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. 56] CHAPTER 3

Role of Staff, Subject Selection, Experimental Site

As specified in the statement of the approach, hypotheses and general design (Chapter 2), the distinctive feature of this study is that subjects interact with one another in activities that appear life-like to them in a natural setting - without being aware that they are being observed while interaction is going on. Therefore, it becomes essential to make explicit (A) the special role of staff members in experimentally introduced problem situations, (B) the criteria observed in the selection of subjects in order to insure adequate testing of hypotheses, and (C) the considerations which determined the choice of an experimental site.

Role of Staff Members in the Introduction of Experimental Conditions

The points covered in this section were instructions given the staff members prior to the experiment. They are presented here in essentially the same form.

In every step of the work it will greatly help coordination of efforts if all participating staff members in the camp realize at every moment that the camp is not one in the usual sense, but is set up as a research project to test definite hypotheses pertaining to group relations, with emphasis upon the intergroup phase. The main features are stated in the outline giving approaches, stages, plan of study and methodological considerations. Conditions and activities are introduced with these objectives in mind (see Chapter 2).

Utmost care will be exercised in ruling out all influences in word, deed, or use of various procedures which are not specified in the methodology and specific characteristics stated for conditions introduced for each stage. No activity is to be initiated, sponsored or encouraged which is not in line with the main criteria specified in the study plan as appropriate for the particular stage at the time. The behavioral effects are to be the outcome of the deliberately introduced conditions and not of verbal means or other usual camp practices.

[p. 57] Therefore, do not use verbal means to influence subjects, do not take initiative to introduce activities on your own accord, and do not try to counsel campers individually. Of course, this does not mean a "hands off" or non-direction policy in any matter which even slightly concerns the whereabouts, safety, health and well being of the campers concerned.

No staff member is to be a leader to the boys during any stage of the study in any of the various activities which are introduced after careful consideration in line with the criteria and hypotheses. In the first stage, every activity is introduced because it is considered to be conducive to interaction among the campers, from which a pattern of status (role) relations, including the leader position, is expected to emerge. You may have to give advice when asked and institute controls when necessary to maintain order, but please refrain from giving direction and initiating action in relation to problem situations. Initiative should come from the subjects under the specifically designed problem conditions of each stage. After they start along some line of action, give them help to carry it out, but do not put yourself in the foreground of on-going activities.

When a problem situation is introduced which demands planning, discussion and execution on the part of the subjects, utmost care should be taken not to show any partiality or preference and not to assign any single camper to take the lead. If the experimentally introduced situation involves common appeal value (motivation), the lead will naturally evolve in the interaction process among the participating campers.

Attention should be especially called to the fact that the participating campers will at times turn to you, as adults, for approval or sanction for carrying out a plan of activity in relation to the experimentally introduced problem situations. Care should be taken to be responsive to such queries or appeals. If the proposals do not run counter to health, safety and well being of the campers, and also if they do not run counter to the criteria specified for the given stage, the boys should be given opportunity to proceed in the direction of their proposed activity.

There will be times at which an ongoing course of action may not be in line with criteria set in the study outline. If such a situation occurs it may be suggested to you by the experimenter [p. 58] that a change be made in the ongoing procedures. In the flow of activities it may be impossible to explain at the particular moment why this suggestion is made. It is expected that the suggestion will be followed, and the reasons for suggesting the change will be explained later at a more appropriate time.

It is fully realized that the end results implied in the hypotheses may be secured in a more short-cut way by using other activities. Activities and procedures introduced in the various stages, especially in the first stage, might appear drawn-out and round-about. Since these successive stages are planned after long and laborious deliberation of existing theories and findings, they constitute an interrelated sequence in the plan of study, and as such, all successive steps are dependent on each other. Therefore, utmost care should be taken not to appeal to short-cuts, but to satisfy the sequence as outlined.

In line with the consideration stated in relation to the rise of leadership among the campers, and of staff not assuming leadership in experimentally introduced conditions, it becomes necessary for staff members not to exhibit special performance skill which may be conducive to focusing popularity and leadership on staff members. This is particularly important for Stage 1. However, it should be repeated that this does not mean that a helping hand should be withheld to the campers after a line of activity is proposed or initiated by them. In line with this consideration, do not wear any clothing, especially shirts, which have insignia or other identifying symbols, e. g., college or camp name. We do not wish subjects to adopt names or associations through adult leader-identification. Do not introduce to the campers nicknames, catchwords, slogans in a way which may cause them to be standardized by the subjects.

All the reports concerning verbal or behavioral observations should be written independently and not as a consequence of discussion with any other staff member. All ratings should also be done independently. Particular care should be taken to observe this procedure in order to secure reliability of results.

