A Theory of Group Structures

Volume I: Basic Theory

Kenneth D. Mackenzie

• Publisher: Routledge
• Year: 2018
• ISBN:
  9781317219996
• Edition: 1
• Topic: Ciencias sociales
• Subtopic: Sociología

First published in 1976, this A Theory of Group Structures is a study of the aggregation of individuals into groups, which cuts across many different social sciences. Volume one attempts to formulate a more rigorous theory of group structures by providing consistent definitions, assumptions, measures, methodology, theory and results. This book will be of interest to students of all social sciences.

CHAPTER 1

INTRODUCTION

Few pleasures compare with the first perception of a new idea. This pleasure flows from a promise of freshness and hope, the rapid reassessments that follow, and the fanciful nature of problem-solving. It is a time for intellectual play. The hard work and troublesome conditions that limit the idea come later. I remember being delighted when I first read about the early work of Bavelas (1948, 1950) and Leavitt (1951) on group structure.

I had been reading economics and some rather turgid tracts in organizational theory and I was bored. While following down a reading list, I came to articles by H. Leavitt and by A. Bavelas. Here was a new idea and a new paradigm: They suggested that a group structure is represented by persons and communication channels that connect pairs of people. The concept of a “communication channel” that would “connect” was abstract and interesting. They proposed to study the effects of group structure on behavior by performing what is now called a communications network experiment. Essentially, a group of subjects (usually five) were allowed to pass messages among themselves to solve a series of problems. Subjects' choices of comnunications channels were constrained to disallow certain selections. The very early results reported by Leavitt were easy to comprehend and seemed very reasonable. Clearly, this line of attack was promising. Immediately one could see potential for new activities: Vary the size of the group; vary the task; vary the type of person; vary the medium of communication; vary the incentives, etc. In addition, it looked like an easy thing to do. The lack of a theory appeared temporary and it was hoped that one could be built up from the sequence of experiments that would flow out of the new idea. I remember feeling very happy with this discovery.

Now, years later, I find long, complex survey articles on this problem area. The, G

anzer and Glazer articles (1959, 1961); Shaw’s (1954a, b; 1964, 1971); Collins and Guetzkow (1964); Collins and Raven (1969) and many others have attempted to survey the related literature. By this time, however, the lack of substantive theory, the inconsistency of methodology, the basic incomparability of research findings, and the mathematically interesting but empirically vacuous models have combined to create a conceptual thicket out of what was once a delightfully simple new growth.

Each of the major surveys of the communications network or group structure literature has mentioned these problems. Collins and Raven (1969), however, made the issues clearer by attempting first to survey the various models and theories before proceeding with the empirical work. They reached the conclusion that:

We are unable to review a significant portion of the empirical data on group structure under the same heading under which we have reviewed the more theoretical treatment of group structure (p. 119).

Later, they also state “…we were unable to devise a theoretical system which would encompass even a small percentage of the studies we felt competent to discuss (p. 184).” That such competent and honest persons plus so many other bright scientists have found the problems associated with providing a theory of group structures so intractable suggests that perhaps there is no solution to these problems as they are currently formulated.

BASIC COMMITMENTS

This work represents an attempt to reformulate a theory of group structures by an extensive sequence of inductive steps. Each step is rather simple and is made by following several intellectual commitments to:

(1) Define concepts and relationships as clearly as possible before proceeding.

(2) Determine how things happen and avoid teleological reasoning whenever possible.

(3) Describe the “how” statements in simple, descriptive mathematical language.

(4) Attempt, whenever possible, to make explanations based upon deductivenomological reasoning.

(5) Keep it simple.

(6) Avoid probability or frequency statements whenever possible.¹

(7) Employ deterministic models at every opportunity.

(8) Use methods based upon strong inference.

