Rethinking the Psychology of Tyranny: The BBC Prison Study

Reicher and Haslam (2006)

The aim of the study was to investigate the factors underlying behaviour in institutional situations where groups have unequal power. This was done to revisit the findings of the Stanford Prison Experiment, where paricipants conformed to their designated roles. Reicher and Haslam's study was also conducted in a simulated prison, with guards and prisoners, but they were less prescriptive about how the guards should behave and used psychometric tests to measure mental health, authoritarianism and other factors during the study. The findings suggest that individual behaviour is not entirely determined by the situation, as both the guards and prisoners fail to conform entirely to role. Personality and the extent to which they internalised their role also played a part.

Background

The dominant research in Social Psychology since the Second World War has been focused on trying to make sense of the genocide of 6 million Jews by Hitler’s Third Reich. This includes important and famous studies such as Milgram (1963), Sherif (1956), Haney et al (1973) and Tajfel (1982). All of these studies consider the individual’s behaviour at a group level, suggesting that it is the situation that determines how an individual behaves. Social psychology often looks at the darker, more extreme aspects of behaviour, such as how people come to hate and discriminate against others, see some as less human / less deserving than themselves, where authoritarianism, dominance and power abuse come from, how tyranny of others is condoned and how people come to act tyrannically.

The Stanford Prison Experiment (SPE, Haney et al, 1973) was intended to run for two weeks, but was stopped after six days, as the guards were seen to abuse their power and the prisoners showed signs of psychological disturbance. Zimbardo and his colleagues suggested that the guards’ behaviour was a natural response to the role they were undertaking. The uniforms worn by the guards and the smocks by the prisoners resulted in ‘de-individuation’, whereby individuals take on a group identity and lose sense of themselves as individuals, leading to extreme anti-social behaviour on the part of the guards.

However a partial replication conducted by Loviband et al (1979) showed that where the guards were trained to respect the prisoners and involved them in decision-making, the behaviours demonstrated were far less aggressive and extreme. There is also only limited data available from the SPE, but some of this shows that both guards and prisoners often challenged their roles. The suggestion is that the guards took their lead from the researchers, who briefed them on their job as being to maintain order in the prison, to “…create feelings of boredom, as sense of fear…that their life is totally controlled by us…no privacy…no freedom of action, they can do nothing, say nothing that we don’t permit.” (Zimbardo, 1989, quoted in Reicher and Haslam, 2002, p4).

The SPE was the culmination of these types of study and due to the ethical issues it raised, social psychological studies since then have been confined to controlled laboratory studies or consider the cognitive / neurological aspects of in tyrannical behaviour.

The main aim of the study was to revisit Haney et al’s (1973) study (better know as the ‘Stanford Prison Experiment’ - SPE). This was considered necessary for a number of reasons:

• Zimbardo suggested that ‘conformity to role’ and the power of the situation were determined the behaviour of both the prisoners and the guards. However, Reicher and Haslam suggest that there were occasions when the participants challenged their roles and that a better explanation could be found by applying social categorisation theory.
• The SPE lacked ecological validity and generalisability to real prisons: “…where the walls of a prison remind the inmates that they must be kept apart from ‘decent’ people, they remind the participants in the SPE that they are honourable participants in adventurous scientific research” (p6). This means that Zimbardo’s application of ‘role theory’ is entirely theoretical as it is impossible to assess whether conformity to role occurs in other situations in the same way as it did in the SPE.
• Reicher and Haslam wanted to test the validity of role theory for these reasons, but also because Zimbardo’s analysis suggests that individuals have little choice over what they do and therefore can not be held accountable. This also discourages any challenges to tyranny.

Social categorisation theory is the way in which an individual adopts a social identity, and is influenced by whether the individual internalises their membership of a group. This relies on them firstly seeing movement within the group as possible, and secondly seeing inequalities between their group and other groups as stable and legitimate. If individuals perceive the group to insecure (unstable or not legitimate) then this opens up ‘cognitive alternatives to the status quo’ – ways in which members of the group can act to change the inequalities.

Aims

• To provide data on the interactions between groups with unequal power.
• To analyse the conditions which lead to group identity and then lead to accepting or challenging group inequalities.
• To examine social, organisational and clinical factors of group behaviour.
• To develop practical / ethical protocols for conducting large-scale social psychological studies.

Predictions

(related to the third aim above and social categorisation theory)
Dominant group members will identify with their group from the start and impose their power.
Subordinate members will identify with the group / challenge the group in relation to the permeability and stability of the group.

