The Adolescent in His Group in Its Setting: I. Theoretical Approach and Methodology Required[1]

Muzafer Sherif and Carolyn W. Sherif

This chapter and Chapter 13 report a research program on natural groups of adolescents which was initiated in 1958 and is still. in progress. The research program focuses on the attitudes, goals, and behavior of individual members in the context of their group and their particular sociocultural setting. It attempts to study interpersonal relations, attitudes, behaviors and misbehaviors when these occur, in numerous interaction episodes over months. The groups under study are groups of the members' own creation or their own choosing, that is, their reference groups.

This chapter summarizes the theoretical guidelines and the empirical basis for a multifaceted research program requiring a combination of psychological and sociological procedures. The intimate relationship between theory and research methods will be articulated through discussion of the choice of data-gathering techniques and the timing of their introduction in the study of natural groups. We have reached the conviction painfully that free or arbitrary choice of methods is not possible in the study of actual groups which possess the essential properties defining "groupness." The nature of the groups sets definite limits on the range of methods that are appropriate.

Chapter 13 will report the methods and operational specifics used in the research program, along with representative findings to date which have general significance for youth problems. Our book, Reference Groups: Exploration into Conformity and Deviation of Adolescents

(Sherif and Sherif, 1964) presents more details of the research program, its findings up to 1962, and the leads derived from them that are applicable to current problems of adolescent misbehaviors and the wastage of talents and energies in socially harmful channels.

( 265)


The individuals studied in the research program are adolescents. As outlined in the Introduction, the dilemmas of this period, in a culture itself undergoing rapid social change, produce a broad and intense motivational base for group formations in all walks of life. The adolescent period is a paradigm for studying the individual-group-society relationship in its clearest manifestation. As such, the issues of theory and research into youthful behavior are essentially the same as those in the study of human social behavior at any period of life.

We start with a premise based on a host of empirical findings. The premise is that adequate study of the behavior of individual group members must include specification of the group properties and the part played by the individual.

Obviously, however, groups do not rise and function in a vacuum. Even though they are units with distinctive patterns of their own, groups are not closed systems defining their own universe. The claims of a group and the goals it pursues are related to the settings in which the group functions. That part of the setting which raises issues and problems about goals and collective efforts to attain them is the sociocultural setting.

Even though our interest is the individual behavior of group members, the components and events comprising the sociocultural setting cannot be reduced to psychological terms. They are out there relative to an individual. They are not reducible to "psychological constructs," as cogently noted by Roger Barker (1963). The sociocultural setting and its various aspects can be and are studied in their own right, by sociology, anthropology, and other social sciences, quite apart from the psychological study of particular individuals immersed in them.

Of course, there can be no denial that a person's psychological world is the world he actually experiences. But this experiential fact should not be stretched into a denial of an equally unmistakable fact: sociocultural products are out there as stimulus conditions for the individual, affecting his behavior and his success or failure in attaining his goals. This is the case even though he may not be aware of their impact. We cannot derive an adequate conception of the sociocultural setting merely from the psychological worlds of individuals in it without falling into a fruitless circle of reductionism.

Therefore, we must conclude that adequate study of individual behavior must include the specification of both the individual's groups and the sociocultural setting in which they form and function. The difficult research task is hitting upon concepts of analysis and procedures appropriate to each domain, namely, the individual, the group, and the sociocultural setting. Necessarily, the research task is interdisciplinary, to use

( 267) a very fashionable term which is often misapplied. In a multifaceted research program which includes the study of individual-group-setting within the same design, the appalling gap between laboratory research and field research may narrow in a relatively short time.

If it is to achieve its purpose, such research must also interrelate its several facets without falling into a bias favoring either an individual group, or cultural approach. Proper operational procedures and the interrelationship of the various aspects will contribute to resolving a number of theoretical or doctrinal conflicts (as represented by the "institutional" vs. "behavioral" controversies in social science [cf. Sherif and Koslin, 1960], and the schisms between proponents of research on small groups and large organizations). In the present conception, individual behavior is seen in the context of groups and larger organizations. Small groups, in turn, are not viewed as units in their own right, but as parts of a larger social system. Our research program is an attempt in this direction.


Mindful of the range of influences on the formation and functioning of human groups and the behavior of members, we designed our research to include operations appropriate for the behavior of individuals, for their groups, and for the sociocultural setting. In specifying the gross characteristics of the setting, as well as the prevailing patterns of values or norms within it, we relied heavily on methods, procedures, and findings of social scientists. After all, social scientists have developed tools and collected a wealth of data on the regularities of the sociocultural setting and the properties of groups. It is wasteful for social psychologists to start from scratch or to improvise tools without discovering those available in other academic disciplines.

The research program has three main facets, which may be summarized as follows:

1. Since the focus of the program is social-psychological, the concentration of procedures and data is on selected small groups of adolescents and their members. Clusters of teen-age boys (13-18 years old) are selected on the basis of their frequent and recurrent association in specified locations. Each group is studied intensively for periods from six months to a year. A combination of techniques is used in collecting data, including observation, behavior ratings by regular observers and independent raters, informal sociometric techniques, situational tests, and case history materials. The distinctive research strategy is that the boys are not aware that they are research subjects or that they are being observed for research purposes.

Behavior in the groups has included both socially acceptable activities, and deeds socially unacceptable to an extent that they would be

( 268) labeled "delinquent" if detected by adult authorities. The basis for selecting groups is their regular and recurrent association, not whether they behave properly or misbehave, or have been labeled socially acceptable or delinquent. For purposes of comparison, groups are selected from settings of low, middle and high socioeconomic rank and from urban neighborhoods with different ethnic populations.

2. A second facet of the research is specification of the characteristics of the particular sociocultural settings where the small groups function. Such characteristics were conceived as stimulus conditions, relative to the attitude and behavior of the individuals studied. Therefore, we could not be content with a blanket, over-all characterization of the setting or with improvised descriptions of it. We relied especially on the Shevky-Bell social area analysis (Shevky and Bell, 1955; Bell, 1958) which yields concise indicators of what is meant by low, middle and high social ranks, and their gradations. (These indicators are presented in Chapter 11 of this volume.)

3. For social-psychological analysis, specification of only major characteristics of the setting is not sufficient. Certain aspects of, the sociocultural setting have greater salience than others, particularly for individuals in the adolescent period. A large research literature shows that, during the adolescent period, individuals are particularly attuned to others in their own age set. Accordingly, special emphasis is given in the research program to collecting data on the bounds of acceptability and on the goals prevailing among representative samples of teen-age youth in each sociocultural setting. These data indicate what youth in different areas regard as "socially desirable" and undesirable, and where they set their sights for achievement in various respects. Thus, cultural values can be compared between areas to ascertain which are common to them all and in what ways they differ. In turn, the concepts and goals prevailing among youth in a given area can serve as a baseline for assessing the relative typicality or deviance of individual members of a group being studied within that area.


