Classics in the History of Psychology

An internet resource developed by
Christopher D. Green
York University, Toronto, Ontario

(Return to index)

Intergroup Conflict and Cooperation: The Robbers Cave Experiment

Muzafer Sherif, O. J. Harvey, B. Jack White, William R. Hood, Carolyn W. Sherif (1954/1961)

[p. 117] CHAPTER 6

Intergroup Relations: Assessment of In-Group Functioning and Negative Attitudes Toward the Out-Group

In order to insure validity of findings and to increase their precision, the plan of this experiment on intergroup relations specified that different methods of data collection would be used and the results checked against each other (Chapter 2). On the other hand, it was noted that excessive interruption of the interaction processes would result in destroying the main focus of study, viz., the flow of interaction within groups and between groups under varying conditions. Therefore, it was necessary to exercise great restraint in introducing special measurement techniques and experimental units to cross-check the observational data.

At the end of Stage 2, special methods were utilized to check observations related to the main hypotheses for the friction phase of intergroup relations. In addition to sociometric techniques, two experimental units were introduced to tap the subjects' attitudes toward their respective in-groups and the out-group. The results of these units are presented in this chapter, along with additional observational data pertaining to the various hypotheses for Stage 2. These results are not intended to test any separate hypotheses, but to provide further evidence to be evaluated in conjunction with the observational data.

Section A summarizes the effects of intergroup friction on in-group functioning, while the following sections deal with end-products of intergroup friction and conflict, and their assessment through judgmental reactions.


Intergroup Friction and In-Group Functioning

The study of in-group structure and functioning was not confined to the first stage of the experiment, which was devoted to experimental formation of in-groups. Several of the hypotheses for Stage 2 specifically concern the effects of intergroup relations (friction, in this case) on in-group structure and functioning. Some [p. 118] consequential effects to the respective in-groups were pointed out in the summary of interaction in the last chapter. Further data will be summarized here. Throughout the experiment the effort was made whenever possible to obtain data by as many methods as feasible without disrupting the ongoing interaction. Checking results obtained by several methods (e. g., observational, sociometric, ratings, judgments of the subjects) leads to confidence in the reliability and validity of the conclusions reached. In considering the hypotheses and data concerning in-group relations in Stage 2, it was necessary to rely heavily on observational data. The more precise techniques of data collection (viz., stereotype ratings and laboratory type judgments) were used for testing the validity of observational findings concerning the negative attitudes (see section B, this chapter).

Because of space limitations, the observational data of this experiment were given in summary form. The danger of selectivity in observation and in reporting observational data is not surmounted by piling example on example. The illustrations chosen are representative of the many available. The conclusions drawn from them are justified by available evidence. They are intended to be suggestive for future research in which observational methods are supplemented increasingly by other more precise techniques of data collection.

At the end of Stage 2, the Rattlers and Eagles were both clearly structured, closely knit in-groups. This is revealed in observational data, observers' ratings, and in sociometric choices obtained at this time from each member individually by the participant observer of his respective group (who appeared as counselor to the subjects).

1. Testing in Terms of Sociometric Indices

The most general criterion on the sociometric questionnaire specified that friendship choices should be made from the entire camp. Table 1 presents the resulting choices for this criterion by members of the Rattler and Eagle groups. In spite of the fact that choices of out-group members were forced somewhat by the manner in which this item was presented, the proportion of choices of in-group members in both groups was approximately 93 per cent, [p. 119] and the differences between choices of in-group members and out-group members are too large to be attributed to chance.

Table 1

Friendship Choices of In-group and Out-group Members

By Rattlers and Eagles

End Stage 2




Choices of





In-group members





Out-group members






Sociograms were constructed for the Rattlers and Eagles using total score on 4 criteria, as a basis for placement of members (see accompanying sociograms). The score on each criterion was the total of weighted choices, first choices receiving a weight of 4, second choices of 3, third choices of 2, and those thereafter a weight of 1. The choice network was based on the most general criterion (friendship); and rejections were obtained from an item included in the interview, but not used as a criterion in computing total scores. The lines on the ordinates of the sociograms represent Q1, Q2, and Q3 of these ranks in ascending order, the lowest line being Q1.

[p. 120]

[p. 121]

Several discrepancies in ratings of group members based [p. 122] on the various criteria were noted. Two of these criteria were concerned with friendship choices (one general, one more specific), and two were concerned with initiative displayed by various members. It is significant that the scores obtained for these two kinds of choice were widely disparate in several cases. For example, Everett (R) ranked second on the friendship criteria but only ninth (out of 11) on the initiative criteria. Craig (E) ranked seventh on the friendship criteria, but fourth on the initiative criteria, as did Hill (R).

Since observers' ratings were made more on the basis of effective initiative than on popularity, it is interesting to compare the status ratings of observers and ranks in total sociometric score (4 criteria). As Table 2 indicates, the rank order correlation for these two rating measures is significant and high. Observers undoubtedly gave greater weight to effective initiative than did the combined sociometric scores. This greater weight given to effective initiative in observer status ratings is revealed in cases of discrepancy between sociometric ranks and observer ratings. For example, Brown (R) ranks fourth in sociometric score, but only eighth in observer's ratings. In this case, the sociometric score as computed from choices does not reveal what the observer knew, and what was also revealed during the sociometric interviews. Although Brown received only one rejection from his group, he was mentioned by six members (more than any other member) as the member who would stand in the way of what most of the group wanted to do. To take another example, Bryan (E) ranked fourth in sociometric score and eighth in observer's ratings. In this case, nothing in the sociometric interview revealed, or could reveal, that Bryan was frightened of physical conflict and that during the closing days of Stage 2 he withdrew from interaction altogether on several occasions (even hiding in the bushes). For this reason, he was rated near the bottom of the group by the observer at this time (in terms of effective initiative and influence in the group). On the other hand, the observer noted that since many eagles were frightened of the Rattlers, they did not (excepting Mason, the leader) impose sanctions because of Bryan's behavior. He participated effectively when physical contact with the out-group was not in the picture. Nevertheless, Bryan had very little influence in the group at the end of Stage 2. Such cases illustrate some difficulties in interpreting sociometric scores based on choice, and point to serious problems of validity when sociometrics are used apart from concrete observational material.

