determine the behavior of a subordinate. The subordinate is powerless to act in any manner different from the manager’s desire. Most observers will testify that a given leadership style will produce different responses in different subordinates or that a given form of communication will be interpreted differently by different people. “So-called ‘information’ does not and cannot instruct the behavior of a living system” (Dell, 1985, p. 6). Information may trigger or select a response but the nature of the response is primarily determined by the structure.

Structural couplings are interactions between and among structurally determined composite unities and/or their mediums (Maturana, 1998, p. 6:3). For example, a manager and a work group are structurally coupled if the manager’s interactions contribute to effective performance of a work group. Maturana asserts that the world is structure determined, which is the “understanding that the operation of every living system, both in its internal dynamics and in its relational dynamics, depends on its structure” (Ruiz, 1996, p. 286). For example, a manager employing an authoritarian leadership style may find that some employees do as the manager wants, but that others overtly or subversively defy the authoritarianism. Maturana’s point is that the employees’ reaction depends more on each person’s structure - beliefs, reasoning, values, experiences, and so forth - than it does on the leader’s action.

Maturana and Varela (1987, pps. 96-97) offer other examples as evidence that structural determinism is operative in the daily lives of people. If people get sick, they go to a doctor for a change in their “structure.” When cars don’t function as expected, people take them to repair shops rather than looking to themselves, as operators of the cars, for the necessary change. An external agent can interact with a composite unity in such a way that a change is triggered in the composite unity, but the resulting change is determined by the composite unity, not the external agent. A composite unity’s structure includes all of the potential changes it can undergo. Moreover, the structure is altered with every interaction. Dell (1985) implies that people can have the psychological experience of causality by effectively coupling themselves to other composite unities or mediums. Such attempts to fit themselves to the situation can result in desired or predicted outcomes, but the outcomes are never guaranteed as in instructive interaction.

Bateson (1972) and others have characterized the foundation of the social sciences as hopelessly confused. A dichotomy is clear between those who push the social sciences more in the direction of the natural sciences and those who believe that the social sciences require a foundation fundamentally different from the natural sciences. Dell (1985) sees Maturana’s (and Bateson’s) work as offering a powerful hypothesis:

We are structure-determined autopoietic unities who operate in structural coupling with our medium. In turn, this hypothesis has proved able to generate (and, thereby, to explain) (a) the relation between the organism and its medium; (b) the nature of the structural coupling of organisms to one another; (c) the nature of social systems; (d) the manner in which language arises; (e) the nature of language; (f) the nature of the observer; (g) the manner in which we, as observers, operate in language, make distinctions, and call forth realities; and thereby, (h) how Maturana himself, as such an observer, has been able to advance the very generative hypothesis which specifies all of the above, including his own functioning as a human being who makes such distinctions and advances such hypotheses (p. 17).

Such a hypothesis provides a better foundation for the social sciences by clarifying the nature of the observer, the nature of cognition, and by contributing a coherent ontology and epistemology.

How then can we operationalize cause and effect dynamics using this different philosophical foundation? Before offering a reconception of the traditional cause and effect model, it is instructive to critique briefly both the theoretical construct of traditional cause and effect and the way in which it routinely gets practiced in the field of organizational science.

One current problem is that researchers analyze an organizational dynamic, sometimes under contrived conditions. They find a correlation and assert causality because if the correlation is between, for example, regulatory conditions and foundings, they assume that foundings couldn’t possibly have an effect on regulatory conditions (often a faulty assumption. See several examples in organizational behavior - Lowin and Craig, 1968; Gombrich, 1960; Hirschman and Lindblom, 1962; Bem, 1967). In the best case, assume causality has been asserted when an 80% correlation was found under a specific set of circumstances. However, little or no sensitivity analysis is done on changing any of the conditions. Chaos theory and other literatures (Epstein and Axtell, 1996) have shown that a change in even one condition, if the situation being studied is in a far-from-equilbrium state, as organizational phenomena frequently are (Vaill, 1996) can result in dramatically different results.  

An Alternative Model of Cause and Effect

In organizational phenomena, cause and effect is often asserted in such a way that person(s) A is taking action(s) X which affects person(s) B in such a way that she produces action Y(s). In the shorthand that is adopted, X and Y are often emphasized. For example, leadership style affects motivation, hotel characteristics affect hotel location, or a strategic planning process affects strategy. The analysis, conclusions, and future research suggested are almost always limited to a discussion of X’s impact on Y.