We sought research methods from the social sciences as well as psychology, and we relied on the empirical findings in both as the basis for concepts and theory. Concepts and theory enter into research from the very beginning, when the investigator formulates the problem of study. They influence what kind of data he gathers and what data he disregards. They affect the selection of data-gathering techniques, the manner in which the data are used, the analysis, and the interpretation of findings. Consequently, it is fitting to examine the theoretical and methodological problems of interdisciplinary research, which the adequate study of youthful behavior must be.

( 269)

In social science and psychology, there are grand theories of the individual-group-society relationship. The present approach, however, did not derive from a grand theory, though that might have greater esthetic appeal. Few, if any, of the grand theories are formulated to encompass the gamut of influences shaping behavior and to guide research operations at three different levels of analysis (individual, group, cultural).

Nor does the present approach stem from theoretical models borrowed (or smuggled) from more established and prestigeful sciences. Formal models of the relationships among the multifaceted data are the goal, so that, ultimately, we can express events and their relationship in five pages instead of five hundred. Because of the current fad, even craze, for models in psychology and social science, we shall be explicit about the limitations placed on data gathering by premature adoption or formulation of a formal theoretical model.

The indiscriminate use of models based on analogy with other sciences has proceeded with a singular lack of concern for the crucial question of the isomorphism between the models, and the events and actualities of which they are supposed to be models. Such concern should be the basis for accepting or rejecting a model. If it is not, the model restricts the range of data collected, conveniently pruning branches of unwanted facts to the point of focusing on tiny and barren twigs of trivia.

The common pitfall in borrowed models was emphasized recently in a presidential address to the Division of Engineering Psychology of the American Psychological Association, a rather unexpected source. Chapanis (1961) observed that attempts are seldom made to validate borrowed models and that those who work with them typically end up being "intrigued with essentially trivial problems."

Model building by making analogies with a more established science is not a monopoly of our time. There was the "mental chemistry" model of Wundt and Titchener, which now lies at dead end. There was the organic analogy of Herbert Spencer, which fared no better. Some mechanical and hydraulic analogies still flourish, but a similar fate for them is inevitable. As Emile Durkheim (1915) and others put it, social life and human value systems are not on a continuum with physical and biological events. They have properties of their own, not to be found through analogy with physical sciences.

The issue at hand, however, is not model building as such, but the proper basis for building them. The first step toward adequate models of social behavior and its social setting starts with formulating the proper questions and defining problems. In order to take this first step, a period of concentration on the actualities of our topics is essential. These actualities must be explored at the level of the complexity which is characteristic of problems of individual-group and sociocultural relationships.

To gain the effective tools for the logical and mathematical formalization so essential for precise model building, we have to define our

( 270) problems and their properties more clearly. Unfortunately, at this erratic stage in the study of human affairs, even the basic problems at stake have not been formulated in stable form.

This is a plea for a disciplined phenomenology as an initial step in formulating problems. This is only an initial step, but one which serves generously as we venture on into research and theoretical interpretation.

As an illustration, consider the study of the formation of social norms in the laboratory through utilization of the autokinetic phenomenon (Sherif, 1935, 1936). This experiment is frequently cited as an example of the precision to be gained from laboratory models (cf. Cartwright and Zander, 1960). Yet it did not originate in the techniques and procedures of the psychological laboratory—it started with a lesson learned empirically in the actualities of social life: that social norms or standards arise when conditions in the lives of men are fluid, uncertain, or in crisis. From this lesson, it was then a matter of devising an appropriate laboratory condition to represent this fluidity and uncertainty; the autokinetic setup was one of several such conditions suitable for the problem. The same concern with observed actualities was the basis for the conception of our experiments on intergroup relations, which have been presented elsewhere (Sherif and Sherif, 1953; Sherif et al., 1961). We are proceeding from such an empirical basis in studies of attitude and attitude change (Sherif, Sherif, and Nebergall, 1965).

The present undertaking on adolescents in their groups presents the investigator with much more challenging and difficult tasks than does experimentation. In the open field of actual life, event follows event, and none can be controlled and manipulated as the investigator chooses.

What guidelines do social science and psychology give us for raising crucial problems and formulating hypotheses? Necessarily, we can give only brief and essential examples of empirical findings that are established with sufficient stability to serve as a basis for research.


From the empirical findings of sociologists, especially from the Chicago school of the twenties and thirties, we learned of the universality of group formations whenever individuals interact with similar motivational promptings. These motivations may be deprivations, frustrations, or desires for earthly goods and political power. This generalization, based on empirical facts, provided the basis for hypotheses both for our experimental studies of group formation and intergroup relations, and for the present research.

As noted earlier, we also learned, primarily from findings by adolescent psychologists, of the motivational dilemmas common to this period. At the same time, we noted in anthropological reports that these dilemmas varied in their severity because of cultural arrangements (cf.

( 271) Sherif and Cantril, 1947). These reports also confirmed the generality and significance of group formation during adolescence.

From studies of both small and large organizations, we learned that human groups have properties more complex than merely an undifferentiated state of interdependence among individuals (cf. Homans, 1950; Blau and Scott, 1962). In actual groups, we learned, the interdependencies can be specified in terms of a differentiated and hierarchical pattern of positions, with decided implications as to what each individual can or cannot do in effectively initiating activities for others in the pattern without being challenged by them.

We learned from the studies of sociologists and anthropologists that every group possesses a set of regulations and standards—that is, a set of social norms—defining a latitude of acceptable behavior, including the ideal, as well as a range of unacceptable attitudes and actions which will bring forth scorn and sanction from others in the group.

Likewise, we have relied on guidelines from empirical findings in psychology. The findings from psychology have been particularly useful in developing behavioral indicators for the individual-group relationship; the hierarchical arrangements within groups, bounded by the end-positions of "leader" and "man at the bottom"; and the way members of an in-group evaluate each other and out-groups. We found invaluable leads in the accumulating empirical facts in the psychology of perception (especially on perceptual selectivity), the psychology of judgment and motivation, and child psychology. We were concerned with including the sociocultural setting explicitly partly because of the unmistakably established findings on background and context effects in the psychology of perception and judgment. Because of their far-reaching implications, these findings will be summarized.

Even the simplest judgment is a comparison process, not immune to the effects of its context or background. As Helson demonstrated, the color judgment of the same patch of material "could be made anything we pleased by appropriate choice of the luminance and hue of the background color" (Helson, 1964, p. 26). Context and background effect are considerations even in a relatively simple judgment of a patch of color or a weight of so many grams. How much more crucial they should be in considering an individual's judgment, perception, and other reactions in his social relations!

