The Robbers Cave experiment: what Sherif's summer camp study actually showed

The Robbers Cave experiment was a field experiment on conflict between groups. The social psychologist Muzafer Sherif and his team ran it in the summer of 1954, at a Boy Scouts camp in Robbers Cave State Park, Oklahoma. Twenty-two boys of about 11, who did not know each other, formed two groups: the Rattlers and the Eagles. A tournament with prizes made the groups enemies. A series of goals that neither group could reach alone then reduced the hostility, step by step. Sherif concluded that groups become hostile or friendly mainly because of how their goals relate: against each other, or dependent on each other.

This article works from Sherif's own report, which first appeared in 1954 and as a book in 1961. The full text is online, and the quotations below keep its original wording, with page numbers. The article also uses archival research by the psychologist Gina Perry and by historians of psychology. Their work shows how much the adults in the camp steered events.

• Robbers Cave was Sherif's third summer camp study. The 1949 study reached the conflict stage, and Sherif stopped the 1953 study early. Only the 1954 study went through all three stages.
• 22 boys went to the camp, and 20 stayed to the end. Two Eagles went home with homesickness at the end of the first stage.
• Shared meals and a film did not reduce the hostility. A series of superordinate goals, which neither group could reach alone, did, slowly. At the end, most friendship choices were still for boys in the same group.
• The report itself describes several steered moments. The staff juggled points in favour of the Eagles, manipulated the treasure hunt and staged a water crisis.
• The boys were never debriefed. Some of them learned from Gina Perry, about 60 years later, that they had been research subjects.

What was the Robbers Cave experiment?

The APA Dictionary of Psychology calls the Robbers Cave experiment "a field study of the causes and consequences of conflict between groups". Its central claim matches what the same dictionary calls realistic group conflict theory, also known as realistic conflict theory. That theory assumes that "intergroup tensions will occur whenever social groups must compete for scarce resources", and that the competition fuels prejudice. If one group can win only when the other group loses, the two groups have incompatible goals.

Sherif was born in Turkey. He earned his PhD in social psychology at Columbia University in 1935, as the historian Jennifer Bazar describes in an article on the camp studies. The 1954 study ran at the University of Oklahoma, with a grant from the Rockefeller Foundation. It was not his first attempt. The table shows the three camps, from Bazar's article and from Sherif's 1954 preface.

According to the report, the 1949 team reduced the tension with a common enemy. The report says this method "was effectively used in our 1949 study because of expediency at the time" (Chapter 7, p. 152). In 1954, Sherif did not use it. A common enemy, he wrote, "implies intergroup conflict on a larger scale" (Chapter 8, p. 202).

Who took part, and how were the boys chosen?

Sherif wanted to make sure that background, personal problems or earlier friendships could not explain the conflict later. His team chose the boys by strict rules (Chapter 3, pp. 59–64):

• The families were Protestant, middle class and long established in the area. Each boy lived with both parents.
• No boy had failed a grade, and no boy had a record of behaviour problems at school or at home.
• All were fifth graders about to start sixth grade. The average age was 11 years and 1 month.
• All came from Oklahoma City, and the team took only one boy from each school, so that the boys did not know each other. Six boys were removed from the list because they knew boys already chosen.

According to Bazar, all the boys were also white. The selection took more than 300 hours. Observers watched fifth graders on school playgrounds and collected about 200 names. The team then contacted the parents of about 50 boys. The goal was 24 boys, but the strict rules gave 22.

The team told the parents that the camp studied "interaction in group activities within teams and between teams". The explanation also mentioned the boys' attitudes "when they competed with another bunch of boys" (pp. 61–62). Parents could not visit, and the camp cost $25.

Gina Perry adds one more detail. In 1954, Sherif's colleague O. J. Harvey "deliberately chose boys with sporting ability". The team expected athletic boys to be more competitive and to identify more with their team (Perry 2014).

The team divided the 22 boys into two groups of 11. It matched the groups on height, weight, sports ability, popularity and camp skills. Then a coin flip decided which observer got which group (p. 65). At the end of the first stage, two Eagles went home because of homesickness. After that, the Eagles had only nine members (Chapter 5, p. 100).

How was the experiment conducted?

The experiment lasted three weeks. The site was a 200-acre Boy Scouts of America camp, "completely surrounded by Robbers Cave State Park", with a fence and "Keep Out" signs (pp. 65–66). The observers appeared to the boys as senior counsellors, and they wrote no word in front of the boys (Chapter 4, p. 71). A hidden tape recorder captured some conversations, and the staff took 1,200 photographs (p. 73).