These commitments are the foundation to this work. It is assumed that knowledge in a field is like a tree and facts are like its leaves. Whenever theory and data are inconsistent, whenever there are many contradictory theories, whenever the facts are in dispute, one should expect to find an unhealthy tree having many dead limbs, dead leaves, overgrowths at its base, and many leaves belonging to other growths. Vigorous cultivation and pruning may be necessary to restore the tree. The stated commitments act as a guide to this extensive gardening.² The two major commitments–emphasizing how events occur and using strong inference–are strategic. The others are tactical derivatives.

To determine how things happen, I decided to act as if I were an alien recently landed on a new planet and privileged to observe the local creatures interact. Assume that in such a role I do not know their language, social system, religion, physiology, or system of government: I would probably be able to remain more objective than if they were fellow humans. Suppose that the ones before me have four legs, bushy antennae, tails, large compound eyes, and three “heads.” About all I can do is to make observations by videotaping their “discussions.” I take these back to my space capsule and attempt to analyze them along with my impressions. First, I notice that when one of them lowers his right antenna and raises his tail and right front leg, the other dips his left antenna and moves to the left. Then I notice that, when this occurs, the first raises his right antenna and lowers his tail and right front leg. If the second does not move to the left, however, the first creature lowers his left antenna, keeps his tail up, and bobs his largest head vigorously. I repeat the observations, study the videotapes, and form possible hypotheses until I begin to perceive patterns of consistent behavior sequences. Although the analogy is somewhat crude, I reasoned that if I could conceptually do this with imaginary creatures, I might be able to do it with small groups if I could only find a powerful method of reducing my bias.

I could, for example, run experiments to control the situation. I could replicate earlier findings. But I was very worried about the unsolved problems of experimenter effects on outcome that Rosenthal (1966) reports. In addition to these effects, being carefully indoctrinated in statistics, I knew that one could not prove a theory by experiments but that one certainly could, with respect to a set of assumptions, conditionally disprove a theory. A procedure for disproving had to be stronger than mere failure to achieve high levels of statistical significance. The easiest form of disproof is the counterexample. So I decided to use a version of Platt’s (1964) strong inference to study group structures. It seemed to me that a clear counterexample to my theory should be sufficient to cause its rejection even if I had many cases wherein the data could not disprove it. The seeking of counterexamples to reject a theory forces a severe discipline on the researcher, for in order to recognize a counterexample he must clearly state what he is doing and have relatively unambiguous criteria to apply to the results.

Since strong inference is uncommon in the social sciences, a few examples of its use may be helpful. In mathematics a theorem is rejected if a single counterexample can be found, despite the fact that it may be true for almost all other cases. The existence of a counterexample forces, at the very least, a restatement of the theorem. In a sense, strong inference is the experimental analogue to the mathematical procedure of disproving by counterexample.

A particularly delightful example comes in an experiment performed by Allen, Francis and Zeh (1971) to disprove a theory about amoebic movement. Apparently, the mechanisms of pseudopod extension and retraction in free-living amoebae are subject to question. According to one theory, the bulk of endo-plastic streaming is due to a pressure gradient. According to another, which I call the “slinky” model (after the toy with a wire coil that can hop down a flight of stairs), each advancing pseudopod tip is the source of contraction; by drawing the viscoelastic endoplasm forward, the pseudopod becomes everted to form the ectoplasmic tube. It is difficult to think of a way of deciding which theory is correct, the pressure gradient or the slinky. In an attempt to limit the possibilities, an experimental procedure was developed which could disprove the pressure gradient theory: They simply sucked one pseudopod of an amoeba into a capillary connected to a partial vacuum. If the pressure gradient theory is correct, there should be no extension of other pseudopods except the one in the capillary. Photographs, however, clearly showed new pseudopods extending in other directions. A strong inference would be that the pressure gradient theory is not correct as stated. One cannot then conclude, however, that the “slinky” theory is correct; the experiment merely disproves one of the theories. The criteria for rejection are clear cut and even a non-specialist can agree with the results.³