In the ongoing activities there is the possibility of an infinite number of events which can be observed and recorded. Therefore, please have the hypotheses for the given stage focal in your mind so that the observations will not be hodge-podge, [p. 59] but relevant to the hypotheses in question. As long as any behavioral items are relevant to the hypotheses, either validating or invalidating the hypotheses, utmost care should be taken to have all of them included. It may not be possible to record all relevant items of behavior, but indisputable recurrences of behavioral items should be recorded.

Subject Selection

Since the hypotheses to be tested require that the behavioral trends and products in in-group formation and the development of positive and negative relations between the groups be outcomes of experimentally introduced conditions and interaction processes within them, certain strict criteria for subject selection were necessary. The criteria that were adopted stemmed from the basic consideration that in-group formation (Stage 1) and the development of negative and positive relations between in-groups (Stages 2 and 3) should not hinge upon similarities or differences in sociocultural background or distinct differences in personal backgrounds and adjustment of the individuals composing the experimental groups. Therefore, homogeneity of subjects as to sociocultural and personal backgrounds was the guiding determinant underlying the establishment of criteria for subject selection.

Subjects were to be normal, well-adjusted boys of the same age, educational level, from similar sociocultural backgrounds and with no unusual features in their personal backgrounds insofar as extreme or prolonged frustrations, broken home life, etc. were concerned. Any potential subject who did not appear normal in terms of these general criteria and the more specific characteristics outlined below was excluded. This meant elimination of all "problem" boys, and of boys who might have suffered unusual degrees of frustration from inadequate sociocultural and personal backgrounds. Equally important was that subjects should not have prior acquaintance before the experiment started. Otherwise it might be said that existing friendship ties influenced in-group formation (Stage 1), and the resulting groups could not be attributed to the experimental conditions introduced. Special precautions, outlined below, were taken to [p. 60] insure that subjects were not acquainted prior to the experiment.

Selection of subjects who met the criteria represented one of the prerequisites for the success of the study. Without adherence to such basic criteria in selecting subjects, many of the necessary conditions presented in Chapter 2 could not have been satisfied. Sociological, psychological, and physical specifications were also set up for subjects in order to insure healthy and well adjusted boys with athletic and other skills sufficient for full participation in the camp activities, which were to be introduced in line with experimental considerations.

More specifically, subjects were to be of established Protestant families (not new in the area), of middle socioeconomic class, living with both parents. (Children from broken homes and foster homes were not accepted. )

Psychological manifestations which precluded the selection of a given boy were any signs of severe homesickness, social isolation, enuresis, failure of one or more grades in school (i. e., subjects had to be of normal educational standing in relation to chronological age), abrupt changes in school performance, temper tantrums, running away from home or truancy from school.

All subjects were to be of normal physical development, and possess no physical deformities or impairments which would limit their participation in the athletic activities that were to be introduced for experimental purposes.

In interviewing parents and teachers and examining school records, information was gathered on certain skills and abilities of each boy that might enter as important factors affecting status positions that would evolve in the interaction at camp. The athletic interests and proficiency of each potential subject were ascertained, as well as musical ability and skit skills, previous camp experience, popularity and number of friends of the boy, membership in youth organizations, etc. In addition, data on attitudes of parents toward the son and his friends, condition of the neighborhood in which the boy lived, how long the family had lived in the area, size and condition of the home and its furnishings, make and model of car were secured.

Methods of selection: As mentioned above, one of the most [p. 61] important criteria of subject selection was that the boys not be previously acquainted with one another. Thus the friendship patterns and intra- and intergroup relationships formed in the experimental setting could not be attributed to existing acquaintances and friendship preferences brought to the experimental situation. This consideration dictated even the city from which subjects were selected. It required a city of sufficient size to have enough schools for children of the appropriate age and grade levels that only one boy could be selected from each school, thus reducing the likelihood of prior acquaintance. Oklahoma City has this many schools, and it was from this city that all the subjects were selected. (Using this method it was necessary to eliminate six boys from the final subject list because they were acquainted with others previously chosen.)

In order to adhere to the criteria as effectively as possible, a rather painstaking procedure was followed in picking out potential subjects and in the final selection of experimental subjects. The city was divided into 3 areas, each containing roughly an equal number of appropriate schools. Each area was assigned to the one of 3 interviewers who knew best that particular section of the city. Schools from sections of the city which had very high and very low income families in large numbers were eliminated from consideration.