[p. 123] Table 2

Comparison of Ranks in Sociometric Scores and Status Ratings

By Participant Observers of Rattler and Eagle Groups

End of Stage 2













2. Testing in Terms of Observational Data

Three of the hypotheses for Stage 2 are concerned with in-group functioning. Data pertinent to them is summarized below.

Hypothesis 2, Stage 2
The course of relations between two groups which are in a state of competition and frustration will tend to produce an increase in in-group solidarity.

Observational data supports this hypothesis with the qualification that in several instances defeat in a contest with the rival group brought temporarily increased internal friction in its wake. This was noted in the Rattler group (Day 3) when they lost the second baseball game. Disruptive tendencies within the group reached their peak when Brown and Allen wrote letters that afternoon wanting to go home, thus threatening to leave the group. Solidarity was achieved shortly afterward through the integrative leadership of Mills, whose joking about these events led the boys to tear up their letters amid rejoicing by all members.

Similar signs of temporary disorganization followed the first tug-of-war, when the downhearted Eagles stared loss of [p. 124] the tournament in the face. In this case, Myers, Clark, and McGraw took the optimistic view that they had to plan tactics which would defeat the Rattlers. After the group joined whole-heartedly in burning the Rattler flag, this view was accepted by the leader as well, and considerable hope was seen for the next day.

A somewhat similar adaptation to defeat was made by the Rattlers at the end of the tournament. In this important instance, group action in raiding the out-group was agreed upon right after defeat, sanctioned and planned by the leader and lieutenants, and was executed soon afterward. The aftermath was self-glorification with reference to the group and all its members. The next morning, when Mills "roughed up" several group members, not one tried to challenge his prerogatives, although any one of them could have whipped him easily. (It should be noted here that the Rattler leader, Mills, was one of the smallest boys in size.) The rest of that day was spent in highly congenial play in which Mills made special efforts to involve the low status members and succeeded in effecting their active participation.

It should be emphasized that the temporary disruptive tendencies following defeat illustrated above did not follow loss of every contest. Some defeats were accepted with remarkably little concern or depression, either because the group in question (both Rattler and Eagle) had decided prior to the event that they probably would not win it or because they felt they did not have to win that particular event to win the tournament.

In instances where temporary disorganization did follow defeat (above), heightened solidarity within the group was achieved through united cooperative action by the in-group against the out-group, and this is in line with the above hypothesis. It should be noted and emphasized that the aggressive actions toward the out-group which followed frustration of group efforts experienced in common by in-group members were taken after they were sanctioned by the leaders (Mills or Mason, the leaders of the Rattlers and Eagles respectively). These aggressive actions were sometimes suggested by high status members (notably Simpson and Mills in the Rattlers and Mason in the Eagles), and sometimes suggested by low status members (e. g., Everett in the Rattlers, and Lane in the Eagles). in no instance did the in-groups engage groups in aggressive action toward the out-group if this had [p. 125] not been approved by the leaders of the respective groups. Other evidence supporting this hypothesis as stated is the recurring glorification of the in-group, recounting of feats and accomplishments of individual members, support and approval given low status members, support given the leader, and intensified claims on areas appropriated as belonging to the group.

The Eagles bragged to each other that they were "good sports" who did their best and who prayed and didn't curse. Later they refrained from bragging in the presence of the out-group since this was agreed to bring bad luck. The Rattlers were constantly telling each other, and all within hearing distance, that they were brave, winners, not quitters, tough, and (naturally) good sports.

After the contests and raids, stories were told over and over of the accomplishments of this person and that person, blisters acquired in the tug-of-war were compared both in winning and losing groups, and these tales of individual feats grew with each telling. (The dramatic reversal by the Eagles of their role in the last raid was noted in the account of that event in Chapter 5). Brown (R) revealed special gifts for recounting such episodes.

During Stage 2, Lane (low status E) was praised for his playing for the first time (by Mason). Lane became more active in in-group affairs and said they must not swim so much in order to save their strength for the tournament, even though he had earlier been a constant agitator to go swimming at every possible moment. Approval was also given to low status Rattlers during games. After the big raid in which Mason (E) had accused several Eagles of being "yellow-bellied" and cowards, he "covered up" for them completely in telling staff of the events. No mention was made of any defection; all Eagles were made to appear heroic.

The leaders (Mills and Mason) were supported by group opinion consistently, especially after the first day or so when Mason was effectively extending his leadership in baseball (elected) to all areas of group life. Mills was supported by the group even during games when he interfered in decisions made by Simpson (baseball captain); and on one occasion he took Simpson out as pitcher and put in another member in his place.

[p. 126] At the end of Stage 1, we mentioned the increased concern of the in-groups over places appropriated as "theirs." Swift (R) even went so far as to object, when he saw fishermen near their swimming hole, that they had no business taking "our fish." The Rattlers talked, near the end of Stage 2, of putting signs on all of "their" places, including the ball diamond and Stone Corral (which was a part of Robbers Cave). The Eagles were extremely concerned over the fact that the Rattlers went to their hideout on the day after the big raid, and claimed they could detect changes there which did not actually exist.

Hypothesis 3, Stage 2
Functional relations between groups which are of consequence to the groups in question will tend to bring about changes in the pattern of relations within the in-groups involved.

The most striking change in relationships within the in-groups as a consequence of the particular functional relationship between them (rivalry and friction) was in the Eagle group. At the end of Stage 1, Craig was acknowledged leader of the Eagles. Mason was elected captain of the baseball team (only) with Craig nominating and backing him. Even after this, Craig informed Mason that he could not play ball if he didn't have the Eagle insignia stencilled on his T-shirt, and Mason submitted to this after some argument. From the first day of the tournament, however, Mason began to extend his leadership to all group activities, while Craig lost ground throughout Stage 2, being in the middle of the hierarchy by the end (fifth in rank). Some of the incidents revealing this alteration in the Eagles' status structure are mentioned in the summary of interaction presented in the last chapter.