Methodology

Design

Field study in conjunction with the BBC documentaries unit. However this was not reality TV, but a documentary of the research taking place. The study was designed, run and analysed by the researchers; the set-up of the simulated ‘heirarchical institution’ and filming was undertaken by the BBC, with some material prepared later for broadcast.

The study was a controlled observational study which took place over 8 days in a controlled environment. The researchers also collected data from the participants via psychometric tests. These tests measured social and clinical factors and were measured on a 7 point scale.

The data from the observations and statistical analysis of the psychometric tests was triangulated (allowing for assessment of reliability). Tests were administered early in the morning, so responses relate mostly to the previous day.

Participants

15 males, all adult, selected through advertisements in the national press and leaflets, who went through three phases of screening:

1. Psychometric tests to measure social variables of authoritarianism, social dominance and modern racism and clinical variables such as depression, anxiety, socia isolation, paranoia, aggressiveness, demontivation, self-esteen, self-harm and druge dependence.
2. A full weekend assessment by independent clinical psychologists.
3. Medical and chracter references and police checks.

The initial pool of 332 applicants was reduced to 27 during screening, 15 of whom were chosen to ensure diversity of age, social class and ethnicity. All men were used (rather than men and women) for ethical reasons.

Participants were split into 5 groups of 3 who most closely matched on the key dimensions of personality implicating tyranny, modern racism, authoritarianism and social dominance. One of each group was randomly selected to be a guard and the remaining 2 were prisoners, resulting in 10 prisoners and 5 guards overall. The 10th prisoner was introduced at a later stage.

Procedure

Five of the participants were invited to a hotel the night before the study and told they were guards. They were then asked to draw up a set of prison rules under headings provided by the researchers. They were told that no physical violence would be tolerated and that the prisoners’ basic rights were to be protected. They had to ensure that the institution ran smoothly and that prisoners performed their tasks. They were also asked to draw up a set of punishments and given a timetable of cleaning chores, work duties, roll calls, exercise and recreation time. They were given no guidance on how they should achieve their goals. The following morning they were taken to the prison in a blacked-out van then fully briefed on the prison layout and resources available (cigarettes and snacks to be used as rewards / withdrawn as punishments, etc.).

The guards had access to an upper level from where they could survey all of the cells. They also had superior uniforms, meals and accommodation.

Prisoners arrived one by one in their t-shirt printed with three digit number, loose trousers and sandals. They had their heads shaved and were given information about the rules and a list of prisoners’ rights (posted in cells).

An initial announcement was made by the researchers to explain that it was possible that a prisoner could be promoted. The guards had also been briefed that they had been selected on the basis of trustworthiness, reliability and initiative and that it was possible that one of the prisoners had been misassigned, so the guards could decide on day 3 to promote someone. After the promotion the researchers announced that there would be no further promotions or demotions.

This introduced permeability and then removed it.

Three days after the promotion the researchers announced that there were no differences between the guards and prisoners, but for practical reasons it was not possible to reassign them to groups.

This removed the legitimacy of the group inequalities.

The following day the tenth prisoner was brought in – he was chosen for this role because of his trade union background. It was thought that he would bring a new perspective on group negotiation and collective / equal rights, as well as providing skills for collective action.

Measurement and Analysis

1. Four channels of video and all audio recorded.
2. Daily psychometric tests were used (but not every day on every participant to minimise fatigue) to measure:
   • Social variables: social identification, awareness of congitive alternatives, right-wing authoritarianism
   • Organisational variables: compliance with rules, organisational citizenship
   • Clinical variables: self-efficacy, depression.
3. Daily swabs of saliva taken to measure cortisol levels as an indicator of physiological stress response.

Ethics

• Participants signed a full consent form, which informed them that they may experience physical and psychological distress, confinement, constant surveillance and stress.
• Two independent clinical psychologists monitored the study throughout and could see any participant at any time or demand that a participant be removed from the study.
• A paramedic as on stand-by.
• Security guards were given detailed protocols on how to intervene if necessary due to participants behaving dangerously.
• An independent ethics committee of 5 people chaired by a member of Parliament also monitored the study and could demand that it was changed or stopped. They published an independent report afterwards.
• Other than minor ailments treated by the paramedic there were no other interventions required.

Results

There were two phases to the results (explained below). For both phases the data comes from:

• Observations
• Statistical analyses based on quantitative measures of individuals within groups (not between groups).

In the first phase the guards did not identify as a group, but the prisoners did after the promotion on day 3 and worked collectively to challenge the guards. This led to the collapse of the prisoner-guard system.

In the second phase the participants decided to continue as a self-governing commune, but this did not work due to being unable to deal with internal dissent and therefore they lost confidence in the system.