If a similar level of clarity can be brought into experiments involving groups, we ought to produce, by successive application to a long sequence of experiments, quite a few rejections.⁴ This would allow the overgrowths to be pruned and help determine which of the branches are healthy. Doing this for a number of years ought to lead to the discovery of a relatively simple theory about group structure. However, that which remains after all of this cannot be interpreted as being correct. It is just not yet disproved. It ought to be interesting, nevertheless, to find out what survives and to continue research by conjuring up devious new experiments designed to reject it. Thus, I use strong inference because I believe that it is an efficient means to improve and develop a theory. I have much less interest in “proving” than I have in improving my theory of group structures.

WHAT WE OBSERVE

A person studying the behavior of groups observes interpersonal interactions. He may also record data about the personal characteristics and attitudes of the subjects. Usually hidden, but quite important, are the intrapersonal decision processes. Intrapersonal decision processes affect the nature, degree, and direction of interpersonal interactions. It is, however, the interpersonal interactions that are observed. In light of the commitments made in the previous section, the theory and results of this book are presented in terms of the interpersonal interaction. We can only make statistically under-identified guesses about the individual decision making process. We observe messages. Our task, albeit a limited one, is the development of a theory of group structures that is based upon the sequences of interactions.

I assume that interactions are purposive and are made to facilitate the person’s satisfaction of his needs. Members of a group interact to determine goals, the activities to be performed, who should perform these activities and who is responsible for their performance, who is the coordinator, how the resources are allocated, what should be the sequence of activities, how the group should adapt to change, what control procedures are necessary, who is to make what decisions, etc. Even the most “trivial” and transitory of groups, the small laboratory group, has much to interact about. The time ordered sequences of interpersonal interactions are the basic observable data. I focus on the interpersonal interactions to gain the clear time sequence that constitutes the group’s “life history.”

Structures are patterns of interpersonal interactions. Structures evolve to satisfy individual and group needs. Structures represent need-satisfying interaction patterns. Structures are considered dynamic, not static. Should any group member believe he is “better off” changing his interactions, he may do so. Structures are measured in terms of the content, direction and number of interactions between all pairs of subjects. They represent the actual pattern of interactions and not just the patterns that “ought to” exist, as in an organization chart.

Usually a group interacts on more than a single dimension. It is quite possible that the group has different structures for different dimensions. For example, a group may be decentralized for exchanging information, centralized on one person for decision making, and have subgroups engaging in non-task-oriented interactions. This is why I speak of group structures throughout the book.

The choice of structures to be measured depends upon the problems the group solves and the purposes and theory of the analyst. The intricate connection between the choice of structures to be studied and the nature of the group’s problem-solving processes is examined in later chapters. I shall argue that problem-solving processes for interacting groups are defined by sequences of group structures. In addition to task-oriented process structures, I shall also examine how the structures are formed and, to a lesser extent, analyze special non-task-oriented structures.

SOCIAL PROCESSES

It is convenient to conceptualize a process in terms of states of a system and the “flow” or progression between pairs of states. The most primitive process is a stimulus-response process. In such a process there is a state describing an entity, an action or stimulus, and the new state that obtains as a response. Many processes are much more complex because (1) more than two states and one event are involved, and (2) the events or “stimuli” involve interactions among separate entities. Most of the small group processes in this book are the more complex type.

In some cases the identification of the states and specification of actions to “flow” between states is determined by logic and unambiguous criteria. For example, in the construction of a house, the land must be cleared, excavation of underground areas completed, the foundation laid, framing constructed, pipes and conduits for heating, water, and electricity placed, walls constructed, roof constructed, insulation installed, plumbing installed, finishing of interior and exterior completed, etc. The process for construction can be set out in advance and criteria for reaching the various states of completion can be readily determined. In fact, there are powerful analytical procedures in the management sciences for planning such processes.

Social processes, especially those i...