Sherif Robbers Cave

TsE Roeenns
Ssnnrp, M. (1956). Experiments in group conflict.
S cientific Americqn, l9S, 51- SB.
Psychology has offered two basic approaches to describe
the phenom_
enon of prejudice. one line of argument sees it as an individual pro_
blem; a sick person model of prejudice. The major influence here
psycho-analysis and the work of Freud, with the root of prejudice
seen to be in childhood experiences that create a damaged
adult p€rson]
ality Adorno et aI. (1950) presented the picture of the authoritarian per_
sonality who projects his or her unresolved childhood conflicts
minority groups. The authoritarian personality is narrow minded,
stickler for rules, inhibited about sex, unquestioningly submissive
authority intolerant of ambiguity, and politically conservative.
basic flaw in this approach to prejudice is the insistence
that prejudice
is a sign of a sick personality and that, by implication,
most people are
not prejudiced.
The other major line of argument sees prejudice as a resurt
of group
membership and group interaction. An example of this approach
is the
work of Sherif and his associates who proposed a conflictmodel
of preju_
dice, Their model suggests that when groups interact
with one another
they inevitably generate attitudes towards each other.
If the groups are
'positively independentl and working
towards common goals, then
good inter-group relations develop and the inter-group
attitudes are
positive (see also Aronson & Bridgeman, 1979, in
this chapter). on the
other hand, if the groups are 'negatiirely independent', in
for scarce resources perhaps, then group conflict develops
and ethnocentric attitudes appear. Sherif and his associates tested
this theory in a
number of field studies.
The basic question under investigation was, could they take a
qlolp of people without any hostile attitudes towards each other,
them into gro*ps, create conflict in the groups through intro1]It9"
ll.,ng competition, and thereby create ethnocentiic attitudes and
TUE Sruov
The field experiments were conducted in 1949, 1953 and 1954, and this
ariicle gives composite firidings, Sherif wanted to study informal
groups so that he could observe the natural and spontaneous development of group organisation and attitudes. To do this the researchers
created an isolated summer camp as an experimental setting.
The subjects were boys aged 11 to 12. According to sherif they were
picked by a'long and thorough procedure' involving interviews with
famiiy, teachers and school officials, and also by the use of school and
medical records, scores on personality tests, and observations of them
in class and at play. The boys ivere unkpown to each other and'ali were
healthy, socially well adjusted, somewhat above average in intelligence
and from stable, White, Protestant, middle-class homes' (p.54). The
sample was deliberately homogeneous to reduce the chances of bringing in established social conflicts (such as class or race prejudice) to the
The boys were unaware that they were part of an experiment on
group relations. The investigators appeared as regular camp staff, and
the boys met the staff and each other for the first time in the buses on
the way to camp. To maintain ecological validity, the experiments were
conducted within the framework of regular camp activities and games.
The researchers made unobtrusive records of behaviour, and, on occasions, used cameras and microphones.
In the first phase of the study, Sherif and his associates observed
the development of group structure. To start with the boys were
housed in one large bunkhouse where they were able to choose their
own 'buddies' (Sherif's term). After a few days they divided the boys
into two groups and took care to separate'best friends' into different
The boys were given a range of challenging activities including
hikes and campouts, and athletics and sports. In each group, the
boys divided up the tasks and organised duties. Leaders and
lieutenants emerged, and each group developed its own jargon,
special jokes, secrets and special ways of performing tasks. They
maintained social control through ridicule, threats and ostracism; for
example, insulting any boy who did not pull his weight at a particular
task. Each froup selected a symbol and a name which was put on their
baseball caps and T-shirts (fashion point: it is interesting to note that in
1954 the style for boys was very close to the style of 1996). The 1954
s,tudy was carried out near to a famous hideaway of Jesse lames called
the Robber's cave (the study is often referred to as the Robbers cave
Experiment), and the groups called.themselves 'The Eagles' and 'The
To test the social evaluations of the boys, the researchers invented a
game of target practice. There were no marks on the target board, and
a judgement of accuracy was made by the watching peers. However,
the board was also secretly wired to give an objective measure of accuracy. The boys consistently overestimated efforts of highly regarded
boys and underestimated efforts of the lowly regarded. The researchers
also made diagrammatic records of group structure, one of which is
shown in Figure 4.1. They asked each boy to name his friends in the
group. The boy who was chosen most times was regarded as having
the highest status, and the boy who was chosen the least was chosen
as having the lowest status.
