Theoretical analysis of the individual-group relationship in a social situation

Muzafer Sherif


I will take the liberty of introducing my analysis of the individual-group relationship in a social situation by mentioning briefly a few thoughts that were triggered when I first received Dr. DiRenzo's invitation to participate in this Symposium. It was stated that the theme of the Symposium was "Conceptual Definition in the Behavioral Sciences"—in its particular relevance to construction of theories and to some persistent philosophical problems. My first reaction was a feeling of trepidation that such an assignment was beyond my depth as an experimental social psychologist. Nevertheless, I gave serious thought to the all-important problem of theory and its relationship to actual research, as I have whenever, inevitably, I have encountered this problem in conceiving, formulating, and operationalizing the various research projects that I have carried out over the past thirty years. Once again I realized that every one of these projects had a theoretical orientation which defined the problem and guided the selection of appropriate methods and techniques for data collection. Illustration of this common thread running through all the projects will lead to the main theme of this paper.

The Relationship Between Theory and Research

Our research in the mid-1930's, for example, on the experimental creation of social norms, utilizing the autokinetic phenomenon, was not conceived through concern over the psychology of vision, though vision was involved.[1] Rather, it was conceived in part on the basis of theoretical leads derived from Durkheim's conceptualization of représentations collectives, or norms, as emerging in interaction situations that are fluid, out of the ordinary, and not encountered in the humdrum of daily routine; and, for the other part, it was conceived from empirical facts and theoretical orientations derived from the fruitful line of research of the Chicago school of sociology under the seminal inspiration of Robert E. Park. The Chicago studies showed in different ways that individuals interacting under conditions that

(48) lack stable anchorages for experience do generate their own norms and that these norms become binding for the individuals over a time span.

The series of experiments on intergroup conflict and cooperation also had a theoretical orientation. They started with the necessity of defining the human group. Such a definition was needed because the then prevailing theories dealt with aggressive, pugnacious, and competitive intergroup behavior, without viewing such behavior within the framework of the membership character of the individuals within their respective groups.[2]

The same is true of a series of studies dealing with attitude assessment and attitude change, some of which were summarized in a joint book with the late Carl I. Hovland, entitled Social Judgment, [3] and also in our recent Attitude and Attitude Change.[4] This line of research, which is still continuing in several current projects, arose from the need for moving toward definitions that would be in line with an established body of research evidence on social perception and social judgment. Prior to this research, the dominant assumptions in the attitude-measurement field had been extrapolated from experiments on psycho-physical judgments devoid of motivational-affective components.[5]

Currently, our research program is confronted with problems of basic definitions on a wider scope; this is particularly the case with our ongoing research on natural groups in differentiated socio-cultural settings.[6] This research involves problems of the group, the place of individual members in it, and specification of the characteristics of the social situation and setting in which a group and its members operate. Therefore, in the last analysis, participating in this seminar with the particular topic of this paper does not leave me with the feeling of uncertainty about plunging into terra incognita.

In order to give concreteness, I have deliberately used my own research experiences as cases in point in order to articulate the relationship between conceptual orientation involving definitions and actual research. (This is perfectly legitimate, since the problem of this relationship is not unique to one researcher; it is representative of all those who devote themselves to research as a

( 49) life work in social science.) For example, the point that my research illustrated concerning the relationship of theoretical orientation and actual research practice involving data collection is, as will become evident, fundamental for the discussion of my topic—analysis of the individual—group relationship in a social situation.

What stood out in this illustrative case can be summarized in a few sentences. The question of whether there is need for theory and conceptual orientation, or whether what is needed is more down-to-earth research unhampered by theoretical problems of the past, is both futile and wasteful. The pious contention "Let the facts speak for themselves" is an absurdity which cannot occur in actual practice, even in the case of empiricism in the raw. Facts are there, all right, whether they are studied or not. This is one thing. But it is altogether another thing to claim that facts speak for themselves. Without exception, there is always a selective process in the collection of data. The span of apprehension or observation has a limited scope—as any standard experimental psychology text, such as Woodworth's, will report, as the work of psychologists from William James, to Oswald Külpe, to Dwight Chapman will verify, and as is proven again and again in the psychology of testimony from Binet and Münsterberg to the present day.

For example, in data collection through observation or interview, the decisions about who will constitute the sample and the time period to be studied immediately determine, in part, the nature of data that will be obtained. Nor is the case of data collection in the laboratory an exception to the general rule of selectivity in the processing of what kind of data will be obtained. By the time the decisions are made about the kind of experimental set-up—subjects, procedures, and techniques selected from so many possible ones; the units of data determined; and units of analysis selected—a great deal of the game already is played. And all of these decisions presuppose the selectivity of the experimenter in deciding what belongs to the domain of the

study, what is to be excluded, and into what categories events will be sliced.

( 50)

Whether or not the researcher in the field or in the laboratory is conscious of the process of selectivity in which he is immersed, these inevitably selective processes in the conduct of research, experimental or otherwise, presuppose assumptions as to the nature of the problem as well as definitions of the universe of discourse (categorizing it into what belongs and what is to be excluded). Whether we are conscious of it or not, research is always guided by assumptions about the nature of the problem at hand and by conceptual orientations which amount to explicit or implicit definitions of the universe of discourse. In short, there is no escape from theoretical and definitional entanglements. This being the case, it will be more productive and fruitful for research if the conceptual or definitional bounds that the researcher utilized are made explicit. What is called theorizing, I suspect, involves this sort of activity, among other things.

