views updated Jun 08 2018


I. The Study of GroupsGeorge Caspar Homans


II. Group BehaviorMorton Deutsch


III. Group FormationMuzafer Sherif and Carolyn W. Sherif


IV. Role StructureA. Paul Hare


V. Group PerformanceBertram H. Raven



In social science the study of groups is often called small-group research. A “group” is defined as a number of persons, or members, each of whom, while the group is meeting, interacts with every other, or is able to do so, or can at least take personal cognizance of every other. This requirement, which cannot be met for larger social units, such as armies, justifies calling groups “small.” Although no investigator has specified the number of mem bers that would make a group “large” rather than “small,” the groups actually studied have seldom had more than fifty members. It is, moreover, often difficult to draw the boundaries between groups, as marginal persons may interact with members of two or more groups.

Investigators working within the field of small groups seldom study the behavior of groups as such. More often they study the characteristics of face-to-face interaction among men, of which the behavior of groups considered as units is only a part. Accordingly, the subject should perhaps be called “elementary social behavior” (Homans 1961) rather than “small groups.” A group so small that an investigator can observe the behavior of each of its members in some detail is a good setting for the study of this subject, but the group usually remains the setting rather than the object of investigation.

For this reason the problem of defining a group or demarcating its boundaries is seldom of theoretical importance. When the behavior of a group as such is to be studied, its boundaries, in practice, are apt to be quite clear.

History of small-group research

The interest in groups has sprung from two main sources, which may be called psychological and sociological after the disciplinary affiliations of the chief investigators. In psychology the interest perhaps began with studies of social facilitation. From the 1930s on, it was stimulated by the experi mental and theoretical work of Kurt Lewin on social influence. Lewin trained a group of students, including R. Lippitt, D. Cartwright, A. Bavelas, and L. Festinger (Cartwright & Zander 1953; compare Lewin 1939-1947). Members of the Tavistock Institute in England largely shared their viewpoint. Although many psychologists in the field were not students of Lewin’s, few escaped his influence.

In sociology much of the work of Georg Simmel (1902–1917) lay in the field now called small-group research. But his influence was not direct: indeed, later research had the effect of reviving interest in him. The other early name of impor tance is that of Charles H. Cooley, who pointed out the influence of certain kinds of groups, such as the family and the children’s play group, which he called primary groups, in forming the attitudes of men (Cooley 1909).

Simmel and Cooley were intuitive observers and theorists. For empirical investigation, Elton Mayo was to sociology what Lewin was to psychology. The researches carried out in the 1920s in the Hawthorne plant of the Western Electric Company by Mayo and his associates of the Harvard Gradu ate School of Business Administration were conceived as studies of human relations in industry (Roethlisberger & Dickson 1939). But several of the researches put small industrial groups under close scrutiny and suggested the possible fruitfulness of more general work. E. D. Chappie and C. M. Arensberg (1940), who developed operational measures of face-to-face interaction, were indirectly associated with the Western Electric re searches; so were sociologists like W. F. Whyte (1943) and G. C. Homans (1950), who have worked in the small-group field, as well as investi gators at the Harvard Business School. Yet there is plenty of such work in sociology that does not stem directly from the school of Mayo.

A third, rather specialized, influence in the 1930s was that of J. L. Moreno (1934) and his associate, H. H. Jennings, who worked at the New York State Training School for Girls. They contributed interesting empirical findings and, more important, the technique they called the sociometric test, which in one form or another is now used by most investigators.

Great Britain, France, and the Scandinavian countries have also made important contributions to the field; in France an early theoretical influence was the “microsociology” of G. Gurvitch (1950).


One justification for speaking of psychological and sociological “schools” is that until recently they tended to use somewhat different research designs. True to the traditions of their discipline, the psychologists emphasized experimental and survey methods. They would bring groups, sometimes groups artificially constituted for research purposes, into laboratories where some of the variables affecting the subjects’ behavior might be held constant and others experimentally manipulated. They developed quantitative measures of behavior and administered standardized questionnaires; they formulated hypotheses relating the independent, experimentally manipulated variables to the dependent ones; they tested the significance of these hypotheses statistically. They also exploited natural situations that provided some approximation to experimental control—for instance, a set of groups all alike in some respect, such as residents of houses of identical design or groups having identical tasks to perform. The psychologists would correlate statistically the variations along a number of dimensions of behavior occurring naturally in these groups.

Instead of studying several similar groups under more or less controlled conditions, the sociologists were apt to concentrate on single, “real-life” groups. Instead of using systematic quantitative and survey methods, they would use the techniques of anthropological field work and nondirective interviewing, which provided, as they hoped, a richer and more rounded understanding. And instead of testing hypotheses statistically, they would work out a quali tative description of the social structure of the group in question.

It is easy to exaggerate the differences between the psychological and sociological methodologies. Although the Western Electric researches are thought of as belonging to the sociological tradition, the study of the Relay Assembly Test Room employed a more rigorous method of measuring output, gathered a greater body of quantitative data, and made a more elaborate statistical analysis than any other small-group study has ever done (Whitehead 1938). Today all investigators use in different mixtures much the same battery of techniques. The sociologists are ready to use survey methods in order to provide support for hypotheses first suggested by field work; and the psychologists will admit both that field work is a necessary preliminary to the design of a survey and that it helps the investigator understand what the findings of his survey “mean”—that is, how they are to be explained. Most investigators now think of them selves as social psychologists rather than as either psychologists or sociologists.

Above all, both methodologies lead to the same end. The first business of any science is to state propositions, or hypotheses, about the relations between variables. The psychologists tend to make their propositions explicit and to pretend they were formulated before the data were gathered. The sociologists are more apt to pretend that they studied their groups with open minds, ready for any relationship to appear in the data, and they often leave their propositions implicit in their descriptions of social structure. But their propositions can always be made explicit, and a proposition supported by data remains a finding, whether formu lated exante facto or expost facto. Nor is it easy to understand why one should put greater confidence in a proposition tested on a set of groups similar in some controlled respect than in one that can be shown, as many of the findings of field research can be shown, to hold good for a number of otherwise quite dissimilar groups. In any event, no one has ever demonstrated that the findings of, for instance, laboratory research are inconsistent with those of field research, once the very different conditions that obtain for artificial and for “real” groups are taken into account. They are not always taken into sufficient account.

Some leading problems

Within the field of small groups numerous problems call for investigation, and there have been a number of surveys and anthologies of research (Cartwright & Zander 1953; Hare 1962; Hare et al. 1955; Olmsted 1959). Here only the chief questions to which research has sought answers are mentioned.

What determines whether a man changes his behavior or opinion under the influence of others? In particular, what determines whether or not he conforms to the norms of a group? What gives some men power over others?

Small-group research deals with all these questions within the context of a general concern about interpersonal relations, which may often be regarded as stabilized resultants of processes of influence. Accordingly, the researcher in this area, starting with a broad interest in the relations between the behavior and opinions of men and their attitudes toward one another, may come to ask more specific questions. Here the ideas that have come to be called “balance theory” (Heider 1958) should be specially mentioned. What are the conditions of status congruence and distributive justice, and what are the effects on interpersonal relations of failure to meet these conditions? What determines an attitude of “respect” for another, rather than a more relaxed “liking”? What determines embarrassment or a “joking relationship”?

Interpersonal relations combine to form larger social structures; consideration of this elementary social fact raises a new order of problems. How do coalitions form in the smallest social groups, such as triads? How do differences in status build up among the members of a group? What are the determinants of different forms of status system? What are the relations between status, influence, and conformity, or between status, the channels of communication in a group, and its division of labor? What are the determinants of conflict between groups, and what are its effects on social structure within groups? Finally, there is the question of social control: What are the conditions maintaining stability in a social structure?

Many studies have been devoted to questions arising from collaboration among group members in attaining a common goal. Is a group more effective than an individual in solving problems? What are the results of competition as opposed to collaboration? How do the motivation of its members, their interpersonal relations, patterns of communication, and division of labor render a group more or less effective in attaining a goal? Are there any necessary phases a group goes through in solving its problems and reaching its goals? What are the effects of success or failure upon its internal structure? What are the determinants of individual satisfaction with collaborative work?

There are also a number of special questions related to what English-speaking people call leader ship. A “leader” may be defined as someone whose orders are in fact obeyed by many other persons, especially by the members of a group to which he belongs. What, if any, are the traits of personality that make a man a leader? What are the relations between the values of members and the behavior of the man who emerges as leader of a group? What is the relation between status and leadership, or between leadership informally won and authority formally assigned to a man by a larger organization of which his group is a part? What makes a leader successful in attaining individual goals or goals assigned to, or adopted by, a group? What are the effects of strong leadership on the other behavior of the leader himself and on the behavior of others toward him?

Many of these questions lend themselves to applied research, which is research designed to provide people with knowledge that might allow them to change social behavior for the better from their point of view. The “better” behavior envisaged has ranged all the way from more productive industrial groups to discussion groups better organized to train the leaders of other discussion groups. In this field, research has been none the worse for being “applied” [seeLeadership].

Codification of research findings

The findings, the tested empirical propositions of small-group research, possess the following characteristics. In substance, they are very varied. They are formulated in many different terminologies: almost every investigator makes up one of his own. And they often appear to have no high degree of generality, to be contradictory or to hold good only in special circumstances.

Since, nevertheless, the findings are all concerned with face-to-face social behavior, their chaotic nature in other respects has stimulated a few investigators to the kind of criticism sometimes called codification (Homans 1950; 1961; Thibaut & Kelley 1959). When carried out deliberately, which it seldom is, codification goes behind the different names given to the variables entering the findings and asks how these variables were actually measured. When the same or equivalent operations of measurement were used in different in vestigations, codification gives these variables the same name. By this means it can sometimes show that the same findings have been reached in different investigations and so reduce the number of findings.

When apparently contradictory findings survive such criticism, codification goes on to examine the conditions under which the findings were established, since two findings are not contradictory unless all the circumstances in which each holds good are identical. If one research, for instance, shows that two persons who interact often with each other tend to like each other and another shows that they do not, an examination of the other features of the two researches may reveal that in the latter the two persons were constrained to interact, whereas in the former they were not.

These first steps in codification immediately suggest the next: an effort to reduce the number of independent findings still further by showing that a number of different findings, including contradictory ones, can be derived from the same set of general propositions under specified given conditions. From an assumption, for instance, about the determinants of sociometric choice (expressions of liking), together with the condition, which is approximately true of some groups, that members use the same criteria for choice, it follows both that a few members will receive many more choices than others do—which is an empirical finding—and that highly chosen members will choose one another— which is another empirical finding.

At this point codification begins to organize and unify the findings by explaining them. In any science, what is to be explained is always an empirical proposition of the general form x varies as y. Explanation is the process of showing that the prop osition follows logically in a deductive system from more general propositions under specified given conditions, themselves stated in the form of propositions. The more general propositions are more general in the sense that they enter into deductive systems by which other empirical propositions are explained; this is their unifying function. The given conditions are not general propositions, nor does the investigator undertake to explain them; he simply accepts them as matters of fact. A characteristic given condition in small-group research is the physical layout of the room in which a group meets.

To explain a phenomenon is to produce a theory of the phenomenon: a theory is nothing if it is not an explanation. Many investigators, especially psychologists, begin their research reports by formu lating a theory from which, they assert, the propositions (hypotheses) they tested were derived. Codification begins by disregarding these theories and looking at the empirical findings themselves. But it must come back in the end to asking how they are to be explained. For this purpose it need not come back to the investigators’ own theories. A theory may be fruitful in suggesting hypotheses to be tested without in the long run providing the most general way of explaining the hypotheses [seeScientific Explanation].

The uses of theory

The process of codification almost forces the codifier to ask himself what his general theory is. Two main types of general theory have been used, not always explicitly, in small-group research. They may be called sociological and psychological, but here these words refer to the characteristics of their general propositions and not to the academic affiliations of the men who use them, since a sociologist may perfectly well use a psychological the ory, and vice versa.

Sociological theory

The sociological type of theory is an offshoot of the “functional” theory with which the name of Talcott Parsons is particularly associated. It is sociological because its general propositions are propositions about social units as such and not about the behavior of individuals. It attempts to state the general conditions under which any social system survives, maintains itself, or remains in equilibrium. From these general conditions it claims to derive the particular features a given social system remaining in equilibrium “must” possess. The empirical findings represent these features.

In small-group research a good example of such a theory is that of R. F. Bales (1950). He argues that if a discussion group is to remain in equilibrium and reach its goal, such as the formulation of a decision, the behavior of the members must collectively display certain characteristics. For instance, the group must pass through certain characteristic phases in discussion, or its members must develop a certain distribution of attitudes of respect and liking for one another. Like other functional theories, Bales’s derives the features of individual behavior from the necessities of group equilibrium, and not vice versa [seeInteraction, article onInteraction Process Analysis].

The difficulties with functional theories of small groups are those of functional theories in general. No social scientist has ever succeeded in formulating a rigorous statement of the general conditions of social equilibrium. The statements actually used are so vague that nothing definite can be derived from them. Moreover, many social groups do not remain in equilibrium, however it is defined, so that the range of phenomena the theory could explain must at best be limited. Finally, it is not at all clear that another type of theory cannot explain the phenomena, one that lacks the disadvantages of functional theory and possesses the further advantage of explaining findings functional theory cannot cope with. Indeed, functional theorists often find themselves inadvertently using the propositions of the other type of theory.

Psychological theory

Most students of small groups use in explanation, not always explicitly, some form of psychological theory. Psychological theories are so called because their most general propositions are propositions about the behavior of individual men, and not about the conditions of equilibrium in groups. They do not in the least deny that the behavior of men is social. What they do, in effect, deny is (for instance) that the general propositions describing the behavior of men when rewarded by the physical environment are different from those describing their behavior when rewarded by other men, although the analysis may be much more complex in the latter case than in the former. Psychological theories also assume there is no characteristic of a group that cannot be explained by the characteristics of the individuals making it up: in this sense the whole is not greater than the sum of its parts. Finally, psychological theories assume that their propositions hold good for all men, that although, for instance, members of different societies learn to behave in concretely different ways because the present circumstances and past histories of these societies have been different, yet the general characteristics of the learning process remain the same for all men.

Behavioral theory. In particular, some of the most prominent codifiers of small-group research have come to find most satisfactory for the explanation of the phenomena one or another of the psychological theories (called “behavioral” or “learn ing” theories) first developed by experimental psychologists working with animals. These theories need not be incompatible with other psychological theories, such as psychoanalytic ones, although the terminologies employed may be quite dissimilar.

The chief propositions of behavioral theory are of the following sorts: A man whose activity (some item of “voluntary” behavior) has been rewarded (or “reinforced”) is apt to repeat that activity. He will repeat it the more often, the more successful it is in obtaining the reward. He is also more apt to repeat it, the more valuable (reinforcing) the reward is to him, and the value of the reward depends on the degree to which he has been deprived of it. Any activity that allows a man to escape or avoid punishment is also by that fact rewarded. A man usually has more than one activity open to him and more than one reward for these activities. The probability of his emitting one of the activities rather than another depends on the relative value of the alternative rewards and the relative probability of the success of the activities in obtaining them. When a particular stimulus situation (some set of cognitive elements) has also attended an occasion in the past when an activity has been rewarded, the presence of some similar stimulus on a new occasion will render it more probable that the activity will again be emitted. Finally, behavioral psychology states various propositions about emotional behavior, such as that when a man’s activity has been regularly rewarded under particular circumstances and then suddenly ceases to be so, he is apt to become angry, and in anger he is apt to find aggressive behavior rewarding. This is sometimes called the frustration-aggression hypothesis.

Behavior is social when the activities of each of at least two men reward (or punish) the activities of the other. Accordingly, the propositions of behavioral psychology may be used to explain the characteristics of social behavior, which is the real subject of small-group research. The variables whose relations are explained are the relative frequency with which each person emits alternative activities, the values to each of the rewards provided by the other, and the variations of stimuli along various cognitive dimensions, including the stimuli each person presents to the other. Since the value and frequency of the activities each emits affect the value and frequency of the other’s activities, the way is at once opened for the explanation of social influence. As the number of persons in interaction increases the analysis obviously becomes very complex, but it can sometimes be handled with the help of simplifying assumptions that may approximate real conditions. Particularly interesting is the situation in which many members of a group set a high value on a scarce reward (scarce in the sense that only a few of the members are able to supply it). In this situation be havioral psychology is able to explain why there should be differences between the members in status and power. It can also begin to explain how a social structure, a stabilized set of social relations, may develop and maintain itself in a group. Indeed, most of the phenomena investigated by small-group research can be explained, at least in a gross way, by the propositions of behavioral psychology [seeLearning Theory].

Since this theory envisages social behavior as an exchange between persons of goods (rewards) in different amounts and of different values, it envisages social behavior as a generalized economy, and the specialized and highly developed economics dealing with prices and markets can be used to sharpen its formulations. Alternatively, behavioral psychology can help to explain the propositions of elementary economics, thus contributing to the intellectual unification of social science.

The need for explanatory principles

Too high hopes should not be held out for detailed explanation and prediction. The problem is partly one of getting the necessary information. Among the crucial variables, for instance, is the relative value to individuals of different rewards.

Yet many values are not innate in men but acquired by them in the course of their past experience. If the investigator knows that a man grew up in a particular culture, he may infer correctly that the man has shared the experiences, and hence the values, of other members of that culture. In other cases, the man’s values may be quite atypical, the precipitate of idiosyncratic experiences. Without detailed knowledge of the man’s past, the investigator, even if he commands the best psychology in the world, may be unable to predict the man’s social behavior. Behavioral psychology itself explains why this should be so.

It must also be remembered that any small group, even one artificially formed, is part of a larger institutional structure. The behavior of its members cannot be wholly explained without reference to that structure. Thus a jury is certainly a small group, but one required by law to reach a unanimous decision in secret. So far as its members conform to the legal norms, their conformity affects their other relations. This does not mean that conformity to institutional norms, or even the nature of the norms themselves, cannot be explained by psychological principles. It does mean that insti tutional norms are always among the given conditions to be used in explaining face-to-face social behavior. Differences in cultural and institutional norms may help explain why apparently similar small-group experiments do not always yield the same results in different societies.