Mason took the group goal of winning the tournament very seriously, giving talks on how to keep from getting rattled, threatening to beat up everyone if they didn't try harder, lecturing on how to win after the first loss in baseball. Although he had not shown interest in keeping the cabin clean before the tournament, he organized cabin cleaning details, and struck Lane (low status) for not helping pick up papers. He had praised Lane's playing in baseball earlier, and the combined effect of Mason's attention was that Lane saw the necessity of reducing the groups' swimming time to "save our strength" - for him a sacrificial act. [p. 127] When captains for the tug-of-war were called, Mason stepped forward, although he had been elected only as baseball captain, and there was no discussion on the point. When the Eagles burned the Rattler flag, Lane first directed attention to it, but Mason took the initial action in trying to tear it up.

Rather convincing evidence of Mason's leadership followed the second tug-of-war, which ended in a tie. Estimates of the time consumed were first obtained individually for each boy, as reported in the last chapter. Subsequently, the boys were asked in a group how long they thought it lasted. Every single Eagle agreed with Mason' s estimate of 45 minutes, although only one other boy had made previously an estimate that high individually.

Craig allowed leadership of the Eagles to slip through his fingers by submitting to Mason' a decisions, perhaps in part because he recognized Mason's superiority as an athlete. (Mills in the Rattler group was not as good a ball player as a few other members; nevertheless he kept control of the group's progress even during baseball games. ) However, Craig fell as far as he did in the status hierarchy because of his defection at several critical points during the tournament. When the group was losing the first tug-of-war, Craig simply walked away from the rope before the contest was over. Afterward he said the Eagles were already beaten in the tournament, and tried to blame others for the tug-of-war loss. However, the Eagles blamed Craig for the loss. When he walked away after the Eagles' loss in tent pitching, the comment was: "He's quit us again. " Craig pretended to be asleep during the first Rattler raid, and kept in the background during the second.

Another shift in the Eagle group which accompanied Mason's rise to the leadership position was Wilson's increasing importance in the group. From a position in the middle of the group's hierarchy (fifth) at the end of Stage 1, Wilson rose to become Mason's lieutenant through his effective playing in sports, his concern with maintaining joint efforts to win the tournament, his support of Mason's decisions. Mason chose Wilson as pitcher in preference to Craig during the first baseball game, and the two figured together in most of the group efforts and activities throughout Stage 2.

The most pronounced changes in the pattern of status relations in the Rattler group during intergroup competition were in [p. 128] the cases of Allen and Brown. After the second baseball game, which the Rattlers lost, Allen was accused of not contributing to the game. He, in turn, accused Martin (higher status) of bragging; but Martin was supported by the other members in the argument. The group members were ruthless in denouncing Allen who cried, wanted to go home, and was talked out of it by Mills (leader). After this, Allen was ignored a good deal, was not chosen to play on the team, and fell from a middle status level to the bottom. Mills' friendship was his chief tie with the group.

Brown, the largest Rattler, slowly slipped downward in the status structure during Stage 2 until just before the second raid on the Eagles. Because of a pronounced tendency to rough up the smaller boys, Brown was subject to group sanctions and fell to the bottom level of the group. During the raid his size so impressed the Eagles that he became something of a hero of that event to the Rattlers, and had attentive audiences of smaller boys in recounting his feats. Intergroup conflict was the medium by which Brown regained a position higher in the group at the end of Stage 2, after having slipped to the bottom level.

Thus, changes in the pattern of relations within the in-groups occurred during Stage 2. These changes are related to the altered contributions of the various members to group activities and efforts as the in-group functioned in a competitive and mutually antagonistic relationship with another group.

Hypothesis 4, Stage 2
Low status members will tend to exert greater efforts which will be revealed in more intense forms of overt aggression and verbal expressions against the out-group as a means of improving their status within the in-group.

The observations relevant to this hypothesis are inconclusive.

As noted in Chapter 2, this hypothesis was not intended to imply that high status members will not initiate and actively participate in conflict with the out-group. In line with one of the major tenets of Groups in Harmony and Tension (1953), its implication should be that intergroup behavior of members consists mainly in participation in the trends of one's group in relation to other groups. Since low status members would be highly motivated to [p. 129] improve their status, it seemed a reasonable hypothesis that they might do so through active participation in the trend of group antagonism and conflict toward the out-group. On the other hand, the establishment and responsibility for such a trend in intergroup affairs rests heavily with the high status members, as does responsibility for sanctioning and conducting affairs strictly within the group. If an upper status member, even the leader, stands in the way of an unmistakable trend in intergroup affairs, he is subject to loss of his standing in the group. This is precisely what happened to Craig, the erstwhile Eagle leader, in the present study. He did not enter into the tournament with sufficiently wholehearted identification with the group's efforts to win; he even walked out on them at critical points when they were losing and "played possum" to avoid conflict (raid) with the out-group.

In view of the necessity to clarify the intent and implications of this hypothesis, it should probably be re-formulated along the following lines: Aggressive behavior and verbal expression against the out-group in line with the trend of intergroup conflict sanctioned by high status members will be exhibited by low status members as a means of improving their status within the group.

This hypothesis could be tested empirically by comparing the reactions of low status members toward the out-group (a) when in the presence of in-group members high in status and (b) when high status members of their in-group are not present. Since the primary concern of Stage 2 in this present experiment was the end products of intergroup friction, this empirical test relating to the behavior of in-group members was not made. As stated in Chapter 2, it was necessary to limit such devices in the present study in order that the interaction processes within groups and between groups would not be complicated by excessive intervention by experimenters. The test situations and more precise methods of measurement which were used during Stage 2 were all devoted to tapping the end products of intergroup friction.

Observational data bearing on this hypothesis cannot be crucial without an empirical test such as that suggested above. The data available are not contradictory of the hypothesis as modified. One consistent finding supports it, namely, that in those instances in which low status members initiated aggressive acts directed toward the out-group, group action followed when the suggestion was approved or taken over entirely by high status members, and [p. 130] particularly the leader. Conversely there were instances of suggested aggressive action toward the out-group initiated by low status members (and even upper status members) which was not carried out because the leader did not assent to it. The number of raids suggested by various group members far exceeded the number actually carried out. In every case the leader decided on the major details of the raid. Mills set the time for both of the Rattler raids, and Mason (who was the moving force in making the suggestion and in its execution) managed the Eagles' morning raid on the Rattler cabin, even though Cutler (low status) led the way to the cabin after the decision was reached.