The two groups shown in the figure have very different structures;
the group on the right is very hierarchical with one clear leader and
lieutenant along with seven low-status boys. The group on the left has
a clear leader but also a number of other boys of intermediate status.
This latter group was reportedly better at a range of tasks.
Sherif's prediction had been that '. . . when two groups have conflicting
aims, their members will become hostile to each other even though the
groups are composed of normal well adjusted individuals' (p. 5Z). In the
second phase of the study the researchers introduced conflict through
games. The tournament started in good spirit but the boys soon started
to call their rivals 'stinkers', 'sneaks' and 'cheaters' (fashion point 2:
clothes style might not have changed but language certainly hasr). The
boys refused contact with the opposing group and turned against their
previous buddies. When they were asked to give ratings to the other
boys in the camp, they gave negative ratings to boys in the other group.
During this period solidarity increased within each group. Name calling, scuffles and'raids' (for example, stealing the other group's flag and
setting fire to it) became the pattern of behaviour.
The hypothesis was supported alarmingly easily and this created a
problem of how to reduce the conflict. The initial attempts involved
bringing the groups together for an activity, but these occasions ended
:l -;&
F d.o 5
P 3* =
.; i.a
i;oB fr ?
E.o € j 5
ji tc
- ootn
s F 5. h6
.: q) .e
cj _d =;iFr..
X -.
ca - i H g:
*i '= !4 ^^ ul
ii o- i:;<
6i o *G
.! h^J
o (!
F.9 do=
I*> a; Ss
o o
: F':
in fights and abuse. So a further hypothesis developed: '. . . working in
common endeavour should promote harmony' (p.57).
In the third and final phase, the researchers had to establish
superordinate goals for the two groups. They created a series of urgent
and natural situations, for example by interrupting the water supply,
and making the camp truck break down on an outing. This latter
'crisis' had to be resolved by the boys pulling the truck. The boys
developed new friendships in the opposing groups as they worked
through these crises, and, according to Sherif, by the end of the camp
the groups were'actively seeking opportunities to mingle, to entertain
and to "treat" each other' (p.57). The boys made far less negative
ratings of the opposing group, and the hostility seemed to have
This is a remarkable piece of research on a number of counts. The
methods are very inventive, with a number of measures of social
behaviour being seamlessly introduced into everyday situations. The
study also has considerable ecological validity which was a feature of
social psychology conducted before 1960 (see Sears, 1986, Chapter 21 of
this volume). The study lacks a certain amount of control, but the trade
off with relevance seems a good one. You cannot help but wonder
whether the level of hostility alarmed the researchers, and whether
they were able to reduce this hostility quite as successfully as they
The study tells us one of the ways of creating inter-group hostility
and, therefore, ethnocentric attitudes and behaviour. This suggests
that we should see, for example, racial prejudice as a consequence of
competition for resources between different ethnic groups. In this case,
we would predict that prejudice would rise when economic hardship
increased, and to a certain extent this is supported by observation. However, the model does not explain why some groups should be singled
out for racism and not others. For example, in Nottingham, where the
authors live, there are a number of large ethnic groups which include
Irish, Ukrainian and Polish, as well as a number of Black and Asian
groups. All these groups are in the same competition for a limited
number of jobs, houses and welfare resources, but the Black and Asian
groups are subject to more racism than the Europeans. Clearly the conflict model does not tell us the n hole storv.
What is the conflict model of prejudice?
what measures of social behaviour and social judgement did the
researchers take2
List some of the (numerous) ethicar concerns with this
what groups in everyday life are in conflict with each other2
what are the advantages of carrying this study out in ,the
Suggested Answers start at p.432