I propose a model which has interaction as the unit of analysis and includes X, Y, A, and B, as well as E (the embeddedness or context of all this) and T (interaction over time). Maturana seems to allude to the greatest knowledge coming from understanding the interaction of X and B, the latter of which plays the primary role in determining B’s action (which is Y). The resulting interaction of X and Y could conform with any of the classes of Schwartz and Ogilvy’s (1979) categorization. Although Maturana and others assert the ontological impossibility of Schwartz and Ogilvy’s first class, I will leave open the possibility that if the interactions of X, A, B, E, and T are carefully specified, that for all practical purposes, a direct, linear causal explanation may be sufficiently useful, in some small number of situations.

The interaction model proposed here is depicted in Figure 1. I hesitate in including a figure because it is necessarily static and simplistic and such representations reinforce the linear mindset I am criticizing here. Figure 1 includes the example of a leader wanting to increase the amount of highly productive work by initiating a program of overtime hours for two months.

What type of information should be provided about A and B? To the extent that the topic of specifying this information is addressed today it is under the heading of generalizability, a subject practiced much more narrowly than what is proposed here. Presently, for organizational behavior phenomena, studies typically identify demographic information such as age, gender, education level, occupation, organizational rank, tenure, marital status, number of children, and salary range (see for example Brockner, et al (1997)). Although some of this information is likely relevant for the structural determination of A or B, what would probably be more useful for this field is information about the “interiors” (Wilber, 1996) of A and B. For example, although the age of B may be useful for understanding how s/he interacts with A who is employing a given leadership style, it may be much more useful to know some of B’s interior attitudes, opinions, values, and/or philosophies. The choice of what aspects of the interior to relate can be informed by prior research. In the example of Figure 1, a sampling of information about A that may be most useful could include: her past experiences in similar situations, her age, whether she is seen as hard working, whether she is seen as fair, whether she is willing to work overtime alongside her work group, and whether she has other idiosyncratic credits with the work group.

For the employees (B), useful differentiating information might include: their level of motivation, the amount of family support they have for working overtime, how the financial incentive is valued, whether they desire to advance in the organization, and how committed they are to the mission of the organization.

The choice of which environmental considerations to exhibit is similar to that of A and B. Although demographic information is commonly provided in many studies, there is no consistent information typically presented about the environment or context of the research. Far too many studies are conducted in settings which have no real world counterparts (Argyris and Schon, 1996; Lindblom, 1959). If the contrived condition or assumption is changed, would the same or similar results hold? Is attention focused in the wrong direction, for example, by studying young college students in unrealistic situations such as laying off employees (Folger and Skarlicki, 1998). The authors contend that their “approach put the participants in the realistic position of having to take conflicting priorities into account” (p. 81). Are the dynamics of a true layoff even remotely present when the students may not be in an employment situation (the article describes them as MBA students, but doesn’t say they are employed), do not have years of day-to-day history together, or do not have the pressures of financial obligations (which may include college tuition for their children, such as students in the study)? Are these students really concerned that someone they “layoff” may become violent with them? Might they, in a fun way, easily layoff their friends, or, on the other hand, choose to retain their friends and easily layoff people they do not know well (perhaps these dynamics do match the real world!)?

For guidance as to what to represent in E, we first turn to Mary Parker Follett’s 1924 book, Creative Experience, which reads like a current book on management and organizations. Although her jargon is different, her writings condemning the construct of linear cause and effect trace many of the arguments her academic descendants of later generations used. She nicely captures the feel of mutual causality.

In human relations... I never react to you but to you-plus-me; or to be more accurate, it is I-plus-you reacting to you-plus-me. “I” can never influence “you” because you have already influenced me... Accurately speaking the matter cannot be expressed even by the phrase used above, I-plus-you meeting you-plus-me. It is I plus the-interweaving-between-you-and-me meeting you plus the-interweaving-between-you-and-me, etc., etc. If we were doing it mathematically we should work it out to the nth power (pps. 62-63).

Follett suggests that for all social science research, researchers should “study the whole a-making; this involves a study of whole and parts in their active and continuous relation to each other...But there is an additional point to be considered: environment too is a whole a-making, and the interknitting of these two wholes a-making creates the total situation - also a-making” (p. 102).

