What Was the Stanford Prison Experiment? A Famous Study That Did Not Hold Up
By the BrainSnail editorial team. How these articles are written and checked, and how to tell us when one is wrong.
In August 1971 a psychologist converted a basement corridor at a university into a mock prison, assigned twenty-four young men at random to be guards or prisoners, and stopped the study after six of a planned fourteen days because the guards had become cruel and the prisoners had broken down. For fifty years it was taught in every introductory course as proof that ordinary people will abuse others if a situation gives them the role. Archive recordings released since 2018 have made that reading very difficult to sustain.
What was set up
Philip Zimbardo advertised for volunteers for a study of prison life, screened applicants for psychological problems, and assigned the twenty-four selected by the flip of a coin. Prisoners were arrested at their homes by real police as a favour to the researchers, booked, stripped, deloused and given smocks, stocking caps and numbers to be called by. Guards were given khaki uniforms, mirrored sunglasses and wooden batons, and told that physical violence was forbidden. Zimbardo himself took the role of prison superintendent rather than standing outside the study, which turned out to matter a great deal. Within two days a prisoner rebellion was suppressed with fire extinguishers, and over the following days guards imposed sleep deprivation, forced exercise, removal of bedding and degrading tasks; one prisoner was released after thirty-six hours in acute distress, and others followed. The study ended early after a graduate student, Christina Maslach, brought in to conduct interviews, objected to what she was seeing.
The conclusion that was drawn
Zimbardo's account, repeated for decades, was that the situation overwhelmed personality. Ordinary young men with no history of cruelty had been transformed by the roles and the uniforms within days, and the lesson was that behaviour is driven by circumstances far more than by character, a position known as situationism. The study reached a vast audience, appeared in textbooks worldwide, was the subject of films and a congressional hearing on prison reform, and was invoked directly in 2004 when photographs emerged from Abu Ghraib, where Zimbardo testified for the defence of one of the soldiers involved. Its appeal was that it appeared to demonstrate in a week what history suggests and no one could otherwise test.
What the archive showed
The original tapes and notes, held at Stanford and examined closely by the French writer Thibault Le Texier and by the journalist Ben Blum in work published in 2018, contradict the published account in several specific ways:
- •Guards were not left to work out their own behaviour. A recording captures the warden instructing a reluctant guard to be tough and telling him the experiment needed the guards to be active and aggressive
- •The guards had been briefed in advance on the conclusions the researchers hoped to demonstrate, including the idea that a prison strips away individuality
- •The most brutal guard, who set the tone for the others, later said he had consciously based his performance on a character from the film Cool Hand Luke and was acting rather than transformed
- •The prisoner whose breakdown became the study's most cited moment said afterwards that he had faked it in order to be released, having realised he could not simply quit
- •Only about a third of guards behaved harshly; the rest were passive or actively kind, a distribution the published summaries did not emphasise
- •No data were ever published in a peer-reviewed journal with a full methods section, and the study had no control condition of any kind
The attempt to repeat it
A partial replication was run for the BBC in 2002 by Steve Reicher and Alex Haslam, under ethical supervision and with the results published in the British Journal of Social Psychology. It produced almost the opposite outcome. The guards were uncomfortable with their authority and failed to form a coherent group, while the prisoners developed solidarity and eventually overturned the regime. Reicher and Haslam's interpretation is that people do not passively absorb roles but identify with groups, and that cruelty requires leadership that persuades followers the harshness is necessary and right. That account fits the archive from 1971 well, since the guards there were told what was wanted, and it fits the historical cases better too, since atrocities are typically organised and justified rather than spontaneous.
Why it is still worth knowing about
The study remains in the textbooks, increasingly as a case study in how research goes wrong rather than as a finding. It illustrates demand characteristics, the tendency of participants to work out what the experimenter wants and supply it; experimenter involvement, since a researcher running the institution he is studying cannot observe it; the absence of a control group; and the way a vivid narrative can substitute for data for five decades. It also sits at the centre of the ethical reforms that followed, since it and Milgram's obedience work drove the creation of the review boards that would not approve either study today. The deeper lesson is not that situations do not matter, which they plainly do, but that the specific claim about automatic transformation by role was never demonstrated, and that a result everybody wants to be true receives less scrutiny than one nobody does.
The takeaway
Twenty-four volunteers were randomly assigned as guards or prisoners in a basement in 1971, and the study was stopped after six days amid guard cruelty and prisoner breakdowns, becoming the standard demonstration that roles transform ordinary people. Archive recordings published in 2018 show guards were coached toward aggression and briefed on the desired conclusion, the most cited breakdown was feigned to secure release, only a third of guards were harsh, and no peer-reviewed data were ever published. A supervised replication in 2002 produced the opposite result.