On the other hand, the inevitable involvement of theorizing is not justification for unbridled and unfounded speculation. Especially when there exists an accumulation of empirical data—as is the case in the social sciences today—with literally thousands of research reports pouring in from the laboratory and field, it is the responsibility of the theorizer or systematist to take stock of the available data, to restrain his theorizing to that which he can legitimately advance, and to square his theorizing with the facts at hand, scattered though they may be. Of course, there is charm and even elegance in grand theorizing. But if it is speculative activity unmindful of the checks from accumulating data, one may find it preferable to satisfy himself with charm and elegance at their best in the works of such masters as Plato and Hobbes.

Nor are the proponents of the current craze for "models" based on the prestige of more established physical and biological sciences exempt from the responsibility to take account of empirical data. The uncritical and hasty extrapolation of models from the physical sciences is not different in character from the unbridled and grandiose speculation against which I have just raised a caution. To be sure, models are needed to formalize a discipline, to give it bounds, and to save it from an aimless

( 51) proliferation of empirical facts in all directions. To be sure, precise and quantitative indices are essential in the process of formalization. And, of course, it is in line with the parsimony of formulation to have, for example, a few quantitative judgments as indices of the status hierarchy of an organization, instead of piles and piles of descriptive data. While it is thus undeniably true that formalization in terms of an internally coherent and consistent model and discovery of short-cut and precise indices (as the hairline of a needle indicates an electric charge) are necessary and desirable goals, nevertheless uncritical adoption of ready-made models without checking their assumptions against accumulated data is not the path that will lead to these goals.

First of all, there is the fundamental problem of the isomorphism between assumptions of the model and the properties of the universe of discourse it purports to represent and to formalize. In the past, there were serious attempts to use models from more established sciences. The model of "mental chemistry" of Wundt and Titchener in the early history of experimental psychology, which attempted to define clearly the psychological counterparts of chemical elements in the form of sensation and image units, utilizing refined instruments, is an example of the maladaptive use of a model that had at the time proved useful in another field. Probably the organic analogy of Herbert Spencer and his followers, whose works would fill a small library, is another case in point. In my opinion, the extrapolations from the so-called dynamic equilibrium models of fluid mechanics or thermodynamics, stemming from the works of Pareto, will suffer a similar fate.

Unfortunately, the problem of isomorphism between assumptions of a model and the properties of the domain it purports to handle is of only minor concern to those psychologists and social scientists who are caught up in the excitement of current fads. It is ironic that some mathematicians regard the problem of isomorphism as basic in scientific activity. Thus, in the very first chapter of a mathematics text for college freshmen, we find the following statement as the first principle of model-building:

(52) "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."[7]

The opposite sequence, much more frequent in current attempts at formalization in the social sciences, is to find a model that has been successful in one of the more established sciences and to transfer it, often uncritically, to some problem area in the social sciences. In an invited address in 1955 to the sixty-third annual convention of the American Psychological Association, the physicist Robert Oppenheimer warned of the pitfalls of the latter sequence of science by analogy in the following words:

between sciences of very different character, the direct formal analogies in their structure are not too likely to be helpful. Certainly what the pseudo-Newtonians did with sociology was a laughable affair; and similar things have been done with mechanical notions of how psychological phenomena are to be explained. I know that when physicists enter biology their first ideas of how things work are indescribably naive and mechanical; they are how things would work if the physicist were making them work, but not how they work in life. I know that when I hear the word "field" used in physics and in psychology I have a nervousness that I cannot entirely account for. I think that, especially when we compare subjects in which ideas of coding, of the transfer of information, or ideas of purpose, are inherent and natural, with subjects in which they are not inherent and natural, that formal analogies have to be taken with very great caution.[8]

Further, Oppenheimer noted a possibility that makes the uncritical extrapolation of a model from the physical sciences an even greater exercise in futility: ".. , it seems to me that the worst of all possible misunderstandings would be that psychology be influenced to model itself after a physics which is not there anymore, which has been quite outdated." [9]

When model-building proceeds without concern for the problem of isomorphism, it may quickly degenerate into a game with

( 53) no objective criterion for deciding when victory has been attained, much less one for whether it was a worthwhile game to play. An observation to this effect comes from a gratifying source—the presidential address to the Society of Engineering Psychologists by A. Chapanis at the American Psychological Association Convention in 1960. In the course of his paper on "Men, Machines, and Models," he commented:

Models are too often not validated . . . or, if attempts are made to validate them, the procedures used are scientifically valueless. . . . I am sometimes frankly appalled by the faith which some model builders have in their own powers of analytical and synthetic reasoning when it comes to making models of human behavior. . . . I will state my bias on this score in no uncertain terms: I will gladly exchange one hundred well-informed guesses at any time for the results of one carefully executed experiment.[10]

"My final criticism of model building," Chapanis concluded,

is that the modeler often becomes so intrigued with the formulation of his models that he constructs them for essentially trivial problems. Having at one's disposal a large electronic computing machine, for example, invites one to try out all kinds of things, because computers are such fun to play with. Considering the state of knowledge within psychology, however, the easiest problems to build models for are essentially unimportant problems.[11]