In the example of Figure 1, possibly important contextual considerations could be: whether there is an organizationally-sanctioned ability of the leader to punish resistors, whether the organization is in crisis, whether there are norms for generally following leaders, whether there are norms for personal responsibility, whether other workers were also asked to work overtime, and whether the request was perceived to have been made in a respectful manner.

A final aspect of the model is the inclusion of time (T).   Many researchers have discovered that what is called “cause” and what is called “effect” often depends on the snapshot of time in which the dynamic is observed. We join with Zaheer, Albert, and Zaheer (1999) in calling for the specification of the relevant time scale, which they argue is as critical as the specification of the unit of analysis. Zaheer, Albert, and Zaheer (1999) suggest that the time periods which need to be specified are the existence, observation, recording, aggregation, and validity intervals. Not only does time scale play a role in the outcomes of a study, it also may change the meaning of concepts or relationships between data.

Sastry (1997) provides an example of how organizational dynamics can change “direction” over time (GIVE DETAIL). In the example of Figure 1, one view is that the leader’s action is causing a change in the work groups’ behavior. Another view is that the work groups’ behavior (perhaps they have not been producing enough) caused a change in the leader’s action. Consequently, in this case it might be important to know: has there been a pattern of such requests (is this the first time or the nth time?), is this request happening early or late in the work group’s formation, and what has been the past performance of the group?

Advantages of the Proposed Model

The interactional model proposed here is more complex than the classic model of cause and effect. In order for a model which is more complex to be accepted when simpler alternatives are available, the complex model must “work” in situations in which the simple model does not. In this section, I will offer one detailed example of the interactional model providing insight in a situation in which the classic cause and effect model would produce a “wrong” finding. For several years, Business Week magazine has been conducting research in order to rank the top business schools in the country. BW takes the “customer’s” point of view and asks students and the companies that hire them to rate their experiences with the business schools. The simple cause and effect explanation is that good MBA programs (X) will result in good opinions by students and companies (Y) which will be reflected in the rankings. The 1998 study, “The Best B-Schools,” is the analysis of Educational Psychology professors David M. Rindskopf and Alan L. Gross at the City University of New York Graduate School and University Center and is reported by Jennifer Reingold.   In a sub-story with Hala Habal, Reingold reports “How We Kept the Data Unsullied,” which discusses a difficulty in this simple X->Y formulation.   “After an investigation lasting several months, BW determined that some students at five schools tried to ‘game’ the system by inflating their responses on the student portion of the Business Week survey” (Reingold and Habal, 1998, p. 94). The reporters suggest that the students “took [the rankings] too seriously.”

A different perspective is that the research designers overlooked the fact that they were studying a phenomenon which included “knowing subjects” (B). These students were aware that their input would largely determine the ranking of their school. Some students decided to “artificially” increase their responses because a higher ranking in the polls is more important for their career entry than their immediate educational experience is. In general, the classic cause-effect construct considers the “target” group to be affected (B) as passive elements of the causal system. What the students have done in this case, and people often do in similar cases, is to take an active role in the process, thereby altering the system under investigation. The researchers and reporters felt that this action affected the “integrity” of the data. Deans at four of the five schools said that the student actions reflect the positive enthusiasm that the students have for the schools. In order to “discount” this “enthusiasm,” the researchers eliminated some responses and gave greater weight to 1994 and 1996 studies.

Researchers have to go to great lengths, many of them questionable, to maintain the simple X->Y formulation. For example, in 1998, they happened to discover the active role of the students. By giving greater weight to the 1994 and 1996 studies, the researchers are assuming that no such action took place then. They are also assuming that such action was limited to the five schools they discovered. Furthermore, BW offered no possibility that the recruiters, who are also surveyed, took an active role in the process. Is it possible that a recruiter would rate more highly the schools of the employees s/he hired, enabling some self-justification by the recruiters to their bosses?

The BW study wanted to ignore the context or environmental (E) factors that are present in such a situation. Such factors would include the student culture in each school, the student-administrator relationships, the role that the findings play in the hiring process, and others. MIT’s Sloan School dropped six places from the previous study, suggest the authors, largely for reasons that could be described as contextual. One reason offered is that the students were sending a message to the administrators about their dissatisfaction with not being able to take a course offered by a popular professor. A second reason offered in the article is that recruiters rated the school lower because they were “spurned” when students accepted entrepreneurial jobs rather than signing up with more established corporations.