Verification of Observational Findings Concerning Intergroup Friction Through Laboratory-type Tasks:
Stereotype Ratings and Performance Estimates

In this stage of negative intergroup relations, Hypothesis 1, Stage 2 is crucial:

In the course of competition and frustrating relations between two groups, unfavorable stereotypes will come into use in relation to the out-group and its members and will be standardized in time, placing the out-group at a certain social distance (proportional to the degree of negative relations between groups).

This hypothesis goes to the core of issues concerning the formation and standardization of prejudice of social distance scales in relation to out-groups that prevail in actual social life.

The main events between groups in this stage were manifested through rivalry and actual conflict. Our emphasis in formulation of the crucial hypothesis was on end products in the form of standardized norms relating to the out-group, rather than on specific events revealing conflict, fights, and rivalry. If negative functional relations between groups are more than momentary affairs and give rise repeatedly to fights and hurling of derogatory terms, the end products will be standardized generalizations concerning the out-group which are expressed in the form of unfavorable p. 131] stereotypes. These standardizations constitute the basis of the institution of group prejudice or social distance. Once standardized, such institutions crystallized in negative stereotypes outlast the actual state of negative relations, henceforth predisposing in-group members to categorize out-group members in the light of unfavorable generalizations even at times when the acts of out-group members are not of an unfavorable character. Therefore, our emphasis in formulating hypotheses concerning negative relations between groups has been on negative generalizations concerning the out-group, i. e., standardized stereotypes, rather than on a syllabus of behavioral items revealing hostility, aggression, and other expressions of intergroup conflict.

As the account of interaction during Stage 2 indicates (Chapter 5), there were many specific examples of conflict, in which members of the two experimental groups had to be separated, much name-calling of the out-group, much use of derogatory terms and ridicule. Briefly, one end result of competition and rivalry in a series of contests and of situations in which the behavior of one group was frustrating to significant aims or goals of the other was a desire manifested by both groups to have nothing further to do with each other. With the assumption that generalizations concerning the out-group and attitudes toward it would outlast the state of actual conflict which engendered them, the two groups were brought within hearing distance of each other after a full day of exclusively in-group association. The result was repetition of the name-calling, derogation, and other manifestations of attitudes revealed during the period of intergroup conflict itself. This was the critical time to tap these end-products of intergroup conflict through more precise laboratory methods to check further the validity of the observations. This was done on the following day, two days after the end of the tournament and the climactic raid by the Rattlers. The two experimental units undertaken at that time are reported on the pages that follow.

1. Verification of Stabilized In-Group and Out-Group Attitudes Through Judgmental Indices: Stereotype Ratings.

At the end of Stage 2, judgmental ratings of stereotypes [p. 132] actually used by the subjects in relation to out-groups were obtained. This unit was carried out to provide a further test of Hypothesis 1, Stage 2, namely, that unfavorable stereotypes will arise in relation to the out-group and its members as a consequence of competitive and frustrating relations between the two groups, and will become standardized in time.

The procedures for this unit were repeated at the end of Stage 3. A comparison of the data obtained at the end of Stage 2 with those obtained at the end of Stage 3 provides a crucial test of the prediction that cooperative efforts in situations embodying superordinate goals will have a cumulative effect in reducing intergroup tensions (Hypothesis 1b, Stage 3). This comparison is presented in Chapter 7 with the summary of Stage 3.

The judgments were obtained in this experimental unit to supplement data from observers' reports and to provide a more clear-cut check of the hypothesis stated above. No new hypotheses are tested separately by this experimental unit. Throughout the entire study, results obtained by as many methods as feasible are brought together to test a particular hypothesis. In this particular instance, judgments of in-group and out-group members are used as further evidence for conceptual products (stereotypes) of intergroup interaction under conditions of competitive rivalry. The hypothesis is to be evaluated in terms of all relevant evidence obtained, including observational and sociometric data summarized in the last chapter and in this chapter, and judgmental data presented below.

In reviewing the problem of prejudice, Sherif (1948) emphasized that prejudice and stereotypes held toward out-groups are products of past or present relationships between the groups in question.

The favorable or unfavorable properties or 'traits' attributed to 'they' groups, and inevitably to their individual members in a rather absolutistic way, are determined by the nature of positive or negative relations between the groups in question. If the interests, direction, and goals of the intergroup relations are integrated or in harmony the features attributed to 'they' groups are favorable. If the activities and views clash while the interacting groups pursue their peculiar interests and goals, the features attributed are negative (p. 357).

[p. 133] The intergroup relations of such small groups as gangs show this in a striking way. In the process of group formation there is a tendency to appropriate certain areas, objects, places, etc., as their "own. " Encroachment or invasion of these private domains by an out-group results in clashes which tend to be accompanied by attribution of unfavorable characteristics to the "intruders." If the relationship of conflict endures for any length of time, derogatory terms for the out-group become standardized which mirror the nature of the underlying attitudes of prejudice or social distance.

When stereotypes become standardized with an in-group, they may and do persist beyond the functional relationships between groups of which they are a product. Existing stereotypes may then be manipulated by powerful members of the in-group or, at times, by other interest groups and extended to groups with whom there has been little or no contact. However, the present experiment on intergroup relationships is concerned with studying stereotypes from scratch, that is, tracing their formation from the time of first contact between groups in conditions of rivalry through a period of intergroup conflict. Therefore, those studies which reveal differential response to contact with out-groups under varying conditions of interaction were especially pertinent in formulating our hypotheses.

A survey of historical studies in various countries reveals that social distance scales do reflect the nature of intergroup relationships in which they evolve and that over a period of time they are responsive to altered conditions of intergroup interaction (Note 1). For example, MacCrone's intensive historical study tracing intergroup relations and social distance attitudes in South Africa during a period of over 200 years (1937) shows "radical alteration" of attitudes of original European settlers, who originally placed natives in the "heathen" category, which offered at least the possibility of salvation. As a consequence of active efforts to utilize native labor, reactions of native groups to these efforts, importation of more docile groups as slaves, and many actual intergroup conflicts, as well as developments external to the area itself (e. g., expansion of imperialism, industrial developments, etc. ), contemporary emphasis on "the white man and his civilization" in contrast to "inferior" native groups began to emerge in the early 19th century.