Detailed explanation and prediction also presents a problem of a different kind. The fundamental propositions of behavioral psychology are very general and refer to the behavior of individuals. How is the investigator to show what the resultant, or synthetic, implications of a set of propositions would be when many individuals are interacting in complicated and varying circumstances? Here the best hope lies in the increasing use of high-speed computers, whose virtue is precisely that of working out quickly the detailed implications of general propositions under varied parametric conditions.

Although students of elementary social behavior will take their basic explanatory principles from behavioral psychology, they will get little other help from the behavioral psychologists. Psychologists who have experimented with animals tend, when applying their principles to human behavior, to explain the learning processes of individuals or to jump to the gross features of institutions like religion and government in the larger society. Yet it may well be that in the long run a more convincing link between psychology and sociology will be forged by those who study at the same time the detailed behavior of individuals in interaction and the simpler social structures that emerge therefrom. They may indeed establish the first theoretical organization of the social sciences on a sound basis. This is the continuing justification of small-group research.

George Caspar Homans

[Directly related are the entriesCohesion, Social; Industrial Relationsarticle onHuman Relations; Sociometry. Other relevant material may be found inFunctional Analysis; Role; Social Psychology; Status, Social; and in the biographies ofCooley; Lewin; Mayo; Slmmel.]


Bales, Robert F. 1950 Interaction Process Analysis: A Method for the Study of Small Groups. Reading, Mass.: Addison-Wesley.

Blau, Peter (1955) 1963 The Dynamics of Bureaucracy: A Study of Interpersonal Relations in Two Government Agencies. Rev. ed. Univ. of Chicago Press.

Blau, Peter M. 1964 Exchange and Power in Social Life. New York: Wiley.

Cartwright, Dorwin; and Zander, Alvin (editors) (1953) 1960 Group Dynamics: Research and Theory. 2d ed. Evanston, III.: Row, Peterson.

Chapple, Eliot D.; and Arensberg, Conrad M. 1940 Measuring Human Relations: An Introduction to the Study of the Interaction of Individuals. Genetic Psychology Monographs 22:3–147.

Cohen, Arthur R. 1964 Attitude Change and Social Influence. New York and London: Basic Books.

Cooley, Charles H. (1909) 1956 Social Organization: A Study of the Larger Mind. In Charles H. Cooley, Two Major Works: Social Organization and Human Nature and the Social Order. Glencoe, III.: Free Press. → Each title reprinted with individual title page and pagination. Separate paperback editions were published in 1962 by Schocken.

Golembiewski, Robert T. 1962 The Small Group: An Analysis of Research Concepts and Operations. Univ. of Chicago Press.

Gurvitch, Georges (1950) 1957-1963 La vocation actuelle de la sociologie. 2d ed., 2 vols. Paris: Presses Universitaires de France. → Volume 1: Vers la sociologie differentiate. Volume 2: Antecedents et perspectives.

Hare, A. Paul 1962 Handbook of Small Group Research. New York: Free Press.

Hare, A. Paul; borgatta, E. F.; and Bales, R. F. (1955) 1965 Small Groups: Studies in Social Interaction. Rev. ed. New York: Knopf.

Heideb, Fritz 1958 The Psychology of Interpersonal Relations. New York: Wiley.

Hollander, Edwin P. 1964 Leaders, Groups, and Influence. New York: Oxford Univ. Press.

Homans, George C. 1950 The Human Group. New York: Harcourt.

Homans, George C. 1961 Social Behavior: Its Elementary Forms. New York: Harcourt.

Hopkins, Terence K. 1964 The Exercise of Influence in Small Groups. Totowa, N.J.: Bedminster Press.

Jennings, Helen H. (1943) 1950 Leadership and Isolation: A Study of Personality in Inter-personal relations. 2d ed. New York: Longmans.

Klein, Josephine 1956 The Study of Groups. London: Routledge; New York: Humanities.

Lewin, Kurt (1939–1947)1963 Field Theory in Social Science: Selected Theoretical Papers. Edited by Dorwin Cartwright. London: Tavistock.

Moreno, Jacob L. (1934) 1953 Who Shall Survive? Foundations of Sociometry, Group Psychotherapy and Sociodrama. Rev. & enl. ed. Beacon, N.Y.: Beacon House.

Olmsted, Michael S. 1959 The Small Group. New York: Random House.

Roethlisberger, Fritz J.; and Dickson, William J. (1939) 1961 Management and the Worker: An Account of a Research Program Conducted by the Western Electric Company, Hawthorne Works, Chicago. Cambridge, Mass.: Harvard Univ. Press. → A paper back edition was published in 1964 by Wiley.

Shepherd, Clovis R. 1964 Small Groups: Some Sociological Perspectives. San Francisco: Chandler.

Simmel, Georg (1902–1917) 1950 The Sociology of Georg Simmel. Translated and edited by Kurt H. Wolff. Glencoe, III.: Free Press.

Thibaut, John W.; and Kelley, Harold H. 1959 The Social Psychology of Groups. New York: Wiley.

Whitehead, Thomas N. 1938 The Industrial Worker: A Statistical Study of Human Relations in a Group of Manual Workers. 2 vols. Cambridge, Mass.: Harvard Univ. Press.

Whyte, William F. (1943) 1961 Street Corner Society: The Social Structure of an Italian Slum. 2d ed., enl. Univ. of Chicago Press.


Although William James, Charles H. Cooley, and George H. Mead had, at the beginning of the century, stressed the importance of primary groups in the development of individual personality, it was not until the 1930s that the small group became a serious focus of scientific attention. A most remarkable independent development of research interest in small groups by a number of creative people in different social science disciplines occurred during this period. Elton Mayo and his colleagues at the Harvard Business School conducted their famous studies of industrial work groups at Western Electric; Jacob Moreno and Helen H. Jennings developed sociometric methods for investigating group structure; William F. Whyte employed anthropological techniques in the study that led to Street Corner Society; and Kurt Lewin, Ronald Lippitt, and Ralph White initiated the experimental study of democratic and authoritarian group leadership. At the same time, Paul Schilder, Samuel Slavson, and Trigant Burrow were doing pioneer work in group therapy.

These early interests and activities collectively seemed to have reached the “critical mass” by the end of World War II, resulting in an explosive and rapidly mushrooming interest in small groups. By 1960, about 2,200 small-group studies had been published, more than 80 percent of which appeared in the decade 1950–1960. Since 1960, articles have been appearing at the rate of more than 250 per year.

No attempt will be made in this article to present the detailed findings of this enormous outpouring of research. Hare (1962) and McGrath and Altman (1966) provide useful summaries of theliterature. Here I shall attempt to provide a framework within which some of the important findings can be integrated. The framework is oriented to such questions as: what is a group? what are the significant ways in which groups differ from one another? what are the effects of such differences?

What is a group? An examination of the different usages of the term “group” suggests that each combines a greater or lesser number of the following distinguishing criteria: two or more persons who (1) have one or more characteristics in common, (2) perceive themselves as forming a distinguishable entity, (3) are aware of the interdependence of some of their goals or interests, and (4) interact with one another in pursuit of their interdependent goals. In addition, some writers, particularly those with sociological backgrounds, indicate that (5) groups endure over a period of time and as a result develop (6) a set of social norms that regulate and guide member interaction and (7) a set of roles, each of which has specific activities, obligations, and rights associated with it. I shall use the term “group” to signify at least the first four of the distinguishing criteria listed above. This usage is consonant with the intuitive notion that a group is an entity that consists of interacting people who are aware of being psychologically bound together in terms of mutually linked interests. A group is thus to be distinguished from an aggregate, class, category, or type, which consist of people who are classified together because of some common characteristic. Also, “group” implies a psychological or perceived bond, not merely an objective linkage, between the members’ interests or goals. Moreover, the psychological linkage has some cohesive feature to it—i.e., members of a group see that in some respects they sink or swim together. This latter statement is not meant to deny that divisive and disruptive tendencies may exist within a group; rather, it is meant to indicate that by definition a group does not exist if its cohesive bonds are not strong enough to contain its disunifying influences.

How groups differ. There are endless ways in which groups differ. It is useful to have some simplifying outline that highlights the central characteristics of groups and that permits a proliferation of detail as this becomes necessary. It is well to recognize that an outline abstracts variables from their contexts and their interrelationships, and thus presents them in somewhat distorted form. The outline that follows is guided by—but not limited to—the criteria of groups listed in the preceding section.

(1) Group size: the number of members in a group.

(2) Group composition: the individual characteristics of the members, including their distribution and patterning.

(3) Group structure: the patterning of member characteristics as perceived by the group members.

(4) The existential criteria of groups: the criteria for recognizing a group’s existence, members, action, property, etc.

(5) Group cohesiveness: the type and strength of the interests binding the members to the group.

(6) Group task and environment: the task confronting the group, and the environment within which the group functions.

(7) Interactional process: the modes and patterns of interaction between members and with the task environment.

(8) Group culture: the norms, standards, role patterns, traditions, and customs operating within the group.

(9) Group effectiveness: the task performance, the viability of the group, the membership satisfaction, and the change within individual members.

It is useful to recognize that any causal arrow connecting an item with any other item in this outline is likely to be bidirectional rather than unidirectional. Consider group size and group composition. It is evident that increasing the number of members in a group will affect the composition of the group—e.g., the more people there are in a randomly composed group, the more likely it is that the group will contain an individual whose intelligence is above or below any specified level. However, the causal arrow also points in the other direction; if a group is composed of a certain kind of members, its size is likely to be affected. Thus, groups of young children are likely to have fewer members than groups of older children. Of course, the causal path in the direction from group composition to group size is longer and more circuitous than the path in the opposite direction: it weaves from group composition to interactional process to group effectiveness (member satisfaction) to group cohesiveness and arrives finally at group size (presumably a size that permits interactions that are satisfying to the children because they are within their cognitive capabilities, and hence, the children are motivated to continue the group).

Some findings from the study of groups

I shall employ the outline presented above to organize the discussion of illustrative findings obtained from the study of groups. The studies of groups have been mainly of temporary ad hoc laboratory groups, created by the investigator, rather than of ongoing natural groups in their native habitats. Hence, there is no assurance that the research findings are generalizable. However, they do not seem inconsistent with everyday observations. Necessarily, the presentation is oversimplified. It largely ignores the evident fact that the effects of any given variable upon another (e.g., of group size upon interaction patterns) are very much influenced by other factors in the situation (e.g., the nature of the group task).

Group size

Hare (1962) and Thomas and Fink (1963) have reviewed the relevant research literature in some detail. The latter have presented a useful framework for viewing the major effects of variations in size. The schema presented here borrows from theirs. It is helpful to make a distinction between the statistical properties of size and the psychological properties of size. The former are those properties of a group that come from taking a given-sized sample of individuals, according to a given procedure, from a population with certain characteristics. The group is considered as an aggregate, and the psychological properties that arise from the compresence and interaction of its members are disregarded.

The statistical properties of size can be appreciated by observing how variations in size affect some common statistical measures: the sum, the mean, intragroup variability, the probability of the occurrence of any characteristic, the probability of concordance of characteristics, the variability of the statistical properties. Consider the resources (perceptual capacity, memory capacity, information, intellectual capacity, physical strength, skills, money, tools, and so forth) available as size increases. Clearly, the total resources, such as the total money in the group, will increase as a linear function of size. However, the usable resources will be determined by the task and environment. In certain tasks it does not help the group to have any duplication of a resource (e.g., to have more than one person who knows how to type; to have several mimeographing machines). Hence, the usable resources will often increase at a slower rate than the total resources and often will, beyond a certain point, not increase at all. If this is so, as size increases, the average usable resource per member will decrease. This reasoning explains why, for certain kinds of tasks, group performance as measured by production per group member decreases as size increases, even as total production goes up: not all of the total resources are usable.

As the size of a sample increases, the probability that any given characteristic will appear (that someone will have red hair, that someone will favor vegetarianism) increases. The probability that at least one individual of a group has some given characteristic clearly depends on the frequency of the characteristic in the population from which the group was formed, the size of the group, and the manner of group formation. If the members of the group manifest statistical independence regarding the characteristic, and if the probability, Pi, that any individual has the characteristic is a constant, then the probability that at least one member of the group has the characteristic is 1 — (1 —Pj)N, where N is group size. Thus, as N increases, the probability increases toward 1. Even for more realistic assumptions about group formation, it is clear that, as group size increases, so does the probability that the group contains at least one individual having any preassigned characteristic. Thus, the larger the size of a group, the more likely it is to have any or all of the following: a very bright person, a very stupid person, a quiet person, a talkative person, a “right-winger,” a “left-winger.”

Similarly, the probability that every member of the group has the given characteristic generally decreases as group size increases, for most reasonable assumptions about group formation. Under the assumptions of independence and constancy, the probability that all group members have the characteristic is P*, which decreases exponentially as N increases. Hence, as the size of a group increases, it is less likely that all individuals will have the same opinion or speak the same language or be equally informed or be equally resourceful. Although heterogeneity is likely to be greater within larger groups, larger groups are less likely to vary from one another in aggregate properties than are smaller groups. (A group of 20 persons is more likely to have the same average IQ as another group the same size than are two groups of three persons.)

Size affects not only the statistical properties of the aggregated resources and other characteristics of the group but also the opportunity to satisfy individual wants—for example, the larger the group, the more time, space, supplies, and facilities it will need to enable all individuals in the group to talk and be heard. Thus, if the total amount of time, reward, and space remain constant for a given task environment, the opportunity for individual participation and reward will decrease as the size of the group increases. On the other hand, the number of potential interpersonal relations increases geometrically as size increases. For example, the number of possible dyadic relations in any group increases with size (N) according to the formula: (N2 — N)/2. Since there appears to be a numerical limit to the capacity to establish close associations with others, a smaller proportion of the possible linkages will be formed as size increases.

From our discussion so far, it should be clear that different-sized aggregates of noninteracting individuals should differ predictably. The larger-sized aggregates should have more resources and more handicaps; more good solutions and more bad solutions; more diversity and difference; more demands for the available opportunities; more opportunity for diversified interpersonal contact but less for repeated contact. If the task environment is such that the presence of resources, good solutions, heterogeneity, and so forth, are more important factors than the presence of handicaps, bad solutions, and homogeneity in determining productivity, then larger aggregates should be more productive than smaller aggregates. But groups are not simply aggregates; they are composed of interacting individuals. A group’s performance may differ from the performance of a comparable aggregate of individuals because the contributions the individuals make in the group will be affected by the group milieu and because the group will combine or assemble the individual performances in a unique manner.

Audience effects. Social influence on individual thinking may reflect either the effect of working before an audience (such as other group members) or the impact of the contributions being made by the other group members. Research results, generally, indicate that there is increased motivation and increased distraction when a person works on intellectual problems before an audience rather than by himself (Kelley & Thibaut 1954). In addition, there tend to be fewer idiosyncratic thoughts, more moderation in judgments, more common associations, more cautiousness, and a general taking into account of the anticipated reactions of the audience. Over a period of time, adaptation to being observed tends to occur (Deutsch 1949), and hence the “audience effects” tend to decrease. Some research evidence (Atkinson 1964) suggests that once motivation to achieve passes beyond a certain moderate level, further increase in motivation tends to result in less effective performance.

Problem solving and productivity. The preceding discussion of “audience effects” suggests the possibility that as the size of a group increases, the intellectual functioning of its members will deteriorate: they will be “overmotivated,” more distracted, and more conventional. But other members of a group do not merely serve as an audience. They also contribute new information and different perspectives; they invoke more aspects of memory and demand greater attention; they provide more material for the individual to think with and about. That is, the contributions of other members may provide new associational starting points, may help the individual to break out of an ineffective set by suggesting new orientations, may fill in gaps or reveal unnoticed errors in the individual’s thinking. On the other hand, the contributions of other members may distract, interrupt a chain of thought, blot out an individual’s own associations, or confuse him by providing too much material for him to assimilate at one time. The meager relevant research indicates that the contribution of others may be more distracting than useful when the task confronting the individual is one that requires sustained, directed attention to a complex pattern where the relations between sequentially ordered parts must be kept in mind (e.g., in “reasoning” problems). In such a task, individuals working alone will hit upon different approaches (or will symbolize the same approach differently), and once they have taken a few steps on their respective approaches, understanding each other may be difficult without going back to the initial formulation of the approach or until an obvious solution has been reached. Hence, with such a task, the larger the group, the more likely it is that it will interfere with the individual’s thinking. On the other hand, when the task is such that the individual is likely to have an initial set that would lead to a clichéd or superficial solution to the problem and would overlook some of its major dimensions, the contributions of other members (starting from different sets, which may also be superficial) may force the individual to go deeper into the problem [seeProblem Solving].

A group solution will depend not only upon the abilities of its members to think within the milieu of the group but also upon the readiness and ability of the members to contribute to the group and upon the way in which their contributions are coordinated, assembled, or weighted to produce the resultant group solution. With regard to the readiness and ability of members to contribute, research indicates that the inequality of participation among the various members of a group increases as the size of the group increases (Stephan & Mishler 1952; Bales et al. 1951). These results suggest that individuals who tend to be shy are unlikely to participate actively in larger groups, although they may contribute much in small groups. On the other hand, individuals who tend to be assertive are likely to have a disproportionately large influence in larger groups as compared with smaller groups.

With regard to the coordination, assembling, or weighting of the contributions that individual members make, investigations have found that the difficulty of keeping track of the contributions of the various members, of coordinating and assembling them, will increase as the size of the group increases. For success in resolving this difficulty, larger groups have to devote more of their energy to activities directed toward coordination than do smaller groups. In addition, it is reasonable to hypothesize that such personality factors as self-confidence, assertiveness, and persuasiveness are more likely to play a significant role in determining the individual’s impact upon the group solution in larger groups than in smaller groups.