Social relations constitute the context of individual attitude and behavior. Background and context intrude even into the carefully contrived confines of an experimental interview or testing situation. There is much more to a contrived laboratory or testing situation than what catches the eye or what the experimenter's instruments can record. The context is not merely the experimenter's stimulus material and special instructions. The experimenter himself, the way he is appraised by the subject, the surroundings, the presence or absence of other individuals

( 272) and the subject's appraisal of them—all these intrude into the shaping of performance.

No wonder, then, that a whole movement on the "social psychology of the psychological experiment" is developing to articulate the hitherto unaccounted—for or neglected ingredients that go into the making of any situation. The trend is well represented in the experiments of Orne (196_ and Rosenthal (1961). Only a few of the recent contributions to this strengthened sensitivity to context and background can be cited here,

That the characteristics of the administrator make a difference on intelligence tests is forcefully brought to our attention once again in recent surveys by Pettigrew (1964) and Martin Deutsch (1964). Even more closely related to our own research is the report by Pearl in Chapter 5 of this book, where he compares interviews with delinquents made by graduate students with those made by other delinquents. "When graduate students conducted the tape-recorded interviews, they confirmed the usual conclusion that lower-class youth were inarticulate. But when lower-class interviewers canvassed the same persons, responses were entirely different. The subjects were animated and highly verbal."

Stevenson and Allen (1964) reported recently that the response rate in a motor task differs according to whether the experimenter is male or female; and Pishkin (1964) showed that errors in concept identification by schizophrenics were significantly affected by whether the experimenter was present or absent. Finally, we note the finding that the context of interpersonal relations in a hospital ward has produced not only behavioral variations, but also significant metabolic alterations (Schottstaedt, Pinsky, Mackler, and Wolf, 1958).

In brief, the context and background of human behavior produce significant variations, whether or not the investigator includes them in his study design. If we would understand the sources of behavioral variation, there seems to be no question about including the sociocultural and group contexts of behavior. Sober consideration of the research findings on context and background effects led to our inclusion of the sociocultural setting and group properties in the design, and to the methods chosen or developed for the study of behavior in groups.


Since our research program focuses on the intensive study of natural groups of adolescents, the major hypotheses and data-gathering techniques for their assessment pertain to the behaviors and attitudes of individuals in such groups. The prerequisite to an adequate study of individual behavior in a group context is a definition of the properties of actual groups and their interaction processes.

Our definition of the properties of groups is based on extensive surveys of empirical findings on group formation and functioning in dif-

( 273) -ferent social spheres, as documented in our own earlier publications (1948, 1953, 1956).

The term "natural group" is used in these chapters to designate the origins of the groups, and carries no evaluative implication. Particularly in the case of adolescent groups, it is convenient to distinguish groups formed through the informal interaction of members from those instituted by adults, a board of officers, or a council.

We are referring to groups in a technical sense, specifying the properties that distinguish a group from casual collective encounters, from temporary "discussion groups," and from experimental groups collected on an ad hoc basis. A formally instituted body may or may not become a "group" in the precise sense to be defined here.

In the first place, a group is a human formation formed over a given time span through interaction of individuals. It is a social unit. Its bounds define who is "in" and who is "not in." The criteria for membership and other properties of the group may or may not be codified explicitly as verbal rules.

The essential condition for the formation of a human group is interaction over a period of time among individuals with similar concerns, similar motives, similar frustrations, or, generally, a common dilemma which is not effectively dealt with through established social channels and arrangements. In other words, people in the same boat of misery or unfulfilled desires love company. They tend to interact with one another.

Repeated interaction in a variety of tasks by individuals with some common striving leads to differentiation of functions in activities, along with the coordination of effort. The differentiated functions are stabilized in time as roles with differing status. This differentiation of roles and statuses over a time span is the pattern of the human group.

Over a time span this patterned give-and-take interaction of individuals produces a set o f rules or norms for the regulation of attitude and behavior within the bounds of the group and toward outsiders. The rules or norms that are salient in the eyes of members are those that pertain to the existence and continuation of the group and to spheres of activity related to the common motivational concerns which initially brought them together.

In summary, the minimum and essential properties of a group consist of (1) a pattern or organization of member roles, differentiated as to status or power as well as to other functions; and (2) a set of values or norms regulating behavior, at least in spheres of activity frequently engaged in by the group.

The "groupness" of a group is, therefore, a matter of degree. It is proportional to the extent that the status and role pattern is stabilized, and to the extent that member behavior is effectively regulated by a set of norms pertaining to their recurrent activities.

Now let us consider some implications that these properties of

( 274) groups have in developing research procedures which will yield valid and reliable data on interpersonal relationships within the group and on the attitudes of individual members.

First, the unit character of a group implies a context for behavior which is, in some degree, "private." The "privacy" of in-group interactions develops from the motivational basis which brings the members together and the prolonged give-and-take among them.

Groups do not form and function for the benefit of an investigator. On the contrary, their very nature militates against his intrusion. The gross intrusion of an outsider into significant ongoing events will affect the usual context for behavior of members. This is particularly true if the outsider is identified as a researcher, an investigator, or anyone intent upon manipulating their behavior. We determined, then, to avoid making members aware that they were being observed and studied. This decision meant that our data had to be obtained in numerous interaction episodes at times and in places not under the observer's control. Observers were instructed to avoid, as much as possible, any interruption of the free flow of interpersonal give-and-take among members.

The definition of the group in terms of differentiation of status and role relationships and a a set of norms did not specify that the individual members could or would report on leader-follower relations or on their conformity to group norms. In fact, we have records in our research of numerous expressions of the following sort: "We have no leader." "We have no boss." "We are all equal." Despite such reports, it was possible to rank these same individuals as first, second, third, down to lowest, in terms of their effective initiative in activities of the group. Effective initiative is an operational index of power in any organization.

In brief, the properties of groups led to inferences about the procedures necessary to study them adequately—namely, that the primary source of data was to be observation of interaction episodes over a period of time, without the awareness of members, and without undue cluttering of the flow of interaction by research techniques. These decisions, in turn, required adjustments to a common problem in field observations, the problem of checking reliability of observations.

In Chapter 13 the operational procedures, including the criteria for assessing group properties, are summarized in more detail. Here it is sufficient to note that the selective bias of a single observer was minimized by using a combination o f methods for data collection. The use of a combination of data-gathering techniques permits findings by one method to be checked against those yielded by another. This is the best insurance against observer bias. The methods included adaptations of techniques ordinarily considered distinctive to the controlled laboratory situation, for example, rating methods and situational tests.