Klineberg's summary (1950) of changing stereotypes of the [p. 134] Chinese by Americans on the West Coast at different periods in American history is particularly illuminating. When there was a great demand for Chinese labor, and thus, conditions of interdependence between Chinese and white settlers, favorable verbal pictures of Chinese were common in journals and newspapers. However, around the 1860's, when other groups began to compete strongly with Chinese for their positions, descriptions of the Chinese began to undergo a radical shift in the negative direction. Whereas they had been described earlier as "thrifty," "sober," "inoffensive, " and "most worthy" adopted citizens, the stereotype held of the Chinese became negative. The Chinese who had earlier been seen as possessing "adaptability beyond praise" now were pictured as "a distinct people" who were "unassimilable," "debased," "servile," etc.

Experimental evidence supporting the view that stereotypes arise as products of functional relationships between in-groups was provided in Sherif's 1949 experiment on intergroup relations (Sherif, 1951, Sherif and Sherif, 1953). The design of the last two phases of that study was the same as that of Stages 1 and 2 of the present experiment (see Chapter 2). The hypothesis tested in the final phase of the 1949 experiment was essentially the same as our main hypothesis for Stage 2. In the course of interaction in intergroup situations of competition and frustration highly derogatory labels were used in relation to the out-group. In time such terms as "pig," "bums," and "cheaters" were standardized for reference to members of the out-group.

Utilizing such leads from the 1949 intergroup experiment, Avigdor (1951) carried out her doctoral study at New York University on the specific problem of "The spontaneous development of stereotypes as a result of a specific type of group interaction." By subtle control of conditions, Avigdor was able to create first a relationship of cooperation between small groups and later to turn this relationship into one of unfriendliness between certain of the interacting groups.

Groups of l0-year-old girls ("friendship clubs") were paired by Avigdor (who became an adult leader of the groups for the purpose of the experiment) in both cooperative and competing activities. The cooperative situation was one in which a compelling common goal existed for each of the paired groups, that of earning enough money to purchase highly desired club jackets. Attainment of this goal required that two groups work together to put on [p. 135] a play. At the height of the cooperative activities each group rated the other cooperating group on 32 characteristics, half favorable and half unfavorable, on a five-point rating scale.

The conflict situation developed when two of the cooperating groups were more successful in preparation of their play than the other groups. Intergroup conflict reached such an intensity that when one group which was lagging behind was brought to a final rehearsal of the play being prepared by two more successful groups, the visitors became objectionable and were forcibly expelled. At this point ratings on the 32 characteristics were obtained from members of the rival group in relation to the two groups which had expelled its members, and from members of those two cooperating groups in relation to the group thrown out.

Among Avigdor's conclusions were that the ratings made after the cooperative interaction were generalized in the favorable direction, "that is, development of favorable stereotypes," while the ratings obtained after interaction involving conflict generalized in the unfavorable direction, "that is, development of an unfavorable stereotype" (p. 65).

In one aspect of an experimental study of negative and positive relations between small groups existing in everyday life, Harvey (1954) obtained results in line with those reported above. He found that when the interacting groups were positively related preponderantly favorable adjectives were attributed to the out-group and its members. But when the relationship between groups was negative, derogatory adjectives were used in relation to members of the out-group.

Before presenting results of stereotype ratings obtained in the present experiment, it should be stressed that negative relations between groups with accompanying patterns of social distance and negative stereotypes do not imply that a similar pattern of relationships prevails among members of the in-group. There is evidence from the present intergroup study, as well as from the 1949 study and sociological fieldwork, to indicate that conflict with an out-group tends to result in increased in-group solidarity with consequent favorable verbal pictures of in-group members. (This significant point was elaborated in relation to Hypothesis 2, Stage 2.)

[p. 136] The results of stereotype ratings by Eagles and Rattlers of their own groups and of the respective out-group provide a critical check of the validity of conclusions based on observational data, namely, that unfavorable stereotypes in relation to the out-group were produced as a consequence of competitive and frustrating relations between experimentally formed in-groups.


At the end of Stage 2, the two experimental groups, Eagles and Rattlers, were asked to make ratings of their own and each others' group. It was explained to the subjects that they were being asked to do this to help the administration find out what they thought of their new acquaintances and how they were enjoying camp.

The stereotype scale contained critical characteristics as well as uncritical or favorable ones, and a five-point rating scale for each of the terms. The five points or categories were the same as those used by Avigdor, via., "All of the (Rattlers or Eagles) are. . .," "Most of the ___ are. . .," "Some of the ___ are. . .," "A few of the ___ are. . .," and "None of the ___ are. . .". The subject made his rating on each characteristic by writing that particular term in the one of five incompleted categories that, in his opinion, was the most appropriate description of the group being rated, in-group or out-group.

The characteristics on which in-group and out-group were rated were not postulated merely on a priori grounds. They were terms that the subjects themselves had used during Stage 2. Thus there was some assurance of the appropriateness of the characteristics chosen. Although more terms were presented on the scale, six were chosen as critical ones. It was thought that these six characteristics were sufficiently standardized in both groups to provide a clear-cut distinction between in-group and out-group ratings. The critical characteristics included three favorable terms (brave, tough, friendly) and three unfavorable ones (sneaky, smart alecs, stinkers).


The frequency of ratings on the six characteristics (brave, [p. 137] tough, friendly, sneaky, stinkers, and smart alecks) within each of the five categories ("All of the ___ are. . .," "Most of the ___ are. . .," "Some of the ___ are. . .," "A few of the ___ are. . .," "None of the ___ are. . .") was determined. The categories were then numbered from 1 to 5, 1 being the most unfavorable and 5 the most favorable category. Thus a response of "All of the ___ are (unfavorable term)" would go into category 1, while a response of "All of the ___ are (favorable term)" would be tabulated in category 5. The results are presented in terms of ratings on the six characteristics combined.