Member satisfaction, also, is affected by the size of the group. Laboratory and field studies both indicate that members of small groups are more likely to feel satisfied with their group, more likely to inhibit expression of disagreement, and less likely to develop cliques and factions. Large groups, on the other hand, are characterized by more absenteeism, more formality, and more internal conflict than are smaller groups.

Group composition

How are the individual members of a group characterized? The answers to this question presuppose that one knows what features of the members may influence the way they interact, interrelate, and function together. There is as yet, however, little systematic knowledge of how group composition—the distribution and patterning of member characteristics—affects group behavior. Nevertheless, it is reasonable to think that a group’s behavior will be affected by the distribution and patterning of such member characteristics as abilities, knowledge, resources, attitudes, interests, personality dispositions, age, sex, and social status. The combined characteristics of the members may be considered in terms of their influence upon the group’s effectiveness in coping with the task confronting it. Or they may be related to the compatibility of the members with one another, the attraction of the group for various members, the likelihood of the formation of cliques, and so forth.

Effectiveness. With regard to group effectiveness, considerable research supports the commonsense proposition that groups whose members have high abilities, training, or experience are more effective than groups whose members are lacking in these respects (McGrath & Altman 1966). However, it is not simply the average level of abilities that is important, but rather whether the kinds of abilities necessary to carrying out the role requirements set by the group’s task exist among the group members and are appropriately distributed. Group composition must be evaluated in reference to the demands confronting the group rather than in a vacuum. Homogeneity of member characteristics is an asset when the various group members are called upon to fulfill the same task functions, but it is a liability when there are varied functions to perform. For example, it is reasonable to assume that a group composed of both “abstract thinkers” and “concrete thinkers” will be more effective in performing in a task requiring both intellectual analysis and action than a group composed exclusively of one or the other type. Hoffman, Harburg, and Maier (1962) report results indicating that groups composed of individuals with dissimilar personalities are more productive than homogeneous groups. Schutz (1958) has theorized that compatibility is likely to be greater when people with different but complementary personality dispositions are paired together (e.g., people who wish to give affection and people who wish to receive it) than when people of similar dispositions are paired together (e.g., two people who wish to dominate). His research indicates that groups composed of members with compatible interpersonal tendencies are more productive than those composed of members with incompatible tendencies.

Patterns of interaction. There is a vast body of research on the effect of the personal and social characteristics of members on the development of attitudes toward the group, interactional patterns within the group, etc. Much of this research is considered in the sociometric literature. In general, research supports the saying, Birds of a feather flock together. People prefer to associate and interact with others who are similar rather than dissimilar to themselves in attitudes, status, background, interests, and so forth. The major exceptions to this generalization occur when similarity enhances competition (e.g., when two suitors are interested in the same girl); personal needs require complementarity rather than similarity (e.g., in heterosexual relations); or task requirements necessitate differentiated functions and statuses.

Individual behavior. There has also been much study of the personality characteristics that affect the performance of members in small groups. Mann (1959), reviewing the literature from 1900 through October 1957, focused on seven personality factors: intelligence, adjustment, extroversion-introversion, dominance, masculinity-femininity, conservatism, and interpersonal sensitivity. He summarized the relations between each of the seven personality variables and each of the following measures of an individual’s status and behavior in groups: leadership, popularity, total activity rate, task activity, social-emotional activity, and conformity. His survey indicates that the best single predictor of an individual’s behavior in the group is his intelligence. Intelligence and also extroversion and adjustment are positively related to total activity rate, leadership, and popularity in the group. In addition, the more intelligent and better-adjusted members are likely to contribute a relatively larger share of their total activity to building up group solidarity and providing emotional support for other members and a relatively smaller share to being critical or rejecting other members. Dominance is related positively to leadership and negatively to conformity; conservatism, on the other hand, is associated negatively with leadership and correlated positively with conformity. Masculinity and interpersonal sensitivity show positive relationships to leadership and popularity. Mann (1959, p. 266) concludes his review with the caution that the magnitude of the median of the correlations between an aspect of personality and performance is in no case higher than .25, and most of the medians of the correlations are nearer .15.

Group structure

Popular conceptions of group and organizational structure have been very much influenced by organizational charts, developed in the military and other large bureaucracies, that stress lines of formal authority. This is too limited a view. It is more fruitful to think of structure in terms of the way members actually relate to one another. In a sense the term “group structure” is a misnomer; there may be many different “structures” within a group—the work structure, the communication structure, the friendship structure, the power structure, the prestige structure, and so forth.

There is often a correspondence among the positions an individual holds in the different structures, so that an individual who holds a central position in one structure (e.g., the communication structure) is likely to hold a central position in other structures (power, friendship, and prestige). The research of Galtung (1964) in Norway indicates that this is the case for Norwegian society: people who are more central on social variables (income, education, occupation, residence, age, and sex) are also more central in the communication and power structures. In the status-equilibration hypothesis, Benoit-Smullyan (1944) and, later, Homans (1961) have stressed the forces that operate to make for similarity in the positions of an individual in different structures. However, status equilibrium is not always achieved. Research by Adams (1953) with air crews demonstrated that lack of congruence on such status dimensions as age, military rank, education, reputed ability, popularity, combat time, and position importance was related to poor morale, less friendliness, and lack of mutual confidence. Exline and Ziller (1959), working with experimentally created groups, found that groups constructed so as to have incongruent status hierarchies manifested more interpersonal conflict and less productivity than congruent groups.

In addition to research on status congruency, research on group structures has investigated such topics as (1) the effects of different communication structures (Guetzkow 1953); (2) the effects of different leadership structures, for example, leaderless groups versus groups with leaders, and of leadership style (Fiedler 1964); (3) the effects of different residential and propinquity structures (Festinger et al. 1950); (4) the effects of similarity or dissimilarity on various social and personal characteristics, such as age, sex, religious belief, attitudes; (5) the Kafkaesque effects of complex organizational structures, which are dimly perceived and little understood by members; (6) the determinants of sociometric structure, leadership structure, communication structure, etc.; (7) methods of classifying and identifying roles within groups (Bales 1950); and (8) mathematical procedures for characterizing different types of structures and different positions within a structure (Coleman 1964).

The existential criteria of groups

Although a vast body of legal principle and practice has been concerned with the conditions under which a “legal personality,” such as a corporation, can come into existence, and with the identifying of those who can act in its name, there has been little social-psychological research that bears upon such related problems as how a group is identified and what determines whether an action of a member is attributed to the group or to him personally. However, drawing upon studies of perceptual organization, it is possible to indicate some general principles concerning the conditions that are conducive to the perception that a collection of individuals or units are part of a system rather than an unorganized aggregate of elements.

As Koffka (1935) pointed out, abrupt discontinuity produces segregating forces between the parts of a visual field that it separates, as well as unifying forces within the separated parts. Further, he indicated that homogeneity tends to produce unifying forces in the visual field. Homogeneity may be based upon (1) the common fate of the elements perceived (they move together); (2) their qualitative or quantitative similarity (they have the same color or the same luminosity); (3) proximity (they occur in spatial or temporal contiguity); (4) a common boundary; (5) past experience or custom that has led to similar responses to the various elements; and (6) set or the expectation that the elements are to be grouped together [seeGestalt Theory].

It seems evident that processes analogous to these determine whether an individual will perceive a collection of individuals as a social group and whether he will perceive himself to be part of the group. Thus, if an individual perceives that he and some others are strikingly different in certain respects from the remainder of the people in their surroundings; that he and the others tend to be satisfied or dissatisfied at the same time or under similar circumstances; that he and the others have similar attitudes or similar backgrounds; that he and the others live or work in close proximity; that he and the others are associated together in other people’s minds or treated similarly by other people —if he perceives any of these patterns, the individual is likely to perceive himself and the others as cooperatively interdependent. I would stress, as does Campbell (1958), the central role of the perception of common fate in determining the consciousness of being joined with others to form a group.

Group cohesiveness

In everyday usage “cohe-siveness” refers to the tendency to stick together; its usage in social psychology is much the same. It refers to the linkages that bind the members of a group together. Deutsch (1949; 1962) has stressed that the linkages among members are cohesive, rather than disruptive, when the goals and interests of the members are cooperatively, rather than competitively, interrelated. Various aspects of these link ages have been the focus of research: the nature of the mutually linked goals or interests—such as friendship, work, money; the strength of the mutually linked goals or interests; the degree of linkage, and the availability of other means of obtaining one’s goals; the forces operating to restrain members from leaving the group; other interests or memberships that are in opposition to continued membership in the group.

Since group cohesiveness is central to the existence of groups, it is natural that its determinants and also its consequences have been studied extensively (Hare 1962; Collins & Guetzkow 1964; Mc-Grath & Altman 1966). Research findings, over-all, indicate that cohesiveness (as measured by interpersonal congeniality, the desire to remain a member of the group, attitudes toward the group’s functioning, or other similar measures) is consistently associated with greater communication between group members, greater readiness of group members to be influenced by the group, more consensus among members on attitudes and beliefs that relate to group functioning, more sense of responsibility toward each other among group members, a greater feeling of personal ease and security within the group by the group members, and so forth. Also, task effectiveness is generally positively correlated with cohesiveness if high accomplishment on the task is valued by the group (some groups restrict performance to achieve their objectives) and if the task is such that its performance is likely to be enhanced by increased group effort. It should be noted that the causal arrow is bidirectional; group cohesiveness not only increases intragroup communication and group success, but group success and intragroup communication increase group cohesiveness [seeCohesion,Social].

Group task and environment

It is self-evident that the task confronting the group and the environment within which the group functions can influence all the other characteristics of a group. Unfortunately, however, the research relating to task and environmental characteristics has been meager, largely because there has not yet been developed any systematic way of characterizing tasks or environments. Nevertheless, certain useful distinctions have been made.

It is possible to characterize many tasks and environments in terms of the type of requirements for success that they impose upon the group. It is apparent that tasks differ in the types and amounts of skills, knowledge, effort, and resources required and in the way these factors have to be interrelated. In other words, the roles within the group, the structure of the group, the size of the group, etc., may vary as a function of the group’s task and environment.

Fiedler (1964), for example, has shown that the effectiveness of different types of leaders is very closely related to the structure of the task confronting the group. In both field and laboratory studies, his findings indicate that controlling, authoritarian leaders tend to be most effective either in very favorable or else in relatively unfavorable group-task situations, while the permissive, considerate, democratic leaders are most effective in situations that are intermediate in favorableness. Fiedler indicates that a situation is very favorable when (1) the leader-member relations are positive, (2) the task is clear and well structured, and (3) the leader has well-defined authority and power to reward and punish. Thus, for example, if a leader has good personal relations with his group and the task is routine and his authority well-defined, he is likely to be more effective if he is “directive,” rather than democratic and permissive. On the other hand, if the task is novel and unstructured but the situation is otherwise similar, democratic, permissive leadership is likely to be more effective [seeLead Ership].

Task structure helps to determine the types and amounts of interaction and communication within a group and also the sequencing and organization of the activities within the group. These, in turn, will often affect the social relations that develop within a group. Much research has supported the proposition (Homans 1950) that people who interact frequently with one another tend to like one another, and vice versa. Thus, tasks that require certain group members to work closely with one another and limit their interaction with other members may, if the task is long-enduring, help to create patterns of friendship that parallel the interactional requirements of the task. Further, the research indicates that if a task places a given individual in a central position because he is able to communicate readily with other members or because he possesses a scarce resource (a skill, specialized information, or a particular tool) that is of critical value to group success, then he is likely to have high status within the group (and high satisfaction with the group).

Tasks differ in the degree to which they permit division of labor and specialization of function. The problem of dividing tasks into subtasks, of sequencing them, and of assigning personnel and resources to them has been of major interest to economists and operations-research analysts, and they have had considerable success in developing rational methods of predicting the effectiveness of different methods of dividing up a task. Here let us note some psychological aspects of the division of labor. Among its possible negative consequences are the loss of one’s identification with the over-all group objectives; the loss of a sense of an over-all significance and meaning to one’s activities; the development of vested interests in one’s specialized activities; the development of specialized languages, values, and modes of thought that interfere with coordination and communication between the various specialized activities. Among the possible positive consequences of specialization is, in addition to increased group productivity and individual economic reward, the greater chance that an individual will be able to find some activity that matches his particular interests and abilities.

Tasks and environments differ not only in their activity and interaction requirements but also in their stressfulness. The term “stress” has been used to refer to a hypothetical state of tension, frustration, or internal conflict induced by such conditions as task difficulty (e.g., a problem without any solution), lack of information about how well the task is being performed, threat of punishment for task failure, danger (e.g., as in combat or survival in the Arctic), intense criticism, time pressure, an unpredictable environment. The results of studies of the effects of such varied conditions are not univocal. The safest generalization seems to be that mild stress often improves group performance and enhances group cohesiveness, while severe stress often has the opposite effects. Optimal stress for a group is presumably higher the more able and cohesive the group is and the more the members see themselves as able and motivated to cope with problems (Deutsch 1959).

Finally, it is relevant to note that environments differ in the probability of reward and the amount of reward they provide for effective group action and also in the manner in which rewards are distributed within a group. Little research has been done on the effects of different “schedules of reinforcement” upon group behavior, yet there is reason to assume that they would influence group performance (Shapiro 1963). On the other hand, a considerable number of research studies (e.g., Deutsch 1949; Raven & Eachus 1963) have demonstrated that whether rewards are distributed cooperatively or competitively within a group may have a striking effect on member behavior. In general, group members who are rewarded cooperatively show more positive response to one another, have greater involvement in the group, are less likely to work at cross-purposes, communicate with one another more effectively, and work more productively together than group members who are rewarded competitively.

Interactional process

The observable transactions between members and their observable transactions with their task environments are lumped together under the term “interactional process.” It is, in effect, what goes on in groups. There are many different ways of characterizing what goes on. Most of them focus on one or more of the following aspects of a transaction: who communicates or does what to or with whom; with what intent or function; how, when, or under what conditions; through what media or channels; with what effects upon whom, as perceived by whom. Each of the italicized terms could be elaborated in considerable detail. For example, if one specified the characteristics of the potential communicators and communicatees (the who and the whom)—their statuses, their personality tendencies, their preexisting attitudes toward one another—one could predict, to some degree, who will talk to whom and how they will talk. (The demeanor of a subordinate making a critical remark to a superior will be rather different from that of a superior criticizing a subordinate.) Similarly, knowledge of the conditions —e.g., what stage in problem solving the group is at—enables one to predict what kinds of content (the what) a transaction is likely to have.

The most widely used system for categorizing interactions is the one developed by Bales (1950). It focuses on the who-to-whom-and-what interaction. His system consists of 12 distinct categories of the content of communication: content that (1) shows solidarity, (2) shows tension release, (3) agrees, (4) gives suggestion, (5) gives opinion, (6) gives orientation, (7) asks for orientation, (8) asks for opinion, (9) asks for suggestion, (10) disagrees, (11) shows tension, and (12) shows antagonism. The categories are grouped in various ways. One major grouping is into task categories (subdivided into questions and attempted answers) and social-emotional categories (classified as positive and negative). More recently Stone and his co-workers (1962) have devised a more generalized system of content analysis called the General Inquirer, which employs a computer to code the actual verbal text of group interaction into 164 categories [seeInteraction, article onInter Action Process Analysis].

Slater (1955) has shown that members who are ranked high as “idea men” by the other members in problem-solving groups initiate interaction more markedly in “attempted answers,” while members who are ranked high in likability participate more heavily in the categories grouped as “positive social-emotional.” His research has also indicated that often the task leader, or idea man, and the social-emotional leader are not the same person; this specialization of interaction function is more evident if the group exists over a period of time. In other words, as Barnard (1938) had noted earlier, the two major problems confronting groups —adaptation to their task environment and provision of personal satisfaction to the individual members—do not necessarily lead to the same emphases within a group.

Why do people interact? Few theorists have gone beyond the common-sense viewpoint that they do so because it is instrumental to a given end or because it is gratifying in itself. Festinger and his associates (1950), however, have suggested that one of the major instrumental functions of interaction is helping to establish “social reality”: the validation of opinions, beliefs, abilities, and emotions in terms of a social consensus. That is, one of the functions of communication within a group is to establish uniform views about reality, so as to provide the members with some confidence in their beliefs and to enable them to coordinate their behavior for effective group action. Thus, group members whose views deviate from those held by the rest of the group will be subject, through communication, to pressures to change their views to conform to those of the rest, or they will be rejected or isolated by the group, so as to eliminate a source of disturbance to the group. Festinger hypothesized that these pressures are greater the more cohesive the group is, the more relevant the belief is to the group, the more discrepant the deviant’s viewpoint is, etc. A considerable body of research is consistent with these hypotheses. Festinger has also suggested that communication may function as a substitute for social locomotion—people who would like to be in powerful positions direct their communication to ward those who hold such positions. The study by Kelley (1951) of communication in experimentally created hierarchies is consistent with this hypothesis. However, Cohen (1958) suggests that up ward communication may be more directly motivated by the desire to receive the benefits that a higher-status person can bestow upon someone of lower status.

Group culture

The members of any group who have had a prolonged experience of interacting with one another tend to develop shared values, expectations, and rules—a normative consensus that helps to regulate interaction between members and between the group and its task environment and that also serves to define the roles of the various members, including their specialized activities, rights, and responsibilities. A normative consensus, or group norm, sets criteria for evaluating the desirability-undesirability, acceptability-unacceptability, of the group members’ activities, beliefs, appearance, etc., and for responding with various sanctions, positive or negative, such as reward-punishment, approval-disapproval, to a member’s conformity with or violation of the norm [seeNorms].

Norms develop about many things, from the type of pronoun to be used in addressing intimates or strangers to the type of wine one should serve on certain occasions. Yet not everything is regulated. Norms tend to develop mostly in areas that are relevant to the group’s functioning, and it seems likely that the more important an area is to the group, the more norms there will be, the more intense will be the sanctions employed to obtain conformity to them, and the smaller will be the range of acceptable behavior. Thus, as Sutherland has shown, the norm of punctuality in appearing at a prearranged time and place is strictly enforced among professional thieves because its violation may endanger an enterprise and lead to arrest (Conwell 1937). Norms are more often developed with regard to overt behavior than private beliefs, not only because the former are usually more important to group functioning but also because beliefs are less controllable, being less observable than behavior. Further, it is apparent that the norms of different types of groups will differ—for example, the norms of a friendship group and those of a work group. It is not uncommon for a person to experience conflict because he belongs to groups that have conflicting norms [seeConformity].