( 275)


From the properties of groups and their formation, several hypotheses about member attitudes and behavior were deduced. The major hypotheses of our research program and a string of auxiliary predictions are spelled out in Reference Groups (Sherif and Sherif, 1964).

Here we have deliberately selected hypotheses of major import for youthful behavior. The research program has collected pertinent data both in the study of natural groups, as summarized in the next chapter, and in laboratory experiments based on the same conceptualization, reported in this chapter.

Here, then, are general predictions deduced from the empirically derived concept of group:

1. To the extent that a group organization (role and status relations) is stabilized, the norms or standards for behavior in activities focal in interaction become binding for individual participants.

("Binding" means here, psychologically binding, such that conformity to the norm occurs because the person considers the norm in question as his personal guideline, not as one imposed on him. One index of the extent to which a group norm is binding for the individual is his behavior in compliance to its bounds when he is out of the reach of other members, and there is no threat of sanctions for deviation.)

2. The salience of various groups for the individual is a function of the extent to which its activities bear on motivational concerns he shares with the membership.

("Salience" refers here to the relative importance of the group as reflected in his behavior, especially in his choices or preferential judgments.)

3. Conformity to an established norm by new participants in the interaction pattern will be a function of the extent to which the conditions giving rise to the norm continue. Conversely, deviation from an established norm is a function of its "arbitrariness" relative to current motivations and circumstances faced by members.

("Arbitrariness" may be defined operationally in terms of the difference between the established norm and a norm stabilized in the current situation by individuals who have had no contact with the established norm.)

4. The extent to which an individual's behavior in group activities is predictable is a function of his position in the status organization and the degree of the group's stability.

(The last section of this chapter will spell out the curvilinear relationship between status and predictability of behavior.)

( 276)


An operational distinction can, be made between mere compliance to the demands, pressures or suggestions of other individuals in an immediate situation, and the concept of conformity to the group.

Conformity, properly speaking, "refers to behavior within the bounds of standards (norms) stabilized in prior interaction situations. The extent to which group norms are psychologically binding for the individual member may be tested by observing his behavior when he is not in contact with other members. It was predicted that the extent to which standards are psychologically binding would vary with the degree of stability in the interpersonal (role) relation among the individuals who participated in their formation.

In a laboratory experiment undertaken with the support of our research program, N. P. Pollis investigated this hypothesis by comparing conformity to standards which had been formed in interpersonal relationships of varying stability. He tested the hypothesis by first establishing standards for behavior (judgment), and then putting the individual in a transitory "togetherness" situation where the others confronted him with conflicting judgments. The question, then, was to what extent would the individual comply with the immediate social influences of others, or conform to his previously established standard in a novel situation.

The experiment was preceded by sociometric study of the sophomore class of a college, conducted under different auspices than the experiment so that the subjects would establish no connection between the two. On the basis of this study, subjects ,were selected from the intermediate range of general social standing on the campus. Certain well-defined cliques (groups) were discovered. It was also possible to specify which individuals were not personally acquainted.

In the first session of the experiment, each of the 144 subjects participated either (a) alone, (b) together with someone he did not know personally, or (c) together with a fellow group member. These three conditions defined variations in the stability of interpersonal relationships.

The task was to judge the frequencies of auditory pulses produced by an Eico audio-generator. A series of pulses was presented, consisting of random arrangements of four different pulse rates. Each pulse rate lasted 3 seconds and was followed by a 6-second rest period.

The task was sufficiently subject to error that Pollis found it feasible to train subjects to three different ranges of judgment by identifying the slowest and fastest rates differently in his instructions. Thus, he established three different judgment ranges, each among one-third of the subjects: low, middle or high. The purpose of this variation was to provide a pool of subjects with differing established ranges for a second session.

In the second session, each subject served with two others, none of

( 277) whom were fellow group members. In the first session, one had stabilized his judgment range alone, one together with another student with whom he was interacting for the first time, and one with a fellow group member. Each of these three had formed a. different range for judgment in session 1 (low, middle, high). The possible effect of the source in speaking judgments within these ranges was controlled by systematically counterbalancing the range established in session 1 and the social situation in which the individual had formed it. In other words, in session 2, the group member in one triplet started giving the high range; in another, the middle range, in another, the low range.

Two measures were used to test the hypothesis. First, the percentage of each individual's judgments in session 2 that fell within his range for session 1 was used as the measure of his conformity to the standard formed in session 1. Second, the percentage of judgments by other subjects in session 2 that fell within his range of judgments was a measure of their compliance with him in the immediate togetherness situation.

Figure 13 presents the findings on conformity to initial standards formed in session 1. If the hypothesis is correct that standards are psychologically binding to the extent that interpersonal relations among those who form them are stable, conformity would be greatest for individuals who formed standards in the group situation. It would be next greatest for standards formed in mere togetherness situations in session 1, and least for individuals who stabilized their own standards alone. The cumulative percentages in the figure show that the hypothesis is supported. Measures computed trial by trial revealed that these significant differences were not merely a result of initial trials.

Figure 14 gives findings on the relative influence in session 2 of subjects who had formed their standards alone, together, or in a group. It shows the average percentage of the judgments made by the two other subjects which fell within the individual's range. In other words, these percentages indicate the relative influence exerted by the subject on others present in the session. It may be seen that those who formed their scales in a group situation both conformed to them more (Figure 13) and exerted the most influence on others, that is, complied least in session 2. Individuals who had formed standards alone were least influential and most compliant in session 2.

This experiment by Pollis verifies in more precise form a recurrent observation in our study of natural groups. Being a member of a group with shared standards for behavior renders the individual less compliant to outsiders in a transitory situation. To the extent that he is part of patterned interpersonal relationships, the standards formed in that pattern are psychologically binding for him. Through conformity to group standards, he is less easily swayed by momentary influences and pressures in a different direction.

The implication that conformity to group standards produces greater

( 278)

igure 13.  Adherence to norms initially formed in situations with varying stability of interpersonal relationships
Figure 13.  Adherence to norms initially formed in situations with varying stability of interpersonal relationships

independence by the individual in outside encounters is paradoxical only when the terms are defined without considering his social background and the context of the immediate interaction situation. The definition of the properties of a group implies definite consequences for individuals belonging over a period of time.


The research program on groups of the adolescent's own choosing has yielded abundant evidence of the salience of these reference groups

( 279)

Figure 14.
Figure 14. Relative influence of others' judgments in a togetherness situation by persons whose norm was formed initially in situations of varying stability

for the members. In making choices between their activities and standards and those of family, church or school, they frequently neglected the latter. As noted in the Introduction, a shift in the salience of age-mate and family reference groups is a general phenomenon of the adolescent period in this society. An experiment by Prado demonstrates behavioral consequences of this shift by a comparison of children's and adolescent's judgments on the competence of their fathers and their best friends.

