Casting Nets    and
Testing Specimens

Two Grand Methods of Psychology

About ordering

Link to Amazon

Link to
Book Deposi­tory

Please have a look at the content of this book
as if you were flipping pages in a store.

Contents

Chapter One    Overview

We all gather information, every day, every moment, squirreling it away in our memories. We gather information whether we are reading, crossing a street, digging in the garden, interviewing a political candidate, painting a picture, or conducting a psychological experiment. Sometimes we use the information in the next split second, sometimes next year.

We use various methods to gather information and so to come to beliefs about the behavior of ourselves and others. Our methods are often very sophisticated when we act as social scientists, and are often rough and ready when we act in ordinary life, but we all use methods of some sort, and we often cite them to justify our beliefs. We say, “I saw her do it,” or “We queried 1,500 randomly selected households.” This book is about the methods we choose by means of which to reach conclusions. It is about how we come to beliefs about the nature and behavior of humans and other living creatures.

Despite the long list of particular methods and the names social scientists have given them, almost all fall into one of two grand classes that I will call the method of relative frequencies (to be explained in Chapters 2 through 8) and the method of specimens (to be explained in Chapters 9 through 12). Both methods can deliver useful information about human behavior. The main point of this book, however, will be to show that for a long time now most social scientists have been using the method of relative frequencies for the wrong purpose—to discover how the human animal, as a species, functions. The method that can do that is the method of specimens. To assert once more, however, my claim that both methods have their suitable uses, I will explain action research,in Chapter 13, as a very useful amalgam of the two methods.

Part I    What the Method of Relative Frequencies Will Do

The method of relative frequencies is the method of choice for estimating the likelihood of finding any arbitrary countable actions associated with any arbitrary class of listable and countable persons—or other entities. It serves that purpose very well indeed. It is also useful for making catalogs, writing histories, and getting hints of internal standards. That is what Chapter 2 will cover.

For the accuracy of its estimates, the method relies unequivocally on random sampling. When sampling is carefully random, the method can estimate in a listed population, with calculable accuracy, not only proportions and means, but also relations among variables. It can even estimate, through the use of causal experiments, the likelihood that certain actions will follow presumed causes. By “likelihood” here, I mean the proportion of subjects (people) who will show the specified action.

Random sampling, however, is sometimes very expensive. Worse than that, it is sometimes impossible. Researchers often feel driven, therefore, to nonrandom samples. Then they lose the power of the method of relative frequencies. But worst of all for the purposes of scientists who want to show forth the manner of functioning of the human animal, the method of relative frequencies is simply incapable of serving them. That incapability is what Chapter 3 will cover.

Chapter Two    Casting Nets

What I am calling the method of “relative frequencies” is the one in which the investigator tallies characteristics or actions of a good many, maybe a great many people—the kind of investigation given the most space by far in most texts on method in social science. It is the method around which has grown up the immense body of technique and lore that includes sampling techniques, statistical inference, regression analysis, rationales about control groups, various classifications of study designs, arguments over quantitative versus qualitative, and all the rest. To clarify the sort of study—and the sort of thinking—in which the method of relative frequencies underlies the conclusions drawn, I will describe in the first part of this chapter some guises in which the method appears. After that, I will explain what I think are the proper uses of the method.

The method of relative frequencies is the preeminent method of finding or locating things, for finding a place or condition where one kind of characteristic or action occurs more frequently than at another, for finding the lode, so to speak, rich in the characteristic or action you want. What proportion of registered voters say they intend to vote for Corcoran? Is the proportion larger in Georgia than in Michigan? With what variable X can we more often find people high on Y by picking people who are high on X? Do we find a larger proportion of tall people, in comparison to short people, in positions of leadership? Will parents who beat their children be found in greater proportion among those who themselves were beaten as children? Suppose a person’s opinion is markedly different from the opinions of the others in a group; will we find that such persons are more likely (if we observe a hundred groups) to receive more communication than other members? What proportion of

Chapter Three    Limitations to Sampling

The method of relative frequencies does not always require sampling. If you want to find out how many members of the office staff want to hold the annual picnic at Jones Beach, simply ask every one of them. If you want to know how many of your students can recognize an aspidistra, put one in the classroom and ask every student to write down its name. If you publish a newsletter and want to know how many of your subscribers care enough about telling you their opinions of your newsletter to return a brief questionnaire to you, send a copy of the questionnaire to all of them.

