Read an Excerpt

Public Management Reform and Innovation

Research, Theory, and Application

The University of Alabama Press

One Hundred Theories of Organizational Change

The Good, the Bad, and the Ugly

Lawrence B. Mohr

My purpose in this chapter is to present and defend some recently derived views on the subject of theory in the social sciences and to illustrate them with reference to the field of innovation in organizations. In a recent publication (Mohr 1996), I claimed to demonstrate that neither universal nor probabilistic laws governing human behavior are possible. The analyses were based on an investigation of certain concepts and processes that seem to me basic to social research, including the definition of causality and the nature of the physiological mechanism that generates intentional behavior. With the treatment of laws added to these investigations, the whole suggests the need for a second look at the goals and methods of explanatory social science research. In this chapter, I concentrate on goals rather than methods, first sketching out the background arguments in this connection and then applying the results to innovation theory.

Expectations for Theory in Social Science

Encounters and Laws

The case against universal laws of human behavior hinges on the notion of "encounters," coupled with the conclusion that "operative reasons" in a certain physiological form are the physical causes of all intentional behavior. By universal law in this context I mean an explanation for human behavior that holds universally (given a very small allowance for measurement error and irreducible random perturbation) for a specified subclass of people under a specified set of conditions. In other words, universal does not mean all people, but any subset as long as it is not left partially vague. Also, the proposed applicability of the law might be narrowed substantially by stipulating that it holds only under certain conditions and not under all conditions, but those conditions must also be specified in order for the proposition to be a valid law. In the end, I argue that it is impossible to specify such conditions for any explanation of any intentional behavior, which is tantamount to saying that there can be no laws.

Encounters

I have also suggested elsewhere that a good deal of prominent scientific work, including both Mendelian and Darwinian theory, is pointed toward the theoretical explanation of encounters rather than of simpler events (see "process theory" in Mohr 1982, 44–70). An encounter, or probabilistic encounter, is a compound event conceptualized as the status relative to each other of two or more free, component objects or events. By "free" I mean the opposite of a fixed frame of reference or a given. For example, when we speak of a motion upward or downward, we mean relative to the earth so that the earth is taken as a fixed frame of reference, and the motion is simply the motion of one object and not an encounter. The earth in this sense does not count because it is considered fixed rather than free. Similarly, whereas the collision of two billiard balls on a table is an encounter, the motion of one of them afterward may be considered the motion of a single object. It does not have to be considered, for example, as an encounter involving the ball and the table because the table may be considered as not free; it may be taken rather as a fixed frame of reference.

An automobile accident—a collision between two cars—is an example of a probabilistic encounter. If we conceptualized an event as two ships passing in the night, that would be an encounter in just the same way. The accident and the passing as such are not motions of one or more objects but rather a juxtaposition of things—a status of things relative to one another. I argue at length (Mohr 1996) that encounters cannot be covered by laws. No law can be specified, for example, that correctly predicts or determines that an accident will take place or that the ships will pass in the night. The compound event is inherently probabilistic. One car might be caught by a red light, or have a flat tire, or swerve at the last moment so that no accident takes place. One might try to specify a lot of initial conditions, including a statement such as "provided there is no red light, flat tire, or swerving," but in order to predict with certainty one would have to account for everything in the universe. Otherwise, in principle, something one did not think of could always come up.

If the accident does indeed take place, one can easily see how each car followed known physical laws under the totality of circumstances that actually prevailed. Moreover, one does not have to name all of the circumstances, such as "sound tires" and so forth, but rather can consider most of them as unidentified background or context. One can have a complete, lawlike explanation of the accident given those circumstances together with the fact that the accident happened. Before the accident does take place, however, the laws cannot tell you that it will.

Events are commonly conceptualized as encounters. "Winning" at roulette, for example, is an encounter between a certain bet and the stopping place of a little ball. Because it is not something that receives a force and that thereby moves, the winning is not covered by the laws of physics. The following are some additional examples: a tie for first place; sole possession of first place; bumping into an old friend; stumbling over a root; getting caught in the rain; agreement; job satisfaction; and so forth.