There is an extensive research literature on the determinants of conformity and deviation to group norms (for a summary, see Symposium on Conformity …1961). It is not a great oversimplification to sum up the findings as indicating that conformity appears to be a function of such factors as the person’s awareness of the norm, the strength of the norm, the strength of the person’s attraction to the group, the likelihood that conformity or deviation will be observable by others, the strength of the sanctions expected for conforming or deviating, personality predisposition (such as dependency, acceptance of authority, self-confidence). Whether a person will conform will depend not only on these factors but also on the strength of the tendency to deviate, which is determined by parallel considerations (e.g., is the tendency to deviate from the norms of one group a tendency to conform to the norms of another group to which the person belongs?).

Groups not only develop norms, which specify the “shoulds” and “should nots”; often they also develop styles, traditions, or customs (and these often become the object of norms), which are the habitual ways of dealing with recurring situations. Thus, a group may develop a unique language (almost every profession and trade develops its own “slang,” its own peculiar abbreviations); distinctive garb, insignia, or appearance, to permit ready identification; a distinctive locale for meeting and engaging in its activities; a special style of inducing emotional responsiveness and of expressing emotion (e.g., distinctive dances, ceremonies), etc. Case-study material suggests that groups are most likely to develop idiosyncratic traditions if they are relatively isolated, as a result of geographical or social factors (due to superior or inferior status); if they are in conflict with other groups; or if their task environment is unique.

Anthropologists have, of course, studied and described in considerable detail the customs and traditions of many simple societies. A similar kind of analysis could be made of the development of customs and traditions in experimentally created or naturally formed groups, to investigate some of the determinants of particular kinds of traditions and customs. Unfortunately, little research has been conducted, apart from the pioneering studies of Sherif (1936), Merei (1949), and Rose and Felton (1955).

Group effectiveness

A group’s effectiveness may be characterized in such terms as (1) task performance—the quality and quantity of the group’s outputs, as measured in terms of external criteria; (2) group viability—the group’s ability to maintain itself as a functioning group under varying conditions; (3) member satisfaction—the desire of the members to maintain their membership and to contribute to the group’s viability and the attainment of group goals; and (4) member change —the change in knowledge, skills, attitudes, adjustment, or personality of the individual members of the group. Although there may be relations between these different types of group effectiveness over the long run, in the short run, however, it is evident that these different types of outcome may vary independently of one another. For example, high member satisfaction may result from or result in high task accomplishment, but high task accomplishment may also occur at the cost of member satisfaction (as when a demanding leader drives the group members on despite their protests).

The bulk of research on group effectiveness has been concerned with the determinants of task performance. This research highlights the importance of the following types of determinants: (1) the strength of the values associated with effective task performance—e.g., the greater the potential rewards for good performance and the more task-oriented the group norms, the more effort the group will be willing to put into the task; (2) the cohesiveness of the group—e.g., the more the members value the group and one another, the more willing they will be to expend effort in compliance with group norms or to achieve group-defined goals; (3) the perceived difficulty of the task—e.g., a task that is perceived to be very easy or virtually impossible is likely to stimulate less effort than a task that is viewed as difficult but attainable; (4) the amount of task-revelant abilities, information, and experience of the group members; (5) the appropriateness of the group structure to the requirements of the task—e.g., how efficient is the particular kind of division of labor for the task? how do the abilities, knowledge, and interests of the role occupants fit the requirements of their roles?; (6) the central role of the group leader and the appropriateness of his leadership style to the task and to the group.

Although there have been many studies of task performance, research in this area has been plagued by the problem of establishing reliable and valid measures of group achievement. There is little evidence to indicate that one group tends to perform reliably or consistently better than another group on a given task. Nor can one predict with much confidence, from a group’s performance on one task, how it will do on another similar task. Research investigators in this area have not yet begun to develop any measures of group achievement that have the usefulness of many of the measures of individual achievement.

The determinants of member satisfaction have been studied as extensively as those of task effectiveness (for summaries, see Collins & Guetzkow 1964; McGrath & Altman 1966). In brief, the relevant research indicates that a member’s satisfaction is affected by (1) the status of the group —its successfulness, its task achievements, its prestige; (2) the interpersonal relations within the group—the attractiveness of the other group members, their attitude toward him, their attitude toward belonging to the group; (3) the member’s role within the group—its prestige, communication centrality, power, significance, interest; (4) the direct rewards and benefits received from membership; (5) the group atmosphere, as determined by such factors as leadership style, group size, group composition; and (6) the nature and desirability of conflicting memberships or activities.

Little research has been done on the determinants of environmental input and of group viability.

There is, however, an extensive literature on member change. Much of the relevant research has been done under the rubric of conformity and deviation (for summaries, see Symposium on Conformity …1961). Also, the growing literature on group psychotherapy (for representative papers, see Rosenbaum & Berger 1963) contains many insightful case discussions, even though the amount of systematic research is still quite small [seeMental Disorders, Treatment of, article onGroup Psychotherapy].

In addition, there is a rapidly developing list of publications dealing with human-relations training groups that are concerned with helping people learn how to function more effectively in groups. Here, too, the amount of published research is negligible. Nevertheless, since the stimulus to this approach to human-relations training came from the theoretical writings on re-education and reaction research of Kurt Lewin (1935–1946), much of the literature on training groups is imbued with social science concepts and suggests research. The major ideas of Kurt Lewin that underlie the training-group approach are that the re-education process basically involves the equivalent of a change in culture, and that for the individual to accept a new system of values and beliefs, he must come to value his membership in a group that has these new values and beliefs as a central component of its culture.

Morton Deutsch

[Other relevant material may be found in Social Psychologyand in the biography of Lewin.]


Adams, Stuart 1953 Status Congruency as a Variable in Small Group Performance. Social Forces 32:16–22.

Atkinson, John W. 1964 An Introduction to Motivation. New York: Van Nostrand.

Bales, Robert F. 1950 A Set of Categories for the Analysis of Small Group Interaction. American Sociological Review 15:257–263.

Bales, Robert F. et al. 1951 Channels of Communication in Small Groups. American Sociological Review 16:461–468.

Barnard, Chester I. (1938)1962 The Functions of the Executive. Cambridge, Mass.: Harvard Univ. Press.

Benoit-Smullyan, Emile 1944 Status, Status Types, and Status Interrelations. American Sociological Review 9:151–161.

Campbell, Donald T. 1958 Common Fate, Similarity, and Other Indices of the Status of Aggregates of Per sons as Social Entities. Behavioral Science 3:14–25.

Cohen, Arthur R. 1958 Upward Communication in Experimentally Created Hierarchies. Human Relations 11:41–53.

Coleman, James S. 1964 Introduction to Mathematical Sociology. New York: Free Press.

Collins, Barry E.; and Guetzkow, Harold 1964 A Social Psychology of Group Processes for Decision-making. New York: Wiley.

Conwell, Chic 1937 The Professional Thief: By a Pro fessional Thief. Annotated and interpreted by Edwin H. Sutherland. Univ. of Chicago Press. → A paperback edition was published in 1960.

Deutsch, Morton 1949 An Experimental Study of the Effects of Co-operation and Competition Upon Group Process. Human Relations 2:199–231.

Deutsch, Morton 1959 Some Factors Affecting Membership Motivation and Achievement Motivation in a Group. Human Relations 12:81–95.

Deutsch, Morton 1962 Cooperation and Trust: Some Theoretical Notes. Volume 10, pages 275-319 in Nebraska Symposium on Motivation, edited by Marshall R. Jones. Lincoln: Univ. of Nebraska Press.

Exline, Ralph V.; and Ziller, Robert C. 1959 Status Congruency and Interpersonal Conflict in Decision-making Groups. Human Relations 12:147–162.

Festinger, Leon; Schachter, Stanley; and Back, Kurt (1950) 1963 Social Pressures in Informal Groups: A Study of Human Factors in Housing. Stanford Univ. Press.

Fiedler, Fred E. 1964 A Contingency Model of Leadership Effectiveness. Volume 1, pages 149-190 in Ad vances in Experimental Social Psychology. New York: Academic Press.

Galtung, Johan 1964 Foreign Policy Opinion as a Function of Social Position. Journal of Peace Research 1:206–231.

Guetzkow, Harold (1953) 1960 Differentiation of Roles in Task-oriented Groups. Pages 683-704 in Dorwin Cartwright and Alvin Zander (editors), Group Dynamics: Research and Theory. 2d ed. Evanston, 111.: Row, Peterson.

Hare, Alexander P. 1962 Handbook of Small Group Research. New York: Free Press.

Hoffman, L. Richard; Harburg, Ernest; and Maier, Norman R. F. 1962 Differences and Disagreement as Factors in Creative Group Problem Solving. Journal of Abnormal and Social Psychology 64:206–214.

Homans, George C. 1950 The Human Group. New York: Harcourt.

Homans, George C. 1961 Social Behavior: Its Elementary Forms. New York: Harcourt.

Kelley, Harold H. 1951 Communication in Experimentally Created Hierarchies. Human Relations 4: 39–56.

Kelley, Harold H.; and Thibaut, John W. 1954 Experimental Studies of Group Problem Solving and Process. Volume 2, pages 735-785 in Gardner Lindzey (editor), Handbook of Social Psychology. Cambridge, Mass.: Addison-Wesley.

Koffka, Kurt 1935 Principles of Gestalt Psychology. New York: Harcourt.

Lewin, Kurt (1935–1946) 1948 Resolving Social Conflicts: Selected Papers on Group Dynamics. Edited by Gertrud W. Lewin. A publication of the University of Michigan Research Center for Group Dynamics. New York: Harper.

Mcgrath, J. E.; and Altman, I. 1966 Small Group Research: A Synthesis and Critique. New York: Holt.

Mann, Richard D. 1959 A Review of the Relationships Between Personality and Performance in Small Groups. Psychological Bulletin 56:241–270.

Merei, Ferenc 1949 Group Leadership and Institutionalization. Human Relations 2:23–39.

Raven, Bertram H.; and Eachus, H. Todd 1963 Co operation and Competition in Means-interdependent Triads. Journal of Abnormal and Social Psychology 67:307–316.

Rose, Edward; and Felton, William 1955 Experimental Histories of Culture. American Sociological Review 20:383–392.

Rosenbaum, May; and Berger, Milton (editors) 1963 Group Psychotherapy and Group Function. New York: Basic Books.

Schutz, William C. 1958 FJRO: A Three-dimensional Theory of Interpersonal Behavior. New York: Holt.

Shapiro, David 1963 The Reinforcement of Disagreement in a Small Group. Behavior Research and Ther apy 1:267–272.

Sherif, Muzafer (1936) 1965 The Psychology of Social Norms. New York: Octagon.

Slater, Philip E. 1955 Role Differentiation in Small Groups. American Sociological Review 20:300–310.

Stephan, Frederick F.; and Mishler, Elliot G. 1952 The Distribution of Participation in Small Groups: An Exponential Approximation. American Sociological Review 17:598–608.

Stone, Philip J. et al. 1962 The General Inquirer: A Computer System for Content Analysis and Retrieval Based on the Sentence as a Unit of Information. Behavioral Science 7:484–498.

Symposium on conformity and deviation, Louisiana state university, 2960 1961 Conformity and Deviation. Edited by Irwin A. Berg and Bernard M. Bass. New York: Harper.

Thoma’s, Edwin J.; and Fink, Clinton F. 1963 Effects of Group Size. Psychological Bulletin 60:371–384.


Groups are composed of individuals. But when and how does a collection of individuals become a group? The phrase “group formation” suggests that something is formed. The task here is to trace the appearance of essential properties or characteristics that distinguish a human group.

Both for theory and research, it is instructive to adopt the strategy of tracing the formation of informal groups rather than of those instituted formally through blueprints handed down by out side authority (such as a committee or board). Despite the limitation of considering groups formed through the interaction of the membership, the implications are broad. Many social institutions and formal organizations (including governments) had informal beginnings. Informal groups are almost invariably found within stable formal structures such as industrial, commercial, political, educational, religious, military, and recreational organizations. Finally, informal groups possess the minimal characteristics essential to any organized association, whether large or small.

Background. The properties characterizing the formation of a group were treated by the nineteenth-century social philosophers for various reasons and with varying emphasis, each using illustrative examples known to him. But the topic was doomed to controversy until data were collected through scientific methods.

In the 1920s, Robert E. Park at the University of Chicago inspired and directed a series of investigations into human groups and their relationships to their settings (for example, Thrasher 1927; Landesco 1929; Shaw 1929; Zorbaugh 1929). Initiated to deal with the problem of homeless and antisocial children, the work of the Soviet educator Makarenko included concrete data on the formation of group properties, which he gradually came to regard as crucial conditions for the effectiveness of his educational efforts (Makarenko 1933; Bowen 1962).

In the 1930s, under the impetus of Elton Mayo of the Harvard Business School, studies in Western Electric’s Hawthorne plant revealed the emergence of groups and their impact on the behavior of workers who had initially been placed together in observation rooms for the purpose of studying the effects of varying illumination and rest periods (Roethlisberger & Dickson 1939). J. L. Moreno (1934) and his co-workers began the systematic sociometric mapping of friendship choices among girls in reformatory cottages and children in classrooms. Using the methods of laboratory psychology, Sherif (1936) tested sociological theories dealing with the formation of social norms in situations of ambiguity and instability, demonstrating the subsequent retention of the norms as personal standards when individuals were alone. Lewin and his students (Lewin et al. 1939) initiated experiments varying the manner of adult supervisors of chil dren’s clubs. Meanwhile, the failure of “person ality” or “intelligence” tests in selecting military leaders led the military in several countries to sponsor studies of what came to be called “leader-less groups” (Gibb 1954; U.S. Office of Strategic Services 1948).


The definition presented here includes only the minimal properties found essential through extensive surveys of empirical and theoretical literature on groups of all kinds (Sherif & Cantril 1947; Sherif & Sherif 1948).


A group is a social unit consisting of a number of individuals who stand in status and role relationships to one another that are stabilized in some degree at a given time and who possess a set of values or norms regulating their behavior, at least in matters of consequence to the group.

By this definition, the “groupness” of a group is a matter of degree. A collection of persons forms into a group proportional to (a) the degree of stability of its organization (consisting of roles and status relations) and (b) the degree to which its particular set of norms for behavior are shared and binding for the participants. The undefined terms in the definition (role, status, norms) will be speci fied through research operations with conceptual relationship to the process of group formation.

It should be noted that this definition includes the properties covered in many modern works, while excluding others. Similar specifications are found in Bales (1950), Blau and Scott (1962), Bonner (1959), Cartwright and Zander (1953), Golembiewski (1962), and Hare (1962). The fol lowing characteristics, included by some investi gators, were omitted here for the reasons specified:

Interaction and communication are not distinc tive to group formation but are essential to any kind of human association of consequence. Shared senti ments, attitudes, and behavior patterns of group members are implied in the normative property; in fact, the extent of sharing is one of the research measures for the degree of group formation at a given time. Many properties of existing groups (for example, morale, solidarity or cohesiveness, and loyalty of members) are dependent upon the conditions of group formation, especially the degree of stability attained.

The properties essential to group formation will aid the reader in evaluating the large body of experimental research on so-called “small groups” conducted in both the United States and Europe since World War II. In summarizing this literature, Golembiewski (1962, p. 47) found that the great majority of these “groups” consisted of strangers exposed briefly in the laboratory to tasks or instructions creating temporary interdependence. Very few studies have allowed a sufficient time span for group properties to form.

Generality of group formation

Beneath the organized forms and routines of societies, the formation and disintegration of groups occur in all walks of life, frequently with important consequences. Informal group formation is well documented with in industrial, military, school, prison, and neighborhood settings (see Hare 1962; Sherif & Sherif 1948). In studying the Near North Side of Chicago, Zorbaugh (1929, p. 192) reported group formations in neighborhoods of all socioeconomic ranks with “an enormously important role in the lives of their members”: exclusive clubs on the fashionable Gold Coast, intimate groups of nonconformists in artists studios, mutual benefit societies in foreign colonies, gangs in slum areas, and cults and sects in the rooming house district.

The extensive documentation on the generality of group formation also shows the striking dependence of the process on other groups and on the material and ideational features and facilities of the environment. The process of group formation is not insulated by the bounds of the membership. The formation that results is not a closed system. The circumstances bringing individuals together initially, their motives in continuing to interact, the particular organization and norms that develop, and the degree of their stability are inevitably dependent upon environmental circumstances. Included in the environmental circumstances are other groups whose activities and aims impinge favorably or unfavorably upon those of the group in formation.

Four essentials of group formation

The essentials of group formation can be traced, starting with the initial conditions for interaction among individuals. The encounter with another person is the most elementary social situation. Even the mere presence of other persons has consequential effects on behavior and task performance. From the time when individuals are merely to gether to the time when the properties of a group begin to appear, we see that the consequential effects on behavior begin to assume regularities. As time goes on, these regularities reflect patterns that are the organizational and normative properties of the group. Accordingly, the essentials in the process of group formation are the following; (1) A motivational base conducive to repeated interaction; (2) formation of an organization (structure) consisting of roles and statuses; (3) formation of rules, traditions, values, or norms; and (4) differ ential effects of the group properties on the attitude and behavior of participants over time.

Motivational base

Any human motive, frustration, problem, or desired goal that an individual cannot handle effectively alone is conducive to his interaction with others who are seen as being in the same plight. The prerequisite for group formation is that persons with motives conducive to initial interaction have the opportunity over a span of time to recognize the concerns they share or reciprocate and to attempt to deal with them in concert.