Finally, considerations of time do not enter into the classic X->Y model. BW would like readers to believe that each bi-annual survey is a free-standing event, that the 1994 results don’t influence the 1996 results which don’t influence the 1998 results. This isn’t often the case when knowing subjects are involved because the people in the system “learn” over time. People know what the past results were, know what the implications of those results were in several dimensions, and factor that accordingly into their input.

Research Critiqued Using the Proposed Model

A second method for showing the value of a model is to consider this model relative to a typical article published in what is commonly recognized as a high quality journal. Such an article usually adds to the organizational knowledge by reporting a finding, such as, “the optimal profile for high-performing groups includes important, moderate task conflicts, no relationship conflicts, little or no procedural conflict, with norms that task conflict is acceptable and resolvable and with little negative emotionality” (Jehn, 1997, p. 552). X, the supposed cause, is carefully specified. Y, the alleged effect (high performance) is also mentioned. The conclusions do not mention A (who/how the cause was produced), B (the nature of the group which will be high performing), E (the environment in which this group works), or T(how all these interactions evolve over time).

Many may suggest that specifying A wouldn’t necessarily be the researcher’s job in this situation, so I will not argue that point here. The article provides some data about the groups in the study (we are told the function they perform in a household- goods-moving organization), but makes no effort to describe the structural dynamics of the group so that others may judge the likelihood of the effect generated by these groups also being generated by other groups. Likewise, there is little or no information about the environment in which these groups operate, and what role environmental features play in the claim. How often do they experience first-order and second-order change? Is the environment highly contextualized? These and many other environmental characteristics could be specified in a more complete interactional model. Several T questions are also raised by the claim above. For example, “no relationship conflicts” could easily be the “effect” and “high performance,” the “cause.” No effort was made to describe the “initial kick” condition, so at the time the snapshot of data was collected and correlations were calculated, the directionality is uncertain.

Consider as a second possibility the relationship between the norm “conflict is acceptable and resolvable” and performance. Suppose the relationship is different after Jehn’s snapshot. Perhaps this norm contributes modestly to performance for a time, but then the more pronounced dynamic becomes that sustained high levels of performance generate this norm.

Model Critique and Future Research

The proposal made here for a reconception of the traditional cause-effect relationship is necessarily crude and perhaps does not fully address the implications of Maturana’s ontology and epistemology. The traditional model has over 250 years of inertia and institutionalization. A different model must overcome the tendency in the field to “over apply” Ockham’s razor (Scott, 1995, p. 2) and strain to fit data into simple, existing (and misleading) models rather than seek more elaborate, accurate models (Harman, 1998, pps. 100-101). In shifting paradigms, it is difficult to move from a highly refined model, even if it is no longer effective, to a coarse model (Kuhn, 1970). If this model is on the right track other theorists will need to suggest elaborations and/or refinements.

This model does address many of the concerns listed in this article. The inclusion and prominent role of the interactions of (E) is consistent with Ackoff’s (1981) call for an environment-ful model. It is also responsive to Follett’s beliefs about interactions, although the model is more static than taking changes to the n^th degree as she theorizes. Argyris and Schon (1996) also take issue with the assumption of the constancy of the definition of the variable. Although the model does include a time dimension, which is a step in the right direction, it does not offer a robust way of handling changes in the definitions (structures in Maturana’s terms) of the variables (composite unities).

The model is sensitive to Argyris and Schon’s (1996) call for including the actor’s intention, because such an intention is certainly part of specifying the actor’s structure. The model can also be used to identify the “initial kick” (or what chaos theorists often refer to as SDIC - sensitivity dependence on initial conditions). However, if the starting point is not included in the timeframe in which the model is used, that information is lost. Still, by focusing on mutually causal possibilities, researchers using this model will be less likely to assert causality in a single direction when the dynamic observed is better explained with knowledge of the initial kick (or conditions).

System dynamics is a field in which a great deal of progress has been made in specifying the environment and representing interactions over time (Meadows and Robinson, 1985). This work is especially important because it has developed new tools for observing and modeling phenomena. Traditional statistical methods such as regression and factor analysis are built upon the assumption of linearity. The work in system dynamics alone, however, is not enough. That discipline makes philosophical assumptions inconsistent with the ontology and epistemology recommended here. System dynamicists often assume objective observation, and do not allow for knowing subjects or self-organization (Dent, 1999).

The model proposed here is substantially more complicated than that of traditional cause and effect as it is commonly practiced today. Yet, even crude research which is conducted this way will advance the field further than the sophisticated mathematical techniques in use today which rest upon a philosophical foundation which leads to dead ends, and misleads to conclusions which are highly questionable.