Reasons

A reason is defined roughly as the combination of a desire and a certain kind of belief: for example, the desire to overcome people's resistance to a change and the belief that getting them to participate in decision making will overcome their resistance. This desire and belief together become a reason for getting them to participate. As used here, a reason is a physiological phenomenon and does not have to be conscious, although it may at times be that, too. We all know that we sometimes do things without being conscious of why beforehand, or even afterward, and indeed it might take a psychoanalysis to figure it out. Nevertheless, the reason was there in unaware, physiological form and operated. An operative reason is the reason that at a given time is the strongest. I have argued (Mohr 1996, chapter 3) that operative physiological reasons are the causes of intentional behavior.

Laws

I hope one can begin to see from this brief characterization of a longer argument why it is that there can be no universal laws governing human behavior. Behavior results from an encounter, namely, the encounter of a reason with other reasons, such that the strongest one wins and the behavior pertinent to it is selected. It might be true, therefore, that a person or an organization adopted an innovation in order to acquire enhanced status, but there can be no law that the desire to acquire enhanced status will lead to innovation. That particular reason for innovating might sometimes be weaker than some other reason for doing something else, and the innovation might not only fail to take place then and there, but it might well never take place in spite of the continued existence of the reason.

We commonly consider that although there are no universal laws governing human behavior, there may be probabilistic ones. There are indeed probabilistic laws in physics, such as the laws of the rate of radioactive decay of the various elements, and in biology, such as the Mendelian laws of segregation and independent assortment. Each of these always depends on a strictly random process that, paradoxically enough, will produce an utterly stable and reliable distribution of outcomes. Intentional human behavior is assuredly probabilistic, as the above paragraphs indicate, because what is done always depends on the probabilistic encounter of reasons of various strengths with other reasons. Furthermore, many social scientists offer explanations for behavior that are probabilistic in the sense that they incorporate a stochastic component. In a regression model, for example, for each combination of values on the independent variables, each possible value of the dependent variable has a stipulated probability of occurrence. This probability is defined by the parameters of the regression surface together with the parameters of the distribution of the random disturbance term.

Nevertheless, such a probabilistic model cannot be a probabilistic law. The reason is that the parameters are not stable across populations and time periods, meaning that outcomes conditioned on any particular values of the independent variables have two sources of variation rather than one—the local parameters as well as the disturbance. The local parameters do not vary randomly, which would imply the absurd result of strictly random human behavior, but neither do they vary in response to specifiable causes, which (by simply including the causes in the model) would imply human behavior with stable conditioned means across populations and time periods. Instead, the parameters vary in response to encounters, which in turn depend on heredity, context, and experience. The parameters, in short, are variable but unpredictable, which negates any claim to lawlike status.

Implications for Theory

The point I want to emphasize is that the categorical disavowal of laws leaves explanatory social science in a challenging position. If we cannot aspire to stable generalizations, to what then can we aspire? Anytime we explain innovation, good leadership, and so forth, whether it be by a quantitative analysis or a case study, it must be with the thought that this explanation cannot be generalized in any systematic way beyond the population and time period studied or sampled. No amount of accumulation of research will change this fundamental fact in any way, and indeed, in areas such as innovation and human relations theory, which represent some of the most relentlessly studied areas in our field, we are as far from stable generalizations as ever. Each study we carry out for the purpose of explaining the behavior of individuals or groups is in this sense "historical," or history bound (Gergen 1973).

The above is not meant to imply that it is pointless to carry out explanatory research, but only that the disavowal of laws calls on us to pause and give some thought to discovering and articulating just what might be the point of such research. Otherwise, we are likely to either continue to operate as though some universal or probabilistic laws were likely eventually to result or be vulnerable to the criticism that we have not been reflective enough about why we do what we do and how, therefore, we can tell the good research from the bad. I wish here to offer one kind of answer to these questions, but the general issue seems to me important enough that one would hope for further proposals and discussion.

Because all explanatory social research must be "historical," I take my cue from the position of the discipline of history—not, I might add, from the raisons d'etre explicitly offered by historians, for these have been few (Carr 1963; Turner 1981), but rather from the apparently prevailing attitudes toward the study of history in society.

Why study history? Why care about the past? The best answer probably involves the observation that human beings are addicted to stories, i.e., to accounts of the causes and consequences of human action, and it seems that these accounts and stories help us, as individuals, to regulate our own affairs. The question might well be raised, of course, whether that vague purpose is enough to justify the activity of large segments of several major academic disciplines—history, anthropology, area studies, and so forth. With some elaboration and specification, I propose that it is indeed enough. Admittedly, it is difficult to say precisely why we are so committed to learning about the way in which certain things happened in the past. Nevertheless, and this is the important point, one cannot foresee the slightest diminution in the inclination to remain so committed. Somehow, we do believe—and I simply accept this as a valid orientation—that an understanding of important past events is valuable enough to receive skilled attention, even when we recognize that such understanding cannot be extended in any systematic way to the prediction, explanation, or control of future events (Scriven 1966, 250).