I have gone into the problems of formal model-building by analogy with the physical sciences because it is a very general problem in the social sciences and psychology and because the effects of uncritical analogy with mathematics and physical sciences have hampered exploration in certain problem areas. One such problem area is attitude research. Specification of measurement units and quantification of data are, I repeat, goals in every science. But they are not ends in themselves. As Oppenheimer said so clearly: "It is not always tactful to try to quantify; it is not always clear that by measuring one has found something very

( 54) much worth measuring . . . and I would make this very strong plea for pluralism with regard to methods that, in the necessarily early stages of sorting out an immensely vast experience, may be fruitful and may be helpful."[12]

Attitude research in this country proceeded for over thirty years as though its major problem were simply the refinement of measurement models based on analogy with the equal-interval scales derived in psycho-physical judgment of neutral stimuli (weights, lengths, and so forth) or, more recently, with the cumulative scales so common in measurement of physical properties of objects. Consequently, the plentiful results from attitude studies had very little to offer toward clarifying theory, or even toward defining the properties of the attitudes that made their study significant for social science in the contemporary world, where one group is constantly competing with another to change attitudes in its direction. On the contrary, the whole effort of attitude research was directed toward assigning a single score to an individual so that his "attitude" could be ranked relative to others.

In association with colleagues and students, I have spent fifteen years trying to break through the mass of convention and sterile assumptions in attitude research. The results of this work are presented in our book on attitudes and ego-involvements.[13] Here, I can indicate only a few of the outcomes.

(1) The assumptions of equal-interval scaling, developed by Thurstone, can he satisfied provided that the individuals do not have strong attitudes—in other words, provided that we ignore the very problem that we are supposed to be studying.

(2) The assumptions of a cumulative scale, as stated in the Guttman model, can be satisfied by taking a very limited range of positions on an issue, and by ignoring others actually upheld by interested groups in society. If we do not choose to put on such blinders, however, then we find that many social dimensions are simply not "cumulative." It is true that six inches will always be included in a twelve-inch measurement, but it is not even typical for an individual who upholds an ardently pro-segregationist

( 55) stand to accept a mildly pro-segregationist statement. Similarly, devoted Republicans or Democrats do not, typically, accept a mild statement of support for their side; they are more likely to reject it. These are phenomena that simply cannot be investigated with a "cumulative" scale, because they violate its assumptions.

(3) A single score representing the positions acceptable to an individual is a most inadequate indicator of his attitude. Nonetheless, this single score is what the prevailing models, on the whole, seek to attain. Much greater predictability of actual behavior can be obtained if we assess an individual's attitude in terms of the range or latitude of positions he accepts, rejects, and toward which he remains non-committal. These findings have led us to define attitude in terms of a set of evaluative categories slicing the relevant stimulus world into a latitude of acceptance, a latitude of rejection, and a latitude of non-commitment. As it turns out, the latitudes of rejection and non-commitment are even more useful in predicting behavior than the single score of acceptance or even the latitude of acceptance. These measurement concepts were not derived from a pre-existing model, but from laborious study of actualities of psycho-social scales in social life.

As Einstein and Infeld stated in discussing major advances in physics, and as Wolfgang Köhler stated in discussing needed developments in psychology: the first, major step toward a breakthrough in any science is a result of formulating pertinent and significant problems, not in thin air, but on the basis of the actualities of the universe with which we are dealing. Operationally, the issue boils down to evaluating the isomorphism between the facts in a universe of discourse and the theoretical scheme or model we propose, as well as the appropriateness of the techniques that we use in gathering data to evaluate our theory.

No model, no theoretical scheme is "right" or "correct" in its own right, no matter how much fun it is nor how intellectually intriguing. No procedure and no technique for data collection are powerful or effective in their own right. The theory should be the guide for fruitful research. The techniques are powerful tools for data collection, if-and only if—they are appropriate in

(56) terms of the nature and characteristics of the problem. And significant problems can he formulated only after gaining substantial familiarity with the universe of discourse, and not before.

An Analysis of the Social Situation

Now I shall pursue this discussion by asking a question fundamental to the entire problem of the individual-group relationship, a question whose answer necessarily determines the conception of that relationship: "What constitutes the social situation for the individual, represented in his interpersonal and group relations?"

One of the most thriving activities in social psychology today is what passes under the name of "small group research." The more I study this thriving activity and the resulting host of publications, the more convinced I become that they are doomed to be a collection of disjointed and incoherent artifacts so long as they are not related to a framework of fundamental orientations concerning significant and persistent problems. An analysis of what constitutes a social situation is certainly among these fundamental orientations. Because of the absence of a unifying orientation, established generalizations in social psychology today are only a tiny fraction of the total output of publications from the universities and the military and industrial establishments, whereas there are hundreds of studies published that are labeled as studies of two-person, three-person, or n-person groups, studies of decision-making, and studies of coalition formation.