Prado selected 25 boys eight to eleven years old and 25 boys fourteen to seventeen years old, who consistently selected their father as the

( 280) most valued and trusted parent. This stringent criterion for selection minimized the possibility that adolescent boys were simply rebelling against unusually authoritarian fathers. Similarly, Prado obtained the boys' sociometric preferences in order to single out their best age-mate friend.

Bringing each subject to the laboratory with his father and his best friend, he had the father and the friend perform a simple eye-hand coordination task (throwing a dart at a target), so arranged that the outcome was indeterminate. Each subject judged, in turn, the performance of his father or friend with scores from 0 to 24.

If appraisal of performance indicates the salience of reference groups, it is reasonable to predict that children would estimate their fathers' performance higher than that of their friends; while adolescents would estimate their fathers' performance lower than they estimated that of friends.

The results showed significant differences. Of the children, 20 of the 25 did estimate their fathers' performance as superior to that of age-mate friends (mean difference = + 3.5 points). Conversely, 19 of the 25 adolescents appraised their fathers' performance as lower than that of their friends (mean difference = - 2.6).

Although the exact outcome was indeterminate, some real differences in the skillsof the fathers and friends could have affected the boys' judgments. But actually, 17 of the 25 fathers of adolescents performed as well or better than their sons' pals, and only 13 of the 25 fathers of children equalled or outstripped their child's friend. In a reanalysis of Prado's data, we have found that, ignoring the direction of error, there are not significant differences in the children's and adolescents' accuracy in judging. In both age groups, average error in the fathers' performance was approximately 4.5 points, and the average error for friends somewhat less (3.5 points).

Considering the errors as overestimations and underestimations of performance, we can compare the difference between errors for fathers and errors for friends as an index of the extent to which the father is favored over the friend or vice versa, taking into account differences in performance. If the difference is positive, the father was favored over the friend. If the difference is negative, the friend's performance was favored over the father's, despite differences in skill.

Figure 15 presents the findings. On the average, children overestimate their father's performance, even with differences in performance controlled (Mean difference = + 2.35). Adolescents underestimate their fathers, with performance level again adjusted (Mean difference = - 3.64). (A test of the significance of these differences between errors for fathers' and friends' performances yielded t - 3.72, p < .001.)

A subsidiary finding in Prado's research supports the hypothesis

( 281)

Figure 15.
Figure 15. Overestimation by children and underestimation by adolescents of parent's performance as compared with best friend's

that the shifting salience of reference groups in adolescence is related to the motivational concerns the members share. He found that adolescents engaged in a significantly greater number of activities with friends than did younger boys, even though the two age groups did not differ in their consistency in choosing the friends in question. Since both children and adolescents in this study actually preferred their father as a parent, the logical inference is that the adolescent's activities and free time are

(282) focused more on common concerns with age-mates than are a younger child's.


If group norms become psychologically binding for individual members, without threat of disapproval, why do an individual's patterns of conformity change? Why do the values or norms of one group persist even when membership changes over successive generations, while those of another group disappear as the last member leaves?

Relative to youth problems, we suggested in the last section that adherence to group norms varies with the salience of the group. This salience shifts from childhood to adolescence, primarily because the dilemmas of adolescence in modern societies are shared with age-mates more than with one's family.

It is equally true that when adolescents gain mature roles, or are thrust into them by circumstances, the salience of age-mate reference groups diminishes. Adults say there is nothing like a job or marriage to "settle down" the wild, rebellious, or "silly" behavior of their offspring and his crowd. His motivational concerns no longer intersect so extensively with those of his age-mates.

But what of the transmission of group norms and customs over time? The transmission of a group, carrying the same identity and similar norms, to a new membership is a distinctive feature of human societies. It is observed in informal social life as well. In the next chapter, we summarize the intensive study in our research of three generations of boys living in the same neighborhood, each of whom belonged to the same group as a teenager.

When a group has established norms for behavior, and new members come as older ones depart, what conditions lead to the maintenance of the group norms in substantially similar form, and which ones are conducive to their alteration?

Our hypothesis predicts that one important variable affecting the transmission of norms and the extent of conformity to them by new members is the degree of their arbitrariness relative to current conditions facing the group. When a norm is transmitted by older members to a new generation of members, its "arbitrariness" can be defined relative to the conditions in which the new generation functions. The norm developing in those conditions without an enculturation process by older members can be termed least arbitrary.

If the arbitrariness of norms is an important variable in conformitydeviation, it will help explain the continuity of groups of boys who grow

( 283) up in the same neighborhood, relative to the changes occurring in the neighborhood and the city of which it is a part.

For these reasons, a laboratory experiment was initiated (as part of our research program) to examine how differing degrees of arbitrariness affected the conformity of succeeding "generations" in the laboratory. Data were collected by Mark K. McNeil with assistance from Michael Lauderdale, both assistants in the program at the time. The autokinetic phenomenon was chosen as the situation to be appraised by the subjects.

As in the earlier studies of norm formation by Sherif (1935, 1936), a norm for behavior in this situation was defined as a common range of judgment around a modal point, stabilized over time by two or more individuals. Individual conformity is defined as a judgment falling within this norm; deviation, a judgment outside the norm.

In this situation, the degree of arbitrariness of a norm could be defined operationally as the extent to which a prescribed and transmitted norm differed from the range and mode of judgments stabilized under the same laboratory arrangements without the introduction of a norm transmitted by "planted" subjects.

The general sequence of procedures followed those used in a study by Jacobs and Campbell (1961), who, through instructions to "planted" subjects, introduced an arbitrary norm of 15.5 inches with a range of one inch. They followed the progress of the norm through successive generations by replacing one of three planted subjects with a naive subject after each block of judgments, and then replacing an experienced subject with a new naive subject at the start of each succeeding block.

Since degree of arbitrariness was our main interest, three degrees were standardized in our laboratory: (I) not arbitrary; (II) an arbitrary norm chosen not to overlap the range of Condition I and with a higher mode ( a range of 9-15 inches around a mode of 12 inches) ; (III) more arbitrary norm not overlapping Condition II and a still higher mode (15-21 inch range and 18 inches mode).

In Condition I (not arbitrary), four naive subjects gave 30 judgments each of the extent of autokinetic movement; then each in turn was replaced by a new naive subject in the succeeding "generations," and so on, through eight generations of 30 judgments each. In Conditions II and III, a preliminary "enculturation" phase consisted of three generations, with the prescribed norm being given by three, then two, then a single planted subject, who in turn was replaced by a naive subject. Eight generations of naive subjects followed, the transmitted norm being traced over these.

