Social Interaction

Process and Products

Muzafer Sherif

• Publisher: Routledge
• Year: 2017
• eBook ISBN: 9781351490269
• Topic: Psychology
• Subtopic: History & Theory in Psychology
• Index: Psychology

Reflecting the many contributions of Muzafer Sherif to social psychology during the past thirty years, this volume presents selections from among Sherif's most widely known essays and provides a systematic overview of his evolving interests, concepts, methods and research findings. Twenty-five essays are divided into five sections according to content; the theoretical and methodological problems at the heart of Sherif's work; the experimental model for interaction process and products; problems of self and reference groups; concepts, attitudes and ego-involvements; and contributions to problems of in-group and intergroup relations through experimental and field research. Though the selections range over a broad spectrum each is characterized by the precise and incisive work techniques Sherif devised as well as by its intrinsic relevance to significant issues. Sherif writes to clarify theory, to define conceptual tools, and to use tools and theory to demonstrate the substantive results of his researches. Each research finding is added to its predecessors as the author advances to his goal of a social psychology that is consistent as it moves from the most basic psychological processes to the complexities of individual involvement in collective activity

Part I

Interdisciplinary Relations and Methodology

One

If Basic Research is To have Bearing on Actualities . . .

My title is somewhat unorthodox. Therefore, I shall start with a few words of explanation.

Logically, “if” should not be in the title. Of course, what is called basic research should have bearing on the domain of events it attempts to investigate and predict. What is called applied should be application of some principles or generalizations that are solidly based on research findings. In fact, the justification for calling research basic is that its findings are the basis for generalization. Logically and in practice, what is called basic and what is called applied should be closely related aspects of the scientific process.

Ironically, the “if” has to be included in the title because there are strong trends in behavioral science toward divorcing the “basic” and the “applied.” Professional meetings and professional journals can almost convince a person that there is a cold war between those engaged in basic research and those in applied fields—when they are not ignoring each other completely. At times, it appears that the twain will never meet.

On the one hand, some psychologists and social scientists in applied fields move ahead to secure data or to settle this or that practical problem, without great concern over how their prescriptions are related to what people are doing as basic research. On the other hand, some representatives of what is called basic or pure research tend toward aloofness from events outside of their laboratory or research center. On both sides, therefore, the essential interplay between basic research and application is blocked. Applied effort ceases to be the application of anything related to scientific study and becomes hand-to-mouth improvisation. Basic research is deprived of feedback from events in real life and stops trying to make predictions about actualities. Yet, in the long run, the interdependence of what is basic and what is applied is essential to the progress of any science.

The stakes are high in the controversy: whether behavioral science is to develop to a point where its theory has predictive value. By this time, you probably recognize that the question I am raising is the much neglected problem of the validity of research findings and generalizations based on them. On the whole, the all important problem of validity has been put on the shelf, despite pious lip service to the contrary. The problem of validity has been merged with the criterion of reliability, to the point where we are in danger of being reliably wrong in our generalizations and predictions. This is why it is important for us to raise the question of how basic research can have bearing on actualities.

In the rest of this paper, I shall try to make what I have said so far more concrete and then suggest some pointers for insuring that basic research is both reliable and valid. Let me begin by considering representative practices in what is called applied research.

A few years ago, I was in a large Washington hotel during the national convention of candy wholesalers. In addition to the free candy bars left at their doors each night, the other hotel guests were offered copies of the association’s official publication. The issue included elaborate analyses of survey data on the public’s taste in candy, on preferred color of the wrappings, and on the kind of people who consume the most candy. The quantity of data and the sophistication of the analyses far exceeded those of research typically reported in psychological journals. Yet all the sophisticated analyses yielded were the facts that some people consume more candy than others and prefer certain packages to others. This may be useful information for some immediate commercial purpose. But the “why’s,” the “how’s,” and the consequences of these preferences remained almost untouched.

Much of today’s research into public opinion and consumer preference has precisely this nature. Nothing is applied except some trappings of the trade—cookbook rules for asking questions that are not “leading,” established sampling methods, and data processing by computers—which convince almost everyone that something scientific is going on. Unfortunately, the use of these tools does not guarantee that the product will have anything to do with behavioral science or its application.