Random sampling

That last example easily raises the question of sampling. The cost of printing and postage to send the questionnaire to all subscribers might be more than you want to spend. In that case, you might settle for a sample. Usually, when we turn to samples, we do so to save both time and money. Often, we want to learn something about very large populations indeed. That is the reason the Census Bureau came into being and then, in the twentieth century, the national polling organizations with their highly sophisticated procedures for random sampling.

Often, time is as much a difficulty as money. You might want to interview people living in sparsely populated areas about their automobiles. But by the time you would drive hundreds or even thousands of miles to interview those scattered people, having to return several times to catch some of them or wait several hours for some of them to come home, some of the people you interviewed first would

Part II    What the Method of Relative Frequencies Will Not Do

In Part I, I asserted that the method of relative frequencies was incapable of delivering the secrets of human nature—the manner in which the species, as a species, functions. In Part II, I will display several features of the method any one of which deprives it of that capability. I will display the features in connection with various sub-activities in social-science research: conducting experiments, interpreting correlations, using words with “subjects” (the people studied) and colleagues, and fine slicing, the latter being my name for what became known by the 1990s as “contingency theory.”

The method of relative frequencies is a metatheory. That is, it requires researchers to act as if they are making certain assumptions about human behavior. The chapters in Part II will pay a good deal of attention to these assumptions. I will show how researchers customarily use those assumptions to replace actual data. The point is the same as that made by E. F. Schumacher in the quotation that appears in the front of this book. The attention to assumptions will continue in Parts III and IV (see also “assumptions” in the index). In Part III, I will show how the linear or straight-line metatheory of causation underlying certain current uses of the method of relative frequencies also vitiates the method. There I will also argue that the circular metatheory is more suitable to studying human functioning.

Chapter Four    Experiments

I do not claim that very many of the ideas in this book are original with me. I hope only that I have brought together some good ideas so that their connection gives them fresh force. What I will call here the substitution of persons, for example, has worried many psychologists for a good many decades. Some psychologists have touched on it when discussing the nomothetic strategy versus the idiographic—that is, the strategy of studying differences among people in the same situation versus studying differences in the behavior of a single person across different situations. As long ago as 1974, Bem and Allen gave a good account of the arguments and the discomfiture. Good recent examples are the chapters by Franck and by Grossman in 1986.

The method of relative frequencies deals with statistics about anonymous people. It does not reveal who falls into some cell in a contingency table, but only how many. It does not reveal who gets a particular pair of scores on X and Y,but only how many people show similar ratios of X to Y. It does not reveal who gets a score above or below a mean, but only whether enough people have high enough scores to bring the mean to a point higher than some other mean. The method of relative frequencies never deals with any one person pursuing an individual purpose, but only with statistics (means, correlations, and so on) about abstracted aspects of behavior tallied over a collection of people.

Treating people as anonymous and interchangeable units works very well in casting nets, making catalogs, and writing histories. The strategy is entirely suitable to those purposes, since the only logic necessary is that of extrapolating statistics. When we try, however, to learn about the nature of the human from correlations and causal experiments, the method requires a string of shaky assumptions. Very shaky.

Chapter Five    Correlations

Since the method of relative frequencies always deals with numbers of people—and with distributions and averages of them—the assumption of the substitutability of persons (“subjects”) runs through every submethod and technique of analysis. It permeates the use of correlations in a very strange way.

Within the method of relative frequencies, we almost always think of a relation as a two-way distribution of people or other entities. We can portray the idea most simply with a four-celled contingency table. Each cell contains the number of people whose categories are those heading the row and column:

                                                      Y

                                Democrats         Republicans

        Easterners                40                      55

X

        Westerners                60                      35

All other relations are elaborations of that basic pattern—tables with more than two rows or columns, tables with more than two variables for categorizing, and even the several varieties of bivariate and multivariate correlation, in which many, even all, of the “cells” contain only one person. For convenience here, I will call all those distributions “correlations.”