The subjects were high school students (ages 16-19). A total of 66 naive subjects participated in two replications of each condition.

The major hypothesis was that degree of conformity over successive generations would decrease (and deviation increase) as a function of the

( 284) degree of arbitrariness of the transmitted norm. In order to test this hypothesis, it was also necessary to demonstrate the following: (a) a norm formed in Condition I would be transmitted with only minor variations as personnel changed, and (b) the enculturation procedures in Conditions II and III did produce conformity to the prescribed norm by naive subjects.

The findings can be summarized briefly as follows:

1. The norm formed by subjects in the first generation in Condition I (not arbitrary) was within an interquartile range of 4-8 inches, with a median and mean of 6 inches. The means and medians of seven successive generations were around 4 inches with an interquartile range from 3 to 6 inches. In short, after the second generation, the norm in Condition I was transmitted with only minor variations.

2. The enculturation by planted subjects in the more arbitrary conditions did result in conformity by naive subjects. In Condition II, 100 per cent of their judgments fell within the prescribed range, and 90 per cent of the judgments conformed in Condition III (most arbitrary), as shown in Figure 16. Similarly, the median judgments of naive subjects in Condition II were very close to the prescribed median of 12 inches (11.7 inches) ; and the median for Condition III during enculturation was 16.5-17.5 inches.

Figure 16.
Figure 16. Conformity through successive generations of laboratory subjects as a function of the arbitrariness of norms


3. The prescribed norm in Condition II was transmitted to the first generation, consisting entirely of naive subjects, 100 per cent of their judgments conforming to the transmitted norm. In the most arbitrary condition (III ), there was some deviation from the prescribed norm by the first generation of entirely naive subjects, but 63 per cent of their judgments conformed to the transmitted norm.

Figure 16 gives the percentages of judgments by naive subjects which fell within the prescribed norm for Conditions II and III through three enculturation and eight transmission generations. Conformity to the prescribed norm was significantly greater at every generation in the less arbitrary condition (II) . The rate of increase in deviation was clearly more rapid in Condition III (most arbitrary).

Figure 17 presents the median judgments of naive subjects, generation by generation. The greatest shift away from the transmitted norm and toward the "natural" norm of Condition I was in the most arbitrary condition, with half of the downward trend occurring by the third trans-

Figure 17.
Figure 17. Conformity through successive generations of laboratory subjects compared to prescribed medians and "natural" norm

(286) -mission generation. The change was more gradual in Condition II (less arbitrary), with the median judgments less toward the "natural" norm even in the last generation of subjects.

The general theoretical inference from this experiment is that conforming behavior by individuals occurs in a context of social interaction which is inevitably related to the conditions, problems and tasks which confront them. The norms stabilized are products of both the interaction process and the conditions in which it occurs. With successive generations of membership, conformity to established norms for behavior is an inverse function of the arbitrariness of the norms. However, deviation by individuals results in new, less arbitrary norms which, in turn, are transmitted to new members (Chapter 13 presents an example of this process in three generations of boys belonging to a neighborhood group in a large city.)


In the research program on adolescent reference groups, we were impressed once again with the generality of group formations in all walks of life during this age period. Whether the groups are labeled cliques, friendship circles, chums, crowds, or gangs seems to depend much more on the social rank of the neighborhood than on their essential properties.

All have differentiated patterns regulating their interpersonal relations. All have rules, customs, fads—in short, norms—regulating behavior in their activities. These are the minimum essentials of a group.

These groups of adolescents were formed, or joined, on the members' own initiative, through interactions among individuals sharing the dilemmas of status and motivation common during adolescence in this society. The status differentiation which developed was not imposed on them; the norms were not considered arbitrary impositions on the members.

The individual had a hand in creating the properties of the group, or had selected it. The group was his—a context where he could have personal ties with others, could amount to something in ways not available elsewhere, could accomplish things as a person. It contained others whose acceptance he wanted, whose yardsticks were his personal gauges for success and failure, whose approval brought inner warmth and whose disapproval left him miserable.

Psychologically, therefore, the basis of group solidarity, of conformity to group norms without threat of sanctions, of the binding nature of group rules even when they conflict with those of parents and officials, lies in the personal involvement of members. The self-image of the individual consists, in large part, of his ties with his reference group and the yardsticks it provides. The continuity of his ego identity from day to day depends to a large extent on the stability of his ties with members and on their consistency in appraising him according to the patterned relation-

( 287) -ships and norms shared in common. They are among the stable anchorages in his world. The disruption of their stability affects him as surely as the sudden lack of stable guideposts in his physical surroundings (cf. Sherif and Harvey, 1952).

Groups of the individual's own choosing are not the monopoly of adolescents. They are formed whenever people see themselves in the same boat, whether this be a neighborhood, a large organization (such as the military, industrial plant, or prison), or in a crisis situation. It is true, as Zorbaugh (1929) reported some years ago, that the slam has its gangs and the Gold Coast has its clubs. These are all human groups with consequences—good or evil—for the attitude and behavior of individuals composing them.


Groups are not formed to be studied by outsiders. They are not formed for the benefit of a reformer. They are not formed to tolerate criticism, advice, or unfavorable evaluations by outsiders. They are formed by individuals with similar concerns. They have designs and ends of their own. They have properties of their own, which are reflected in the attitude and behavior of members toward each other and toward outsiders.

Therefore, if we would study individuals who belong to groups, the cardinal research strategy must be a design that places the behavior of members within their groups, and places the groups, in turn, within their sociocultural setting. The selection of methods, procedures and research models is not a matter of convenience or the arbitrary preference of the investigator. If they are to yield valid data, the methods and procedures must accommodate the properties of groups as they actually form and function.

In terms of the properties of groups, as spelled out in this chapter, certain methods are appropriate and certain are inappropriate. Furthermore, the appropriateness or inappropriateness of methods depends on the time at which they are used. The properties of groups require a procedural sequence of steps.

The issue is not merely whether or not data can be collected, for some kind of data can almost always be obtained. The issue at hand is the validity of the obtainable data. This issue is bound to be confronted more and more seriously as social scientists and psychologists are called upon to contribute to baffling human problems ranging all the way from delinquency and school dropouts to segregation, prejudice, and other threatening problems of intergroup relations.

The basic criterion for securing valid data on groups and their members is that procedures and techniques be timed so that members will not view the event as an unwarranted intrusion or imposition by an out-

( 288) -sider. This is the reason why, from the beginning, a special effort was made in our research not to arouse the group's awareness that an observer was studying them and rating their behavior.