This state of applied activities is not confined to survey research. Some clinicians are convinced that the effectiveness of clinical practice rests entirely upon the personality of the therapist. By their definition, clinical practice is not application of anything based on research. And, I should add, social psychologists are not exempt from the tendency to rush into application—for example, on such matters as leadership and productivity—with little concern over what they are applying.

The claim of behavioral science to a place under the scientific sun must rest on conduct of basic research, not on ventures into practical affairs of the moment. The development of any science presumes research activities that are more than dated studies of immediate practical concerns. Louis Pasteur’s studies for the French wine makers would not be remembered outside the wine industry if he had confined his interest only to spoiling wine or, for that matter, if bacteria flourished only in such a spirited environment.

What is the distinctive character of research defined as “basic”? Formally, it starts with a problem for study that defines the domain of events to be investigated, including the essential independent variables. The problem defines the domain of study in a generalized way. The concepts used to define the problem are generalizations representing the major variables and are based on initial familiarity with the problem area at hand and prior research findings. Units of analysis, appropriate to the variables in question, are defined. Measurement techniques are selected to secure indicators of these variables. A plan or design is formulated to include the major variables, controlling some and varying others in the combinations necessary to eliminate alternative explanations of the results. Data are collected by procedures that can be communicated to others and reproduced by them. Finally, statistical analysis and inference are used to evaluate the data.

Clearly, basic research involves abstraction from actualities at every step. There can be no escape from the abstraction process, nor should there be. In abstraction lies the power of science to predict a great variety of apparently discrete and unrelated events. Our question here is what kind of abstractions give this power. Abstraction becomes a game if divorced from actualities; it becomes abstraction for its own sake or for the impression it may make on one’s colleagues in the profession. In this case, abstraction becomes inner gymnastics for a select group of people who are “in” on the secret and exclusive lingo.

The history of psychology includes many graveyards filled with abstractions that were insulated from actualities. Let me give a few examples, necessarily in short-cut and somewhat dramatized form. Here are some generalizations that have been proposed seriously on the basis of what is called basic research.

Frustration always leads to aggression. Among other things, this generalization was used to explain why poor white southerners indulged in lynching Negroes. Unless “frustration” is a completely abstract concept, it does not explain why the even more deprived Negroes seldom behaved aggressively toward whites during the same period.

People change their attitudes more in response to a communication opposite to their own than to a view nearer their own stand, particularly when the matter is of considerable personal importance to them. If this generalization bears on actualities, it is very difficult to see why, for example, we find no mass defections from liberal groups to the Ku Klux Klan nor from White Citizens Councils to the civil rights groups.

Judgments of value statements are not influenced by the attitudes of persons doing the judging. This assumption was considered basic to scaling attitudes by the method of equal-appearing intervals. In fact, judgment is influenced by relevant attitudes, as shown, for example, by judgments made by civil rights workers and by white supremacists of a moderate position on the desegregation issue. To the civil rights worker, a moderate position is pro-segregation; to the white supremacist, it is support for the desegregation movement. For the civil rights worker, the pace of desegregation is too slow; for the white supremacist, the pace is too fast.

I have deliberately chosen examples from social psychology. They could have been selected with equal ease from other research areas in behavioral science. What is called “pure” research has at times been so pure that it omitted essential independent variables that operate in real life and hence had very little bearing on events outside of the particular laboratory. We must ask, therefore, how pure we want to be. Do we want to be so pure that our generalizations are violated every day by events in life going on around us? Is basic research to be merely a stunt to impress those who are part of the “in crowd”? Can we afford to refuse the feedback from actualities?

When basic research becomes aloof from the correctives that can be derived from such feedback, the risk we take is a risk to basic research itself. We do not have to think that behavioral science has a duty to participate in the solution of social problems or even to insure the future of its application. Without close familiarity with actualities at the outset, basic research risks leaving out some of the essential variables or of taking them so much for granted that they are never included in theory.