In writing an article like this, I am trying to be structurally coupled with you, the reader. I (A) have written an article (X) which you (B) are reading and my intention is that you will find merit in the interactional model (Y). All the considerations of the model suggested apply here. The examples I have offered, or the style in which I have written affect how you perceive the content. (B) factors possibly include your academic discipline, your set of experiences, your own research and others. The acceptability of such a model in the community of scholars is a contextual (E) factor, as would be its complexity relative to the simpler, classic model. A time (T) factor is that the model being proposed here is happening centuries after the classic model. Consequently, a huge “investment” exists in the prevailing paradigm, and it has all of the advantages of inertia and incumbency. It will be interesting to see how all of these factors interrelate as this alternative model receives consideration.

References

Ackoff, Russell L. (1994). The democratic corporation: A radical prescription for recreating corporate america and rediscovering success. New York: Oxford University Press.

Ackoff, Russell L. (1981). Creating the corporate future: Plan or be planned for. New York: John Wiley & Sons. 

Argyris, Chris and Schön, Donald A. (1996). Organizational learning II: Theory, method, and practice. Reading, MA: Addison-Wesley Publishing Company.

Argyris, Chris (1958). “Some Problems in Conceptualizing Organizational Climate,” Administrative Sciences Quarterly. Vol 2. No. 4, pp. 501-520.

Argyris, Chris (1957). “The individual and organization: Some problems of mutual adjustment,” Administrative Sciences Quarterly. Vol 2. No. 1, pp. 2-24.

Bateson, G. (1972). Steps to an ecology of mind. New York: Ballantine Books

Begun, James W. (December 1994) "Chaos and complexity: frontiers of organization science" Journal of management inquiry. Vol. 3, No. 4, 329-335.

Bem, Daryl J. (1967). “Self-perception: the dependent variable of human performance,” Organizational Behavior and Human Performance 2:105-121.

Brand, M. (1976). The nature of causation. Chicago: University of Illinois Press.

Brockner, Joel; Siegel, Phyllis A.; Daly, Joseph P. ;Tyler, Tom; and Martin, Christopher (1997). “When trust matters: The moderating effect of outcome favorability,” Administrative Science Quarterly. Vol. 42, pps. 558-583.

Bunge, M. (1959). Causality. Cambridge, MA: Harvard University Press.

Caws, Peter (1993). Yorick’s world: Science and the knowing subject. Berkeley: University of California Press.

Clemson, B. (1984). Cybernetics: a new management tool. Tunbridge Wells, Kent, England: Abacus House.

Dell, Paul F. (1985). “Understanding Bateson and Maturana: Toward a biological foundation for the social sciences,” Journal of Marital and Family Therapy. Vol. 11, No. 1, 1-20.

Dent, E. B. (1999). “System science traditions: Differing philosophical assumptions,” Systems Research and Behavioral Science, under review.

Dent, E. B. (1997). The design, development and evaluation of measures of individual worldview. Dissertation, The George Washington University, The School of Business and Public Management.

Dill, William R. (1958). “Environment as an influence on managerial autonomy,” Administrative Sciences Quarterly. Vol 2. No. 4, pp. 409-443.

Drath, Wilfred H. and Palus, Charles J. (1994). Making common sense. Greensboro: Center for Creative Leadership.

Engel, S. Morris (1980). With good reason : An introduction to informal fallacies. New York : St. Martin's Press.

Epstein, Joshua M. and Robert Axtell (1996). Growing artificial life: Social science from the bottom up. Washington, DC: Brookings Institution Press.

Folger, Robert and Skarlicki, Daniel P. (1998). “When tough times make tough bosses: Managerial distance as a function of layoff blame,” Academy of Management Journal. Vol. 41, No. 1, pps. 79-87.

Follet, Mary Parker (1924). Creative experience. New York: Peter Smith.

Giddens, Anthony (1976). New rules of sociological method : a positive critique of interpretative sociologies. New York : Basic Books.

Gombrich, Ernst Hans (1960). Art and Illusion: A study in the psychology of pictorial presentation. New York: Pantheon Books.

Guba, Egon G. (1985). "The context of emergent paradigm research." Organizational theory and inquiry. Lincoln, Y. S. (ed.) Beverly Hills: Sage Publications.

Harman, Willis (1998). Global mind change: The promise of the last years of the twentieth centuryIndianapolis, Individual. : Knowledge Systems.