The common motive or motives faced in the initial stages of interaction may be one or several of those found in any society—for example, hunger, sexual desire, desire for recognition or power in some respect, and fear or anxiety in the face of threat. They may be culturally defined—for example, desire for material possessions or prestige through particular activities, or pursuit of political goals. Here only a few points can be considered:

The common problem, motive, or goal conducive to repeated interactions is necessarily dependent on environmental circumstances, both in its occurrence and in attempts to deal with it. Whatever its nature, the motivational base for group formation invariably affects the activities and tasks engaged in by the group’s members and the kinds of personal qualities that become prized by them. When a set of norms takes shape, those most binding are typically related to the motives or problems that initially brought the persons together. One reason why the many controversies over problems of conformity-nonconformity are inconclusive is that many theorists pay scant attention to the relationship between the particular norms of a group and the initial motivational base underlying them.

However, to the degree that group formation achieves a degree of stability over time, new sources of motivation and new goals, arising from the existence of the group itself, are generated among the members. These may even take precedence over those that originally brought the members together. Thus, the hungry person may refrain from eating until he can share with his starving fellows; the politician may spurn an advantageous political bargain out of loyalty to his supporters; the member of a group struggling for equal opportunity and freedom from fear may undergo great deprivation and bodily injury to secure recognition of his group.

Formation of organization or structure

As in dividuals interact over a period of time in activities related to the common problems that brought them together, their behavior and their expectations for one another’s behavior assume regularities from which a pattern can be constructed. Here we shall define certain features of these regularities that appear to be crucial in any group formation. Heavy reliance will be placed upon findings from three experiments on group formation and relations be tween groups, each experiment lasting several weeks (summarized in Sherif & Sherif [1948] 1956, chapters 6 and 9; and Sherif 1966), and from more recent studies of naturally formed groups (Sherif & Sherif 1964). In each case the experiments started with unacquainted persons divided into collections of 10-12 as similar as possible in composition. All of these studies were conducted under naturalistic conditions, and data were collected without constant awareness by the individuals that they were being investigated.

The development of organization has been de fined in terms of role and status relationships among a number of individuals. Role denotes reciprocities in the treatments and expectations of individuals, each for the others. Unlike well-defined occupational or sex roles, definite prescriptions for behavior are lacking when unacquainted persons first meet. Reciprocities among them must be built on the basis of performance in the activities engaged in, the reactions of others to a person, and his reactions to them. The typical finding at early stages of interaction is that individual contributions to task performance differ from one activity to the next (U.S. Office of Strategic Services 1948; Gibb 1954). Thus, observers rating behavior in the different situations find that the degree of participation and prominence of the individuals differs from one task to the next, according to individual differences in skills, abilities, temperament, or physical resources and tools relative to the activities in question.

The single most salient feature of group formation is that over time the various member roles become differentiated, not merely with regard to task performance or personal qualities but according to the evaluation of the roles by the members themselves. Members are accorded differing degrees of prestige and respect by their fellows. The member roles acquire different degrees of relative power to initiate and control activities important to all of the individuals in the group.

A member’s position (rank) in a developing power structure is his status in the group, defined in terms of the relative effectiveness of his actions in initiating activities, making or approving decisions affecting the group, coordinating interaction, and invoking correctives for deviation.

Power, defined as effective initiative, is not identical with influence, in the limited sense that person A affects the actions of person B. Influence of this kind may occur with little or no relation to the effectiveness of person A’s actions in the group. Power is implemented by sanctions, while influence is not.

Status (rank in power) is not identical with popularity or degree to which the person is liked. In fact, status and popularity may be poorly correlated (see Hare 1962, p. 115; Sherif & Sherif 1964, chapter 6). Nor should status be confused with the use of force or aggression. As Whyte (1943) showed, even in a street corner group whose members valued masculine toughness, the best potential fighter was not necessarily highest in status. Status was rooted in “mutual obligations” incurred among the group members over time and the reliability with which a member lived up to his obligations.

Since it is defined as effective initiative, status in a group is necessarily hierarchical. The highest status represents the leadership role. Especially in societies or situations where social equality is emphasized, the operational leader, defined by observation of his effectiveness over time, may not be openly chosen “leader” by the group’s members. The accompanying figure is a diagram of the stabilization of statuses, based on experimental findings for six groups. At time Tt (the top) the individuals in two collections have first encountered one another. The circles represent the individuals and indicate that, at that time, independent observers do not agree on regularities in the relationships among them from one situation to the next. Instead, their ratings of effective initiative are different in various activities. (See Figure 1.)

Just below, at time Tf, the observers’ ratings be gin to agree (from one activity to the next and one day to the next) that the highest and lowest

positions (represented by triangles) are stabilized. Both in the experiments and in real-life groups, the leader position typically stabilizes earlier than other high positions. This does not imply that group formation consists of the “search for a leader.” On the contrary, leadership is subject to change. As Hofstatter correctly pointed out (1957, especially p. 24), tracing group organization over time is necessary to clear up many glib formulas propounded without sufficient evidence.

At time Tk in the figure, observers are able to agree on the positions most members occupy, except in the middle of the diagram. Again, this is a typical finding. In part it reflects attempts by those in the middle to improve their standing or to align themselves with those higher in status. At time Tl in the diagram the status relationships are stabilized, all observers agreeing on the status structure, which is also revealed in members’ perceptions of it as manifested by sociometric choices obtained from them.

The diagram is intended to be representative. The different patterns of status in the two groups are intended to suggest that there is no predetermined form to the “steepness” or “flatness” of the hierarchy. Group formation represents an ideal occasion to study factors affecting the organization of groups, but little research has been done on this problem.

The rate of stabilization varies. In the experiments, the groups stabilized within about a week of continuous living together. Other investigators have reported the discernible beginnings of group structure among individuals meeting in the same location with similar activities within three to five meetings of a few hours duration (Merei 1949). Environmental events are as important as internal events in affecting the speed of stabilization. The stability of the pattern is sensitive to the introduction of new members, to changes in location and facilities, and to outside threat or emergency.

In particular, the stabilization of group structure is never independent of relationships with other groups. Prolonged competition between groups for mutually incompatible goals is particularly effective in quickly stabilizing a structure. Important intergroup confrontations, especially those resulting in defeat or humiliation, produce changes in the internal organization of a group (Sherif 1966).

The leader of a group, although most powerful, is still a member subject to loss of status. When the group structure is stabilized in some degree for the time, no person within it is free to ignore its regulation. It defines for members the bounds of “we” or the in-group, as compared to others who are not members. If sufficiently stabilized, the group can continue after a leader’s departure with little disruption (see Toki 1935). This persistence of a group structure and the effects of the group even on the leader are clearer in terms of the normative property of group formation.

Formation of group norms

As a group structure takes shape, members come to prefer certain ways of going about their important activities. They may adopt a group name. They set up standards for the ways members should and should not behave among themselves and with outsiders. “Norm” is a general term to refer to such results of interaction which produce regularities among group members.

Unlike the “norm” on an examination or a test, a group norm does not necessarily refer to the average of individual behaviors. It designates what is expected as proper, as moral, or even as ideal. Yet a group norm seldom denotes a single action as the only way to behave. A range for individual variations is permissible in any group. A norm de notes the range ofbehaviors that members come to deem socially desirable and acceptable (latitude of acceptance) and a range of behaviors condemned as objectionable (latitude of rejection).

A norm implies, therefore, an evaluative scale (measuring rod) defining for individual members a latitude of acceptance and a latitude of rejection, to regulate their behavior in matters of consequence to the group (Sherif et al. 1965). Not all social behavior is regulated by clear-cut norms, particularly when groups are in formation.

How can a group norm be detected? There are at least three objective ways:

(1) By observing similarities and regularities in the behaviors (words and deeds) found among one set of persons but not another set in a similar situation.

(2) By observing correctives (sanctions) for certain behaviors and praise or reward for others. Reactions to deviations are among the best evidence of the bounds of acceptable behavior. These may range from disapproval, frowns, and correctives to threats and actual punishment.

(3) By noting the increasing similarity or convergence over time in the behaviors of individuals who initially behaved differently. For example, the entrance of a new member into a group provides an opportunity to detect the existence of its norms.

When groups are in the process of formation, as in the experiments, one of the best indicators of their stability is the degree of consensus among members on the correctness of their norms and the degree to which the latitude of acceptance is binding without direct social pressure or threat of sanctions. Stabilization of the set of norms is indicated when members privately regulate their own behavior within the latitude of acceptance. The person’s own conception of how he should behave and how others should act comes to fall within bounds defined by the norms. Especially when the individual has had a part in creating the norms as a group member, they become aspects of his self concept relative to others. He experiences personal guilt or shame if he violates them.

The personal acceptance of group norms during group formation accounts in large measure for the tenacity of tradition once established. Merei (1949) demonstrated this tenacity by permitting play groups to develop procedures and rules and then introducing a new child who was older and had evidenced leadership skills in other situations. In every group the new child quickly found that his superior skills in coordinating play were not effective, for no one paid attention. In every case, the new child was then absorbed into the group, adopting its traditions and rules.

Sherif and Sherif (1964) present evidence that the stringency of norms and resistance to their change varies according to the importance of the norm for the group. Violations in major activities or in dealings with outsiders, exposure of group secrets, or other behavior jeopardizing the maintenance of the group were unerringly responded to by strong sanctions such as expulsion, threat, or physical punishment. Even leaders whose actions exposed the group or its members to humiliation, embarrassment, or danger were chastised.

In less important activities the range of tolerance for individual differences was much wider, particularly for the leader and higher-status members. In matters of daily routine or amusement strictly within the group, leaders were free to innovate and sometimes engaged in behavior that would not have been tolerated in lesser members. In these fairly stable groups, the great bulk of conforming behavior occurred without direct social pressure or threat of sanctions, particularly behavior by members of moderate or high status.

Differential effects on behavior

The formation of a group structure and norms has consequences for the attitudes and behavior of individuals within its fold. These consequences may be referred to as the differential effects of group formation.

Any social situation provides a context for behavior which differs from that in a solitary situation. The context includes the other people present, the activities and tasks undertaken, the physical site and its facilities, and the person’s relationship to all of these. Experiments have repeatedly shown the differential effects of various aspects of the social context on behavior as compared to behavior when alone.

The formation of a role system and norms during interaction among persons over time brings about alterations in the relative contribution of the task, activity, setting, and individual reactions. When the persons are at first simply together, with out stabilized reciprocities, their personal charac teristics and skills relative to the tasks and those of other people are important determinants of be havior (Gibb 1954; Hare 1962). As the process of group formation starts taking shape, the developing organizational and normative schemes become more and more binding on members. As a result, over time, the characteristics of the task and location— in short, immediate situational factors—recede in relative importance, and behavior increasingly reflects the person’s role in the group, the roles of others, and the emerging norms.

The group formation experiments of Sherif and Sherif ([1948] 1956, chapters 6 and 9) traced the development of a “we-feeling.” It was found that sociometric friendship choices became almost exclusively concentrated within the group, even though initial choices before group formation had been given predominantly to persons placed (deliberately) in another group. In one experiment it was shown that estimates made by members of one another’s performance became significantly re lated to the member’s status, the relationship being closer when the structure was more stable. Per formance by high-status members was overestimated, and that of low-status persons was minimized.

In another experiment the groups formed separately, then competed for a series of mutually exclusive goals. As predicted, norms developed in each group justifying hostility to the other group. The performance by members of the other group in a novel task was estimated to be significantly lower than performance by members of the ingroup, revealing the prejudicial norm in the judgments of individual members (Oklahoma, University …1961).

In proportion to the significance a particular group has in a person’s life, his membership in it affects his attitude and behavior. As the group formation stabilizes, his sense of identity becomes tied to being a member of that group, proportional to its scope and importance in his daily living. For this reason the socialization of the person is incom pletely described by reference only to his acquisition of formal prescriptions from family, school, and other official institutions. From early childhood through adolescence, groups formed among age-mates exert compelling impact upon the person’s conceptions of what is desirable for him, what is acceptable in others, and what is right and wrong (Campbell 1964; Sherif & Sherif 1964). In other words, they become aspects of the individual’s own conscience.

Recognition of the consequences for the self concepts and attitudes of participants has led to attempts in various countries to utilize group formation for corrective and therapeutic purposes. The varied outcomes reveal both gaps in the knowledge of group formation and lack of familiarity with the knowledge available on the part of many practitioners (Rosenbaum & Berger 1963, especially pp. 1-32).

The emphasis earlier in this article on the motivational base of group formation and on the importance of environmental alternatives suggests fruitful lines of inquiry. The significance of the motivational base was revealed in a study of group formation among “emotionally disturbed young adolescents of poor prognosis” by Rafferty (1962, p. 263). They interacted rather freely in a wide range of activities for five hours daily, five days a week for nine months. They lacked motivation toward the institution’s aim of changing their behavior, and their personal disturbances hindered any kind of stable interpersonal relationships. However, they did unite with incipient group formation in activities reflecting a motivation genuine to them: defiance of the hospital staff in forbidden activities.

This instructive finding raises the issue of predicting or controlling the character of the structure and norms during group formation. Here the importance of the environmental setting and the behavioral alternatives it encourages or permits becomes evident. In the group experiments referred to frequently here, solidary groups devoted to constructive activities were formed simply by placing unacquainted persons in situations of high appeal to them, with facilities and conditions so arranged that coordination of activity was the only way to secure individual satisfaction. Subsequently, con flict and hostility between the groups were pro duced, followed by their reduction through coop erative efforts by the groups. These effects were achieved merely by varying the facilities available, the other persons present, and other conditions external to both groups. Future research on group formation and its applied implications might prof itably focus on the effects of varying the envi ronmental alternatives and facilities available to groups, including other groups and persons, upon the character of the organization and norms that develop.


The material briefly presented above warrants the following conclusions:

Whenever individuals with similar motives, similar frustrations, and similar personal concerns for acceptance, for recognition, and for stabilizing their perception of themselves encounter one another, and when these goal-directed concerns are not effectively dealt with through the established channels of custom and law or the routine of prevailing arrangements of social organization—these indi viduals then tend to interact among themselves.

Repeated interaction in some common striving is conducive to differentiation of roles or functions to be performed toward the common end. And differentiation of roles and functions among the participating individuals, over a time span, is the pattern or formation that can be designated as the group. Every such human formation creates its own set of rules or norms to stabilize the regulation of behavior and the attitudes of members within its bounds.

In a natural group, as in any other group, the rules or norms that count and have salience in the eyes of the members are generally the ones that pertain to the existence and perpetuation of the group and to the spheres of activity that are related to the common motivational concerns that were initially conducive to repeated interaction among the individuals in question.

The main properties of the group thus formed are an organization (structure) of roles and statuses and a set of rules or standards (norms) for its activities toward the common ends. The “or ganization” (which need not be formally recog nized) and the set of norms (which need not be formally written in blueprints) define the sense of “we-ness” cherished within the group and upheld by its members in their dealings with outsiders.

In time, the standards or norms shared in the feeling of “we-ness” become personally binding for individual members. The members who are worthy and true make their judgments and justify or condemn events within the sphere related to their “we-ness” in terms of their sense of identification with the group. Proportional to the importance of the group in the lives of its members, the person’s self picture, his sense of personal accountability, his loyalty, and the “do’s” and “don’ts” of the group become parts of his conscience. Hence, group for mation has broad implications for the regulation of individual attitude and behavior with and with out external sanctions and controls.

Muzafer Sherif and Carolyn W. Sherif

[Directly related are the entriesCohesion, Social; Conformity; Leadership, Other relevant material may be found inAttitudes; Friendship; Interaction; Norms; Role; Self Concept; Social Psychology; Social Structure; Socialization.]


Bales, Robert F. 1950 Interaction Process Analysis: A Method for the Study of Small Groups. Reading, Mass.: Addlson-Wesley.

Blau, Peter M.; and Scott, W. Richard 1962 Formal Organizations: A Comparative Approach. San Fran cisco: Chandler. → Contains an extensive bibliography.

Bonner, Hubert 1959 Group Dynamics: Principles and Applications. New York: Ronald Press.

Bowen, James 1962 Soviet Education: Anton Makarenko and the Years of Experiment. Madison: Univ. of Wis consin Press.

Campbell, John D. 1964 Peer Relations in Childhood. Volume 1, pages 289-322 in Martin L. Hoffman and Lois W. Hoffman (editors), Review of Child Development Research. New York: Russell Sage Foundation.

Cartwright, Dorwin; and Zander, Alvin (editors) (1953) 1960 Group Dynamics: Research and Theory. 2d ed. Evanston, 111.: Row, Peterson.

Gibb, Cecil A. 1954 Leadership. Volume 2, pages 877-920 in Gardner Lindzey (editor), Handbook of Social Psychology. Cambridge, Mass.: Addison-Wesley.

Golembiewski, Robert T. 1962 The Small Group: An Analysis of Research Concepts and Operations. Univ. of Chicago Press.

Hare, Alexander P. 1962 Handbook of Small Group Research. New York: Free Press.

Hofstatter, Peter R. 1957 Gruppendynamik: Die Kritik der Massenpsychologie. Hamburg (Germany): Rowohlt.

Landesco, John 1929 Organized Crime in Chicago. Pages 823-841 in Illinois Association for Criminal Justice, The Illinois Crime Survey. Chicago: The As sociation.

Lewin, Kurt; Lippitt, R.; and White, R. K. 1939 Pat terns of Aggressive Behavior in Experimentally Cre ated “Social Climates.” Journal of Social Psychology 10:271–299.

Makarenko, Anton S. (1933) 1951 The Road to Life: An Epic of Education. 3 vols. Moscow: Foreign Lan guages Publishing House. → First published as Peda-gogicheskaia poema.

Merei, Ferenc 1949 Group Leadership and Institution-alization. Human Relations 2:23–39.

Moreno, Jacob L. (1934) 1953 Who Shall Survive? Foundations of Sociometry, Group Psychotherapy and Sociodrama. Rev. & enl. ed. Beacon, N.Y.: Beacon House.

Oklahoma, University of, Institute of Group Relations 1961 Intergroup Conflict and Cooperation: The Robbers Cave Experiment, by Muzafer Sherif et al. Norman, Okla.: University Book Exchange.