This orientation provides us with a clue in our quest for a perspective on the role of explanatory social science: good research and ideas are apparently those that impart this sort of understanding of the past, with the proviso that such understanding must be useful. The idea is to impart through the research a consummate causal understanding of the past behavior investigated such that all pertinent challenges and questions regarding that behavior and our explanation of it are answerable from the data and their logical extensions. How nearly this is accomplished will in turn be affected primarily by the set of events or variables we choose to investigate and the creativity and quality of the way in which we conceptualize them, as well as the quality of measurements, analyses, and presentation. Understanding as a goal of research is a venerable idea; the well-known verstehen school has an institutionalized position in the philosophy of social science, often seen as a rebuttal and alternative to positivism (a brief summary pertinent to the present discussion may be found in Warnke 1984; see also Scriven 1966, 250–54). What is different in the present treatment, without attempting a philosophical exposition, is that understanding here embraces at its very core the notion of causation, rather than setting itself up in opposition to it. This is accomplished primarily through the role of operative reasons, which are at the basis of the understanding of human behavior and which at the same time are causally related to it.

As noted, moreover, we commonly think that understanding the background of a certain behavior in depth teaches something valuable in terms of future conduct; therefore at best something is to be learned from such research products in terms of behavioral expectations in similar circumstances. When we know enough of just the right facts about a situation to be able to understand thoroughly the behavior of the actors, we believe that the information we tuck away has a good chance of being reused with profit at some future time. This form of memory suggests a process of creative-selective generalization. Generalizability of this sort is indefinite, but it is also common, and this creative-selective sort is, I suggest, the best kind of theoretical generalization we can aspire to in social science (recognizing that probability sampling permits no generalization beyond the population sampled and the past time period studied).

Let us now view the two notions of consummate causal understanding and creative-selective generalization together as complementary criteria of quality in explanatory research and introduce one more concept that forms a bridge between them. Simply establishing a causal connection between organizational size and innovation in a piece of research, for example, might be a reasonable start toward understanding that relation, but if it were left at that, the nature of the connection would be hidden away in a black box, and creative-selective generalization would therefore be difficult. The idea is that particular research on size and innovation should be instructive in thinking about the same or a similar issue in new contexts. Perhaps in the next few time periods, the same relation might reasonably be expected to hold in the same population of organizations, but as we change populations, or leave the time period far behind, then whether we can apply these findings at all, and whether fully or partially, will depend on relevant similarities between the two contexts, and just which dimensions are relevant is in turn likely to be revealed only through a thorough understanding of the modus operandi of size on innovation in the context studied. Thus, in social science, the particular program of research does not induce or even test a strict regularity, nor does it present a partial law or contribute to the eventual development of a law. What it does at best, instead, is present us with a significant possibility (Scriven 1966, 246–51)—a way in which behavior frequently does indeed unfold, even if not always (possibility), and a finding that is both aesthetically appealing and important for the world to know about (significant).

In the use of the term "significant possibility," the word "significant" may readily be seen to connote importance. That is, private individuals may be drawn to stories and accounts that are of interest mainly to themselves, but academic disciplines, as large social investments, should devote their explanatory side to behaviors that have broad importance. I mean the idea of significance to have a further connotation as well, however, namely, to suggest the aesthetic side of the academic enterprise. Social science is a kind of craft. It is not strictly a trade, such as carpentry, or an art form, such as sculpture, but is rather a mixture of the two, analogous to cabinet making. It takes its value in being both useful and beautiful. The aesthetic side has always suggested characteristics such as parsimony, the intriguing juxtaposition of ideas, and the unusual and skillful use of methodological tools. I will use terms such as "the aesthetic dimension" and "intriguing" to denote this meaning of "significant."

---

Excerpted from Public Management Reform and Innovation by H. George Frederickson, Jocelyn M. Johnston. Copyright © 1999 The University of Alabama Press. Excerpted by permission of The University of Alabama Press.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.

---