In recent years, more researchers have become seriously concerned about the state of affairs that tends to produce artifacts instead of valid results and generalizations. Among the correctives in this respect is the experimental orientation represented under the title, “the social psychology of the psychological experiment.” This movement is among those shaking the complacent assumption that the only variables that count in shaping the response are those immediately salient to the naked eye.

Some researchers have only recently discovered variables that would have been essential in their study designs had they not been indoctrinated to view the laboratory as an insulated environment, free of influences that pervade the simplest social situation. Evidence is rolling in to show that the laboratory, too, is a social situation, no matter how circumscribed the set-up is. The important variables need not be confined to those the experimenter happens to choose and writes up in his procedures, without due regard to the laboratory situation, his own role in it, and others who may be there.

Martin Orne of the University of Pennsylvania has shown that the laboratory experiment has potent “demand characteristics” involving differential authority and responsibility for experimenter and subjects. Like any people in any social situation, research subjects have expectations about the laboratory and about the experimenter. Experiments by J. D. Frank and by Orne have shown that research subjects will engage without question in boring and meaningless activities (such as adding random numbers and systematically destroying each sheet of addition) to the point of boring the experimenter.

Subjects have considerable faith in the experimenter as a responsible person and place value upon their own contribution to the scientific effort. Thus, most of Orne’s subjects who were instructed to simulate hypnosis quite willingly reached for a poisonous snake, plunged their hand into a bowl of acid, and threw the acid at the experimenter when instructed to do so. Stevenson and his co-workers succeeded in having adult subjects perform the ridiculous task of placing marbles into holes at rates beyond their endurance to maintain—simply because it was an experiment. Over 80 per cent of Stanley MilgranTs subjects administered a level of shock labeled highly dangerous, merely because they had been paid as subjects and the experimenter ordered them to do so.

It should not be thought, however, that the research subject is merely reacting passively to the experimenters instructions. On the contrary, as persons in any social situation will do, subjects try to appraise what the researcher is up to, and they differ in the accuracy of their assessments. As Stevenson and his co-workers have shown, a task that may appear to the experimenter as extremely simple may become highly complex for the subject who has an overly sophisticated hunch about the purpose of the procedures. Dahlke and his students have recently shown that the spread of affective value from an emotionally toned word to a neutral word in paired-associate learning occurred only among those subjects who correctly sized up the experimenters intent and also desired to appear as a “good subject” in his eyes. Those who did not catch on to the basis for pairing, and those who were annoyed at the idea of conforming to his intent, did not increase their ratings of the affective value of neutral words.

MilgranTs research, referred to earlier, demonstrates clearly the complexity of variables in a laboratory situation. The level of shock administered by subjects and the proportion of subjects refusing to administer high levels decreased with the distance between the subjects, according to whether the experimenter was physically present or communicating by telephone and even according to whether the experiment was performed in the Yale laboratory or in an office building in Bridgeport.

Finally, the experimenter himself and his own hypotheses about the outcome undeniably affect the results he obtains, as Rosenthal has shown in a series of experiments. A few years ago, Rosenthal thought that the solution to this problem was to randomize assignment of experimenters, none of whom knew the correct hypothesis. This suggestion has been discarded in his recent writing, because even uninformed experimenters seem to develop their own hunches, right or wrong.

There are, of course, some ways to circumvent these problems of experimenter bias and of the demand character of research situations. We have been incorporating the most obvious of these into our research since 1947: the use of a combination of independent measures obtained by different techniques to minimize bias and the development of research designs that do not alert the subject to the fact that his behavior is being studied.

The studies of the psychological experiment as a social situation, of which there are a good many, point to a larger lesson that should concern us seriously when we define problems for research. These show that it is impossible to escape actualities, even in the laboratory. Actualities, in fact, constitute the backdrop or context for the operation of those variables that the investigator has singled out for experimental manipulation. True, the context of the laboratory represents actualities of a particular kind, not necessarily representative of many other social contexts. However, the logical conclusion is that our definition of research problems, concepts, and research designs must include specification of the variables in this context, not arbitrarily isolated from it. This is an enormous task, but necessary if findings and generalizations based on them are to have points of contact with actualities outside of the laboratory.

