The Stanford Prison Experiment Reread
The Stanford Prison Experiment is taught as proof that roles make ordinary people cruel. Sorted into three claims, the evidence supports the first, weakens the second and leaves the third contested.
Answer to the question
What did the Stanford Prison Experiment show?
It showed that student volunteers given guard roles, and instructions from the research team, behaved harshly within days. Archive material and later work suggest the instructions mattered more than the role, and the one real replication produced a different outcome. It does not show that a role alone makes ordinary people cruel.
Key takeaways
- Sapolsky (2017) reports that the Stanford Prison Experiment has been cited more than 58,000 times, against more than 27,000 for Milgram and more than 4,000 for Asch.
- Le Texier (2019), working from the archive and interviews with 15 participants, found that guards were briefed on the aim of producing powerlessness and were told, "We have total power in the situation. They have none."
- Haslam, Reicher and Van Bavel (2019) argue from archive tapes that the leadership the guards received, not the role, pushed reluctant guards towards toughness.
- Carnahan and McFarland (2007) found that volunteers for a study advertised in the 1971 way scored higher on aggressiveness, authoritarianism, Machiavellianism, narcissism and social dominance, and called the implications for the 1971 study speculative.
- Reicher and Haslam (2006) ran the BBC Prison Study over eight days; its guards did not adopt tyrannical roles and were overpowered by the prisoners.
Three claims travel under the name of the Stanford Prison Experiment. The first is that situations shape behaviour. The second is that handing people a role is enough to make some of them cruel. The third is that ordinary people given power become tormentors. The Stanford Prison Experiment ran in a Stanford basement in 1971, with student volunteers sorted into guards and prisoners, and was stopped after six days of a planned two weeks. Robert Sapolsky counts it among the most cited studies in the field, at more than 58,000 citations against 27,000 for Milgram and 4,000 for Asch (Behave, p. 464). The three have not fared equally.
The first claim holds, and it holds in a modest form. Sapolsky states what is left of it plainly: "a far higher percentage of perfectly normal people than most would predict succumb and do awful things," and yet "there are always those who resist" (Behave, pp. 468-469). That is a claim about how badly people forecast their own conduct. It needs no particular rate of cruelty, and the Stanford Prison Experiment cannot supply one. The second and third claims depend on what happened in the basement, and the archive now allows that to be checked.
What were the guards told?
Thibault Le Texier examined the Stanford Prison Experiment through the Stanford archive and through interviews with 15 participants (American Psychologist, 2019). At the guard orientation, Philip Zimbardo, who died in 2024, described the aim of producing boredom, frustration and powerlessness and said, "We have total power in the situation. They have none" (p. 4). The orientation also covered counting routines and humiliation tactics (p. 5).
The guards were not told they were research subjects; they took themselves to be helping run the study (p. 7).
The participants' own words point the same way. One guard wrote, "I was always acting" (p. 7). A prisoner wrote that "the one thing that kept me there was the money" (p. 8), and pay came at the end, so it depended on staying. Only two of the 11 prisoners showed clear signs of breakdown (p. 9). Sapolsky adds that nearly all the volunteers had said at the start that they would rather be prisoners than guards, and that many in both groups were deeply distressed afterwards (Behave, p. 465).
Haslam, Reicher and Van Bavel reached the same ground on the Stanford Prison Experiment from the same tapes with a different frame (American Psychologist, 2019). Cruelty, they argue, becomes likelier when leaders persuade people that their actions serve a noble cause, and they read the guards' toughness as a response to leadership from the research team. In their Scientific American account of the paper (30 May 2019), the authors report that the warden used collective pronouns 57 times in one short conversation with a guard, about once every 30 words. Winning politicians manage once every 79 words and losing candidates once every 136.
Sapolsky had made a related point earlier, from the design alone. Zimbardo acted as superintendent, told guards they could make prisoners feel afraid and helpless, and was, in Sapolsky's words, "someone whom you'd very much wish to please" (Behave, pp. 466-467). Some guards also resisted the abuse, and the authors ask for as much attention to resistance as to leadership. On the second claim, the record favours instruction over role.
The advertisement and its volunteers
Carnahan and McFarland recruited students with an advertisement close to the original and compared them with volunteers for a generic psychology study (Personality and Social Psychology Bulletin, 2007). Volunteers for the study advertised in that way scored higher on aggressiveness, authoritarianism, Machiavellianism, narcissism and social dominance, and lower on empathy and altruism. The authors call the implications for the Stanford Prison Experiment speculative.
The finding shows who volunteers for such a study, not that those volunteers would have been cruel without instruction. Sapolsky adds a qualification of his own: if prisoners had the same makeup as guards, it is unclear why that would tilt the outcome toward brutality (Behave, pp. 466-467).
A replication that went another way
Reicher and Haslam ran the BBC Prison Study, filmed in December 2001 and broadcast by the BBC in 2002, and published the academic account in 2006 (British Journal of Social Psychology; Nature reported the filming in May 2002). Men were randomly assigned as guards or prisoners over eight days. The guards did not come to identify with their role, were reluctant to use authority and were eventually overpowered by the prisoners, who set up an egalitarian system. When that proved unworkable, a harsher regime was imposed and the prisoners offered only weak resistance. The authors read this as identification and the failure of groups making tyranny seem acceptable.