Chapter Six    Using Words

Social scientists rely heavily on words. All scientists rely on words to report results and in the hope of influencing other scientists. But social scientists rely heavily on words when coaxing people to let themselves be observed or to answer questions; they rely on words when giving instructions to subjects; and they often rely on words to carry information from subject to researcher. Some studies can be done without words—for example, watching the movement of crowds from the top of a tall building or ascertaining the proportion of people who violate a sign reading “Do not enter.” But most studies rely on words. Here I will describe some difficulties and complexities in using words to get information about people. I will hang my disquisition on an example.

Example

Suppose we have a supervisor and five workers. The six of them might work in a factory, a school, an insurance company, an automobile repair shop, a city bus line, or some other kind of organization. Suppose the supervisor is finding that things are not going right. Things do not get done when he thinks they should. Sometimes workers do things not in their job descriptions. Workers make complaints that puzzle him. He cannot figure out what the workers really want.

The supervisor asks the workers what is troubling them. He can understand some of the things they tell him, other things he cannot. Some workers mention things that seem to him too small to matter to anybody. He has the feeling that

Chapter Seven    Fine Slicing

The underlying conception of causation in the method of relative frequencies has often been symbolized as S-O-R,for stimulus-organism-response, where the stimulus (S) is a cause and the response (R) is an effect. The conception is one of input causing output, as when you frighten (S) a cat (O) and the cat defecates (R) or as when you put a coin (S) into a vending machine (O) and it puts out a cup of coffee (R). I will discuss that conception of linear causation at greater length in the next chapter. Here, my point is that studying human action as if it can be understood in separate episodes or slices encourages researchers to believe they can understand more and more by slicing the episodes thinner and thinner.

With the theory (or metatheory) of S-O-R,researchers would be delighted to find that at one level of input variable X, all the observed people produce output behavior Y,and at another level of X,they do not. That delightfully simple hope is represented by the simplest contingency table:

                                  Y

                          No        Yes

          High           a           b

X

          Low             c           d

If that hope were borne out, all the data would fall into cells b and c. With more values of X and Y (more rows and columns), all the data would fall into the cells lying along the diagonal. With continuous variables, all the data would fall on a line; that

Part III    Causation

Along with its conception of independent and dependent variables, a metatheory of straight-line, episodic causation underlies the method of relative frequencies. That view of causation has served well in the physical sciences, though some physicists do worry about it.

Living creatures behave very differently from lifeless things. Unlike a rock, a human does not just sit until something bumps it. Writing that, it startles me that I feel it necessary to make a point of such an already pointed fact.

The method of specimens makes it much easier to examine the behavior of living creatures, where causes arise simultaneously from the environment and from within the creature. The internal processes by which living beings deal with those causes are ceaseless, seamless, and circular.

Chapter 8 will review linear, straight-line causation. Chapter 9 will prepare the way for the method of specimens by describing the interactions of internal and external causes.

Chapter Eight    Linear Causation

The most pervasive and fateful assumption underlying the great bulk of the writings in social science (and underlying most lay thinking, too) is the assumption of linear, straight-line causality. Not only is that assumption not the only available one, but it is the wrong one with which to learn how living creatures manage to keep living.

The underlying conception of causation in the method of relative frequencies has often been symbolized as S-O-R,for stimulus-organism-response, where the stimulus (S) is a cause and the response (R) is an effect. The conception is one of input causing output, as when you kick (S) a can (O) and the can sails (R) to a new position—and stays there—or as when you prick (S) a person (O) and the person cries (R) “Ouch!”

In the most naïve interpretation of the S-O-R,the conception is that the organism sits passively until some environmental event causes it to move. Then it does something, after which it sits waiting for another environmental event. On the face of it, that view of things seems too simple for anyone to believe. Nevertheless, you can find reports of studies in almost every issue of the professional journals of almost every social science that assume that simple scheme and no more. Furthermore, a great many practical people also hold to that simple theory. There are teachers who believe that students learn nothing until instructed to do so by teachers. There are managers who believe that workers will not work until prodded to do so by rewards and threats of punishment.