Stage 1: forming the groups

The first group arrived on 19 June 1954, and the second on 20 June. The two groups lived in separate cabins and at first did not know about each other (p. 75). They hiked, camped and cooked, and each group developed leaders and fixed rules. The groups chose their names only near the end of this stage, which lasted eight days for the Rattlers and seven for the Eagles.

The rivalry started before the first contest. After work on their baseball field, the Rattlers declared: "Now this is our diamond." Then the staff told them that another group wanted to challenge them. The answer was: "They can't. We'll challenge them first... They've got a nerve..." (p. 79).

Stage 2: competition and hostility

The staff announced a tournament. The prizes were "a trophy, medals, and 11 four-bladed knives", for the winners only (p. 98). The tournament had two kinds of events (Chapter 5, p. 99):

• events whose result the boys could check themselves: baseball, tug-of-war, touch football and tent pitching
• events that the staff judged: three cabin inspections, skits and songs, and a treasure hunt

After the Eagles lost the first tug-of-war, they took down the Rattlers' flag and burned it. Then they hung the burnt remnant back up (p. 104). Raids on the cabins and new insults followed. In a questionnaire at the end of this stage, the boys rated the other group on traits such as brave, tough, sneaky and smart alecks. Most of these ratings were unfavourable (Chapter 6, pp. 136–138).

At the end of this stage, the staff ran a bean toss contest. Each group collected beans. Then the staff said that a projector showed the beans of each boy, and the boys estimated the number on each image. In fact, every image showed the same 35 beans. Both groups still judged the performance of their own members significantly higher than that of the other group (Chapter 6, pp. 144–146).

Stage 3: contact first, then superordinate goals

In the third stage, Sherif first tested whether contact alone helps. The groups ate together, watched a film together and took part in the bean toss together, and the staff moved away. The groups used these meetings to insult each other. At shared meals, they threw rolls, napkins and mashed potatoes (pp. 153–158).

Then the team introduced several superordinate goals. Sherif defined them as "goals of high appeal value for both groups, which cannot be ignored by the groups in question, but whose attainment is beyond the resources and efforts of any one group alone" (p. 202):

1. 1.Water: the whole camp suddenly had no drinking water. Both groups searched the pipeline and met at the water tank. There, the Rattler leader found a sack stuffed into the faucet. Boys from both groups took turns on it for more than 45 minutes, and then the staff finished the job with wrenches (pp. 161–163).
2. 2.Film: the boys wanted to see "Treasure Island". The film cost $15, and the camp could not pay all of it. The groups agreed that each group would pay $3.50 (pp. 165–166).
3. 3.Truck: on a camp-out at Cedar Lake, the truck with the food did not start. The boys pulled it together with "our" tug-of-war rope (pp. 170–171).

The effect did not come at once. After the first shared tasks, the old lines often came back. On the evening after the water task, the groups threw leftovers, bottle caps and paper at supper, and "the throwing continued and had to be stopped" (p. 164). Only the series of tasks changed the situation. On the last day, most boys asked to go home on one bus. On the way, the Rattlers used their $5 bean toss prize to buy malted milks for all 20 boys (p. 182).

What were the results?

The questionnaires at the end of the second and third stages measured the change. The table compares the two points in time, from Tables 1 and 2 and from the text of the report.

At the end of the competition, about 93% of all friendship choices were for boys in the boy's own group (pp. 118–119). After the superordinate goals, the choices for the other group rose clearly. But in the report's words, "friendship choices were still predominantly for in-group members" (p. 187).

The ratings changed more. At the end, 86.3% of the Rattlers' ratings of the Eagles were favourable, and 68% of the Eagles' ratings of the Rattlers. The report notes that the Eagles' ratings "did not change as much in the favorable direction" as the Rattlers' ratings (p. 192).

What Sherif concluded

In the final chapter, Sherif drew four conclusions (Chapter 8, pp. 207–210):

• The relation between the groups' goals decides. In his words, "the limiting condition determining friendly or hostile attitudes between groups is the nature of functional relations between them, as defined by analysis of their goals". When only one group can win, hostility and unfavourable stereotypes follow.
• Cooperation inside a group does not transfer to the other group. Solidarity inside both groups was strongest when the hostility between them was at its peak.
• Contact without dependence on each other does not reduce tension. The groups used such meetings "for overt acts of hostility".
• Superordinate goals work, but only as a series. The hostility "did not disappear in one stroke".