Rafferty, F. T. 1962 Development of a Social Structure in Treatment Institutions. Journal of Nervous and Mental Disease 134:263–267.

Roethlisberger, Fritz J.; and Dickson, William J. (1939) 1961 Management and the Worker: An Ac count of a Research Program Conducted by the West ern Electric Company, Hawthorne Works, Chicago. Cambridge, Mass.: Harvard Univ. Press. → A paper back edition was published in 1964 by Wiley.

Rosenbaum, Max; and Berger, M. (editors) 1963 Group Psychotherapy and Group Function. New York: Basic Books.

Shaw, Clifford R. 1929 Delinquency Areas: A Study of the Geographic Distribution of School Truants, Juvenile Delinquents, and Adult Offenders in Chicago. Univ. of Chicago Press.

Sherif, Carolyn W.; Sherif, M.; and Nebergall, R. E. 1965 Attitude and Attitude Change: The Social Judg ment—Involvement Approach. Philadelphia: Saunders.

Sherif, Muzafer (1936) 1965 The Psychology of Social Norms. New York: Octagon. → A paperback edition was published in 1966 by Harper.

Sherif, Muzafer 1966 In Common Predicament: Social Psychology of Intergroup Conflict and Cooperation. Boston: Houghton Mifflin. → A British edition was published by Routledge as Group Conflict and Co operation: Their Social Psychology.

Sherif, Muzafer; and Cantril, Hadley 1947 The Psychology of Ego-involvements, Social Attitudes and Identifications. New York: Wiley; London: Chapman & Hall.

Sherif, Muzafer; and Sherif, Carolyn W. (1948) 1956 An Outline of Social Psychology. Rev. ed. New York: Harper.

Sherif, Muzafer; and Sherif, Carolyn W. 1964 Ref erence Groups: Exploration Into Conformity and Deviation of Adolescents. New York: Harper.

Thrasher, Frederic (1927) 1963 The Gang: A Study of 3,313 Gangs in Chicago. Abridged and with an introduction by James F. Short. Univ. of Chicago Press.

Toki, K. 1935 The Leader-Follower Structure in the School Class. Japanese Journal of Psychology 10:27–56. → A discussion of Toki’s article appears on pages 8-10, 604, 608, 613, and 631-635 in Fundamentals of Social Psychology, by Eugene L. and Ruth Hartley.

U.S. Office of Strategic Services, Assessment Staff 1948 The Assessment of Men. New York: Rinehart.

Whyte, William F. (1943) 1961 Street Corner Society: The Social Structure of an Italian Slum. 2d ed., enl. Univ. of Chicago Press.

Zorbaugh, Harvey W. 1929 The Gold Coast and the Slum. Univ. of Chicago Press.


“Role” refers to a set of “expectations” for inter action between a person who holds one position in a group and another person who holds a reciprocal position. In other words, there can be no “leader” role without a “follower” role. Since the person playing the role has a personality (or self) which he brings to the situation, the behavior of a person in a role will be some combination of the tendencies of his personality and the expectations of his role.

Roles in discussion groups, which are the focus of much of the research on groups, are usually not fixed. Any aspect of an individual’s behavior that is initially an expression of his personality can come to be expected by other group members and thus become part of his role. In general, the dimensions of role are the same as the dimensions of interaction and of personality. Roles have a form and a content, where the form includes the frequency of interaction and the communication net work and the content includes task and socialemotional behavior. Within the task area, the expectations refer to problem-solving behavior and within the social-emotional area, according to one recent formulation, to behavior along at least three dimensions: dominance-submission, positive-negative, and joking-serious (Couch I960; Hare 1962).

Patterns of role differentiation. Natural groups which have been in existence for some time will probably have a greater degree of role differentiation than laboratory groups (Sherwood & Walker 1960), unless the group organization allows only one leader who is expected to perform all functions. Thus, in some industrial and governmental conference groups a “sharing” of leadership is re sisted by the members (Berkowitz 1953). Over the life of a group there may be more differentiation in roles when functional problems of the group become more acute, for example, when the group is under stress or the task is complex (Bales 1953). Similarly, members in cooperative groups are more apt to be differentiated in their functions than members who are competing (Homans 1961. p. 135).

The small military unit or industrial work team is perhaps the best example of the simplest form of role differentiation. In each case one or two members may have clearly defined functions with little overlap in their “expectations,” while the remainder of the members play undifferentiated roles of “soldier” or “worker.” In discussion groups (and especially in laboratory groups) it is not always clear whether a single person plays one role or many roles, or whether these roles are assumed at the same or at different times.

In some research all members are viewed as specializing in some content area, with those with the highest interaction rates identified as the leaders (Heinicke & Bales 1953; Grusky 1957). In other research some individuals may be described as playing a variety of roles and others none (Davis et al. 1961; Cloyd 1964). In either case the authors are not usually concerned with the rights and duties associated with each role. They rather use the terms “expectations” in the sense that group members can predict or “expect” that a group mem ber will act in a given way on the basis of past performance. There is no sense that the member is “obliged” to act in this way to fulfill a function in the group or that he is entitled to any special privileges for performing this function [seeEthics, article onEthical Systems and Social Structures].

In some research, the evidence that a person is playing a given role is that other members associate a certain set of behavioral characteristics with him; these behavioral characteristics are presented as a cluster without any special rationale concerning their interconnection (Bates & Cloyd 1956; Davis 1961; Cloyd 1964). In other research a per son is described as playing a role because he has been nominated as a frequent contributor of some specified type of behavior, for example, giving in formation or giving opinion (Bales 1958; Borg 1960). An even less functionalist view is taken in research in which some members are described as playing individual or self-oriented roles that pre sumably are extraneous to the group task (Benne & Sheats 1948). Another variation on the use of the term “role” is for the experimenter to describe someone as playing a role such as “newcomer” (Mills et al. 1957) or “doctor’s assistant” (Margolin 1952) without actually assessing the perceptions of the other group members.

Most of the research on small groups has so far not used a definition of role as a set of rights and duties but rather as a description of recurring pat terns of behavior. But the former, more limited definition of role still appears to be theoretically useful even though it is not widely used. For in stance, in those studies that report the interaction in initially “leaderless” groups, the expectations for the average member are probably most clearly re lated to the directions given by the experimenter at the beginning of the session. Often, indeed, the experimenter will test to see if the subjects have been as friendly or as competitive as he directed them to be, but he will not refer to these instructions as “role expectations” (Olmsted 1954).

The power of an experimenter’s instructions is evident in an experiment in which one member in each of several discussion groups was given more information about the task than the other mem bers. In some of these groups the experimenter announced, “Some of you may have more information than others.” In other groups he gave the im pression that all members were equally informed. It was observed that group members reacted negatively to the best-informed man unless they had been led to “expect” that he would play a different role (Shaw & Penrod 1962).

Although early research on groups, especially with children, often described “individual” or “self-oriented” roles as if the only role of some indi viduals was to satisfy their own needs in the group (Benne & Sheats 1948), later formulations by Redl (1942), Bion (1961), and Stock and Thelen (1958) suggest that all roles in the group serve some function. However, some of these roles, par ticularly the ones that allow members to deal with emotional themes, may not be recognized as part of the “official” group structure. For example, in a training group composed of teachers who were being led by Bion’s methods of “interpretive group discussion,” the group members persisted, without success, in trying to induce the leader to act as a therapist. Finally six members did not appear at the scheduled group session and a spokesman for the absentees sent in a paper on truancy to be discussed by the remaining members. Thus, by taking the “role” of absent members, some of the group acted out the group’s need for flight from the task (Herbert & Trist 1953).

Task versus social-emotional roles. The most common division of roles that has been described in small discussion and work groups, as well as in families viewed as small groups, is specialization in the task and social-emotional areas (Bales & Slater 1955; Bales 1958). This has been shown most clearly in small laboratory groups in which mem bers have a high degree of consensus at the end of a meeting on the relative amount of interaction in each of these areas exhibited by group members. These groups appear to recognize two kinds of role specialists: one an “idea man” who concentrates on the task and plays a more aggressive role; the other a “best-liked” man who concentrates on social-emotional problems of group process and member satisfaction, giving emotional rewards, and playing a more passive role. However, in groups similar to these in which there is less con sensus on the status of members, a third type of person appears to be present, one who talks a great deal but who is not well liked or highly rated on his task ability. He has been referred to as a “deviant.” In addition, researchers have found a more passive task specialist and a “popular” person (Slater 1955).

The so-called “deviant” who overtalks is probably expressing the group’s anxiety about the discussion task. To please the experimenter, the group allows a member to fill the time with “discussion” even though he may not be the most effective at the task. This tendency to have a high or low inter action rate has often been reported as the first of three factors or dimensions that may be used to describe behavior, the other two being task be havior and sociability (Carter 1954). However, the frequency of interaction should probably be con sidered an independent dimension because it de scribes the “form” of interaction and not because it represents a separate area of “content.”

In families, the father has been identified as the task specialist, the mother as the social-emotional specialist (Parsons & Bales 1955). A similar dichotomy has been reported among caseworkers in a welfare agency where some colleagues were respected and sought out for consultation on cases and others were attractive because of their sociable companionship (Blau 1962). However, a study of therapy groups reported that the distinction was not evident in this setting, since the “task” of the group was to deal with social-emotional problems (Talland 1957).

When the same individual is required to play the role of both the task and the social-emotional leader, he may find some aspects of the roles in compatible and thus experience “role conflict” (See-man 1953). Officers in small military units, for example, may be required to consider the personal problems of their men as well as to be task leaders. The second role requires distance, since some assignments must be made without regard to personal feelings, yet the first role requires closeness and intimacy (Hutchins & Fiedler 1960).

The joker. In addition to these two roles, a third role has been identified in some groups which actually has a long history in many cultures. This is the role of the clown or the joker. Just as the English court jester’s costume set him apart from the group, so the joker tends to take a somewhat marginal position with regard to the task. He tends to look at things differently, providing both a source of humor and of new ideas. Under the cover of wit he is able to introduce ideas that the group might otherwise find unacceptable.

In its literary form, comedy has been described as “an escape, not from truth but from despair: a narrow escape into faith” (Fry 1960, p. 27). One must be able to grasp the tragic nature of life before one can go on to grasp its comic nature. Thus, the comic person in a play or in a real-life small group is often one who gives special insight into the problem. This is evident from a study of a series of “great-books” discussion groups where the joking role was most highly correlated with the role of providing “fuel” for the discussions in the form of new ideas and opinions (Davis et al. 1961). In a study of laboratory discussion groups the following cluster of behavioral characteristics was identified as composing a role: jokes and makes humorous remarks, is liberal, challenges other’s opinions, gets off the subject, is egotistical, is cynical, and interrupts others (Cloyd 1964). Within the dimension of “joking versus task-serious behavior” there may be different styles of wit. Thus, sarcastic wit may be perceived as powerful but unpopular in a group, while clowning wit may be seen as popular but powerless (Goodchilds 1959).

The member. The fact that simple membership in a group carries with it a set of rights and duties is evident in the group’s concern for the “silent member.” Although silence may be functional if it means that more able persons are being allowed to solve problems (Homans 1961, p. 136), group members are usually dissatisfied with a member’s performance if he does not participate. This dissatisfaction may be reduced if it is made clear at the outset that certain members will not participate at all (Smith 1957). On the other hand the group will be more concerned if a member appears indifferent and neglectful in his role (Rosenthal & Cofer 1948).

Another type of member role that has received some attention in the literature is that of the “new comer.” Group members will have an easier time assimilating the newcomer if they have been told to expect change (Ziller et al. 1961), provided that the newcomer is not seen as too different from other group members (Ziller et al. 1960) and that group members have already had a pleasant time with each other (Heiss 1963). In any event there will probably be a minimal alteration in the role patterns of the old members at first (Mills et al. 1957).

Other roles. Other roles may be found in groups with special tasks. In therapy groups a “doctor’s assistant” may fill the group’s need to keep the discussion going when the therapist is playing a rather passive role and a “help-rejecting complainer” may give the group case material to discuss (Frank et al. 1952; Margolin 1952). Somewhat similar to the “doctor’s assistant” may be the “feeder-to-leaders” identified in a sociometric study of a home for girls (Jennings 1947). These were girls who would have their ideas accepted after they had been en dorsed by highly chosen leaders, although they were not highly chosen themselves [seeLeader Ship].

The classic study of authoritarian and democratic group atmospheres by Lewin, Lippitt, and White does not focus on role differentiation; how ever, two types of roles are mentioned in addition to the leaders. In the “democratic” groups the authors describe two boys who are allies of the adult leaders, and in the “autocratic” groups they describe a scapegoat who receives the aggression of the group (White & Lippitt 1960, pp. 160-186). Thrasher (1927) in his study of gangs in Chicago gives a longer list of special roles, to which he gives such names as brains of gang, funny boy, sissy, show-off, and goat. Redl (1942) lists ten types of roles of central persons, which may be grouped into three categories: identification ob jects, objects of drives, and ego supports.

In addition to roles which may arise naturally in groups, some authors have suggested certain roles which should be introduced if a group is to operate with maximum efficiency. For example, Jenkins (1948) suggests that effective discussion requires attention to such mechanics of operation as awareness of direction, goal, and rate of prog ress. He proposes that groups appoint a “group pro ductivity observer” who will report at the end of each meeting.

Where role differentiation occurs in a group, in dividuals accustomed to playing the same roles in other groups will probably continue to play them in the new situation (Strodtbeck & Mann 1956; Davis et al. 1961). In addition, differences in age, sex, social class, and occupation between members will result in role differentiation in small discussion groups even when these differences are not related to the task at hand (Maas 1954; Torrance 1954).

Six roles in discussion groups. From the review of the literature presented above it is evident that a variety of roles have been identified in discussion groups. The number of roles reported may be small or large depending upon the level of analysis, the task, the group size, and the length of time the members have been together. Although the dichot omy of task versus social-emotional roles is ade quate to describe the basic differentiation in some groups, in others there are probably more distinct roles that have been grouped under these two more general headings. A framework for the description of more distinct and independent roles is provided by Couch’s (1960) factor analysis of categories of interaction in five-man laboratory discussion groups. He finds that interpersonal behavior can be described by the following independent dimen sions: dominance versus submission; positive ver sus negative; task serious versus joking; influence attempts versus receptivity; and surface acqui escence versus resistance. If we assume that be havior at the extremes of each dimension may identify a role, we find a rather good fit with six role patterns identified by Cloyd (1964) without reference to any particular dimensions and with out actually assigning them names.

At the dominance end of Couch’s first factor we find a set of behaviors similar to a cluster identified by Cloyd that includes “aggressive, self-confident, and gets things started.” This role might be called the “high talker.” At the submissive end of Couch’s first factor is Cloyd’s cluster of traits: “modest, shy, and ill at ease.” This role might be called the “silent member.”

In a similar way one could match dimensions and clusters of traits to identify a “supporter” who is friendly and objective, a “critic” who is idealistic and argumentative, a “serious worker” who is de pendable and constructive, and a “joker” who makes humorous remarks and challenges others’ opinions. Cloyd does not provide examples that would fit the last two of Couch’s dimensions; how ever, the last dimension suggests the roles of “con formist” and “nonconformist.”

From this example it will be seen how small group experiments, although independently de signed by researchers using different concepts, can be said to deal with a recurrent, scientifically identifiable class of phenomena. Indeed, such is the wealth of suggestive findings in this area that the organization of our present knowledge has be come both an indispensable and a rewarding preliminary to future progress.

A. Paul Hare

[See alsoInteraction, article onInteraction PRocess Analysis; Leadership; Sociometry.]


Many of the articles listed below are discussed in more detail in Hare 1962, while 11 of them have been reprinted in whole or in part in Hare, Borgatta & Bales 1964. Refer ences to other introductory texts or readers will be found in the bibliographies to Groups, article on THE STUDY OF GROUPS; and Sociometry.For the layman who has no training in statistics but would like to savor some of the interest and importance of the field, Bion 1961; Homans 1961; and Mills 1957 should prove rewarding.

Bales, Robert F. 1953 The Equilibrium Problem in Small Groups. Pages 111-161 in Talcott Parsons et al., Working Papers in the Theory of Action. Glencoe, 111.: Free Press.

Bales, Robert F. 1958 Task Roles and Social Roles in Problem-solving Groups. Pages 437-447 in Eleanor E. Maccoby, T. M. Newcomb, and E. L. Hartley (editors), Readings in Social Psychology. 3d ed. New York: Holt.

Bales, Robert F.; and Slater, Philip 1955 Role Dif ferentiation in Small Decision-making Groups. Pages 259-306 in Talcott Parsons and Robert F. Bales, Family, Socialization and Interaction Process. Glen coe, 111.: Free Press.

Bates, Alan P.; and Cloyd, Jerry S. 1956 Toward the Development of Operations for Denning Group Norms and Member Roles. Sociometry 19:26–39.

Benne, Kenneth D.; and Sheats, Paul 1948 Functional Roles of Group Members. Journal of Social Is sues 4, no. 2:41–49.

Berkowitz, Leonard 1953 Sharing Leadership in Small Decision-making Groups. Journal of Abnormal and Social Psychology 48:231–238.

Bion, Wilfred R. 1961 Experiences in Groups, and Other Papers. New York: Basic Books. → Seven of these papers were published in Human Relations from 1948 to 1951.

Blau, Peter M. 1962 Patterns of Choice in Interpersonal Relations. American Sociological Review 27:41–55.

Borg, Walter R. 1960 Prediction of Small Group Role Behavior From Personality Variables. Journal of Ab normal and Sociu! Psychology 60:112–116.

Carter, Launor F. 195i Evaluating the Performance of Individuals as Members of Small Groups. Person nel Psychology 7:477–484.

Cloyd, Jerry S. 1964 Patterns of Role Behavior in In formal Interaction. Sociometry 27:161–173.

Couch, A. S. 1960 Psychological Determinants of Inter personal Behavior. Ph.D. dissertation, H, arvard Univ.