These are, of course, perfectly legitimate topics of study. But can they be studied adequately without first raising the question, "What is the social situation?" For example, is it just that the number of individuals varies—and so we have dyads, or triads, or quartets? Experiments have shown that the number of people is an important factor, but it is not the only significant factor. More important for the behavioral outcome may be such considerations as the prevailing relationships among the individuals as friends, or as enemies, or as collaborators for mutual benefit in a temporary crisis situation. The degree of ego-involvement that individuals

(57) have in the problem or task at hand can also be more important than the number of individuals participating.

To take another illustration: is a social situation characterized by the task or problem at hand, for example, as it is introduced by an experimenter in the laboratory? Can the experimenter even adequately describe the task or problem as an isolated factor? Here the evidence is contradictory and even conflicting, as one might expect. The task or problem introduced is always a task or problem for individuals with particular experiences, particular involvements, and particular skills.

In the study of any social situation—whether the problem is the effects of the number of individuals, whether the problem is the setting or the task, or whether it is the decision-making process—the factors under study have to be taken as parts of the social situation at hand. The need for considering different aspects of a social situation as part-processes within their appropriate framework is a problem of cardinal significance and not only in social psychology. It is also of cardinal importance in the study of all psychological problems, including those traditionally considered most basic and most elemental, such as psychophysics. The point was illustrated well by Harry Helson in his address in Philadelphia in 1963 upon receiving the APA Distinguished Scientific Contribution Award:

Let me begin by recalling an episode in the early 1930's when I gave a demonstration before the Optical Society of America of the inadequacy of the CIE (Commission Internationale de l'Eclairage) method of color specification. In this demonstration, although the stimulus qua stimulus did not change with change in surround, its color could be made anything we pleased by appropriate choice of the luminance and hue of the background color. In the discussion that followed, the late Selig Hecht, perhaps the leading worker in visual science at that time, arose and said: "Why do you complicate the problems of color vision by introducing background effects? Why can't you wait until

( 58)

we have solved the simpler problems before we go on to the more complicated ones? "[14]

Helson continued his address with a list of subsequent discoveries in color vision that were possible only because he did not wait for the solution of so-called "basic" problems, but continued investigating the problems that he had formulated as basic—the problems of background and illumination in color vision. This work led him to develop a "frame of reference psycho-physics" with a fruitful line of experiments.

Helson's point is of great relevance to our present discussion. Let us pose the question: If there is a necessity, as demonstrated by Helson and his co-workers, for studying visual perception of a single patch of color within the context of its background, how much more compelling is the same principle in studying social behavior. How much more important it becomes to consider the social situation for a two-person or n-person group beyond the sheer presence of the two persons or the n-persons.

What is the background and the stimulus context for individual behavior in a social situation? Obviously the most concrete aspect is the presence of other individuals in face-to-face relations, and for some social psychologists, as we have said, their domain of study consists only of the study of person-to-person relations in face-to-face interaction. Of course these are important.

However, persons cannot be studied in isolation. At no time are persons in a socio-cultural vacuum, even in their moments of greatest intimacy. They have peculiar customs of inhabiting certain places and having certain histories. If we restrict our concept of the social situation to consist only of other persons here and now, we commit a fundamental error. There is a long tradition in experimental psychology that we specify the independent variables or experimental conditions. We take it for granted that an experiment on vision or audition will specify the light and sound frequencies presented, their intensity, and whether the experimental set-up permitted other light and sound frequencies to enter. We would severely criticize an experiment

( 59) on learning as a function of drive and reward that did not specify how many hours the organisms were deprived and the quantity and kind of reward to be found in the goal box, as well as the dimensions of the paths leading to the goal object.

Applying the same standards, how can we possibly say that we, as social psychologists, are interested only in "personality" variables and that this is all we shall consider? We may temper this statement by saying that "sociologists" or "anthropologists" will study the rest of it, and that we are sophisticated enough to know that there are "variations." This amounts to saying: "I know that the behavior I am studying is affected by influences that are not apparent to my naked eye nor detected by my instruments and which I, as a researcher, am not controlling. But I am only interested in what I see and detect here and now." Unfortunately, much of the theory and research on social interaction among different combinations of individuals is about as nearsighted as this statement suggests. It is usually justified as being a study of "basic" features of personality or "pure" interaction process.

Unfortunately, however, personality is not an isolated phenomenon; "interaction" is not "pure process" devoid of content, devoid of context and background. Person-to-person interaction must be placed in a context and background, in turn. Interpersonal relations, among friends, or lovers, or enemies, do not take place in a vacuum. They take place in a restaurant, a bedroom, or an office, or a church, or a convention hall, or on a city street; or they take place in a psychological laboratory. In recent years, our attention has been called to the fact that the subject's appraisal of a situation and the setting can affect the outcome even of a psychological experiment.[15] In any culture and in any specific situation, every social situation forms a pattern or context for individual behavior. What is being done to specify this context in contemporary psychology?

One of the noteworthy developments is the simple awareness of the variety of factors that may affect social behavior. Saul Sells has emphasized the point by his attempt to list the stimulus factors that may contribute to behavior variance under general

( 60) headings; he has tried, in other words, to indicate the possibility of a taxonomy of the stimulus situation. His published volume of papers on this topic, The Stimulus Determinants of Behavior,[16] provided, in my opinion, a sorely needed contribution to our field. But the sheer hulk of his listings is sufficient to show us that a "taxonomy" in the form of a syllabus is not a solution to the problem, valuable as it may be as a methodological device. At latest count, there were over 250 factors in his list. I do not think that his purpose in compiling it was to suggest that we should permute these 250-odd factors in factorial research designs.