Chapter Nine    Specimens and Circular Causation

The key idea in circular causation is that internal causes and causes from the environment operate simultaneously, not sequentially, not in tandem or in episodes. The internal operations affect what will be perceived from the environment, and what is perceived affects what is done internally, and all that goes on continuously and seamlessly.

Circular causation implies a boundary separating the internal from the external, and that separation implies that the internal thing has characteristics that make it in some way a unitary thing with a characteristic way of functioning. In the next chapter, Figure 10.2 will show circular causation diagrammatically. There, however, circular causation will be mentioned only as a feature of a theory; here, I will argue the necessity of the concept.

Some of the features of the functioning of the human creature are the same in every member of humankind. All of the visible surface features of humans vary from person to person, and all of the visible actions of humans vary from moment to moment and person to person, but my claim is that when we look deeply enough into purposes and patterns, we will find some features of internal functioning that are the same in all of us. If that were not so, we could not transmit our humanness through genes. Furthermore, we can maintain our physical and operational integrity, in our constantly changing environment, only through the lively and immediate responsiveness that comes from making use of circular causation.

Part IV    What the Method of Specimens Will Do

In Parts I and II, I wrote about what the method of relative frequencies can and cannot do. It can locate densities of kinds of behavior. It cannot find out how a specimen of humankind functions—the rules by which it interacts with its environment. In Part III, I explained why I think circular causation, not linear causation, must provide the metatheory for the study of specimens.

In Chapter 10 of Part IV, I will describe a theory suitable for the study of specimens—a theory that rests on circular causation and the postulate that we act to control our perceptions. In Chapter 11, I will recount a few experiments that demonstrate how very well that theory serves to discover invariances. In Chapter 12, I will set forth some implications of the theory for social psychology.

Chapter Ten    Perceptual Control Theory

I have been lucky enough to come upon a theory and a growing body of experimentation that shows brilliantly what can be done with a theory firmly melded with the method of specimens. The theory is called “perceptual control theory” and is credited primarily to William T. Powers. I will cite his writings later.

Control as fact

Theory arises to explain observed facts. Control theory begins with the observation of behavior as control—that is, the observation that living creatures act to control what happens to them, that one of the consequences of any action is always the control of some energy the creature can sense. Examples abound. When you drive along a highway, you control the position of the car between the painted lines—a position you sense with your eyes. When you draw a horizontal line across a page, you control the distance you see between the line and the top or bottom edge of the paper. When you walk, you control the upright position of your body—a position you sense with your semicircular canals. When you take a sip of tea, you control the distance between cup and lip. As I type these words, I control the direction, momentum, and distance of the movements that I feel with my fingers.

Our bodies maintain their internal temperature within a very small range. We often act on the environment to aid our bodies in doing that. To feel warmer, we put on more clothing, go into heated buildings, eat more food, and so on.

Chapter Eleven    Testing Specimens

Here I will tell about four experiments using the method of specimens and designed as investigations of perceptual control theory. The first three test for visual invariants at the lower levels of the hierarchy. The last tests a hypothesis about an invariant at one of the highest levels—the maintenance of a self-concept.

What is the person doing?

Whatever people are doing, they are pursuing purposes and controlling perceptions to do so. Within control theory, indeed, “pursuing purposes” and “controlling perceptions” are almost synonymous. But researchers who do not habitually look for control of perceptions in behavior often have difficulty in believing it even when they see it. All experiments designed according to control theory exhibit the fact of control, but Powers devised a demonstration that is especially dramatic in showing that people control the perceptual consequences of their acts. People initiate acts and guide the progress of their actions, but only to control the perceptual consequences that result. As I write, Powers has not yet published an account of this demonstration, though he described it at the meeting of the American Society for Cybernetics in St. Gallen, Switzerland, in March of 1987.