Davis, James A. 1961 Compositional Effects, Role Systems, and the Survival of Small Discussion Groups. Public Opinion Quarterly 25:575–584.

Davis, James A. et al. 1961 Great Books and Small Groups. New York: Free Press.

Frank, Jerome D. et al. 1952 Two Behavior Patterns in Therapeutic Groups and Their Apparent Motivation. Human Relations 5:289–317.

Fry, Christopher 1950 Comedy. Adelphi Third Series 27:27–29.

Goodchilds, Jacqueline D. 1959 Effects of Being Witty on Position in the Social Structure of a Small Group. Sociometry 22:261–272.

Grusky, Oscar 1957 A Case for the Theory of Familial Role Differentiation in Small Groups. Social Forces 35:209–217.

Hare, A. Paul 1962 Handbook of Small Group Re search. New York: Free Press.

Hare, A. Paul; Borgatta, E. F.; and Bales, R. F. (1955) 1964 Small Groups: Studies in Social Interaction. 2d ed. New York: Knopf.

Heinicke, Christopher; and Bales, Robert F. (editors) 1953 Developmental Trends in the Structure of Small Groups. Sociometry 16:7–38.

Heiss, Jerold S. 1963 The Dyad Views the Newcomer: A Study of Perception. Human Relations 16:241–248.

Herbert, Eleonore L.; and Trist, E. L. 1953 The In stitution of an Absent Leader by a Students’ discussion Group. Human Relations 6:215–248.

Homans, George C. 1961 Social Behavior: Its Elemen tary Forms. New York: Harcourt.

Hutchins, Edwin B.; and Fiedler, Fred E. 1960 Task-oriented and Quasi-therapeutic Role Functions of the 288 GROUPS: Group Performance Leader in Small Military Groups. Sociometry 23:393–406.

Jenkins, David H. 1948 Feedback and Group Self-eval uation. Journal of Social Issues 4, no. 2:50–60.

Jennings, Helen H. 1947 Leadership and Sociometric Choice. Sociometry 10:32–49.

Maas, Henry S. 1954 The Role of Member in Clubs of Lower-class and Middle-class Adolescents. Child Development 25:241–251.

Margolin, Joseph B. 1952 The Use of an Interaction Matrix to Validate Patterns of Group Behavior. Human Relations 5:407–416.

Mills, Theodore M. et al. 1957 Group Structure and the Newcomer: An Experimental Study of Group Expansion. Studies in Society No. 1.

Olmsted, Michael S. 1954 Orientation and Role in the Small Group. American Sociological Review 19:741–751.

Parsons, Talcott; and Bales, Robert F. 1955 Family, Socialization and Interaction Process. Glencoe, 111.: Free Press; London: Routledge.

Redl, Fritz 1942 Group Emotion and Leadership. Psy chiatry 5:573–596.

Rosenthal, David; and Cofer, Charles N. 1948 The Effect on Group Performance of an Indifferent and Neglectful Attitude Shown by One Group Member. Journal of Experimental Psychology 38:568–577.

Seeman, Melvin 1953 Role Conflict and Ambivalence in Leadership. American Sociological Review 18:373–380.

Shaw, Marvin E.; and Penrod, William E. Jr. 1962 Does More Information Available to a Group Always Improve Group Performance? Sociometry 25:377–390.

Sherwood, Clarence E.; and Walker, W. S. 1960 Role Differentiation in Real Groups: An Extrapolation of a Laboratory Small-group Research Finding. Sociology and Social Research 45:14–17.

Slater, Philip E. 1955 Role Differentiation in Small Groups. American Sociological Review 20:300–310.

Smith, Ewart E. 1957 The Effects of Clear and Un clear Role Expectations on Group Productivity and Defensiveness. Journal of Abnormal and Social Psychology 55:213–217.

Stock, Dorothy; and Thelen, H. A. 1958 Emotional Dynamics and Group Culture: Experimental Studies of Individual and Group Behavior. New York Univ. Press.

Strodtbeck, Fred L.; and Mann, Richard D. 1956 Sex Role Differentiation in Jury Deliberations. Sociometry 19:3–11.

Talland, George A. 1957 Role and Status Structure in Therapy Groups. Journal of Clinical Psychology 13: 27–33.

Thrasher, Frederic M. (1927) 1963 The Gang: A Study of 1,313 Gangs in Chicago. Abridged and with an in troduction by James F. Short. Univ. of Chicago Press.

Torrance, E. P. 1954 Some Consequences of Power Dif ferences on Decision-making in Permanent and Temporary Three-man Groups. Washington State University, Research Studies 22:130–140.

White, Ralph K.; and Lippitt, Ronald 1960 Autoc racy and Democracy: An Experimental Inquiry. New York: Harper.

Ziller, Robert C; Behringer, Richard D.; and Good-childs, Jacqueline D. 1960 The Minority New comer in Open and Closed Groups. Journal of Psychology 50:75–84.

Ziller, Robert C; Behringer, Richard D.; and Jansen, Mathilda J. 1961 The Newcomer in Open and Closed Groups. Journal of Applied Psychology 45: 55–58.


It was a practical interest in group perform ance which led to the massive literature dealing with experimental studies of groups. The industrial revolution and the growth of the large factory raised such practical questions as: How can groups of workers become more productive? How might the performance of the individual worker be affected by the presence of others? When is a group more effective than the sum of its individual members? Thus, the typical study of group performance will vary some factor and measure differences in quantity, quality, speed (number of units produced per unit time), or efficiency (number of units produced per man per unit time). Some studies might concentrate on homogeneity or similarity of behavior and production of indi viduals in the group as compared to that of in dividuals working alone. The original industrial interest spread to other areas, such as military situations, research laboratories, and classrooms, where the problems were often quite different. Reviews of studies of group performance may be found in Hare (1962), Kelley and Thibaut (1954), and Lorge and his associates (1958).

Studies of group performance have been grouped into four major problem areas: (a) group produc tivity—e.g., studies of how many relays can be wired by a group in a given period of time, or how many addition problems can be completed; (b) group problem solving—e.g., studies of the solution of puzzles, construction of a complex design involving thought and planning, creativity in determining unusual uses for everyday implements; (c) group judgment—e.g., group estimation of the number of parachutes seen in an air photo of military exercises, estimation of number of beans in a jar, judgment of lengths of lines or movement of lights; and (d) group learning and retention—e.g., group memorizing of a complex story, learning a maze or series of words, military units learning complex strategy as a unit. Although these categories are convenient, it is obvious that a given task will often involve several of the above problems in combination: the group must form some judgments in learning and retention, learning and retention are involved in problem solving, group productivity may be improved through creative problem solving in the reorganization of group operations, etc.

Individual versus group performance. Early observations suggested that workers on production lines would tend to increase their level of activity when working together. Floyd H. Allport (1920) considered such “social facilitation” in terms of classical conditioning. The sounds of tapping from the shoemaker’s hammer become conditioned stimuli for the muscular movements which move his hammer. When he works in the presence of others, the tapping sounds from his co-workers serve as additional conditioned stimuli and thus increase further the rate at which the shoemaker pounds his shoes. Allport found, in fact, that subjects given word-association tasks, vowel-can cellation tests, or multiplication tests performed more rapidly when working in the presence of others. This was true even when rivalry was diminished by using instructions and by not per mitting knowledge of the scores of others. [See the biography ofAllport.]

Early studies of problem solving also indicated the superiority of groups over individuals. It was suggested that this resulted from erroneous solutions being rejected in group discussions (Shaw 1932). It would follow that the superiority of the group over individual problem solving would be greater in tasks which permitted a greater variety of responses, with elimination of incorrect solutions being more crucial. Such was, indeed, demonstrated by Thorndike (1938). [See the biography ofThorndike.]

In one of the few clear-cut experiments on group versus individual learning, subjects were asked to learn lists of nonsense syllables. Some learned cooperatively, in groups of three, and then alone; for others the order was reversed. It appeared that, generally, subjects learned more rapidly in groups and after the group experience could learn better on their own. Some superior subjects, however, did not benefit from the group learning experience (Perlmutter & de Montmollin 1952).

There is evidence that groups can inhibit the be havior of individuals. In an early study most children asked to turn a crank to move a flag to a finish line tended to show “social facilitation” when oper ating together. However, some operated less effectively when coacting with others. This decrement was attributed to overstimulation (Triplett 1898).

While the studies of individual versus group performance indicated some superiority of group per formance, they also suggested limitations on any such conclusions. The problem can be reconsidered in terms of the factors which lead to variations in group performance.

Size of group. In an early study by Moede (1927), a man was given a rope to pull as hard as he could, then a second man was added and given the same instructions, then others, until eight men were pulling, with the strength of pull measured. The first man pulled, on the average, 63 kilograms; the two together pulled 118; three, 160; and eight, 248 kilograms. Clearly the total amount of pull in creased with each man, but the work per man was reduced with additional members—each new mem ber reduced the work per man by approximately 10 per cent. This early study of group productivity is a prototype of later research on size of group in group performance. The results of later studies seem to be the same. A study of problem solving in the game “Twenty Questions” found four-person groups asking fewer questions than dyads, but the four-person groups were less efficient (Taylor & Faust 1952). Thomas and Fink, summarizing the literature, say that generally “both quality of performance and group productivity were positively correlated with group size under some conditions, and under no conditions were smaller groups su perior. In contrast, measures of speed showed no difference or else favored the smaller groups” (1963, p. 373).

Cohesiveness and morale. In 1927 the Haw thorne plant of the Western Electric Company initiated a series of experiments to raise productiv ity and morale. A group of female workers en gaged in wiring relays was placed in a separate room and, in the tradition of efficiency experts Frederick W. Taylor and Frank B. Gilbreth, changes in working conditions were introduced which were calculated to improve efficiency—changes to piece work production, increased rest periods, hot meals, earlier quitting times, etc.—and with each change productivity increased. However, when all these im provements were removed, productivity rose still higher. It was pointed out that, clearly, the physical factors were not as crucial for increasing produc tivity as cohesiveness and morale [seeCohesion, Social; Industrial Relations; see also Roethlis-berger & Dickson 1939, p. 86].

Later studies in various types of work groups, including railroad maintenance workers, office staffs, and machinery workers, also indicated that productivity increased with attraction to the work group (Michigan, University of …1954, p. 7). Other research indicated similar relationships in military crews (Berkowitz 1956). In a careful experiment, Van Zelst, working in conjunction with the supervisor and foreman of a large construction project, made up some work groups of carpenters and bricklayers according to their personal choices of co-workers and some of members who were not mutually attracted to one another. The former groups showed greater job satisfaction, lower labor and material cost, and lower turnover rates (1952).

There are, however, limitations to any generali zation that cohesiveness increases productivity. Mutual friendliness, when it shows itself in high sociability and horseplay, can decrease productivity (Horsfall & Arensberg 1949). Thus, group goals and group norms must also be considered.

Interdependence, cooperation, and competition. Patterns of interdependence can have a great effect on group performance. The individual worker on the production line may be under great pressure to work rapidly either through the “pull” of the man next in line, who is waiting for the next piece, or through the “push” of the preceding worker, whose rapid rate of work leaves work piling up on his successor. In the British coal mines, a change to a “longwall method” of mining led to each shift of workers be coming less aware of other shifts and less aware of the extent to which each was dependent upon the other. Accidents increased because a preceding shift departed without allowing proper safeguards for those that followed. For similar reasons, pro ductivity decreased. A major remedy involved in creasing both formal and informal communications between shifts and, through other means, making shifts and workers aware of their interdependence (Trist & Bamforth 1951). Thomas (1957) indicates the importance of considering both interdependence with respect to means and with respect to goals. He found greatest productivity in a group with division of labor where both types of interdependence ob tained, though there were signs of greater emotional tension as well.

Early studies indicated that rivalry, or competition, tended to increase rate of performance, though sometimes at the expense of quality (Triplett 1898). However, later studies found more rapid performance among persons who are cooperating.

Part of the answer lies in the extent to which there is also interdependence with respect to means. When each person works on a parallel task, as in the Triplett experiment in which each subject turned his own crank, with speed being the basis of success, then competition will likely be more effective in increasing performance. When each person depends on others for his own movement— for example, three persons raising or lowering cor ners of a triangular board so as to center carpen ter’s levels (Raven & Eachus 1963)—then competition will be detrimental. With competition and interdependence of means, each person is concerned about his activity helping others. There is less division of labor, and a person may not wish to give information to a co-worker who might thus be helped to defeat him (Deutsch 1949).

Division of labor and information. While divi sion of labor may be beneficial to a cooperative work group, mutual interdependence also imposes additional pressures on the individual, resulting in greater pressures to produce but also greater ten sion and greater loss of individuality and crafts manship. This has been observed in industrial settings. It has been demonstrated that such resulting tensions can be alleviated somewhat, with greater group cohesiveness, greater acceptance of influence from co-workers, and less hostility, through clarifying for the individual the nature of the final product and the ways in which the individual worker’s task fits in with that of his co-workers (Raven & Reitsema 1957).

The question arises as to how information should be distributed for maximal group retention. If, for example, a complex military tactic must be learned by a squad, giving all the information to each member would increase learning time and lead to an increased rate of forgetting. On the other hand, giving each person a nonoverlapping portion, while reducing redundancy, increases the possibility of a piece of information being permanently lost if only one member forgets it. Zajonc and Smoke (1959) sug gest a mathematical model for dealing with this problem, but extensive empirical testing is still lacking.

Group norms and decisions. In the Hawthorne study, while it was found that group morale contributed to higher productivity, it was also observed that norms developed which prevented production from either exceeding a given level (such a worker would be considered a “rate-buster”) or falling be low (being a “slacker”). Similarly, the study by Seashore found that increased cohesiveness did not lead to greater productivity but rather to greater homogeneity of performance in a work group (Michigan, University of …1954). It has been demonstrated in experimental situations that a highly cohesive group will not only be better able to influence a member to increase his production but also to decrease it (Schachter et al. 1951).

Lewin suggested that changing the level of pro ductivity of a work group involves a process of “unfreezing” an old norm, then change, and “re-freezing” at a new normative level (1951, p. 228). A reduction in productivity, due to a change in methods of production, may lead to establishment of a lower group norm, which will remain long after necessary relearning has taken place. Members who exceed the norm may be subject to strong group pressure to work more slowly. This suggestion led to studies which indicated that involving the work group in the re-evaluation of the job, with establishment through group discussion of an appro priate group standard of production, will ultimately lead to a higher level of production with less ten sion and higher morale among the workers (Coch & French 1948).

With respect to group judgments, group norms might also have a deleterious effect. Though a group would be more likely to have the combined skills necessary to make a proper judgment, it may sometimes happen that a majority is wrong. In that case, there will be strong pressures for an individual who might otherwise judge correctly to refrain from uttering the correct answer and, instead, to go along with the group. A properly trained leader can sometimes assist a group in group judgment or problem solving through encouraging the expression of minority opinion and bringing out the useful ideas and contributions of members who are otherwise restrained by conformity pressures (Maier&Solem 1952).

Communication networks. An application of principles of topology to group behavior by Alex Bavelas (1950) led to a series of studies of the effects of restriction of communication channels on productivity and morale. In an early study, five-man groups were given a problem of finding which symbols were held in common by all members. The communication possibilities were varied. The most effective group was that in which four members could communicate only with the one “central per son, a pattern called a “wheel.” A “circle” was least effective, with a “chain” and a “Y” being interme diate. The more involved the person was in the net work, the happier he was with the group task. The “circle,” while least effective, had the greatest over all morale (Leavitt 1951). The initial findings indicate that restriction of communication leads to greater efficiency. It is suggested that this may be true for the simple problems in which information must be collated to find a correct answer. The restricted pattern of the “wheel” saves the group from spending time on organization for effective problem solving. However, it also lessens flexibility, making for less efficiency when the tasks are changed or become more complex (Guetzkow & Dill 1957). When the problem is complex and involves a number of alternative paths to solution, the less restricted network is more efficient (Shaw 1954).

Personality factors undoubtedly play a role. Workers high in acceptance of authority will be relatively more efficient in a centralized group structure; those low in acceptance of authority will be more effective in a decentralized structure (Shaw 1959). It might also follow that an increase in intelligence and skills would lead to increased effi ciency for the decentralized structure relative to the centralized network.

Leadership and supervision. Clearly, leadership and supervision play a very important part in group performance, since the leader can often affect and control many of the other critical variables discussed above. The style of leadership has been given particular attention in the literature. In a pioneer experiment with groups of boys, it was found that the greatest productivity and morale occurred with a “democratic” leader, who encouraged participation in decisions, gave a clear picture of the group activities and the reasons for his requests, took an active but not overactive role in the group’s activi ties, etc. A “laissez-faire” leader, who allowed com plete freedom and assumed no active role, achieved the least productivity. An “authoritarian” leader, who was very active in issuing commands without giving reasons, was effective in raising productivity only insofar as the group was under his immediate surveillance. Morale was lowest in the last case (White & Lippitt 1960). Other studies in industrial settings confirm that productivity and morale are higher with a participatory leader or supervisor who assumes an active role in the group, gives support to his workers, delegates authority, and maintains an optimal degree of supervision (Kahn & Katz 1953). There is evidence that similar group-oriented roles on the part of the teacher may also lead to improved learning.

Characteristics of the leader have also received some attention in relationship to productivity. For example, Fiedler (1958), in investigations in a large variety of settings, has found that productivity is highest in teams where the leader maintains an optimal distance from his team and where he per ceives clear differentiation between his best and worst co-workers [seeLeadership].

Training and group process. Given a knowledge of factors which contribute toward improved group performance, it should follow that training and education in the application of these principles would be a logical next step. Indeed, there have been many attempts to introduce training for improved group performance—either through training the groups themselves or the leaders and supervisors. It has been demonstrated that a discussion leader who is trained to encourage expression of minority opin ions will effect superior problem solving and decision making (Maier 1953; Maier & Solem 1952). It is generally assumed that educators can be taught to improve learning in the classroom.

