Experiments in Intergroup Discrimination

Henri Tajfel (1970)

This article starts with an ironic reference to a Slovene friend of Tajfel. This friend had once spoken about the stereotypes that existed about immigrant Bosnians, who came from a poorer region of Yugoslavia [note that this article is written in 1970, long before the present civil war]. A parallel is drawn with the stereotypes that sadly persist for our own immigrants and their British born off-spring.The paper continues by questioning the reasons usually attributed to these stereotypes {it is assumed that we are speaking of negative stereotypes, and where these stereotypes influence our judgements, we could substitute the word prejudice].

Conflict between groups can be explained with social and psychological factors. What may start as economic competition can turn into psychological prejudice. Such prejudice perpetuates from generation to generation. A study of six and seven year olds demonstrates that there is high agreement amongst them, when they are asked to place four countries in order of preference. Social norms are defined as the way that we think others expect us to act. We therefore conform, and to conform we take on board the prejudices of those close to us. We talk of ‘we’ and ‘they’, when referring to members of the in-group and members of the outgroup, respectively. It should be noted that we could belong to many different in-groups simultaneously. Because of the conflict between groups we tend to think that the groups are in competition with each other, perhaps over limited resources. This may not be the case, because to bring order into our ‘social construction of reality’, [I would prefer to say ‘for cognitive efficiency’] we might simply automatically discriminate against the outgroup. [Notice we have spoken first of stereotypes, then of prejudice and now of discrimination.] Make sure you understand how these terms inter-relate.] If it is true that we automatically discriminate against the outgroup, then it can be predicted that we can discriminate against people for no reason, even though we have not quarrelled with them in the past and before we have had a chance to form a prejudicial attitude against them.

The present study concerns the behaviour of the individual towards both other in-group members and outgroup members. It is claimed that the subjects are presented with a ‘clear alternative to discriminating against the outgroup’. The subjects were 64 boys of 14 and 15 years of age. They were from a comprehensive school in Bristol. They were tested in groups of eight. All the boys in each group came from the same form and house, so knew each other well. The experiment has two distinct parts. The first was to establish an intergroup categorisation. The second part was to assess the effects of this categorisation on intergroup behaviour. In the first part the boys were brought together in a lecture theatre and asked to estimate the number of dots flashed onto a screen. There were forty trials of varying sizes of dot clusters. The boys were told that the experimenters were interested in the study of visual judgements. In one condition, the boys were told that there are two categories of people, overestimaters and underestimators, but there was no difference in accuracy. In another condition the boys were told that people were either accurate or not accurate. The experimenters pretended that they had marked the boys’ answers, and had placed them in the appropriate category. In fact, the boys had been randomly allocated to their groups. The boys were then told that to take full advantage of their help, the experimenters would like to run another experiment. The groups, as set up in the previous experiment, would be used, for the sake of efficiency. The boys were asked to fill in matrices.

Figure 11.1

 The numbers in the matrices represented units of 1/10 of a penny. In one condition the top row of the matrices represented the amounts that could be allocated to a fellow group member. The bottom row referred to amounts available for allocation to another member of the in-group. Never could money be awarded to himself. The subject also did not know the identity of any member of either group. In another condition the subjects awarded amounts to two different members of the outgroup. In a third condition, the main experimental condition, in half the trials the top row represented the amount to be awarded to another in-group member, and the bottom row represented the amount to be awarded to an outgroup member. In the other trials the groups were switched around for both rows. There were six matrices, repeated three times; one for each of the three conditions. Figure 11.1 gives the matrices used. Note that each box within a matrix forces the subject to favour one boy over another; there is no box that allows equal amounts to be given. It should be noted that for each box, within the matrix, there was another that held its inverse. In making their Intergroup choices, a large majority of the subjects allocated significantly more points to their in-group compared to the amount allocated to the outgroup. In the other two conditions the points were distributed fairly.

In the second experiment, the experimenters are interested in the strategy adopted by the boys, when allocating points. The boys could allocate for maximum joint profit, or for maximum profit for the in-group, or for maximum difference between the points allocated for one group compared to the other. In this experiment, the groups were randomly allocated to two groups after the boys had judged 12 paintings by two ‘foreign painters’. The groups were labelled the ‘Klee group’ and the ‘Kandinsky group’; being named after the actual painters whose work had been shown. This time the matrices consisted of 13 boxes, and were designed to facilitate the use of any one of the three strategies, mentioned above. In the centre of the matrices was a box with either 13 or 17 points in both the top and bottom rows, allowing for an equal allocation. Towards the ends of the matrices a choice could be made that would help to maintain maximum joint profit, maximum ingroup profit or the maximum difference in amounts allocated between the two groups (see figure 11.4).

Figure 11.4

The results demonstrated that when the boys had the choice between maximising the profit for all and maximising the profit for their own group, they chose the latter. Even more interestingly though, the boys were found to be more concerned with creating as large a difference as possible between the amounts allocated to each group (in favour of their own group), then in gaining a greater amount for everybody, across the two groups. Tajfel points out that this last finding is blatant discrimination caused by categorising the boys into meaningless groups. He compares his result to that of Sherif. In his experiment, with groups of boys, Sherif went to the trouble of building up competition between the two groups, before observing the resultant prejudges between them. Tajfel believes that in everyday life ‘fairness’ is interpreted according to how the situation is viewed in terms of ‘groupness’. Interpretations would be according to the perceived group norms.