Table 3 presents the ratings of out-group members by each group at the end of Stage 2 (friction stage).

Table 3

Stereotype Ratings for the Out-Group on Six Characteristics

(Combined) by Members of Rattler and Eagle Groups

End Stage 2


Rattlers' Ratings of Eagles

Eagles' Ratings of Rattlers






























* Most unfavorable category.
** Most favorable category.


[p. 138] These results confirm observational data indicating that members of both groups tended to rate the out-group unfavorably following the stage of intergroup competition and friction. Fifty-three per cent of the ratings made by the Rattlers were negative and 24.9 per cent favorable. The Eagles' favorable picture of the out-group is even more accentuated. Their ratings of the Rattlers were 76.9 per cent unfavorable and only 15. 4 per cent favorable.

Table 4 presents a composite picture of the ratings six characteristics made by Rattlers and Eagles of their groups (in-groups) at the end of Stage 2 (friction).

Table 4

Stereotype Ratings for the In-Group on Six Characteristics

(Combined) by Members of Rattler and Eagle Groups

End Stage 2


Rattlers Ratings of Rattlers

Eagles' Ratings of Eagles






























* Most unfavorable category.
** Most favorable category.

From these results it can be seen that at the end of Stage 2, in which conditions gave rise to intergroup friction, members of both the Rattler and Eagle groups were rated favorably by other members of their respective in-group. While the tendency was to rate out-group members unfavorably at this stage (Table 3), there was an even more pronounced tendency to rate in-group members favorably (100 per cent favorable ratings of in-group by Rattlers and 94.3 per cent favorable ratings of in-group by Eagles). [p. 139]

The accompanying figure presents the main findings in graphic form. Ratings of the out-group by both Eagles and Rattlers are significant in the unfavorable direction; ratings of in-group by both groups are significant in the favorable direction. And, as noted above, ratings of in-group and out-group differ significantly in direction.

Thus, competition and rivalry between the groups led to attribution of unfavorable characteristics to the out-group, while this same pattern of intergroup relations was accompanied by a marked tendency to see members of one's own group in a highly favorable light. This finding is relevant to the prediction of intense in-group solidarity under conditions of intergroup competition, rivalry and hostility (Hypothesis 2, Stage 2). It constitutes further evidence that intergroup relations do not necessarily [p. 140] follow the same pattern as in-group relations, particularly when the relationship between the interacting groups is one of rivalry and antagonism.

On the basis of both observational and sociometric data, and evidence in this unit, Hypothesis 1, Stage 2 is supported. During the course of competition and frustrating relations between the experimentally formed groups, unfavorable labels were assigned to the out-group and its members and were used and shared to varying extents by members of the in-group. Social distances crystallized in these standardized derogations were great enough that for a time members of each group expressed a strong desire not to associate in any way with the out-group or its members (see Chapter 5).

2. Verification of Stabilized In-Group and Out-Group Attitudes Through Judgmental Indices: Performance Estimates.

At the end of Stage 2 (intergroup friction), a second experimental unit was introduced to obtain further evidence of the products of prolonged negative interaction between the two experimentally formed in-groups. In this experiment, direct judgments of a numerical nature were obtained in such a way that they might reflect the character and intensity of intergroup relations after a period of competition, rivalry and friction between the two groups. Members of both experimental groups made judgments of items presumably accumulated by members of their own in-group and by members of the rival out-group while performing a task for which the winning group would be rewarded.

It was proposed that negative intergroup relations would produce in time derogatory conceptions (stereotypes) of the out-group accompanied by intensified in-group solidarity and a highly positive picture of the in-group. Further it was proposed that the deprecatory picture of the out-group and flattering picture of the in-group would be internalized by individual members as negative attitudes toward the out-group and positive attitudes toward the in-group, and that these would be revealed

(a) in their ratings of in-group and out-group on significant characteristics embodied in the stereotypes (see [p. 141] preceding section), and

(b) in judgments of the performance by members of the in-group and out-group on a relevant task.

The task chosen for the latter purpose was a bean toss contest, judgments of the number of beans presumably collected by each individual being made after the contest.

Specifically it was predicted that as a consequence of intergroup competition, rivalry, and hostility:

In-group members will tend to overestimate the number of items purportedly obtained by in-group members and underestimate the number of items attributed to out-group members (Hypothesis 1 a, Stage 2).

The data from this experimental unit are to be evaluated in conjunction with observational findings showing increased glorification of the in-group and its members and deprecation of the out-group as a consequence of intergroup rivalry and conflict (Note 2). The judgmental indices obtained should reflect this state of affairs reported by observers and are intended to supplement their findings, not to replace them.

The plan to obtain experimental measures of attitudes formed toward the groups by the end of Stage 2 represents an extension to the level of intergroup relations of a basic psychological principle underlying the conception of this entire study, namely, that all psychological activity is determined by the frame of reference within which it occurs (Chapter 2). The frame of reference consists of the totality of functionally related factors, external and internal, that operate interdependently to determine the psychological reaction at any given time. The relative weights of the external and internal factors in determining psychological activity are not necessarily the same in different instances. The relative weight of these factors varies with the degree of stimulus structure and the nature and intensity of internal states at the time. When stimulus conditions are compelling in structure, the effects of internal (e. g., motivational) factors in patterning perception and behavior are not readily apparent; but under conditions of minimum stimulus structure, of ambiguity or flux, internal factors operating at the time may be clearly reflected in the subsequent behavior. Therefore, in attempts at studying [p. 142] motivational factors through their influence on the patterning of perceptual and judgmental responses, stimulus conditions should be both appropriate and sufficiently unstructured that the nature of the motivational factors can be revealed through the resulting behavior.

Since the presentation of the foregoing formulation (Sherif, 1935), there have been a number of studies investigating various motivational factors through their influence on such processes as perceiving and judging. The distinctive feature of the present experiment is that judgments are used as indices of the relationship between experimentally produced in-groups.

In Sherif's 1949 intergroup relations experiment, in-groups and negative relationships between them were produced through controlling conditions of interaction in essentially the same way as in this 1954 experiment. It was observed that as a consequence of the negative relationship between groups, in-group members extolled and maximized the performance of in-group members while deprecating and belittling the performance of members of the unfriendly out-group. Attention was called to the feasibility of obtaining precise experimental indices of intergroup relations through their differential effects on the perception and judgments of individual members. Judgments and perceptions of individual members will reflect the influence of membership and participation in the on-going activities of the group.