By the end of the study participants were becoming increasingly open to accepting the new, more draconian and unequal system that some of the participants wanted to impose.

Phase 1: Rejecting Inequality

Social Identification
From the start prisoners’ social identification was as predicted and they were clearly dissatisfied with their inferior conditions. However, initially many tried to improve their conditions by displaying the qualitites necessary for promotion. There was no shared identity among the prisoners and no consensus about how they should behave.

After the promotion the prisoners developed a shared social identity and more consensual norms. Before the promotion JE and KM (Cell 2) worked conscientiously. After the promotion all 3 occupants of Cell 2 recognised that they needed to change the system to improve their situation and discussed how they coud do this. If anyone expressed doubts they were reminded of the need to resist as a collective. Therefore the promotion and the end of permeability led to a shift from individual to collective identification.

The guards’ pattern of social identification went again the predictions, which were in line with Zimbardo (Haney et al, 1973) in suggesting they would identify with their high status. However after some evidence of this on the first day some guards did not appear to internalise this identity and the group never reached consensus about norms and priorities.

The use of behavioural observations and statistical analysis from the psychometric tests allowed for triangulation to test reliability of the findings.

The social identification of prisoners and guards was measured every day on two 3-item scales:

1. I feel strong ties with the prisoners / guards
2. I identify with the prisoners / guards
3. I feel solidarity with the prisoners / guards.

Tests were administered early in the morning, so responses relate mostly to the previous day.

An ANOVA was used to analyse the results for guards and prisoners for days 1 to 6. This showed that group and time both had significant effects on social identification. (p<0.5). The social identification of prisoners significantly increased over time, whereas it decreased for guards, but not significantly.

Security of intergroup relations

After the promotion the consensus between prisoners and the lack of consensus between guards undermined the legitimacy of the inequality as the guards were not displaying the qualities on which they had apparently been selected.

On day 4 the three prisoners in Cell 2 challenged the guards to see how they would respond. One threw his plate down and demanded better food. The guards tried to intervene and the other two prisoners demanded smoking rights and treatment for a blister. The guards were divided in how they dealt with this, one gave in and gave one of the prisoners a cigarette and afterwards when the guards returned to their mess hall, they discussed the consequences of what had occurred. Meanwhile the three prisoners were celebrating and congratulated each other, exchanging high fives.

The legitimacy intervention was not used as the situation created a lack of legitimacy by itself. However, the tenth prisoner was still brought in on day 5 (then withdrawn on day 6). He did not suggest alternatives to the status quo, instead questioning the legitimacy of the study. He suggested to his cellmates and the guards that they should unite a challenge against the researchers. This in part came from the excessive heat in the prison.

The guards were pleased to accept this new communal order as it seemed viable, even if it meant surrendering some of their heirarchical advantage.

The quantitative data confirms this, as the awareness of cognitive alternatives measures administered on days 1, 3, 4 and 6 asked four questions and showed that the awareness of congitive alternative increased as the study progressed (p<.05).

1. I cannot imagine the relationship between guards and prisoners being any different (reverse scored)
2. I think that the guards will always have more privileges than the prisoners (reverse scored)
3. I think that the relationship between prisoners and guards in likely to change.
4. I think that it would be possible for the prisoners to have more power than the guards.

Acceptance of the unequal regime (Compliance and Organisational Citizenship)

Compliance was measured on days 1, 3 and 5 on a 2 item scale:

1. I try to do what the guards want.
2. I try to comply with the guards rules.

A significant decline in prisoners compliance was seen after the promotion on day 3.

Organisational citizenship was measured on days 2, 4 and 5 on a 3 item scale:

1. I am willing to do more than is asked of me by the guards.
2. I will do whatever I can to help the guards.
3. Whenever possible I will try to make the guards’ work more difficult (reverse scored)

Again, the guards did not vary significantly across the three days, but the prisoners significantly declined over time (p<.01).

Collective self-efficacy and mental health

The effectiveness of the prisoners in collectively pursuing their goals led to an improvement in mood, whereas the guards became despondent.

Collective self-efficacy was measured on days 2, 4 and 6 through a 5 item scale:

1. My prison group is confident that we could deal efficiently with unexpected events.
2. My prison group can remain calm when facing difficulties because we can rely on our coping abilities.
3. My prison group can always manage to solve difficult problems if we try hard enough.
4. When my prison group is confronted with a problem, we can usually find several solutions.
5. My prison group can usually handle whatever comes our way.