Hirschman, Albert O. and Lindblom, Charles E. (1962). “Economic development, research and development, policy making: Some converging views,” Behavioral Science 7:211-222.

Jehn, Karen A. (1997) “Qualitative analysis of conflict types and dimensions in organizational groups,” Administrative Science Quarterly. Vol. 42, pps. 530-557.

Kitcher, Philip (1991) “The pragmatics of explanation.” The philosophy of science. Boyd, Richard, Gasper, Philip, and Trout, J.D. (eds.) Cambridge: The MIT Press.

Kuhn, Thomas S. (1970). The Structure of Scientific Revolutions. 2nd edition. Chicago: The University of Chicago Press.

Lindblom, Charles E. (1959). “The science of ‘muddling through,’” Public Administration Review. Vol. 19, No. 2, pp. 78-88.

Lowin, Aaron and Craig, James L. (1968) “The influence of level of performance on managerial style: An experimental object-lesson in the ambiguity of correlational data,” Organizational Behavior and Human Performance 3:440-458.

Maruyama, Magoroh (1968). “The second cybernetics: Deviation-amplifying mutual causal processes,” pps. 304-338 in Modern systems research for the behavioral scientist, Walter Buckley (ed.) Chicago: Aldine Publishing Co.

Maturana, Humberto R. (1998). “Ontology of observing: The biological foundations of self consciousness and the physical domain of existence.” http://www.inteco.cl/biology/ontology/

Maturana, Humberto R. and Varela, Francisco J. (1987) The tree of knowledge: The biological roots of human understanding. Boston: Shambhala.

Meadows, Donella H. and Robinson, Jenny M. (1985). The electronic oracle: Computer models and social decisions. Chichester: John Wiley and Sons.

Rakover, S. S. (1990). Metapsychology: Missing links in behavior, mind, and science. New York: Paragon.

Reingold, Jennifer (1998). “The best B-schools,” Business Week, October 19, pps. 86-92.

Reingold, Jennifer and Habal, Hala (1998). “How we kept the data unsullied,” Business Week, October 19, pps. 94.

Rossi, Peter H. (1958). “Community decision making,” Administrative Sciences Quarterly. Vol 2. No. 4, pp. 415-443.

Ruiz, Alfredo B. (1996). “The contributions of Humberto Maturana to the sciences of complexity and psychology,” Journal of Constructivist Psychology. Vol. 9 No. 4 (Oct-Nov), pp. 283-302.

Russell, Bertrand (1948). Human knowledge: Its scope and limits. New York: Simon and Schuster.

Sastry, M. Anjali (1997). “Problems and paradoxes in a model of punctuated organizational change,” Administrative Science Quarterly. Vol. 42, pps. 237-275.

Schwartz, P. and Ogilvy, J. (1979). The emergent paradigm: Changing patterns of thought and belief. Menlo Park, CA: SRI International.

Scott, Alwyn (1995). Stairway to the mind: The controversial new science of consciousness. New York: Springer-Verlag.

Singleton, Royce and Straits, Bruce C. and Straits, Margaret Miller (1993). Approaches to social research (2nd ed.). New York: Oxford University Press.

Slife, Brent D. and Williams, Richard N. (1995). What’s behind the research: Discovering hidden assumptions in the behavioral sciences. Thousand Oaks, CA: Sage Publications, Inc.

Vaill, P. B. (1998). Spirited leading and learning: Process wisdom for a new age. San Francisco: Jossey-Bass.

Vaill, P. B. (1996). Learning as a way of being: Strategies for survival in a world of permanent white water. San Francisco: Jossey-Bass.

von Bertalanffy, Ludwig (1968). “General system theory - A critical review." Modern systems research for the behavioral scientist. Buckley, Walter (ed.) Chicago: Aldine Publishing Co.

Weick, Karl E. (1995). Sensemaking in organizations. Thousand Oaks, CA: Sage Publications

Weick, Karl E. (1985). "Sources of order in underorganized systems: Themes in recent organizational theory." Organizational theory and inquiry. Lincoln, Y. S. (ed.) Beverly Hills: Sage Publications.

Wilber, Ken (1996). A brief history of everything. Boston: Shambhala.

Zaheer, S., Albert, S., and Zaheer, A. (1999). “Time scales and organizational theory,” Academy of Management Review. Vol. 24, No. 4, pps. 725-741.