On the assumption that group ideation can be more productive than individual ideation and that a group may sometimes inhibit the expression of unusual ideas which might be modified so as to be fruitful, Alexander F. Osborn introduced training in what he called “brainstorming.” In this problem-solving situation, members are given a problem, such as how to increase the attractiveness of teaching as a profession, and then are asked to give whatever ideas come to mind, no matter how out landish. Thus, group norms are established to en courage rather than discourage strange suggestions. These are then gradually selected and evaluated, and a unique practical solution may be developed from an initially bizarre suggestion. Though the basis appears sound, one of the few careful evaluative experiments suggests that “brainstorming” inhibits rather than facilitates creative thinking (Taylor et al. 1958). [SeeCreativity; Problem Solving.]

The National Training Laboratories were established at Bethel, Maine, for the purpose of pro viding summer workshops in interpersonal relations. Educators, military leaders, supervisors in industry, and others participate in programs in volving some theory sessions, but largely free in teraction in groups, in order to become aware of how they affect others and others affect them. The original program spread widely and may now take place in sessions lasting from a few hours to months. “Sensitivity training” is said to have pro duced dramatic changes in the effectiveness of or ganizations. Unfortunately, the problems of evalu ating the effectiveness of “sensitivity training” are immense, but the clearest evidence to date of its effectiveness consists of testimonials from members of organizations which have participated and the fact that many such organizations have continued to train their supervisors and employees in this way (Tannenbaum et al. 1961, pp. 233-238).

Another recent approach to training is that de veloped in the System Development Corporation. In “system training,” the working unit (in this case, most work has been done with air defense crews) is considered as a “man-machine system” involving more than the time-motion aspects emphasized in an earlier period and the interpersonal relationships of a later stage. In practice, an air defense crew at its home site is presented with a series of problems which are simulated on the crew’s own radar scopes, with detailed flight plans, etc. After the problem, a careful record is available of all relevant interactions, and the crew holds a debriefing session in which members analyze their faults and attempt to make necessary corrections. One goal is to devel op greater flexibility in the crew, allowing for rapid changes in task assignment so as to provide for effective adjustment to all sorts of problem situations. Data indicate that there is, indeed, improvement in ability to deal with the simulated problems. It is assumed that such improvement also applies to future problems which the crew might encounter in real life as well (Gagne 1962).

Bertram H. Raven

[See alsoField Theory.]


Allport, Floyd H. 1920 The Influence of the Group Upon Association and Thought. Journal of Experi mental Psychology 3:159–182.

Bavelas, Alex 1950 Communication Patterns in Task-oriented Groups. Journal of the Acoustical Society of America 22:725–730.

Berkowitz, Leonard 1956 Group Norms Among Bomb er Crews: Patterns of Perceived Crew Attitudes, “Ac tual” Crew Attitudes and Crew Liking Related to Aircrew Effectiveness in Far Eastern Combat. Sociometry 19:141–153.

Coch, Lester; and French, John R. P. jr. 1948 Over coming Resistance to Change. Human Relations 1:512–532.

Deutsch, Morton 1949 An Experimental Study of the Effects of Co-operation and Competition Upon Group Process. Human Relations 2:199–231.

Fiedler, Fred E. 1958 Leader Attitudes and Group Ef fectiveness. Urbana: Univ. of Illinois Press.

Gagne, Robert M. (editor) 1962 Psychological Prin ciples in System Development. New York: Holt.

Guetzkow, Harold; and Dill, William R. 1957 Fac tors in the Organizational Development of Task-ori ented Groups. Sociometry 20:175–204.

Hare, Alexander P. 1962 Handbook of Small Group Research. New York: Free Press.

Horsfall, Alexander B.; and Arensberg, C. M. 1949 Teamwork and Productivity in a Shoe Factory. Human Organization 8, no. 1:13–25.

Kahn, Robert L.; and Katz, Daniel 1953 Leadership Practices in Relation to Productivity and Morale. Pages 612-628 in Dorwin Cartwright and Alvin F. Zander (editors), Group Dynamics: Research and Theory. Evanston, 111.: Row, Peterson.

Kelley, Harold H.; and Thibaut, John W. 1954 Experimental Studies of Group Problem Solving and Process. Volume 2, pages 735-785 in Gardner Lindzey (editor), Handbook of Social Psychology. Cambridge, Mass.: Addison-Wesley.

Leavitt, Harold J. 1951 Some Effects of Certain Com munication Patterns on Group Performance. Journal of Abnormal and Social Psychology 46:38–50.

Lewin, Kurt 1951 Field Theory in Social Science: Se lected Theoretical Papers. Edited by Dorwin Cart-wright. New York: Harper. → A collection of papers first published between 1939 and 1947. A British edition was published in 1963 by Tavistock.

Lorge, Irving et al. 1958 A Survey of Studies Contrasting the Quality of Group Performance and Individual Performance: 1920–1957. Psychological Bul letin 55:337–372.

Maier, Norman R. F. 1953 An Experimental Test of the Effect of Training on Discussion Leadership. Hu man Relations 6:161–173.

Maier, Norman R. F.; and Solem, Allen R. 1952 The Contribution of a Discussion Leader to the Quality of Group Thinking: The Effective Use of Minority Opinions. Human Relations 5:277–288.

Michigan, University of, Survey Research Center 1954 Group Cohesiveness in the Industrial Work Group, by Stanley Seashore. Ann Arbor: Univ. of Michigan Press.

Moede, W. 1927 Die Richtlinien der Leitungspsychol-ogie. Industrielle Psychotechnik 4:193–209.

Perlmutter, Howard V.; and De montmollin, Germaine 1952 Group Learning of Nonsense Syllables. Journal of Abnormal and Social Psychology 47:762–769.

Raven, Bertram H.; and Eachus, H. Todd 1963 Co operation and Competition in Means-interdependent Triads. Journal of Abnormal and Social Psychology 67:307–316.

Raven, Bertram H.; and Reitsema, Jan 1957 The Ef fects of Varied Clarity of Group Goal and Group Path Upon the Individual and His Relation to His Group. Human Relations 10:29–45.

Roethlisberger, Fritz J.; and Dickson, William J. (1939) 1961 Management and the Worker: An Ac count of a Research Program. Conducted by the Western Electric Company, Hawthorne Works, Chicago. Cambridge, Mass.: Harvard Univ. Press. → A paper back edition was published in 1964 by Wiley.

Schachter, Stanley et al. 1951 An Experimental Study of Cohesiveness and Productivity. Human Relations 4:229–238.

Shaw, Marjorie E. 1932 A Comparison of Individuals and Small Groups in the Rational Solution of Complex Problems. American Journal of Psychology 44:491–504.

Shaw, Marvin E. 1954 Some Effects of Problem Complexity Upon Problem Solution Efficiency in Different Communication Nets. Journal of Experimental Psychology 48:211–217.

Shaw, Marvin E. 1959 Acceptance of Authority, Group Structure and the Effectiveness of Small Groups. Jour nal of Personality27:196–210.

Tanneneaum, Robert; Weschler, Irving R.; and Mas-sarik, Fred 1961 Leadership and Organization: A Behavioral Science Approach. New York: McGraw-Hill.

Taylor, Donald W.; Berry, Paul C; and Block, Clif ford H. 1958 Does Group Participation When Using Brain-storming Facilitate or Inhibit Creative Thinking? Administrative Science Quarterly 3:23–47.

Taylor, Donald W.; and Faust, William L. 1952 Twenty Questions: Efficiency in Problem Solving as a Function of Size of Group. Journal of Experimental Psychology 44:360–368.

Thomas, Edwin J. 1957 Effects of Facilitative Role Interdependence on Group Functioning. Human relations 10:347–366.

Thomas, Edwin J.; and Fink, Clinton F. 1963 Effects of Group Size. Psychological Bulletin 60:371–384.

Thorndike, R. L. 1938 On What Type of Task Will a Group Do Well? Journal of Abnormal and Social Psychology 33:409–413.

Triplett, Norman 1898 The Dynamogenic Factors in Pacemaking and Competition. American Journal of Psychology 9:507–533.

Trist, E. L.; and Bamforth, K. W. 1951 Some Social and Psychological Consequences of the Longwall Method of Coal-getting. Human Relations 4:3–38.

Van zelst, Raymond H. 1952 Validation of a Sociometric Regrouping Procedure. Journal of Abnormal and Social Psychology 47:299–301.

White, Ralph K.; and Lippitt, Ronald 1960 Autoc racy and Democracy: An Experimental Inquiry. New York: Harper.

Zajonc, Robert B.; and Smoke, William H. 1959 Re dundancy in Task Assignments and Group Perform ance. Psychometrika 24:361–369.


views updated May 29 2018








A group is a collection of individuals with some degree of interdependence and some element of common or shared identity. Most often, membership in a group involves face-to-face interaction with other members, although such interaction is not a necessary component of a group. (For example, Internet groups, which do not involve face-to-face interaction, are becoming increasingly important in the twenty-first century.) Scholarly interest in groups began with sociologys birth. German sociologist Ferdinand Tönnies (18551936) made an early contribution when he developed the influential concepts of Gemeinschaft and Gesellschaft to depict two different types of groups related to different kinds of economic and social structures. Gesellschaft refers to instrumentally-based groups or communities in which social relations are formal and little consensus exists. Gemeinschaft refers to small communities in which social interactions are based on friendship or kinship.

The importance of groups and community were also a focus of the French sociologist Émile Durkheims (18581917) writings. For example, in The Elementary Forms of Religious Life (1915), Durkheim discussed the significance of religion for ones sense of belonging. Religion might take different forms in different societies, but all religions stress the importance of social control and cohesion of the group. In this way, the rituals associated with religion serve as visible evidence of the power of the community as they are enacted in a group and for the group.

The American social psychologists Charles H. Cooley (18641929) and George H. Mead (18631931) both frequently addressed the central importance of groups, especially small groups, and made important contributions to the construction of a framework for understanding symbolic interaction. Cooley, for example, developed the idea of primary groups, that is, groups that are important in intimate interaction. Meads theoretical conceptualizations in Mind, Self, and Society (1934), including the centrality of groups and the importance of role-taking, were influential in the development of the field of group dynamics.

The German sociologist Georg Simmel (18581918) detailed many concepts and perspectives that have preoccupied group theorists since the early twentieth century. His discussion of conflict and exchange framed issues in terms of a dynamic. Simmel argued that exchange was pervasive and that what might at first appear to be purely individual acts are actually influenced by others. To use one of his examples, a teacher teaches, but this involves an exchange with students in part because the teacher considers the students as an audience. People are not isolated agents, acting on their own; rather they act in response to groups.

The study of groups and their structure is sometimes termed group dynamics or group processes. While there are many areas of study within group dynamics, some of the most important are group identity, status, cooperation and competition, exchange, justice, and legitimation.


The intensity of identity as a group member varies according to the members common fate. If membership within a particular group seems to predict important outcomes, group identity may be strong. For example, minorities are likely to feel a strong identity as minority group members because it is evident that their categorization both comes from and creates common experiences.

Collectivities can have elements of common fate, but members may not have group identification and as a result are not usually considered groups. For example, Karl Marx (18181883) clearly delineated class interests based upon the means of production; membership in a class was fundamental to life chances. The bourgeoisie controlled the means of production, while the proletariat did not. However, at times, members of those groups may not conceive of themselves as sharing commonalities; in fact, as Marx argued, although members of the proletariat had powerful class interests uniting them, they often lacked the critical recognition of membership in this group. This lack of recognition served the purposes of the bourgeoisie and was an important means of control and power.

Group identity can also be created solely through common experiences, even where no categorization preceded the experience. Experiences such as surviving a hurricane or cancer, for example, can create a powerful group identity among the survivors. Research, especially in the social identity tradition, also indicates that, under some conditions, the slightest form of categorization can function as a type of group identitystrong enough to create in-group favoritism. Research in the area of social identity theory, which developed within European social psychology, emphasizes how individual cognition and group identities are related.


Status is usually defined as a position in a social network. Importantly, status involves status beliefs, beliefs about the social worth of the individuals who occupy positions within a network such that a person who occupies one position is better than a person who occupies another position (Sewell 1992). Early status studies examined the concept of leadership and different forms of group rules that corresponded to different political processes (such as democratic rules versus autocratic rules). In the 1950s, attention turned more to the internal dynamics of groups. The social psychologist Robert Bales (19162004) and his associates developed interaction process analysis (IPA), a categorization technique that, in different forms, shaped much of group analysis. In particular, these researchers were interested in how the behavior of one group member conditioned the behavior of others. This idea that status was relative to the group was a central insight and formed the impetus for thinking about characteristics once viewed as fixed (such as sex or ethnicity) as varying in intensity and salience depending upon context.

One of the most developed research programs in the analysis of status is expectations state theory. The theory has several subsets; one of these is status characteristics theory, which concerns how status generates and then sustains inequality of power and prestige within groups. It is posited that this process, called the burden of proof process, is so strong that unless some event or some information intervenes, it organizes interaction consistent with prior evaluations of the status characteristics. In other words, the cultural stereotyping associated with status characteristics are reproduced in different settings and within different groups. Dissolving status hierarchies is difficult because there are layers of group interactions that support and uphold the status quo. However, under some conditions, hierarchies can be dissolved; particularly noteworthy are a series of applied studies in school settings (Cohen 1993).


Related to issues of status are issues of legitimation, the process through which a principle or set of rules is adhered to even in the absence of incentives. There have been studies of differing sources and processes of legitimation, and particularly promising is research concerning how the granting of power within groups is affected by and reinforced by referential belief structures, or socially validated beliefs at the cultural level that are imported into the local setting.

Other research has addressed how conflict between different sources of legitimation might affect interactions and the establishment or disruption of norms and routines. There is a large literature that aims to answer how individuals make assessments of justice based on their own and others benefits. The sociological contribution to justice is the extension of the justice concept beyond the specific individual. Referential belief structures, for example, serve as an external comparison by which to judge the fairness of local settings.


The social aspects of exchanges are central to the study of groups. Some of the early sociological formulations were patterned after economic models, while others were patterned after behavioral psychology. Sociologist Richard Emerson (d. 1982), relying on behavioral models as a foundation, developed a conceptual framework that viewed the exchange, rather than individual actors, as the unit of analysis. This formulation took the power-dependency relationship among actors or groups as the determining factor in dictating interaction. For a given relationship, the more powerful the actor (whether that actor is an individual or a group), the less dependent the actor. According to Emerson, this power dependency leads to a continual balancing, so that the actor who has the most power uses it (because it is to that actors advantage to do so), but such use of power leads to some loss of power. This shift in power leads to balancing or sets of strategies by which actors try to retain their power. Theories developed after this initial formulation sometimes refuted the idea that power was lost. This seems to be especially true in settings in which the social network provides some actors with particularly advantageous positions.

Work on coercive power indicates that in small groups (and perhaps in large groups as well), coercive power (in the sense of punishing others) is a risky strategy because it can decrease the possibilities of future exchanges. Of course, the risk to the coercer is related to the alternatives present for the coerced. Relatedly, the conflict spiral, a theory about bargaining processes, predicts that unequal power, even without punishment, can produce negative emotion.


One of the longest traditions in the study of groups is the investigation of cooperation and competition. The form and type of incentives that encourage or discourage competition have been extensively examined. These incentives might be material rewards, such as money, or social rewards, such as honor or friendship. Even when the incentives are structured so that all might be better off cooperating, cooperation does not always obtain. This is because even in simple settings, coordination can be problematic.

Social dilemmas, settings in which individual and group incentives conflict in some way, are prominent research areas because they are pervasive in many different aspects of life. Such dilemmas range from the small and intimate (e.g., how to maintain a clean house), to the large and relatively anonymous (e.g., how to maintain biodiversity). Many solutions to social dilemmas involve changing the incentives and thereby changing the structure. These might involve punishments for not cooperating or rewards for cooperating.

Other solutions have focused upon social factors arising from group interaction. Two powerful such factors are group identity and trust. If, for example, one actor trusts another to cooperate and then acts on this basis, the dilemma can sometimes be solved. Group identity, as mentioned, can arise from cooperation. Once born, group identity can also lead to cooperation.

SEE ALSO Collective Action; Competition; Cooperation; Durkheim, Emile; Ethnicity; Gemeinschaft and Gesellschaft; Group; Groupthink; Identity; Marx, Karl; Mead, George Herbert; Nationalism and Nationality; Race; Social Exchange Theory; Social Identification; Sociology


Berger, Joseph, Morris Zelditch Jr., Bo Anderson, and Bernard P. Cohen. 1972. Structural Aspects of Distributive Justice: A Status Value Formulation. In Sociological Theories in Progress, vol. 2, eds. Joseph Berger, Morris Zelditch Jr., and Bo Anderson, 119146. Boston: Houghton Mifflin.

Cohen, Elizabeth. 1993. From Theory to Practice: The Development of an Applied Research Program. In Theoretical Research Programs: Studies in the Growth of Theory, eds. Joseph Berger and Morris Zelditch Jr., 385415. Stanford, CA: Stanford University Press.

Durkheim, Émile. [1915] 1965. The Elementary Forms of Religious Life. New York: Free Press.

Lovaglia, Michael J., Elizabeth A. Mannix, Charles D. Samuelson, et al. 2005. Conflict, Power, and Status in Groups. In Theories of Small Groups: Interdisciplinary Perspectives, eds. Marshall Scott Poole and Andrea B. Hollingshead, 139184. Thousand Oaks, CA.: Sage.

Mead, George Herbert. 1934. Mind, Self, and Society: From the Standpoint of a Social Behaviorist. Chicago: University of Chicago Press.

Molm, Linda D., Nobuyuki Takahashi, and Gretchen Peterson. 2000. Risk and Trust in Social Exchange: An Experimental Test of a Classical Proposition. American Journal of Sociology 105: 13961427.

Sewell, William H., Jr. 1992. A Theory of Structure: Duality, Agency, and Transformation. American Journal of Sociology 98 (1): 129

Simmel, Georg. [1907] 1971. On Individuality and Social Forms: Selected Writings. Ed. Donald N. Levine. Chicago: University of Chicago Press.

Jane Sell