In order to get best access to school records and to be permitted freedom in observing potential subjects at first hand in the school situation, the principals of appropriate schools were contacted. After presentation of credentials from higher school authorities, a brief explanation of the purpose of the visit was given. It was explained to the principal that an experimental camp under the auspices of the University of Oklahoma was being conducted. The announced purpose of the camp was the study of interaction in group activities within teams and between teams. The statement of purpose was informally worded but uniform. It was pointed out that one of the main things that would be studied was how team members assumed and carried out initiative and responsibility under adult supervision, what would be the attitude of the boys as they participated in activities toward common goals they wanted to attain and also the attitudes that would occur when they competed with another bunch of boys. It was explained that another item for study was how the boys take it when they win or lose in various activities, when things are [p. 62] not going their way, when they feel others are being good or bad sports or unfair, when situations are felt as more or less frustrating, as well as how the boys pull together and cooperate toward common goals.

In order that school principals and teachers would have little opportunity to recommend favorite boys for the camp, the interviewer explained to the principal that he would like to go out on the school yard where the fifth graders were playing so that from first-hand observation he could pick out some candidates who best seemed to meet the criteria. This served the dual purpose of allowing first-hand observation of the boys in usual circumstances of interaction by observers who knew the specific criteria for selection of a boy, and preventing the principals and teachers from trying to have their favorite boys chosen.

The observer then went to the playground, and when he saw a boy that seemed to meet the criteria he asked the name of that boy from the playground supervisor. At all times the interviewer tried to be as inconspicuous as possible in order not to arouse too much curiosity on the part of the boys. After getting the names of from 5 to 10 candidates who seemed to satisfy the criteria best, the interviewer then found out all he could about each of these boys from the playground supervisor. Then he went to the homeroom teacher (if he or she was not the playground supervisor) and obtained school records and further information on each boy. From school records, information was secured concerning I.Q., grades, adjustment in school, social attitudes, and from the teacher, the boy's relationship to teachers and to other children, status, popularity, and membership in school cliques. The 5 or 6 potential subjects who remained after this first screening procedure were ranked by the interviewer in terms of the extent to which they satisfied the criteria.

The next step in selecting subjects was to contact the parents of those boys who best met the criteria up to this point. All the parents were met by the same individual, the one who was to appear at the camp as the camp director in the eyes of the boys.
This policy was followed so that when subjects were divided into two groups, neither of the participant observers (who had served as interviewers of teachers) would have a particular personal preference for any given boy, that boys in neither group would go to camp already knowing their participant observer, and that [p. 63] boys in neither group would know the participant observer of the other group when the two groups first came into functional contact (Stage 2). (The participant observers appeared to the boys at camp as senior counselors.)

Parents were contacted in the order that their sons were ranked by the interviewers in terms of the criteria. In interviewing parents, the same explanation of the purpose and aim of the study was given as had earlier been given school principals and teachers. It was stressed to the parents that no visiting of the boys would be permitted. The explanation given was that such visits would contribute to problems of homesickness which would be detrimental both to the enjoyment of the boys at camp and the success of the study. They were also told that there was a nominal fee (twenty-five dollars for the entire period) and that a doctor's examination and permission for the boy to participate in all camp activities would be required, in addition to their own permission. (The fee, which was a nominal device for making mutual commitments, was lowered for a few boys who met the criteria but to whose family this seemed a sizable expenditure for recreation.)

Altogether roughly 200 names of potential subjects were obtained in the manner described above. Of this number, school records and interviews with homeroom teachers were completed for almost half. In the selection of the final experimental subjects the parents of approximately 50 boys were contacted and/or interviewed. The original goal was 24 subjects who met the criteria in every respect, but strict adherence to the criteria resulted in the procurement of 22 experimental subjects. Altogether more than 300 hours were spent directly in selection of subjects, in addition to the numerous hours spent in establishing criteria and setting up the procedures.

Final experimental subjects: The 22 subjects who were finally selected were relatively homogeneous in terms of the major criteria outlined above. All were from established Protestant families. All were well adjusted both in school and at home, according to observations, school and home interviews. According to school records, all the subjects were doing average or above school work (none was failing or had a history of failures). All were fifth graders about eleven years old who were promoted to the sixth grade for the next school year. This age level was [p. 64] selected so that none of the boys would have reached puberty, which could have been an important additional factor in determining the status positions that would emerge in group interaction. All were taken from the same grade for similar reasons, namely, that no one should have a status advantage because of a more advanced grade in school.

The median income of the subjects' families was $4,900 a year. The income of eleven families was below $5,000, the lowest being $3,200, and only two were (slightly) above $7,000. However, on the basis of occupation, education, home, neighborhood, etc., the subjects' families can be characterized as "middle-class" on the whole.

The average (median and mean) age of subjects was 11 years and 1 month. Five boys would have their 11th birthdays shortly after camp, and only one would be 11 well after the school term started. One boy had reached 12; all others were eleven.

I.Q.'s were available for 18 of the 22 boys. The median I.Q. for these 18 boys was 112. Four boys had I.Q. scores between 90 and 105, and only one above 120. Thus, by and large, these boys were above average in intelligence test scores, 11 scores of the 18 available being between 110-120. The boy with the lowest I.Q. was doing satisfactory work in all school subjects and was rated by his teacher "at the low end of the upper one-third" of his class in school achievement.