Time had a significant effect on prisoners’ self-efficacy (p<.05 – Reicher and Haslam suggest this occurred linearly over time, although the graph indicates it peaked at day 4). For guards it decreased over time, but not significantly.

Depression was measured on a 7 item scale:

1. In general, how has your mood been over the last few days?
2. Do you ever feel low or depressed?
3. Do you feel hopeless about the future?
4. Do you have difficulty dealing with everyday problems?
5. Are you self-confident?
6. Do you think that you are a worthwhile person?
7. Do you think about harming yourself?

Prisoners’ scores decreased significantly over time (p<.01) whereas guards’ scores increased, but not significantly.

Summary

Over time the prisoners became aware of and confident in their collective identity, wherease the guards’ sel-efficacy declined. The prisoners executed a plan to destroy the guards’ regime. On the evening of day 6 the prisoners in Cell 2 broke out and occupied the guards’ quarters.

Phase 2: Embracing Equality

All but two participants decided that they wanted a new communal system, but those who had been at the forefront of the challenge to the old system refused to perform their tasks and the new system had no way to deal with this. On the second day it was clear that the new system was in crisis. By chance the breakfast was of very poor quality and the participants took this as a sign that the researchers did not approve of the commune system. The opponents of the system (one guard and three prisoners) took the opportunity to formulate a harsher system. The supporters of the commune were passive in their response and later commented in the debriefing that they were less opposed to it than they had been previously and would have accepted a strong social order where someone took responsibility for making it work.

An 8 item scale was used (taken from a 30 item scale) to measure right-wing authoritarianism:

1. Things would go better if people talked less and worked harder.
2. It is better to live in a society in which the laws are vigorously enforced than to give people too much freedom.
3. People should always comply with the decisions of the majority.
4. You have to give up an idea when important people think otherwise.
5. There are two kinds of people: strong and weak.
6. What we need are strong leaders that the people can trust.
7. Our social problems would be solved if, in one way or another, we could get rid of weak and dishonest people.
8. People should always keep to the rules.

Two separate findings came from this. Firstly both prisoners and guards increased linearly over time, but guards scores were higher than prisoners, even though they were both the same to begin with. Secondly, based on their roles at the end of the experiment, the new guards were consistently higher in right-wing authoritarianism than the new prisoners, but their scores increase gradually over time also.

Discussion

Four major criticisms have been made of this study:

1. The role of television: it has been argued that the presence of the TV cameras will have led to the participants acting unnaturally, although the researchers suggest that it would be impossible for the participants ‘act up’ for nine days, plus it is usual for people to be under constant surveillance of others in their daily lives.
2. The role of personality: it has been suggested that personality differences influenced the outcomes of the research. Again, the researchers highlight that they measured various personality traits and then matched participants in the two groups on these. They also point out that over the course of the experiment participants scores changed on these personality factors, indicating that they are transient and were not a fixed difference to begin with. They recognise that individual differences do play a role in this study.
3. The reality of inequality and power: critics have suggested that the study failed to provide a real situation whereby guards failed to assert their authority because the situation was not engaging and also because they had no power to assert. The researchers suggest that the guards were wary of demonstrating authoritarianism (at their first meal they complained about the inequalities and gave prisoners leftovers), but also point out that all participants did engage with the situation.
4. The impact of interventions and key variables: this criticism relates to the absence of experimental design and clear independent variables which could be manipulated and compared to a control conditon. The researchers point out that it is necessary to consider how various factors (including confounding variables) influenced the outcome and point out that it is difficult to come up with alternatives ways of investigating this subject.

The researchers conclude that role theory, as offered by Zimbardo, does not explain their results. Nor does social identity theory, although the study demonstrates some aspects of this, such as the permeability of group boundaries and whether the role was positively or negatively valued also influenced the extent to which it was internalised. However, one of the guards was a successful businessman outside of the study and it was this which stopped him being a successful guard. Therefore the researchers highlight that the context is important as it was the individual’s situation outside of the research which affected their role within the prison. This was clearly influenced by the televising of the study.

Like Zimbardo, the researchers argue that understanding collective conflict and ryranny means analysing group processes and intergroup relations. However, the findings indicate that groups are not an uncontrollable mob. Instead the behaviour of members of groups is determined by the norms and values of their social identity and may be anti- or pro-social.

It would seem that it is failing groups which create the problems, as they may adopt extreme solutions offered by others. This is when an authoritarian order may be tempting, as it is the breakdown and powerlessness of the group which leads to a lack of choice.

These findings can be applied to both Nazi Germany and to the SPE, which is not to suggest that this one small study is the final word, but that it should lead to further research and debate, as it shows that individual behaviour is not determined by the situation alone.