This being the case, the effects of the group situation, and the changes brought about in attitudes toward the in-group and the out-group and their respective members, can be studied in terms of precise laboratory experiments, such as the currently accumulating judgment and perception studies. This will constitute a significant advance in method over observation of actual behavioral events alone. The actual behavioral events are more difficult to observe with precision and present baffling problems in their ordering along definite dimensions. If the psychological significance can be epitomized and measured in terms of representative judgmental and perceptual situations, we shall be achieving a methodological gain close to the laboratory level. (Sherif, 1951, p. 422).

This proposed method of studying group relations through [p. 143] judgmental indices was applied to a study of status relations within groups that already existed in everyday life (Harvey, 1953). It was found that judgments of future performance of group members provided an index of their relative status positions within the group. Owing to the differential expectations that had become standardized for each status position, performance of high ranking members tended to be overestimated, while that of lower ranking members was estimated significantly less, even to the point of underestimation of actual performance.

The same methodological approach was applied in our 1953 experiment on intergroup relations to the study of status relations in experimentally produced groups. Going a step further, that study tapped the differential expectations for members occupying positions in a status structure which was itself experimentally produced (Note 3). Members of the experimental groups judged the performance of in-group members on the task of throwing handballs at a target board which was designed so that there was little indication of actual performance. The status positions that had evolved during the period of group interaction were reliably revealed in the subjects' judgments of other members' performance made immediately following each throw. The performance of higher ranking members was judged significantly higher than that of lower ranking members.

More recently this technique was extended to the study of negative and positive relationships between small groups existing in everyday life. Harvey (1954) showed that the relationship prevailing between interacting groups is revealed in the judgments of group members under appropriate stimulus conditions. Performance actually achieved by each subject, i. e., the names of cities written under conditions of distraction, were projected on a screen too briefly for actual count, and the number was judged by both in-group and out-group members. When intergroup relations were negative, the tendency was to judge the performance of in-group members at a significantly higher level than that of out-group members.

The present experimental unit is concerned with obtaining judgmental indices of the relationship between two groups which were experimentally produced and which came into conflict as a consequence of experimentally introduced conditions of competition and frustration.

[p. 144] Procedure:

Members of the two groups (Eagles and Rattlers) participated in a bean toss contest under strongly competitive conditions and then made judgments of the number of beans collected by each in-group and out-group member as the purported performance of the particular individual was projected on a screen by an opaque projector.

The contest was held after the tournament and various raids of Stage 2 (see Chapter 5). Social distance between the groups was sufficiently great that neither wanted to be in a situation with the other. They entered into this contest when told that the staff members of their respective groups had made a wager on the outcome and as they began to anticipate the five dollar reward offered to the winning group. Before the contest, the leader of the Eagle group predicted darkly, "It will turn into a gang fight." In spite of this initial resistance, both groups participated in the contest with considerable zeal once it was underway.

The beans were spread in equal density in two marked-off areas of similar size. The Rattlers picked up beans from one area, while the Eagles gathered beans from the adjoining area (see Figure). Separate areas were used to prevent pushing and shoving of out-group members. The time allowed for picking up beans was one minute. Pre-tests with comparable subjects before the experiment showed that one minute permitted the collection of 25-40 beans. To prevent the subjects from exceeding this range and to limit possibilities of their counting the beans collected, each was given a small brown paper sack, the open end of which was gathered around a hollow rubber tube with a half-inch opening. Subjects were instructed to pick up only one bean at a time and put it in the sack through the half-inch opening. Thus speed of performance was at a premium. They were instructed not to count the beans, that this would be done later, and were told that everyone would judge the performance of everyone else. The judgmental aspect of the task which was actually the crucial one for this unit, was presented as a regular part of the bean toss contest.

After the beans had been collected, the subjects went to an experimental room (large recreation hall) where the beans collected of each member of both groups were purportedly projected by an opaque projector and judged by both in-group and out-group

[p. 145] members. Actually the same number of beans (thirty-five) was projected every time as the performance by every individual in the two groups. The number chosen had been found in pre-tests prior to the experiment to be the optimum number for the brief exposure time used (5 seconds). It was necessary that the number of items projected should exceed the perceptual span (should be too great to count in 5 seconds), but at the same time be few enough that the subjects would feel that if they had tried just a little harder they could have finished the count. If the actual number could have been accurately established by the subjects, obviously there would have been no indication of motivational factors in their judgments.

The experimenter made it appear that he was putting in new beans to be projected each time, but the same ones were retained for every projection. The experimenter moved the location of some of the beans each time only slightly. Therefore, the form of the projection (circular) remained essentially the same, but the pattern was slightly different from projection to projection.

By a toss of a coin, it was determined that the performance of members of the Eagle group should be projected and judged first. Before the purported performance of each boy was projected, the experimenter called his name and the boy stood up so that both in-group and out-group members would know exactly whose performance was being judged. The subjects wrote their judgments on small pads of paper. Each boy wrote down the name called out. When the biggest of the Rattlers was called off, "Red Brown, "one of the Eagles (Wilson, a lieutenant) said, "I'm just going to put Red Bum on mine."


To test the hypothesis, it was necessary to ascertain the extent to which the performance of each in-group and out-group member was over- or underestimated and to compare these mean differences. Since the same number of items (35) was projected as the performance of each subject, this constant was subtracted from each judgment of performance. Means of these differences were then computed for judgments of performance by in-group members and by out-group members. The differences between the means were subjected to the t-test, using the formula for correlated means. Results of this analysis are given in Table 5.

[p. 146] Table 5

Comparison of Mean Discrepancies between Judgments of

Performance and Number of Items Projected (35):

In-group and Out-group Members

Group Judging

Group being Judged

Mean Differences

















From this table it can be inferred that the performance of in-group members was judged significantly higher than that of members of the out-group. The Rattlers' mean discrepancy for judgments of performance by in-group members was 3.404 and that by members of the out-group (Eagles) was -- .293. Thus while members of the Rattler group overestimated the performance of in-group members, they tended to underestimate the performance of negatively related out-group members. (When the beans supposedly collected by the leader of the Rattlers were projected, a member of his group whistled appreciatively.)