Later, Muzafer Sherif and his wife Carolyn summarised the result in their 1969 book Social Psychology. An observer who arrived during the conflict, they wrote, "could only have concluded on the basis of their behaviour that these boys (who were the 'cream of the crop' in their communities) were either disturbed, vicious or wicked youngsters". The social psychologists Stephen Reicher and Alexander Haslam quote this 1969 passage (Reicher and Haslam 2014). The point is that the hostility came from the groups' situation, not from the boys' character.

Three things summaries often get wrong

Short summaries of the study often contain three errors.

The first error is about friendships. Some popular summaries say that by the end of camp, about two-thirds of friendship choices crossed group lines. The report's Table 2 shows 36.4% for the Rattlers and 23.2% for the Eagles. In both groups, most friendship choices stayed inside the group (p. 187).

The second error is about the design. Some accounts say the boys first lived together and made friends, and then the staff split them to separate friends. That was the design of the earlier studies. In 1949, the staff watched "budding friendship groupings" among 24 boys and then divided them "in order to split the budding friendship groupings" (Chapter 1, p. 9). For 1954, the report says that this stage "is eliminated" (Chapter 2, pp. 38–39). The two groups arrived on different days and lived apart from the start (p. 75).

The third error is about the outcome. The APA Dictionary entry says of the third stage: "the result was a complete reconciliation". The ratings did change a lot. But 22.6% of the Eagles' ratings of the Rattlers were still unfavourable at the end, and most friendships were still inside each group. A large change is not a complete one.

How much did the staff steer the result?

Sherif's report describes in several places how the team steered events:

• Points: the staff-judged events "were included to enable the experimenters to juggle points". Both groups were meant to hope for a win until the end. The report adds: "This juggling had to be done in the present study in favor of the Eagles", because they lost on the first day of contests (p. 99).
• Treasure hunt: the tournament depended on the last event, the treasure hunt, which "could be manipulated by the experimenters". The Eagles won it in 8 minutes 38 seconds, against 10 minutes 15 seconds. The team feared that the Eagles, with only nine members left, would fall apart after a defeat. The report calls "the decision to tip the scales in favor of the Eagles" sound (p. 100). A later page names the method: "manipulation in plotting the routes" (p. 110).
• The burnt flag: the flag burning was the Eagles' own idea. But the staff arranged for the Rattlers to reach the field first the next morning, so that they found the flag (p. 105).
• Water crisis: the staff closed the valve on the tank, put two large boulders on it and stuffed sacking into the faucet. They told the boys that vandals had damaged the water system before, "in order that the blame for their thirst not be placed on the camp staff" (pp. 160–161).
• Truck: the truck at Cedar Lake "just would not start (as planned)" (p. 171). The water tank, the joint camp-out and the truck were already in the plan as possible superordinate goals before the camp started (Chapter 2, pp. 51–52).

An interview by Gina Perry confirms this. In 2010, O. J. Harvey told her about the Eagles: "So we deliberately let them win the competition" (Perry 2014). According to Perry, "artificially inflating the scores of one team over another to increase rivalry was common in both the 1953 and 1954 studies".

Every experiment creates its conditions, and Sherif reported these interventions openly. But they change how the results should be read. The adults planned part of what looks spontaneous.

Middle Grove 1953: the camp that failed

Sherif wrote little about the 1953 study. The 1954 preface says that it was planned in four stages and that "we succeeded in completing only two stages", which covered the formation of the groups (1954 preface). Chapter 1 gives the reason as "various difficulties and unfavorable conditions, including errors of judgment in the direction of the experiment" (Chapter 1, p. 18).

Later archival work gives a fuller picture. Herbert Kelman was on the 1953 team and wrote detailed notes that were never published. According to Reicher and Haslam, the notes describe how the team tried again and again to cause competition and conflict. "But the more they tried, the more the boys became suspicious that the experimenters were the real outgroup", and the more this united the boys. Sherif ended the study because more interventions would only confirm this suspicion (Reicher and Haslam 2014).

Gina Perry, who had earlier studied the archives of Milgram's obedience experiment, gives more details (Perry 2014):

• One boy asked the staff what the microphones in the mess hall were for.
• Both teams accused the staff of bias and favouritism.
• A rumour went around that the camp was an experiment in which the boys had to fight each other. Then a boy found a staff notebook with detailed observations, and the rumour was confirmed.
• Some staff members pulled down a tent so that one group would blame the other. A staff member who was dismayed by this told Perry: "It was supposed to be like watching a rat in the maze. But you don't push the rat."