Sapolsky reads the same study differently. The participants who ended up running the final regime had scored high on authoritarianism beforehand, which "suggests bad apples rather than bad barrels" (Behave, p. 468). He compares the sequence to the French and Russian revolutions, in which a hierarchy collapses into a commune and a harsher order then takes over (pp. 467-468). We side with neither reading entirely. Each leaves the Stanford Prison Experiment without a replication that supports it, and a single study of this kind cannot decide between leadership and disposition. Sapolsky adds that little of the data appeared in professional journals, so there has been only one attempted replication (pp. 466-467).
Where does the dispute stand now?
The people closest to the study now agree on part of it. A statement of 28 August 2018, signed by Zimbardo and Craig Haney, who ran the Stanford Prison Experiment, and by Alexander Haslam and Stephen Reicher, welcomed the new information. It said the information suggests that traditional accounts understate how much the experimenters intervened, that their leadership helped drive the guards' brutality, and that people may not slip into roles as easily as once thought. The statement asks for open and respectful debate that stays with ideas, and points to the public online archive so that anyone can take part. The statement does not set out where the signatories still differ, so that part of the debate remains open.
The same question of how much one famous result can carry runs through [What Forty-One Participants Buys](https://thewhyweproject.com/articles/what-forty-one-participants-buys), on a priming study, and [The Psychology You Learned Didn't Replicate](https://thewhyweproject.com/articles/the-psychology-you-learned-didnt-replicate), on the wider audit. [Everyone Thinks Everyone Else Wants It](https://thewhyweproject.com/articles/everyone-thinks-everyone-else-wants-it) takes up conformity from the side of what groups actually believe.
Set out by claim, the position looks like this.
| Claim | Where it stands | Main evidence |
|---|---|---|
| Situations shape behaviour | Holds, without Stanford | Sapolsky, Behave |
| A role alone produces cruelty | Not supported | Le Texier; Haslam, Reicher and Van Bavel; BBC Prison Study |
| Ordinary people become tormentors | Opposing views | Carnahan and McFarland; BBC Prison Study |
One question stays open. The archive records what the guards were told, and it cannot say how many of them would have been harsh without the briefing. The BBC Prison Study had no such briefing, but it also had a different design and a different era, so it cannot close the gap either.
Common questions
What did the Stanford Prison Experiment show?
It showed that student volunteers given guard roles, and instructions from the research team, behaved harshly within days. The archive and later work suggest the instructions mattered more than the role, so it does not demonstrate that a role alone makes ordinary people cruel.
Was the Stanford Prison Experiment ever replicated?
The closest attempt is the BBC Prison Study, run by Reicher and Haslam, filmed in December 2001, broadcast in 2002 and published in 2006. Its guards did not adopt tyrannical roles, though a harsher regime emerged later. Sapolsky describes it as the only real attempted replication.
Is the Stanford Prison Experiment still worth teaching?
It is worth teaching as a case of how a study's public account can diverge from its archive, and of how leadership and instruction shape conduct. It cannot be taught as a settled demonstration of what roles do to people.
Sources
Sapolsky, R. M. (2017). Behave: The Biology of Humans at Our Best and Worst. New York: Penguin Press. Page references to the Vintage 2018 edition, pp. 463-469.
Le Texier, T. (2019). 'Debunking the Stanford Prison Experiment.' American Psychologist, 74(7), 823-839. doi:10.1037/amp0000401
Haslam, S. A., Reicher, S. D., & Van Bavel, J. J. (2019). 'Rethinking the nature of cruelty: The role of identity leadership in the Stanford Prison Experiment.' American Psychologist, 74(7), 809-822. doi:10.1037/amp0000443
Carnahan, T., & McFarland, S. (2007). 'Revisiting the Stanford Prison Experiment: Could participant self-selection have led to the cruelty?' Personality and Social Psychology Bulletin, 33(5), 603-614. doi:10.1177/0146167206292689
Reicher, S., & Haslam, S. A. (2006). 'Rethinking the psychology of tyranny: The BBC Prison Study.' British Journal of Social Psychology, 45(1), 1-40. doi:10.1348/014466605X48998
Adam, D. (2002, 16 May). 'Reality TV show recreates famed social study.' Nature, 417, 213. doi:10.1038/417213a
University of St Andrews (2018, 28 August). 'Researchers call for open and respectful debate around Stanford Prison Experiment.' https://news.st-andrews.ac.uk/archive/researchers-call-for-open-and-respectful-debate-around-stanford-prison-experiment/
Van Bavel, J. J., Haslam, S. A., & Reicher, S. D. (2019, 30 May). 'Rethinking the Infamous Stanford Prison Experiment.' Scientific American, Observations blog. https://www.scientificamerican.com/blog/observations/rethinking-the-infamous-stanford-prison-experiment/