To the extent that a group is at odds with established routines and channels for behavior, they not only have designs of their own, but also secrets about these designs. Whether or not their behavior is socially acceptable to adults, most adolescent groups do have plans which they consider private, especially from grownups, who seldom "understand" them. (See, for example, the Introduction to this book.)

To the extent that group members consider their affairs "private," they erect walls of resistance to any outsider. The appropriate timing and sequence of research procedures are, therefore, contingent upon steps to bring down first these walls of resistance and then the walls of secrecy.

In short, the timing and sequence of research procedures must be planned in terms of the degree of rapport established with group members, particularly those of higher status who are most obligated to protect the group's privacy. Results from the more usual research procedures—interviews with direct questions, questionnaires, sociometric choices, laboratory procedures requiring the cooperation of subjects—are less than worthless if introduced before rapport is firmly established. They are misleading. Very early in our research, this realization was forced upon us by an incident involving the untimely use of sociometric techniques.

The observer had had considerable give-and-take with the group for several months. In securing sociometric choices, he talked individually to each boy, inquiring whom he liked to "hang around with" most. The answers seemed free and courteous, but revealed the futility of such a procedure when rapport is no deeper than surface courtesy. The boys were all telling the observer that their best pals were boys about ten or twelve years old, most of them younger brothers.

The cycle of procedures for study of groups, as developed in our research, and problems of developing and assessing rapport, are discussed in the next chapter.


A serious pitfall in research is the tendency to select a particular mode of behavior, often because it is socially unacceptable, and then to "explain" it through postmortem analysis without specifying the background and context of the events in question. In experimentation, this tendency leads to laboratory models with little bearing on the actualities of the field. In the field, it leads to failures in predicting behavior and, practically, to views impeding the development of preventive measures.

In studying youthful behavior, we have been equally interested in

(289) those attitudes and actions which are socially acceptable and those which are not. There cannot be one social psychology for acceptable behaviors and another, entirely different, social psychology for unacceptable ones. In both cases, general psychological principles must be sought which take into account the background and context of behavior.


A research design to accommodate all the influences shaping behavior and to secure valid, reliable indicators of behavior as it actually occurs is essential in developing a base for making predictions. We have secured data on planning, decision-making, and actions that members considered secret and would not have exposed to any adult with authority. Frequently, the events had consequential outcomes for the individuals.

A case in point was our prediction of two school dropouts some weeks before they occurred. The prediction was based on extensive evidence that the two boys were highly committed members in a group in which the high status positions were held by boys who had left school. The two continued in school under pressures from parents and teachers. They were not failing. But in their group, derogation of school as a place for "sissies" was common. They had no close personal ties with schoolmates. Being in school deprived them of time when other members were doing "interesting" things. First one, then the other boy stopped attending school, a few months short of graduation.

The study of behavior in its group context has been severely hampered by a moralistic ,view which condemns group influence as a sign of individual weakness or even pathology. This view is blind to the fact that group influence on individual behavior may be for good as well as for evil.

In our research, we encountered numerous instances of an official tendency by community authorities (school, recreation leaders, parents) to minimize even the existence of groups within the area of their work. Apparently the tendency is nationwide in scope, as indicated in a report for the Committee on the judiciary of the U.S. Senate in the following words

We further encountered difficulties when confronted in many communities with the attitude that the mere existence of any gang or gang problem constituted a failure of responsibility of the agencies, so that the type of reports made by these agencies were less than complete. In several cities we found, based on our own investigations, discussions, and contacts made with gangs, gang workers, police on the beat, teachers and citizens, a wide discrepancy between their off-the-record statements and the official reporting regarding the existence of gangs, their antisocial behavior, and the degree of problem they created. (U.S. Senate Report, 1961, p. 12)

Yet, ignoring the group context of behavior will continue to lead, as

( 290) it has in the past, to failures of prediction in research and to continued "surprises" at a community level. For example, the following type of news item is not unusual:

The uncovering by police (Sunday) of a ruthless, pistol packing gang of 20 or more youthful LeDroit Park Ramblers (in Washington, D.C.), who are suspected of engineering over 80 "hustles" in 6 months has left area youth workers shocked.... Playground officials, the Recreation Department's roving leaders, the junior Police and Citizens Corps, the LeDroit Park Civic Association, and the Youth Aid Division of the police department had all been unaware even that a group of youth using the name of "Ramblers" existed. (U.S. Senate Report, 1961, p. 12)

The deeds and misdeeds of youth cannot be predicted solely on the basis of class membership, nor solely on the basis of family relations, nor through any theory advocating a single sovereign cause. It should be sobering for both theorists and practitioners to read a recent report ascribed to the Pennsylvania Chiefs of Police Association (in the same Senate report) which concludes that "the broken home and unemployed and socially handicapped youth can no longer be solely blamed for juvenile crime." Several examples are cited of serious violations by youth from comfortable neighborhoods.

An equally strong case could be made that theories advocating oversimplified etiology cannot account for socially acceptable behavior. The etiology of behavior, whether acceptable or not, must include the context of membership in natural groups, in whose formation and functioning many influences participate. These influences stem from the individual and his background, from the sociocultural setting in which he develops, and from the prevailing images of success and achievement set by people who count in his eyes, whether these people are immediately present or are reflected from the mass media. All of these contribute to the natural reference groups of age-mates in which youth moves and to their directions, which may be for good or for evil.


The greatest promise for a theory to enable prediction lies in identifying general principles that are equally valid in the laboratory and in the rich context of the field. The experiments summarized earlier in this chapter are examples of research utilizing general principles of judgment with relevance for behavior in social actualities. Thus, Pollis' study showed that the tendency toward compliance with others' judgments is modified by the individual's background, specifically by whether his standards were formed in a group or in a less stable interpersonal context. Prado used the general finding of systematic variations in judgments of performance to index shifts in the salience of reference groups from childhood to adolescence. As a final example from this research program, we will mention

( 291) continuity of a basic principle of judgment discovered in the interaction patterns of adolescent groups in the field.

The interpersonal relations among contemporaries are patterned affairs. To some extent, an individual regulates his behavior relative to others, and to that extent his behavior becomes predictable in a variety of situations. In comparing the ratings made by observers with the sociometric choices made by members of the relative power (effective initiative) in a group, we have found that predictability of behavior is related to the individual's position in the group structure. But the relationship is not a linear function of relative power; it is curvilinear. As Chapter 13 will show in more detail, consistency and confidence in ratings are invariably greater regarding the leader and other high status persons, followed by ratings of those with lowest status. Variability in rating is greatest in the intermediate ranks.

Thus, we see in the patterning of group interpersonal relations a phenomenon which has been extensively studied in the psychophysical laboratory—namely, end-anchoring. The extreme representatives of a set of stimuli are singled out most readily and used as standards for assessing the others. In natural groups, the leader position is typically the most potent anchor; the lower positions are rated with less variability than intermediate levels.