Bazar reports a close reading of Sherif's original footnotes by Frances Cherry in 1995. In his first accounts, Sherif admitted that the boys had discovered the ruse. Later accounts named only "unfavorable conditions" and "errors in judgment". In public summaries, Sherif merged the three camps and talked mainly about 1954 (Bazar 2026).

Criticism: ethics, sample and generalisation

Ethics

The parents knew that the camp was research. But the team described it to them as a "wholesome cooperative living experience". It did not say that the team planned to cause conflict on purpose. The boys themselves were never told about the study (Bazar 2026). Perry writes that "it never occurred to me" that nobody had told the boys later. So she found herself explaining the experiment to the men about 60 years after it took place (Perry 2018).

According to Perry, the observers did not step in when a boy was bullied or excluded. They recorded it as a sign of a developing group hierarchy. The team did not discourage food fights and name-calling. Two cooks threatened to quit unless Sherif took a firm stand on "cussing" and the throwing of food.

None of the men Perry interviewed remembered the camp as a happy time. The staff's notes mention bedwetting, running away and homesickness (Perry 2014). Perry also says that her interviewees are not a representative sample: "I make no claim to a representative sample" (Perry 2018).

The sample and the setting

The boys were white, Protestant, middle class and about 11 years old. The study had no girls, no adults and no other cultures. The two groups had no shared history before the camp. Real conflicts between groups, such as between nations, often have a long history. The social psychologist Michael Billig made one more point in 1976: the camp had three groups, not two. The third group was the team of adults (as cited by Perry 2014).

Does conflict need competition?

A test with scout troops whose members already knew each other well found a different pattern. Andrew Tyerman and Christopher Spencer reported in 1983 that solidarity inside the groups did not increase during the camps. It decreased. Two other hypotheses, on ego involvement and on responsiveness to appeals for cooperation between groups, had substantial support (Tyerman and Spencer 1983).

Henri Tajfel and colleagues showed in 1971 that a division into groups is enough for people to favour their own group. Their participants favoured their own group when they divided real rewards. The only difference between the groups was "fairly irrelevant classification". Neither self-interest nor earlier hostility could explain the behaviour (Tajfel and colleagues 1971).

At Robbers Cave, too, the rivalry started before any prize was announced. The Rattlers wanted to challenge the other group as soon as they heard about it (p. 79). So competition can make hostility stronger, but it is not the only reason why people favour their own group.

Does contact between groups reduce prejudice?

From Robbers Cave, it is easy to conclude that contact between groups does not help. That goes too far. The APA Dictionary defines the contact hypothesis as "the proposition that interaction among people belonging to different groups will reduce intergroup prejudice". In 1954, the psychologist Gordon Allport named four conditions under which contact helps most. Elizabeth Levy Paluck and colleagues list them as follows (Paluck, Green and Green 2019):

1. 1.equal status between the groups in the situation
2. 2.common goals
3. 3.cooperation between the groups
4. 4.support from authorities, law or custom

The contact situations at Robbers Cave had no common goal and no cooperation. The superordinate goals had both. Thomas Pettigrew and Linda Tropp combined 515 studies, with 713 samples, in a meta-analysis in 2006. Contact between groups typically reduced prejudice. Allport's conditions made the effect stronger, but they were "not essential" (Pettigrew and Tropp 2006).

A later meta-analysis by Paluck and colleagues included only 27 studies with random assignment and a delayed measurement. It also found that contact typically reduces prejudice. But the effects on ethnic or racial prejudice were substantially weaker, and no study in the set addressed adults' racial or ethnic prejudice (Paluck, Green and Green 2019). So Robbers Cave shows something narrower: short, pleasant meetings without a shared task did not end an open conflict between two groups.

Keeping the terms apart

Realistic conflict theory, superordinate goal, contact hypothesis, in-group and out-group, and the minimal group paradigm belong to the same topic. They are easy to confuse in an exam. Psychology World is a dictionary app with more than 18,000 psychology terms. You can save terms like these to one list. Then you can drill the list as flashcards, as multiple-choice questions once it has four or more terms, or by typing the answer.

What each study mode does is set out on the home page. Our guide to memorising psychology terminology explains which study methods have the strongest evidence.