Procedure

The subject sits before a computer screen with joystick in hand. Perhaps you have not operated a joystick. It is a small lever poised vertically and moored at the bottom

Chapter Twelve    Social Psychology

When I have written about behavior or theory in this book, I have usually written about the individual human. I brought in control theory because it illustrates so well what can be done with the method of specimens. If, following control theory, we take life’s constant activity to be that of maintaining internal standards against threats from outside, then social life looms large, because other people comprise a very large part of the environment of most modem humans. Sometimes other people obstruct our purposes; sometimes they give us help. And often both at the same time.

The study of social life now becomes the study of the ways in which two or more people, all acting to maintain their perceptions according to their internal standards (purposes, goals, interests, values, and so on), sometimes find them disturbed by other people and sometimes find other people useful in counteracting the disturbances. The study of interaction among people becomes the study of overlapping feedback loops.

Interaction

From the viewpoint of control theory, an event or condition in the environment affects behavior if (and only if) the event disturbs a perceptual input the person is maintaining—that is, controlling. Everybody else, therefore, is simply a part of the person’s environment and either a possible source of disturbance or a possible resource for opposing a disturbance. Sometimes another person is a help, sometimes a hindrance. That is not news.

Part V    Other Methods and Summary

I have devoted this book to two methods of research—two methods of getting information that helps us get ready for future experience. I have argued that most social scientists—and most of the rest of us, too—expect the method of relative frequencies to deliver a kind of information it cannot deliver. As the right alternative, I have described the method of specimens.

Those two methods cover a vast amount of information-getting. They do not, however, cover everything. I agree with Lindblom and Cohen’s (1979) book about Professional Social Inquiry. They tell how PSI (of which academic social science is a part) is only one of many ways of getting useful knowledge and can be no more than that. Also useful, they say, are ordinary knowledge, social learning, and interactive problem solving. But though I agree with them, I naturally prefer my own categories. In Chapter 13, I will describe action research and what I call the method of possibilities. Chapter 14 will summarize everything.

Please note that although the method of action research touches on the method of specimens, it does not deliver scientific findings in the sense of ascertaining invariances. I include a brief description of it in this book, along with the method of possibilities, to make more clear the domains in which we gather useful information about living creatures. On the one side, we gather useful information, often quantitative and systematic, but transitory and limited in scope, by the methods of relative frequencies, action research, and possibilities—and, if you wish, the methods described by Lindblom and Cohen. On the other side, we gather information about invariances throughout a species by the methods of specimens. And I will say once more that when I speak of information that is transitory and limited, I do not mean to belittle it. You and I, too, are transitory and limited.

Chapter Thirteen    Action Research and Possibilities

Most books on social science portray generalizing as turning one’s attention from a smaller number of people in a sample to a larger number of people in a population. That kind of generalization is useful in casting a net. If, however, we want information generated from research to be useful locally, with a particular organization, group, or person, then generalization must go the other way—from what is true of a lot of people to what is true of these few or this one.

Generalizing

Research, at least some research, should give us some information useful later, outside the research setting. It should help us get ready for future events.

Sometimes in research, formal or workaday, we observe a few instances and extrapolate from them. We get ready for a lot of later instances by supposing they will be similar to the lesser number of instances we have already experienced, as in observing a sample from a population. We reason, that is, from the particular to the general. At other times, we observe a lot of instances and generalize from them to the very next particular instance about to happen. So it is when we study trends in the stock market and then purchase particular stock, and when we study the scientific literature on managing organizations and then start up a series of “quality circles” in our own company. We go, that is, from the general to the particular.

Chapter Fourteen    Summary

Theory makes a crucial difference in method. I am not saying merely that theory affects the variables you will choose to control in an experiment or the questions you will choose to put in a questionnaire. I am saying that theory affects the very kinds of events and successions of them that you will believe to be possible and impossible in any investigation.

If you believe that stimuli are connected in some systematic, regular way to responses, inputs to outputs, if you believe that input conditions can by themselves cause output actions and that output actions do not simultaneously cause inputs, if you believe that people have qualities or traits or attitudes inside them such that “more” or the “stronger” of them make certain actions more likely (or more rapid or intense) or if you believe that stimuli have “strength” or “drive potential” that makes certain actions more likely or rapid or intense, and, in particular, if you believe that you can tell whether you have guessed right about your variables only by counting the people whose actions you predict correctly (almost as if you were counting votes in your favor), then you will be convinced of the logic of the method of relative frequencies.