Therefore, instead of a syllabus, I venture to repeat a simple classification of sets of factors that, as sets, enter into any social situation, even though particular sub-sets and their members may be present or absent in different cultures, and even though the range of the sets may differ in different cultures (which, in fact, it does). Here are the sets of factors in brief form.

(1) The set of factors pertaining to individuals who participate in the social situation. These include:

(a) The characteristics of the individuals, such as the number of persons, their ages, their sex, their educational, occupational, economic, and social attainments. (b) The composition of the total participants in the social situation in terms of their similarities and differences in age, sex, homogeneity and heterogeneity as to religion, class, and so on.

(c) Relations among the participating individuals. Are they strangers, friends, rivals, and in what combinations? To what extent are their relations stable or subject to change? This is a crucial sub-set of factors. To dramatize it, we may note that it makes a big difference whether a distribution of opinions—say, 40 per cent pro, 50 per cent con, and 10 per cent undecided—represents a collection of unrelated individuals or the member ship of two well-organized groups and potential adherents.[17]

(2) The set of factors pertaining to the task, problem, or activity at hand. Man's activities are, after all, of some importance to the study of his behavior. Is his task new or familiar, simple or

( 61) complex, habitual or calling for creative efforts? Is it structured or unstructured in some degree in the structured-unstructured gradation?

(3) The set of factors pertaining to the setting and the circumstances surrounding it. These include the place, the material culture of that place, the objects and tools available, the facilities, the presence or absence of other people not involved in the task or problem at hand, and notably they must include the cultural and value orientations of the setting. There are, after all, appropriate settings for work, for problem-solving, for religious conversions, and for romance. These cannot be defined apart from a background of cultural values and status and role relationships.

(4) The set of factors pertaining to each individual participant's particular relation to the above three sets of factors. These include, among other things, his proficiency in the task or problem at hand, the degree of his enduring involvement in the problem, his attitudes toward other participants, his feelings of ease or discomfort in the situation, and so on.

One way of attempting to deal with these four sets of factors would be to vary the components of each set. If you are prepared to undertake this venture, you also should be aware that the relative weights or contribution of each to the behavioral outcome may vary in the following ways:

(1) The relative weights of different sets of factors may vary in different cultures. In fact, in different cultures the number of different factors and the range of different factors do vary.

(2) Within any given culture—including within the microculture of the psychological laboratory—the weights of different factors may vary. Thus, the task may be very important if the individuals are strangers and unfamiliar with the task; on the other hand, it may be a minor detail if they are comfortable companions and if the task is within the range of their experience.

(3) Finally, the relative weights, or significance, of the individuals participating, of the task or problem, and of the setting in which events take place vary according to the temporal sequence of events. This is a problem that we, as social psychologists, have

(62) left to the developmental area. But it is equally crucial in terms of the development of interpersonal, intergroup, and intercultural relationships. What may be crucial today—the old patterns of interaction and the old conceptions of tasks—may be "old hat" tomorrow. The problems of getting acquainted among strangers, for example, may give way to the efforts of interested parties and rivals to gain the day. Unlike "pure interaction" or "pure psychological process," the patterning of social life, and thus the patterning of the social stimulus situations for its participants, do not stay put with fixed and immutable weights.

I have, quite deliberately, concentrated so far on the various sets of factors and their variable weightings in social situations, because the overwhelming tendency among psychologists has been to over-simplify the context in which the individual-group relationship takes place. The traditional and sharp dichotomy between individual versus group appears hopelessly naive when we engage in the effort of analyzing the context of the social situation.

However, I should not like to leave the impression that the variability of the social situation means that no regularities are exhibited, nor the impression that prediction of behavior is therefore impossible. On the contrary, the social sciences have already, even in their infancy, demonstrated that certain variables .are persistently weighty ones and that knowledge about these variables has striking predictive value for the student of individual behavior. Here I shall mention only two such variables and associated concepts. One pertains to the first set of factors that I have listed, namely the individual's place or position relative to others who, perceptually or conceptually, count for him. The second refers to factors included in the third set, especially the characteristics of the setting and its socio-cultural background.

Let me illustrate the first variable concretely. When several strangers meet, by coincidence or circumstance, their presence does affect the behavior of each in some way, no matter what the task or setting. A thousand times over, it has been demonstrated in the psychological laboratory that the mere presence of other

( 63) persons does affect experience and behavior in definite and measurable ways.

Now, let us define the presence of several individuals who are unacquainted and who are not related by any past history of association as a togetherness situation. Think now of what happens over a period of time if these individuals are faced by some common problem, some dilemma, or seek a common goal that requires their joint participation. We have replicated this situation experimentally a number of times, and the results are in accord with numerous field studies. Over a period of time, relationships develop among these former strangers. And, if the problem or activity is of any consequence at all to the individuals, they do develop modes of procedure, customs, private jokes, and other shared products which are, to some extent, respected and even cherished by each individual.

In short, under these circumstances over a period of time, a human group will take shape. When it does, the properties of the group provide a powerful basis for predicting the behavior of individual members. But, suppose that you are doing research starting with unacquainted individuals, and that you must decide when you have a group, and not just a togetherness situation in which the variations among individuals, in the task, and in the settings produce profound shifts in behavior. How can you decide when you have a group?