Commentary

Aim and Nature

Tajfel later used the term minimal group to describe groups that form under these baseline conditions. This minimal group paradigm is thought to explain the conformity demonstrated by Asch, and the obedience demonstrated by Milgram.

Context and Background

Minimal group experiments provide evidence to support Social Identity Theory. An alternative explanation of prejudice is proposed by Adorno et al (1950). They believed that prejudice had a genetic (or dispositional) component by proposing that some people are possessed of an Authoritarian Personality. Such people are characterised by being strict and unloving, demanding obedience from those below them (most often their children), yet displaying deference and obedience to those above their status. Brown (1985) proposes that there is a link between the two explanations as the authoritarian personality conforms to society’s norms. He continues by suggesting that people in Nazi Germany were more likely to be overtly anti-Semitic because of the social norm rather than because of their personality.

Schiffman and Wicklund (1992) give the following criteria for minimal groups:

  1. No face-to-face interaction;
  2. Personal identity of group members should not be known;
  3. There should be no particular advantage of belonging to a particular group, nor logical reason for holding a negative attitude against the group (e.g. group membership based on colour of eyes).
  4. There should be no advantage or gain for the individual as a result of his making a particular response.
  5. The strategy employed when differentiating between groups should conflict with a more ‘rational’ strategy (e.g. only allowing adults, not children, to drink alcohol).
  6. The responses should be real decisions about the distribution of rewards or penalties.

Tajfel’s experiment fulfils all of these criteria. Concerning point 3, Billig and Tajfel (1973) found that even when group members knew that membership had been decided randomly (by tossing a coin or drawing lots), the results were just the same. Further to point 5 above, it should be noted that subjects are not always aware of their strategy. Interestingly, subjects award more to their own group members in the in-group/in-group condition, then they do to the outgroup members in the outgroup/outgroup condition.

Evaluation

  1. The act of categorisation is sufficient to elicit in-group favouritism.
  2. Schiffman and Wickland (1992) have pointed out the narrowness of minimal group research. The research does not take into account other psychological factors, such as ego involvement or needs. When tackling complex behaviour, it would be valid to identify one factor, and to study that alone. However, this reductionist approach would have to be recognised when formulating the aims of the experiment. Tajfel can not claim to have informed us about the determinants of Intergroup behaviour, because he has identified only one factor. It could be claimed that he has produced the circular argument ‘differentiation between groups is dependent on differentiation between groups’!
  3. Could the results be an artefact of the forced-choice nature of the matrices used? Discrimination could be the most appropriate strategy, with this style of presentation. It would seem not, judging by Locksley et al (1980). American subjects were given the freedom to distribute poker chips as they wished. The results were just the same as Tajfel’s.
  4. Demand characteristics could be playing a part here. Schiffmann and Wickland (1992) believe that the test instructions suggest that the boys should discriminate. Further, Brown (1986) points out that boys of this age, when set tasks by adults, might think that they are expected to compete with each other. This last point, doesn’t seem to be Right, because other population samples still produce similar results. St. Claire and Turner (1982) designed an experiment to test the demand characteristics explanation. One group underwent the standard procedure, another group was asked to predict how this group would react, and a third group underwent the standard procedure, but was informed [by a printed title] that the research was about prejudice. Group two was unable to predict the performance of group one, and group three performed no differently to group one. This would suggest that demand characteristics do not play a part. However, Schiffman and Wicklund point out how inaccurate people were at predicting Milgram’s results! Also, many of the subjects in group three, were unable to say at the end of the test what the experiment was about, even though they had been given the title at the beginning of the test! Nonetheless, group one subjects said that they had not followed a rule, and that they tried to be fair.
  5. In a meaningless situation (as in these experiments) people try to impose their own order on matters. Even though a toss of a coin decides who is in their group, subjects will still attribute similar characteristics to members of their group. According to the similarity-attractiveness hypothesis (Byrne, 1969), everything else being equal, similar others are preferred to dissimilar others.
  6. Social Identity Theory (SIT) (Tajfel & Turner, 1979) suggests that we divide the world up into manageable categories to simplify matters. Also our sense of identity depends upon our various group memberships. Because of this categorisation and identification, discrimination, ethnocentrism and hostility can be (partly) explained. We should criticise the ecological validity (mundane realism) of the minimal group paradigm. SIT assumes that the subjects are content to join any group they are allocated to, and will eagerly defend their group. Real life is much more complex than this. How can we attribute ‘social identity’ to a subject’s attitude to his group? Remember, the experimenter, assigned him to that group, without verifying that he wanted to belong to it. How can SIT be used as an explanation, if we can not be sure a subject has identified with the group first? (Schiffman and Wicklund 1992).

Exercises

  1. Define prejudice.
  2. Give an example of a dispositional determinant of prejudice, according to psychologists.
  3. Give an example of a situational determinant of prejudice, according to psychologists.
  4. What was the important difference between Sherif’s and Tajfel’s experiments.
  5. Describe Tajfel’s most minimal ‘minimal’ group.
  6. What IV’s and DV’s were there in Tajfel’s experiment?
  7. How would a subject behave in the allotting of points, if he was aiming at maximum fairness?
  8. What were the results of Tajfel’s experiment?
  9. What is the main point that Tajfel’s work makes about the way prejudice arises?



Return to Gary Sturt's Homepage