If you believe that some people will find a certain environmental event disturbing, but others will not, if you believe that among those who act against the disturbance, some are acting to maintain one kind of input, some another, and some several, if you believe that internal variables and bodily action are connected not by correlations between the one and the other but by discrepancy between an internal preferred value or level and the value or level of an incoming perception, if you believe that people are never moved to a particular action by either internal motivation or by external

References

Aoki, Ted T. (1984). Interests, knowledge, and evaluation: Alternative curriculum evaluation orientations. In T. T. Aoki (Ed.), Curriculum evaluation in a new key. Edmonton, Alberta, Canada: Department of Secondary Education, Faculty of Education, University of Alberta.

Baum, W. M., Hayne W. Reese, and W. T. Powers (1973). Behaviorism and feedback control. (An exchange of letters.) Science,181(4105), 1116, 1118–20.

Bem, Daryl J., and Andrea Allen (1974). On predicting some of the people some of the time: The search for cross-situational consistencies in behavior. Psychological Review,81(6), 506–20.

Bohannan, Paul, W. T. Powers, and Mark Schoepfle (1974). Systems conflict in the learning alliance. In L. J. Stiles (Ed.), Theories for teaching,pp. 76–96. New York: Dodd, Mead.

Breger, L., and C. Ruiz (1966). The role of ego-defense in conformity. Journal of Social Psychology,69, 73–85.

Bronfenbrenner, Urie (1958). Socialization and social class through time and space. In E. E. Maccoby, T. M. Newcomb, and E. L. Hartley (Eds.), Readings in social psychology,pp. 400–424. New York: Holt, Rinehart, and Winston. Condensed in H. Proshansky and B. Seidenberg (Eds.), Basic studies in social psychology,pp. 349–65. New York: Holt, Rinehart, and Winston.

Brown, Daniel J. (1975). Mirror, mirror. . . . Down with the linear model. American Educational Research Journal,12(4), 491–505.

Brown, J. Larry (1987). Hunger in the U.S. Scientific American,256(2), 37–41.

Bullock, R. J., and Daniel J. Svyantek (1987). The impossibility of using random strategies to study the organization development process. Journal of Applied Behavioral Science 23(2),255–62.

Cairns, Robert B. (1986). Phenomena lost: Issues in the study of development. In Jaan Valsiner (Ed.), The individual subject and scientific psychology,pp. 97–111 New York: Plenum.

Chase, Ivan D. (1980). Social process and hierarchy formation in small groups: A comparative perspective. American Sociological Review,45(6), 905–24.

Name Index

Allen, Andrea, 33
Altman, Irwin, 80
Aoki, Ted T., 158
Baum, W. M., 119
Bell, Warren E., 173
Bem, Daryl J., 33
Bertalanffy, Ludwig von, 155
Bohannan, Paul, 119
Breger, L., 151
Bronfenbrenner, Urie, 15
Brown, Daniel J., 82-83
Brown, J. Larry, 11
Bullock, R. J., 22
Cairns, Robert B., 2
Chase, Ivan D., 182
Clark, R. K., 119
Cohen, David K., 165
Cook, Stuart W., 169
Corder-Bolz, Charles R., 52, 71
Cotton, John L., 11, 15
Cronbach, Lee J., 25-26
Crutchfield, Richard S., 163

Subject Index

Action research, 168, 184; examples, 171. See also Method of action research
Acts: highly predictable, 152; predicting particular, 142; requisites for, 147-51
Additive correlational models, 81-83
Assumptions, 34, 41, 49, 51-52, 57, 81, 179; in causal experiments, 34, 37; of control theory, 116-117; linear causality, 77; in method of specimens, 122, 126, 136, 139; of S-O-R, 78
Behavioral illusion, 104
Calibration, 70
Casting a net, 13
Cataloging, 14
Category, 113
Causation, 3; circular, 92-99, 126; cycles in, 80; linear, 77-84
Cause and effect, 92
Causes, simultaneous, 107
Change, statistical reliability of, 71
Compendium, eventual, 65-68