The boys did, of course, differ within limits in size, manner and other personal characteristics, ability in various games, hair coloring, etc. However, these individual variations were within the range of normality for boys in the schools, grades, neighborhoods, types of families, etc., which had been chosen in terms of the criteria. Home and school interviews and school records did not offer a critical indicator of status that would be achieved in the new camp situation. The status structures developing during in-group formation and the nature of relationships between the groups as they came into functional contact during the experiment cannot be accounted for on the basis of differentiation of individuals or between clusters of individuals based on characteristics stemming from different sociocultural backgrounds, from atypical personal backgrounds, or previously existing relationships.

[p. 65] In dividing the 22 subjects into 2 groups of equal size prior to the experiment, great care was taken to match subjects signed to the 2 groups, so that they would be composed as similarly as possible. Matching was carried out in terms of certain personal characteristics of the boys. Considered in their order of relevance for interaction in the camp situation, they were: height; weight; sports ability (general); sports skill (special); popularity (in neighborhood and school groups); other skills relating to camp such as musical and skit skills, cooking ability, etc.; swimming; and previous camp experience. After the two groups had been matched as closely as possible in terms of these characteristics, a coin was flipped to determine which group of subjects went to what participant observer. This was a final precaution to rule out possible effects of any personal preferences of participant observers for particular boys.

Experimental Site

Certain characteristics were necessary in the experimental site. It had to provide isolation from the outside world during the experiment so that extraneous influences would not enter and the results would be mainly a function of conditions deliberately introduced. There had to be separate facilities for two groups to be handled in isolation from each other during in-group formation (Stage 1), so that group formation would be the consequence of conditions introduced and interaction within the in-group, without contact with an out-group. Also, the physical characteristics of the camp and surrounding area had to be of a nature allowing flexibility in choosing and planning in-group and intergroup problem situations by providing numerous circumstances conducive to the arousal of common goals of high appeal value and to a variety of activities.

The site finally chosen after inspection of a number of camps was a densely wooded area in the Sans Bois Mountains of southeastern Oklahoma about seven miles north from the small town of Wilburton which is on U. S. Route 270. This is a 200 acre Boy Scouts of America camp which is completely surrounded by Robbers Cave State Park (See Figures 1 and 2). It was available exclusively for purposes of the experiment for the three week [p. 66] period. The nearest large town - McAlester, Oklahoma - is about 40 miles distant.

Since terrain and facilities were utilized as part and parcel of stimulus conditions throughout the experiment, it will be helpful to specify them. For the readers' convenience, places of functional importance in in-group and intergroup activities are indicated on the accompanying map at the end of this section which gives approximate distances between points mentioned (Figure 2).

Effective isolation of the camp was made possible by a surrounding fence with "Keep Out" and "Restricted" signs posted and by the heavy foliage which screened the camp area from a park road running some 100 yards outside the fence. Functional isolation of the groups from each other during in-group formation was made possible by the terrain of the area, and by careful timing of their coming and going. The cabin used by each group was beyond sight and hearing distances of that of the other group, and duplicate facilities were available for both groups (bath houses, swimming, boating and campfire facilities, etc. ). Both groups used the mess hall which was about equidistant from the two cabins. However, it was not visible from the cabin at the south end of the camp because of a hill, and its entrance could not be seen from the north cabin because of intervening buildings and trees.

Because of the characteristics of the experimental site itself, the surrounding park, and the mountainous areas within a sixty mile radius, it was possible to plan activities of high appeal to the subjects for both in-group and intergroup stages. Within easy walking distance from each cabin, and in opposite directions, were swimming, boating and camping areas which were available for the exclusive use of each group. Campfires could be held near the cabins, at the "hideout" areas, or in a natural stone corral which was near Robbers Cave on the hill above camp. A very isolated reservoir in the hills above the camp supplied its water and offered facilities within hiking distance for overnight campouts. An athletic field was located across the park road, outside camp property, and nearest to the north cabin. The field was accessible by two different routes for the two groups. Thus, when and where contact between groups would take place during competition situations could be controlled.

[p. 67] Lake Carlton, a part of the state park area, was located about three miles from the experimental site and provided excellent swimming and picnic facilities. A number of camping areas were located on lakes and rivers within 20 to 60 miles of the camp. These could be used to advantage in increasing interdependence of the groups (Stage 3) through cutting off their usual sources of food, housing, etc. Of these various possibilities, Cedar Lake, about 14 miles from Heavener, Oklahoma and 23 miles from the Arkansas state line, was utilized in Stage 3.


The campsite (cabins, mess hall, etc.) is situated on the flat stretch of land below this ledge. [p. 68]