Henri Tajfel 1970

Henri Tajfel, Michael Billig, Robert Bundy & Claude Flament 1971

AIMS: Henri Tajfel wanted to test the idea that prejudice and discrimination can occur between groups even when there is no history between the groups and no competition. Having found prejudice and discrimination between such minimal groups, Tajfel’s team wanted to investigate the possible causes. In order to study discrimination as well as prejudice, it was important to have an experimental situation that involved real behaviour. Therefore, they aimed to generate a situation in

which members of a group had to act in some way in relation to another group.

Tajfel carried out two experiments. The first (1970) created groups from judgements about how many dots were in an image and the second (1971) created groups from an apparent preference for the artists Wassily Kandinsky and Paul Klee.

PROCEDURE (METHOD) - 1970 EXPERIMENT: The participants were 64 boys aged 14 and 15 years from a comprehensive school in Bristol. They were tested in a laboratory in 8 separate groups of 8 boys. The boys in each group knew each other well.

The experimenters set out to:

• establish in-group categorisation (the formation of the groups)
• assess the effect of the formation of the groups on behaviour

To form the two groups the boys were taken into a lecture room and told that the study was about visual judgements. 40 clusters of varying numbers of dots were flashed onto a screen. The boys had to estimate the number of dots and record their estimates on score sheets.

• In condition 1, after they had estimated the number of dots, they were told that people constantly overestimate or underestimate the number.
• In condition 2, they were told that some people are more accurate than others.

The judgements were then ‘scored’ by one of the experimenters and the boys were told that they could help the researchers with something else. They were told that for this task they would be put into groups according to the judgements they had made about the number of dots. The boys were allocated randomly to the groups and told which group they were in. In condition 1, they were said to be ‘overestimators’ or ‘underestimators’. In condition 2, they were either better at making the judgements or worse. There was no face-to-face contact between any of the boys once they had recorded their estimates of the numbers of dots.

The boys were told that the task used real money for rewards and punishments. Individually they would have to decide whether or not to allocate money and they would know the code number of the boy receiving the reward or punishment, and which group he was in. They had to choose how much to reward or punish another boy in either their own group or the other group. The experimenters showed the boys the type of matrix that they would use. (As in the example below, each matrix had two rows each with 14 numbers, with each number in a box. Some numbers had minus signs, which meant that amount of money would be taken away from that person; the other numbers represented the amount of money allocated to that person.

Tajfel’s query was whether they would favour boys in their in-group and award them more than boys in their out-group.

In the baseline condition, it was clearly in the boys’ interests to show favouritism to their in-group as they were told points earned prizes for their group.

Tajfel varied the grid to create conditions where it was not in their best interest to show in-group favouritism. He manipulated the grids so that the maximum number of points for prizes the boys could give to their in-group meant the out-group member automatically got more points.

For example, with the grid to the left, a Klee Group member could follow one of 3 strategies:-

• Choose 19 for the Klee Group which would give the in-group most points for prizes - maximising the in-group profit. However, this strategy would give the out-group even more profit!
• Choose 13 so that each group does equally well - maximum fairness and a good strategy if the boy has a friend in the Kandinsky Group.

• Choose 17 which would give the Kandinsky Group 25, thus, ensuring the maximum number of points for prizes was shared between the two groups - maximum joint profit.
• Choose 7 - the smallest gain in points for prizes for both group but the one which would maximise the difference in favour of the in-group.

FINDINGS (RESULTS) - 1971 EXPERIMENT: In the baseline condition, the boys generally awarded more points to members of their in-group, demonstrating strong in-group favouritism.

Even when the grid was manipulated (as in the second example), the boys generally opted to maximise the difference in favour of the in-group rather than score maximum points for prizes which would have benefited the out-group at least as much.

CONCLUSIONS: Categorising boys into essentially meaningless groups led them to identify with their in-group and engineer a positive social identity by giving their group more points. Even if it meant they scored fewer points for prizes overall, it was more important to ensure their group had more points than the out-group.

Intense competition was not needed to produce this rivalry.