We must have a definition, and the definition must specify certain units for assessing whether the definition is met. Cutting a very long story of study and research very short, let me give you a definition of a group that does resemble groups in real life, and that does help to predict the behavior of individual members: a group is a social unit that consists of a number of individuals who (1) at a given time, stand in status and role relationships to one another, stabilized in some degree; and who (2) possess a set of values or norms of their own, regulating the behavior of individual members, at least in matters of consequence to the group.

Status positions in a group can be measured in terms of observed differences among the individuals with respect to the

( 64) effectiveness of their initiative over a time span. Role relationships are more complex, and require a composite measure relative to the activities of the individual and the personal contributions of members in interaction. The second major property, values or norms, can be measured by defining the limits within which behavior is acceptable to group members, and beyond which some form of disapproval or punishment occurs—in short, by observing when sanctions are invoked.

To sum up, the organizational and normative aspects of a human group are, invariably, weighty determinants of individual behavior. Knowledge of them does permit prediction of individual attitude and action. Together, these organizational and normative aspects of a human group, along with any tools and procedures that it may develop, constitute its culture. These aspects are analytically separable, but they are all part-processes in the group's culture. The separation of the organizational and cultural, or the social and non-social aspects of culture, is tenable only for purposes of closer analysis of part-processes.

The second, and more comprehensive, set of factors that lend regularity to the social situation, and individual behavior within it, pertains to the socio-cultural background. Since Professor White will inform us on its patterned properties in a later chapter, I shall not dwell on them. Instead, I shall simply emphasize that social organization, cultural values, and the man-made arrangements of the material environment do have distinctive properties that must be included in any adequate account of a social situation.

But let me articulate further one of the points just made. It is the social structure, the organization or social system of the group, and also it is the norm or value system which, together with other group products, constitute the culture of the group. Both social organization and norm systems, along with the other material and non-material products, are of the group. If we subscribe to the notion that sociology is primarily concerned with organization or social systems, and cultural anthropology with other cultural products, it will be puzzling to make sense of

( 65) one of the major preoccupations of cultural anthropologists, which has been with kinship systems.

In social psychology, there is a widespread assumption that any influences of the socio-cultural setting are revealed in immediate situations, here and now, through the words and actions of individuals. This may ultimately prove true, but certainly not unless the investigator is prepared to study socio-cultural influences and has a prior grasp of their patterned properties.

The attempt to boil down the patterns of social organization and cultural products to items of individual behavior is one form of "reductionism." It is still strong in psychology and in the larger trend in social sciences which adopts, by preference, the label "behavioral sciences." The reduction of social organization and culture to a question of individual behavior is, ultimately, an untenable solipsism. By the same logic, everything in science could be reduced to psychology, including chemistry and physics. To be sure, if there were no human beings to interact, there would be no culture, no social organization, no means of communication or transportation or production. But once such social products come into existence, they do become significant parts of the social situation in which any individual functions. Man makes machines; we can also say that machines, in turn, make man. Man creates social organization; we can also say that social organization recasts man. Man is the beginning of all of these social things, but his products are not man himself. His products do have properties that are patterned and distinctive and that cannot be studied with sole reference to the behavior of single individuals.

This account of what constitutes a social situation, and of the significance of its organizational and normative aspects, suggests a sequence of study that is most appropriate for assessing the individual-group relationship. If we do not know anything about the perceptual and conceptual relationships among individuals in a social situation, we are like blind men trying to understand their actions. In brief, the first—not the last—step in research is specifying the properties of social organizations and their cultural

(66) patterns, within which the situation is a part-process. Then, and only then, are we in a position to specify the importance of the immediate situation and the individuals within it, including their uniquely personal characteristics, in as great detail and shading as possible.

Analysis of the Factors in a Social Situation

With such questions in mind as, "What constitutes a social stimulus situation?" we initiated in 1958 a project to study the behavior of individuals as a function of their group membership in differentiated socio-cultural-socio-cconomic settings. We have reported the formulation and summary of the first five years of this research in a book entitled Reference Groups: Exploration into Conformity and Deviation o f Adolescents."' Here, of course, I can only hope to indicate in a general way how the various sets of factors composing a social situation have been translated into research operations.

(1) First, different socio-cultural settings are selected. Here we encounter problems still not solved in social science. How "low" is "lower class" and how "high" is "upper class"? What do these designations mean for individuals growing up and living their lives in these settings? We have used indicators developed by sociologists (particularly Eshref Shevky and Wendell Bell) that can specify the rank of urban areas as low, middle, or high in terms of socio-economic level, urbanization, and ethnic composition. These indicators have the great advantage that they permit comparisons in terms of gradation of low and high, rather than merely arbitrary divisions into lower, middle, and upper ranks. In addition, we have studied the settings through local data and block surveys, in order to specify what it means to live, work, be educated, and play in a setting of a particular social rank.

(2) Within the areas designated as low, middle, or high social rank—in terms of indicators that are specified and can be checked —we have surveyed the pervading patterns of values and goals among representative samples of high-school boys who are similar to the boys of high-school age whose behaviors are to be

(67) studied intensively over a period of time. Therefore, the indices of value orientations-labeled "Self-radius" and "Goals" indices—are obtained from the normal high-school age-mates in the same urban areas as the members of the natural groups of boys, whom we study intensively.