The prevailing intellectual heritage has provided a ready-made escape route from this huge task. Following the models of earlier work in chemistry and physiology, some psychologists concluded that the essence of science was to reduce complex phenomena to their simplest units and that efforts should be concentrated on study of these elemental units. It was further assumed, of course, that the elements would be combined, but meanwhile, basic research meant study of the simplest possible units of analysis. This was the model for the great founders of experimental psychology, including Wundt and Titchener with their “mental chemistry” model, and for the early behaviorists with their model of a punctiform stimulus and response, as Arthur Melton (a behavioristic psychologist himself) aptly characterized it.

There need be no quarrel at all with those who wish to make it their life work to study the nerve synapse rather than the nervous system, visual discrimination rather than perception of the visual field, or the learning of words rather than sentences. It must be emphasized, however, that in each of these cases, the simpler unit is in no sense more basic than the more complex, if our aim is prediction of behavior in the actualities of a social context. The properties of a pattern are as real as the elements whose relationships compose the pattern, as shown by experimental work by Gestalt psychologists, by Bartlett, and by Helson, among others. What is basic is defined by the problem at hand, not necessarily by what is arbitrarily chosen as elemental.

In his recent revision of A Textbook of Psychology, the experimentalist Donald Hebb made the point so clearly that I shall quote him at some length. Near the end of a book offered as a basic text without a single chapter on an applied topic, Hebb discussed the same problem of what are to be considered basic units of analysis, in relation to psychology and physiology.

It seems on occasion to be thought that neurological entities are somehow more substantial, more “real,” than psychological entities: that the study of nerve impulses is a more scientific affair than the study of anxiety or motivation. This is entirely mistaken. It may be that the “probable error” of a psychological conception is larger than that of the neural conceptions of anatomy and physiology; our conceptions, that is, may need more revision and sharpening, but they are not less related to reality. The wood is as real as the trees; a shower of rain as much an entity as the drops that compose it. There must be different levels of analysis in natural science, from the microscopic (or submicroscopic) to the large-scale macroscopic. At any given level, “reality” consists of the unanalyzed units whose existence is taken for granted as the basis for analyzing the next higher level of complexity. Otherwise we should have to deny the reality of the raindrop as well as of the shower, for the drop is “only” a group of molecules, and such reasoning would lead us to the ultimate conclusion that the only fit objects for scientific discourse are the subatomic particles of nuclear physics—this page would not exist as an entity, nor would the student who is now reading it (Hebb, 1966, pp. 319-20).

If Hebb’s conclusion is correct—and I have been supporting such a view strongly for years—then our task is to consider how those behavioral scientists who are interested in human perception, learning, problem-solving, and decision-making in social contexts can select units of analysis and design research that do have a bearing on the actualities of human behavior and human relations. There is no escape from the first step.

The concern at the outset should be close familiarity with the events in the problem area as well as with techniques for studying them. Ironically, the need for close familiarity and accurate description of actualities is sometimes recognized more clearly by people in the more established sciences than by those in behavioral science. For example, in one of the basic texts in college mathematics, Allendoerfer and Oakley (1959) state the first principle of model building as follows:

The first step in the study of any branch of science is that of observing nature. When enough facts have been collected, the scientist begins to organize them into some pattern (p. 19).

The plea for the importance of first-hand familiarity through observation and descrip...

Table of contents

• Cover
• Title Page
• Copyright Page
• Contents
• Introduction to Aldine Transaction Edition
• Introduction
• Part I: INTERDISCIPLINARY RELATIONS AND METHODOLOGY
• Part II: EXPERIMENTAL MODELS FOR SOCIAL INTERACTION
• Part III: THE SELF AND REFERENCE GROUPS
• Part IV: CONCEPTS, ATTITUDES, AND EGO-INVOLVEMENT
• Part V: EXPERIMENTAL AND FIELD RESEARCH: MAN IN IN-GROUP AND INTERGROUP RELATIONS
• Works Cited
• Index