The extent to which end-anchoring of the group structure occurs, the stability of the structure, and the number of status categories (or levels) from top to bottom vary considerably in natural groups, as they do in larger organizations. The variations, in turn, are affected in a crucial way by the relation of the small group to its setting. Thus, the stability of the leader position, which is the upper anchor, as well as the extent of differentiation in the pattern, are strikingly dependent on what the members engage in and on what their dealings are with others in the setting—both other groups and adult authorities. The changes in the status patterns which we have observed as they happened resulted from external changes in the settings—changes in the opportunities faced by the members or exposure of their most private activities by adult authorities.

But, whatever the variations, differentiations according to the effectiveness of initiative and patterns of deference among members was the rule in these natural groups. This phenomenon is the invariant consequence of prolonged interaction among individuals facing common problems. Which individual will occupy what status position, and which individual will succeed in changing his position, rests on unique personal characteristics of individual members—their contribution relative to the demands of group activities in which certain personal characteristics matter. These are fascinating problems of individual differences to be explored in the research program as it continues.



The properties of natural reference groups do limit the range of procedures and the sequence of their application in collecting valid data. Much more effort and patience are required than in laboratory research. But within the approach outlined here, precise and reliable measurement is within the grasp of field research. Laboratory methods of assessment can frequently be adapted for this purpose. And with greater specification of the variables, findings in the field become the proper basis for laboratory research.

When small groups and their members are studied relative to their sociocultural settings, and the setting studied relative to the groups within it, the dichotomy between small group research and research on large organizations will disappear. The "psychological" and the "sociological" study of social behavior will supplement one another, instead of being monopolistic preferences of their respective disciplines.

The contributions of particular individuals, for good or for evil, can be studied to any desired degree of elaboration within their appropriate behavior settings. The appropriate behavior settings are patterned affairs, consisting of the individual's reference groups and the sociocultural setting of which these reference groups are parts. Personal skills and qualities are not contributed in the abstract. Unique individuality shows in the ways an individual interacts with others in activities and situations important in their concerted undertakings. These interactions provide the context for his own attitude and behavior.

On the basis of the theoretical and methodological background presented here, the research on adolescents in their groups in their settings is summarized in the next chapter. It is our conviction that the approach outlined is equally viable for the study of behavior at other age periods, taking into account the changes in the setting and their salience throughout development.


Barker, Roger G., 1963. On the nature of the environment. Kurt Lewin Memorial Address, J. soc. Issues, October, 1963, 17-38.

Blau, P. M., and W. R. Scott, 1962. Formal Organizations. (San Francisco: Chandler.)

Bell, W., 1958. The utility of the Shevky typology for the design of urban sub-area field studies. J. soc. Psychol., 47, 71-83.

Cartwright, D., and Zander, A., 1960. Group Dynamics: Research and Theory. (Evanston, Ill.: Row, Peterson.)

Chapanis, A., 1961: Men, machines, and models. Amer. Psychologist, 16, 113-131.

( 293)

Deutsch, Martin, 1964. Guidelines for testing minority group children. J. soc. Issues, 20, No. 2.

Durkheim, E., 1915. The Elementary Forms of Religious Life. (London: G. Allen.)

Helson, H., 1964. Current trends and issues in adaptation level theory. Amer. Psychologist, 19, p. 26.

Homans, G. C., 1950. The Human Group. (New York: Harcourt, Brace and World.)

Jacobs, R. C., and D. T. Campbell, 1961. The perpetuation of an arbitrary tradition through several generations of a laboratory microculture. J. abnorm. soc. Psychol., 62, 649-58.

Orne, M. T., 1962. On the social psychology of the psychological experiment: With particular reference to demand characteristics and their implications. Amer. Psychologist, 17, 776-783.

Pettigrew, T. F., 1964. Negro American intelligence: A new look at an old controversy. J. Negro Educat., Winter, 1964, 6-25.

Pishkin, V., 1963. Experimenter variable in concept identification feedback of schizophrenics. Percept. & Motor Skills, 16, 921-922.

Rosenthal, R., 1961. On the social psychology of the psychological experiment: with particular reference to experimenter bias. (Paper read to Annual Convention, APA, New York City.)

Schottstaedt, W. W., Ruth H. Pinsky, D. Mackler, and S. Wolf, 1958. Sociologic, psychologic and metabolic observations on patients in the community of a metabolic ward. Amer. J. o f Medicine, 25, 248-257.

Sherif, Carolyn, M. Sherif, and R. Nebergall, 1965. Attitude and attitude change. (Philadelphia: W. B. Saunders.)

Sherif, M., 1935. Study of some social factors in perception. Archives of Psychol., 187.

———, 1936. Psychology of Social Norms. (New York: Harper and Bros.) 1948. Outline of Social Psychology. (New York: Harper and Bros.)

Sherif, M., and H. Cantril, 1947. Psychology of ego-involvement. (New York: Wiley.)

Sherif, M., and O. J. Harvey, 1952. A study in ego functioning: Elimination of stable anchorages in individual and group situations. Sociometry, 15, 272-305.

Sherif, M., O. J. Harvey, B. J. White, W. R. Hood, and Carolyn Sherif, 1961. Intergroup conflict and cooperation: The Robbers Cave experiment. (Norman, Oklahoma: University Book Exchange.)

Sherif, M., and B. Koslin, 1960. The "institutional" vs. "behavioral" controversy in social science, with special reference to political science. (Norman, Oklahoma: Institute of Group Relations, University of Oklahoma, mimeographed.)

Sherif, M., and Carolyn Sherif, 1953. Groups in Harmony and Tension. (New York: Harper and Bros.)

———, 1956. Outline of Social Psychology. (New York: Harper and Row.)

———, 1964. Reference Groups: Exploration into conformity and deviation o f adolescents. (New York, Harper and Row.)

Shevky, E., and W. Bell, 1955. Social area analysis. (Stanford: Stanford University Press.)

Stevenson, H. W. and Sara Allen, 1964. Adult performance as a function of sex of experimenter and sex of subject. J. abnorm. soc. psychol., 68, 214-216.

U.S. Senate, 1961. Report of the Committee on the judiciary, 87th Congress, 1st Session, Report 169, April 18.

Zorbaugh, W. W., 1929. The Gold Coast and the Slum. (Chicago: University of Chicago Press.)


  1. This chapter is based on the senior author's invited address to Division 9, American Psychological Association, Los Angeles, September 6, 1964. Since 1961, the Research has been supported by grants from the National Science Foundation.



Valid HTML 4.01 Strict Valid CSS2