CRITICISMS (EVALUATION):

• How ecologically valid was Tajfel’s experiment? Would the simple act of categorisation be enough to create discrimination in a ‘real life’ situation? Much r’real life’ categorisation carries with it some inherent degree of competition.
• It has been argued that, although there was no intention to create competition between the two groups, the very nature of  being divided into two groups by adults and the forced choice nature of the task implied competition led the boys to think in terms of ‘teams’ and discriminate in favour of their own in-group.

However,  Ann Locksley, Vilma Ortiz & Christine Hepburn (1980) created explicitly random groups - where the members knew their assignation was random - removed the forced choice task and got them to divide poker chips amongst anonymous members of the groups. They still found strong in-group favouritism. This supports the validity of Tajfel’s work.

• Billig & Tajfel (1973) themselves tried a study in which the participants were told that group membership had been determined randomly - yet they still found in-group favouritism.
• It has also been proposed that the boys were affected by demand characteristics. They were aware that they were involved in an experiment in an unfamiliar university setting so they may have tried to behave in a way they thought the experimenters wanted.
• The study was very reliable. With strict controls over the information the boys were given and the experience they had - for example, the boys only knew other in-group/out-group members by a code number so there was no face-to-face interaction between group members - the study was replicable. Tajfel himself carried out a number of studies of this type and consistently obtained similar results.
• The tight controls mean cause-and-effect can be inferred.
• The study has been accused of lack of population validity due to the limited nature of the sample. However, other researchers using different populations - eg: adults in Cardiff (Braithwaite & Jones 1975), female adults in California (M B Brewer & Myron Rothbart 1980) and German soldiers (Hanns Dann & Willem Doise 1974) - have all found the same minimal group effect of Tajfel’s original research.
• Rupert Brown (1988) criticised Tajfel for the way he interpreted his results. He suggested that, although the boys’ behaviour was biased towards their own in-group, the bias was not very extreme and seemed to be moderated by a sense of fairness.

The boys could not allocate money to themselves; at the end, they would receive the money allocated to them by the other boys. They worked through a booklet of 18 matrices. Each time they were told either, “These are rewards and punishments for member XX of your group” or “These are rewards and punishments for member XX of the other group”. They had to choose which column (pair of numbers) to allocate. One row referred to one boy and one to another boy, so they had to choose which pair they wanted that would affect both boys.

The important part of the study was that the boys had to make decisions about the rewards and punishments that they would impose. They were faced with three types of decision:

• ‘in-group/in-group’
• ‘in-group/out-group’
• ‘out-group/out-group’

If they allocated as much as possible to one boy, this was given a score of 14 (because there were 14 decisions for each row of each matrix). If they allocated as little as possible to a boy, this was given a score of 1. For each decision they were allocating to two boys. Therefore, a fair score would be 7 because this would mean that they had allocated rewards (or punishments) equally.

FINDINGS (RESULTS) - 1970 EXPERIMENT:

• When decisions involved two boys, one from each group (an ‘in-group/out-group’ decision), the average score was 9 out of 14.
• When boys were making ‘in-group/in-group’ or ‘out-group/out-group’ decisions, the average score was 7.5.

It seemed that decisions about boys in the same groups were fairer than decisions when one boy was in the same group as the boy making the judgements and one boy was in the other group. A large majority gave more money to their own groups and showed in-group favouritism. This was found in all trials in the study.

PROCEDURE (METHOD) - 1971 EXPERIMENT: A new sample of 48 boys split into 3 groups of 16 was shown 12 slides of paintings, 6 by Kandinsky and 6 by Klee without the artists’ signatures; and the boys were asked to express their preferences. They were then seemingly allocated into one of two groups on the basis of this - either the Kandinsky Group or the Klee Group. The groupings were, in fact, purely random.

They were then given a rewards allocation task where they were each asked to award points to two other boys (one from each group) at a time. The only information they were given was what group they were in. The point scores for each boy were tied together, so that when a participant chose a particular score for one boy, the other boy automatically got the score tied to it.

For example, if the participant was asked to allocate points from the grid below and chose to give 14 points to Klee Group member 25, then Kandinsky Group member 2 got 1 point.