These Self-radius and Goal indices of the representative group serve as the baseline against which the self-picture, the attitudes, and behaviors of the members of the intensively studied group are evaluated.

(3) Within each of the areas, the behaviors of individual members belonging to groups of their own choosing are studied intensively over a period of time ranging from six months to a year. These groups are reference groups for the individuals. The intensive study of small groups is carried out under field conditions, and it is the focal aspect in the study. Yet it is fruitless and meaningless without the background and context of the ecological and socio-cultural situation specified in the previous two phases of the research design.

In studying behavior as a function of membership in informal groups, and as a function of groups in different settings, we have developed procedures to avoid arousing the individual's awareness of himself as being a research "subject." At the same time, through careful specification of a sequence of observation, through introducing independent observers, and through adapting measurement techniques to the field situations, we have found it possible to combine the validity of facts in our universe of discourse with the precision and reliability associated with the psychological laboratory. The study focuses, successively, on organizational and normative properties of the group and then upon particular individuals within it.

This research program attempts to integrate: (1) the study of the characteristics of the socio-cultural setting; (2) the study of prevailing values of the comparable age-mate population in the same neighborhood, which serves as a baseline for evaluating the behavior of particular individuals; and (3) the intensive study of behaviors and interaction processes among individuals within groups chosen by the members. This intensive study involves

( 68) specifying the characteristics of each interaction episode, and through them, the characteristics of the particular groups.

This multi-faceted research program exemplifies the integrative research orientations and methods that can be derived from the analysis of the facets of the social situation.

Concluding Remarks

I began my analysis of the individual-group relationship in a social situation by pointing out that an active researcher cannot help seeking conceptual and theoretical orientations to guide his research operations. This led to a consideration of the relationship between theory and research, with emphasis on the all-important problem of isomorphism between theoretical analysis and the domain of events that analysis purports to treat.

What kind of theorizing serves to provide effective conceptual tools? What kind is ineffective and futile in the conduct of fruitful research? These questions led to a brief consideration of some current practices in theorizing and model-building.

Then, an analysis of the individual-group relationship proceeded by placing the analysis within the framework of the social situation. The problem of what constitutes a social situation is a sorely neglected one. Yet experimental psychology definitely has established that a single item of individual experience or behavior has very little significance unless it is considered as part of the context and background within which it takes place. As the context and background of individual behavior, the social situation was analyzed in terms of interdependent sets of factors. Among these, the presence or absence of stabilized reciprocities among individuals, conceived in gradations from mere togetherness to group situations, and the socio-cultural setting within which interaction takes place, were singled out for special articulation.

The analysis of the social situation has inescapable implications for psychology, sociology, and interdisciplinary efforts—which should supplement one another instead of being a source of bickering over the prerogatives of the various social sciences.


Articulation of the sets of factors in the social situation points to a definite and orderly sequence in the study of individual-group relationships. It eliminates a sharp dichotomy between individual and group.

The conceptualization of a social situation provides for due emphasis on the socio-cultural setting for interaction. It provides for the analysis of groups as groups, without falling into the pitfall of "reductionism" which is prevalent, on the whole, in what are given the blanket label of "behavioral sciences." At the same time, the analysis places proper emphasis on the unique individual or personality characteristics, which can be studied in as fine detail of idiosyncratic factors and their shadings as can be desired by proponents of personality nuances.

The sets of factors that should be included in analysis of the social situation (and that therefore constitute the framework of the individual-group relationship) were indicated in broad outlines, making place for socio-cultural, group, and idiosyncratic personal factors. The last section of the paper was devoted to a brief summary of our research program since 1958, whose aim is to operationalize the theoretical analysis of the individual-group relationship in a multi-faceted research design and in concrete research procedures carried out in successive phases.


