Last Updated on August 21, 2026 by Karl Thompson
The Stanford Prison Experiment (1971) is one of the most famous studies in social science and one of the most disputed. This post covers:
- What Philip Zimbardo actually did, and what he concluded
- Why it is a laboratory experiment and not a field experiment
- The practical, ethical and theoretical problems with it
- Why the findings are no longer accepted as they once were
- How to use it in A-level sociology, which is not the way most textbooks suggest
In this famously notorious study, college students volunteered to take on the role of either prison guards or prisoners and spend time in an artificial prison. It was meant to last two weeks. It was stopped after six days, because the guards had become abusive and the prisoners were showing signs of severe distress.
The study belongs to the research methods component of A-level sociology, and it is also useful for crime and deviance.
What Zimbardo did
In 1971 the psychologist Philip Zimbardo and his colleagues set out to examine what happens to people when they become a prisoner or a prison guard. They built a mock prison in the basement of Stanford University’s psychology building and recruited 24 male undergraduates through a newspaper advertisement asking for volunteers for a study of prison life.
The volunteers were screened for psychological health and then randomly assigned to be either prisoners or guards. This is the study’s one genuinely strong design feature, and it is worth understanding why. Because guards and prisoners were not different sorts of people to begin with, any difference in how they subsequently behaved cannot be explained by personality. It has to come from the role.
Prisoners stayed in the mock prison around the clock. Guards worked eight-hour shifts in teams of three and went home in between. The researchers observed both groups using hidden cameras and microphones.
What happened, and what Zimbardo concluded
Interactions between the two groups became hostile and then degrading. The guards behaved with increasing aggression towards the prisoners; the prisoners became passive and depressed. Five prisoners experienced distress severe enough — crying, acute anxiety — that they were released from the study early. Zimbardo ended the whole thing on day six of a planned fourteen.
His conclusion was that the situation had overwhelmed the individual. Ordinary, screened, psychologically healthy students, placed in a position of power, behaved in ways they never would in the rest of their lives; students placed in a position of powerlessness became passive and depressed. Personality had not predicted behaviour. Position had.
That is a genuinely sociological claim, and it is why a psychology study ended up on a sociology syllabus.
It is also, on the current evidence, largely wrong. More on that below.
Is the Stanford Prison Experiment a field experiment?
No, and this trips up a lot of students.
A field experiment takes place in a naturally occurring social setting — a real school, a real workplace, a real street. Zimbardo’s prison was built by researchers in a university basement. It did not exist before the study and it did not exist afterwards. That makes this a laboratory experiment, and more precisely a simulation: an artificial environment constructed to imitate a real one.
The confusion is understandable, because a fake prison feels more real than a room with a one-way mirror. HOWEVER, the test is not how realistic the setting looks. The test is whether the researcher built it. Zimbardo built it, controlled who entered it, and could end it whenever he chose — which he did.
Get this wrong in an exam and you lose the mark. If you need a genuine field experiment on institutions and behaviour, use Rosenthal and Jacobson or the Hawthorne studies instead.
Evaluating the Stanford Prison Experiment
Theoretical strengths
Control over variables. Zimbardo decided who was a guard and who was a prisoner, and assigned those roles at random. As above, that randomisation is what allows any causal claim to be made at all.
Reliability, in principle. The procedure was documented in enough detail to be repeated, and it has been — see the BBC study below.
It asked a question sociologists care about. Whether behaviour is driven by the individual or by the position they occupy in a social structure is the argument between Interactionism and structural theories in miniature.
Practical limitations
A sample of 24 male American undergraduates. The advertisement asked specifically for volunteers for a study of prison life, so the sample selects for people drawn to that idea in the first place. Twenty-four students in California in 1971 are not a basis for claims about human nature.
No control group. There was no second mock prison run differently against which to compare the outcome, so there is nothing to isolate the effect of anything.
Ethical limitations
Serious harm to participants. Five prisoners were released early with acute distress. This is the textbook case of a study failing the principle that participants should not be damaged by taking part, and no ethics committee today would approve it. The British Sociological Association’s statement of ethical practice makes that reasonably clear.
Consent was given, but not informed. Volunteers agreed to a study of prison life. Nobody consented to what actually happened.
Zimbardo was not an observer. He ran the mock prison as its superintendent while simultaneously being the researcher studying it. He later accepted this was a mistake. It is hard to think of a cleaner example of a researcher’s own values contaminating the thing being researched — which is the standing objection to the idea that research can ever be value free.