Members of the Eagle group greatly overestimated the performance of in-group members (mean discrepancy of 11.802) and overestimated the performance of out-group members significantly less, although not to the point of underestimation reached by the Rattlers in their judgments of the Eagles' performance. (Perhaps the Eagles' recent victory in the tournament over great odds had something to do with this difference.)

For both groups, the performance of in-group members was judged significantly higher than that of out-group members. The results cannot be explained in terms of differences in actual performance:

(a) Since speed was the crucial factor each bean had to [p. 147] be inserted through a small opening in the bag, individuals did not have an opportunity to observe each others' performance.

(b) The number of items projected as the performance of each individual in both groups was identical, in all cases being thirty-five.

These results are concordant with observational findings concerning the valuation of in-group members and deprecation of members of the out-group, and with sociometric findings which revealed a preponderance of in-group choices. Therefore, it can be concluded that the results reflect the solidarity within groups and negative attitudes toward out-groups at the end of Stage 2. Data from this unit also support the hypothesis that, as a consequence of intergroup competition and conflict, in-group members will tend to overestimate the performance of members of their own group and will tend to deprecate the performance by members of a rival out-group.

The findings indicate that conceptions of the in-group and out-group can be tapped experimentally through judgments obtained from individual members of the functionally related groups. Their larger significance bears on the individual - group relationship. They point again to the fallacy of dichotomies between the individual and the group. Here the relationship between groups, a phenomenon at the group level, has consequences not only for the formation of values or social norms within the group, but for the perception and judgments of individual members as well.


Summary of End Products of Intergroup Friction

As a consequence of repeated interaction between the two experimentally formed groups in competitive and reciprocally frustrating situations, and of the cumulative intergroup friction thus engendered, negative attitudes toward the out-group were formed by members of each in-group. These negative attitudes toward the out-group, crystallized in unfavorable stereotypes, were manifested by name-calling, derogation of the out-group, and the explicit desire to avoid association with the out-group (see observational data summarized in Chapter 5). In order to [p. 148] check these observations and to ascertain that they indicated more than merely momentary reactions in intergroup conflict situations, an entire day was devoted to in-group activities following the close of actual competition and conflict between the groups. On the day following this, attitudes toward in-group and out-group were tapped through (1) sociometric choices, (2) stereotype ratings of the in-group and out-group, and (3) judgments of performance by in-group and out-group members in a competitive task (bean toss contest). Analysis of these results confirms in a more precise way the observational findings. The methodological importance of this correspondence between observational and judgmental data is two-fold. First, it provides a check on the validity of conclusions reached. Second, it indicates the possibilities of utilizing judgments of in-groups and out-groups as indices of the state of affairs prevailing on the level of group relationships.

The methodology stems from the approach to the study of the complex processes of interaction in intra- and intergroup relations stated in Chapters 1 and 2. On the one hand, it is feasible to set up the flow of interaction processes in a life-like, natural way; on the other hand, validity and precision can be insured through obtaining data by observation (looking from without), by sociometric techniques (looking from within) and by the introduction of precise laboratory-type experiments at choice points. In our opinion, only through the use of such a combination of methods, applied to the flow of interaction process without chopping it into disjointed pieces, can we hope to attain generalizations with some bearing on the persistent group problems of actual life situations. If we follow such an approach in testing hypotheses, we are less likely to be troubled with the problems of validity which plague so many current studies in this area.

From a theoretical point of view, the results of these experimental units, carried out after actual intergroup conflict had ceased, indicate that events occuring between groups have consequences at both a group level (norms relating to the out-group) and at a psychological level (formation of negative attitudes toward the out-group), and that these consequences outlast the intergroup events themselves.

The crucial question which remains to be answered is whether or not changes in the character of relationships between groups, wrought through altering the conditions in which the groups interact [p. 149] from those conducive to friction to those of interdependence, will result in reduction of friction between groups and changes in negative attitudes and stereotypes standardized in relation to the out-group. This question was the point of departure for Stage 3 of this experiment, to which the next chapter is devoted.


1. Representative historical studies of social distance scales in formation are presented in M. Sherif and C. W. Sherif, Groups in Harmony and Tension, New York: Harper, 1953, Chapter 5.

2. It had been planned to repeat this experimental unit and the stereotype ratings at the end of Stage 3 to secure further evidence of the differential effects of intergroup cooperation toward superordinate goals. However, it became evident that a repetition of this unit and of the stereotype ratings too would yield most direct evidence with least cluttering of the on-going trend of Stage 3.

3. The plans for the 1953 study called for utilizing the same methods to tap attitudes toward in-group and out-group after a period of intergroup friction. However, this unit was not undertaken until the present 1954 study.


Avigdor, R. The Development of Stereotypes as a Result of Group Interaction, on file in the Library, New York University, 1951.

Harvey, O. J. An experimental approach to the study of status relations in informal groups. American Sociological Review 1953, 18, 357-367.

Harvey, O. J. An Experimental Investigation of Negative and Positive Relationships Between Small Informal Groups Through Judgmental Indices. Doctoral dissertation, University of Oklahoma, 1954.

Klineberg, O. Tensions Affecting International Understanding, New York: Social Science Research Council, Bull., 62, 1950, 114-115.

MacCrone, I. D. Race Attitudes in South Africa  London: Oxford University Press, 1937.

Sherif, M. A Study of Some Social Factors in Perception, Archives of Psychology, No. 187, 1935.

Sherif, M. An Outline of Social Psychology, New York: Harper, 1948.

Sherif, M. Chapter 17 in Social Psychology at the Crossroads, Rohrer, J. H. and Sherif, M. (edits.), New York: Harper, 1951.

Sherif, M. and Sherif, C. W. Groups in Harmony and Tension, New York: Harper, 1953, especially Chapter 8.

Sherif, M., White, B. J. and Harvey, O. J. Status relations in experimentally produced groups through judgmental indices. American Journal of Sociology, 1955.