  1. Muzafer Sherif, The Psychology o f Social Norms (New York: Harper and Brothers, 1936; and Harper Torchbook Series, 1966).
  2. Muzafer Sherif and Carolyn W. Sherif, Groups in Harmony and Tension (New York: Harper and Brothers, 1953); Muzafer Sherif, B. J. White, and O. J. Harvey, "Status in Experimentally Produced Groups," American Journal of Sociology, LX (1955), 370-379; Muzafer Sherif, O. J. Harvey, B. J. White, W. R. Hood, and Carolyn W. Sherif, Intergroup Conflict and Cooperation: The Robbers Cave Experiment (Norman, Oklahoma: University Book Exchange, 1961).
  3. Muzafer Sherif and Carl I. Hovland, Social Judgment (New Haven: Yale University Press, 1961) paperback edition, 1965.
  4. Carolyn W. Sherif, Muzafer Sherif, and R. E. Nebergall, Attitude and Attitude Change: The Social Judgment-Involvement Approach (Philadelphia: W. B. Saunders Company, 1965).
  5. L. L. Thurstone and E. J. Chave, The Measurement o f Attitude (Chicago: University of Chicago Press, 1929); Carl I. Hovland and Muzafer Sherif, "Judgmental Phenomena and Scales of Attitude Measurement: Item Displacement in Thurstone Scales," Journal of Abnormal and Social Psychology, LIV (1952), 257-261; and Muzafer Sherif and Carl I. Hovland, "Judgmental Phenomena and Scales of Attitude Measurement: Placement of Items with Individual Choice of Number of Categories," Journal of Abnormal and Social Psychology, XLVIII (1953), 135-141.
  6. Muzafer Sherif and Carolyn W. Sherif, Reference Groups: Exploration into Conformity and Deviation of Adolescents (New York: Harper and Row, 1964).
  7. C. B. Allendoerfer and C. O. Oakley, Fundamentals o f Freshman Mathematics (New York: McGraw-Hill Book Company, 1959), p. 19.
  8. R. Oppenheimer, "Analogy in Science," American Psychologist, XI (1956), 133-134.
  9. Ibid., p. 134.
  10. A. Chapanis, "Men, Machines, and Models," American Psychologist, XVI (1961), 130.
  11. Ibid., pp. 130-131.
  12. Oppenheimer, op. cit., p. 135.
  13. Sherif, Sherif, and Nebergall, op. cit.
  14. Harry Helson, "Current Trends and Issues in Adaptation Level Theory," American Psychologist, XIX (1964), 26.
  15. M. T. Orne, "On the Social Psychology of the Psychological Experiment: With Particular Reference to Demand Characteristics and Their Implications," American Psychologist, XVII (1962), 776783; R. Rosenthal, "On the Social Psychology of the Psychological Experiment: With Particular Reference to Experimenter Bias," paper read at the 1961 Annual Convention, American Psychological Association, New York City; and B. L. Kintz, D. J. Delprats, D. R. Mettee, C. E. Persons, and R. H. Schappe, "The Experimenter Effect," Psychological Bulletin, LXIII (1965), 223-232.
  16. B. Sells (ed.), The Stimulus Determinants of Behavior (New York: The Ronald Press, 1963).
  17. Muzafer Sherif and B. Koslin, "The `Behavioral' vs. `Institutional' Controversy in Social Science with Special Reference to Political Science" (Norman, Oklahoma: Institute of Group Relations, University of Oklahoma, 1960). Mimeographed.
  18. Sherif and Sherif, Reference Groups: Exploration into Conformity and Deviation of Adolescents.



Allendoerfer, C. B., and Oakley, C. O. Fundamentals of Freshman Mathematics. New York: McGraw-Hill Book Company, 1959.

Chapanis, A. "Men, Machines, and Models," American Psychologist (1961), 16:113-131.

Helson, Harry. "Current Trends and Issues in Adaptation Level Theory," American Psychologist (1964), 19:26.

Hovland, Carl I., and Sherif, Muzafer. "Judgmental Phenomena and Scales of Attitude Measurement: Item Displacement in Thurstone Scales," Journal of Abnormal and Social Psychology (1952), 54:257261.

Kintz, B. L., Delprats, D. J., Mettee, D. R., Persons, C. E., and Schappe, R. H. "The Experimenter Effect," Psychological Bulletin (1965),63:223-232.

Oppenheimer, R. "Analogy in Science," American Psychologist (1956), 11:127-135.

Orne, M. T. "On the Social Psychology of the Psychological Experiment: With Particular Reference to Demand Characteristics and Their Implications," American Psychologist (1962), 17:776-783.

Rosenthal, R. "On the Social Psychology of the Psychological Experiment: With Particular Reference to Experimenter Bias." Paper read at the 1961 Annual Convention, American Psychological Association, New York City.

Sells, S. B. The Stimulus Determinants of Behavior. New York: The Ronald Press, 1963.

Sherif, Carolyn W., Sherif, Muzafer, and Nebergall, R. E. Attitude and Attitude Change: The Social Judgment-Involvement Approach. Philadelphia: W. B. Saunders Company, 1965.

Sherif, Muzafer. The Psychology of Social Norms. New York: Harper and Brothers, 1936; Harper Torchbook Series, 1966.

———, Harvey, O. J., White, B. J., Hood, W. R., and Sherif, Carolyn W. Intergroup Conflict and Cooperation: The Robbers Cave Experiment. Norman, Oklahoma: University Book Exchange, 1961.

———, and Hovland, Carl I. "Judgmental Phenomena and Scales of Attitude Measurement: Placement of Items with Individual Choice of Number of Categories," Journal of Abnormal and Social Psychology (1953), 48:135-141.

———— and ————. Social Judgment. New Haven: Yale University Press, 1961.

———and Koslin, B. "The `Behavioral' vs. `Institutional' Controversy in Social Science with Special Reference to Political Science," Norman, Oklahoma: Institute of Group Relations, University of Oklahoma, 1960 (mimeographed).

————, and Sherif, Carolyn W. Groups in Harmony and Tension. New York: Harper and Brothers, 1953.

————, and ————. Reference Groups: Exploration into Conformity and Deviation o f Adolescents. New York: Harper and Row, 1964.

————, White, B. J., and Harvey, O. J. "Status in Experimentally Produced Groups," American Journal of Sociology (1955), 60:370-379.

Thurstone, L. L., and Chave, E. J. The Measurement of Attitude. Chicago: University of Chicago Press, 1929.



Valid HTML 4.01 Strict Valid CSS2