Theoretical limitations
Very low ecological validity. Real prisoners have been sentenced by a court, and real prison officers are trained, paid, unionised and answerable to an institution that outlasts them. Students who go home at the end of a shift are not prison officers. Whatever this study measured, it was not prison.
Demand characteristics. Participants knew they were in an experiment and knew broadly what it was about. Some later said they had worked out what Zimbardo wanted and provided it. The Hawthorne effect applies here with unusual force.
The findings are now seriously disputed
This is the part most revision material has not caught up with, and it is where the marks are.
In 2019 the French researcher Thibault Le Texier published an analysis in American Psychologist based on the study’s own archives at Stanford and Akron — recordings, notes, and interviews with fifteen of the original participants. What he found undermines the standard account:
- The guards were given fairly specific instructions on how to treat prisoners, rather than sliding into cruelty on their own. Recordings show the research team coaching them to be tough.
- Data collection was selective and incomplete.
- Participants were often aware they were performing. The guard usually described as the most brutal later said he had treated the whole thing as an acting exercise.
- The design was not original. It drew closely on a prison exercise run by students in one of Zimbardo’s own classes a few months earlier.
Put those together and the finding inverts. The study does not show that ordinary people become cruel when handed power. It shows that people do what the person running the experiment makes fairly clear they are supposed to do — which is a finding about research methods, not about human nature.
Zimbardo, who died in October 2024, defended the study to the end. Some psychologists have argued it should be dropped from textbooks altogether. Others — and I think this is the better position — say it should stay, but be taught as a lesson in how research goes wrong rather than as a fact about people.
The BBC Prison Study
Steve Reicher and Alex Haslam ran a version of the study for the BBC, broadcast in 2002 and published in the British Journal of Social Psychology in 2006. Crucially, neither took a role inside the prison and neither gave the guards any guidance.
The result was almost the reverse of Stanford. The guards did not form a coherent tyranny; the prisoners eventually organised and overturned the guards’ authority. Reicher and Haslam concluded that people do not simply absorb the roles they are handed — they decide whether to identify with a group, and what that group should do.
That is a far more sociological finding than Zimbardo’s, and it fits much better with what Interactionists say about how people negotiate the roles they are given.
How to use this study in A-level sociology
Not as evidence that power corrupts. That reading is no longer defensible, and an examiner who has read anything published in the last five years will not be impressed by it.
Use it as an evaluation point instead. It is the best worked example on the specification of why sociologists distrust laboratory experiments: artificial setting, tiny unrepresentative sample, researcher inside his own study, participants performing for the camera, serious harm to those involved, and a set of findings that did not survive contact with the archive. If a methods question asks you to evaluate laboratory experiments, this study answers it on its own.
It is also useful on crime and deviance. After the abuses at Abu Ghraib prison in Iraq came to light, Zimbardo argued that situations rather than individuals produce cruelty, and testified in the defence of one of the guards on that basis. Whether wrongdoing is best explained by the person or by the institution is a live argument in the sociology of crime, and this study sits in the middle of it — even if, on the current evidence, it does not settle it.
And it is worth knowing as a story about sociology itself. A study can be famous for fifty years, appear in every textbook, and still turn out not to show what everyone said it showed. That is an argument for exactly the sceptical habits the methods topic is trying to teach you.
Sources: Le Texier, T. (2019) ‘Debunking the Stanford Prison Experiment’, American Psychologist, 74(7), pp. 823–839. Reicher, S. and Haslam, S. A. (2006) ‘Rethinking the psychology of tyranny: The BBC prison study’, British Journal of Social Psychology, 45(1). Zimbardo’s own account of the study is at prisonexp.org.
Signposting and Related Posts
This post has been written for students studying research methods as part of the AQA’s A-level sociology course, and is also relevant to crime and deviance.
Since this is a laboratory experiment, the place to start is Laboratory Experiments in Sociology, which covers the practical, ethical and theoretical strengths and limitations of the method in full.
Field Experiments in Sociology covers the other main type, with seven worked examples — and explains why this study is not one of them.
Milgram’s Obedience Experiment is the other famous laboratory study wheeled out in sociology, and it raises very similar ethical questions.
Bandura’s Bobo Doll Experiment is a third classic, relevant to media effects and to crime.
For the basic logic of experiments — hypotheses, independent and dependent variables — see Experiments in Sociology: An Introduction.
For everything else on this topic, see the main research methods page.
Please click here to return to the homepage – ReviseSociology.com