The question
Stanley Milgram, a young assistant professor at Yale, opened his first paper on obedience with the Holocaust: millions of people had been killed on command, and that could only happen because very many people carried out orders. That framing has been debated ever since (more below). The question he actually tested was narrower. If a person in a lab coat tells an ordinary adult to keep hurting a stranger, at what point does that adult say no?
How it was done
Men from New Haven and nearby towns answered an ad for a paid study of memory and learning ($4.50, theirs whatever happened).
At the lab, each met an apparent fellow volunteer, and they drew slips for the roles of “teacher” and “learner”. Both slips said teacher. The learner was an actor, a 47-year-old accountant trained for the role, and the stern experimenter in a grey coat was a high school biology teacher.
The learner was strapped into a chair in the next room with an electrode on his wrist. The teacher sat at an impressive shock generator with 30 switches, from 15 to 450 volts in 15-volt steps. Groups of switches carried labels rising from a mild shock to a severe, dangerous one, and the last two were marked only with an ominous triple X. The teacher first received a real 45-volt sample shock.
The teacher read word pairs over an intercom and tested the learner. For every wrong answer he had to give a shock one step higher. The learner never received a real shock; his answers, and later his protests, followed a script. In the first published version, he pounded on the wall at 300 volts, again at 315, and then fell silent.
If the teacher hesitated, the experimenter used a fixed series of four prods, from a polite request to go on, through statements that the experiment required it, to the last: “You have no other choice, you must go on.” If the teacher still refused after the fourth, the session ended.
What came out
Of the first 40 men, reported in 1963:
- 26 (65%) went all the way to 450 volts.
- None stopped before 300 volts.
- The 14 who defied the experimenter stopped between 300 and 375 volts.
Before the results were known, 14 Yale psychology seniors had been asked to predict what 100 people would do. On average they expected 1.2% to reach the end.
Obedience did not look calm. Milgram reported that many men showed signs of intense tension. Fourteen of the 40 laughed nervously, and three had uncontrollable seizures of laughter.
Between August 1961 and May 1962 Milgram ran 23 conditions with 780 people, changing one thing at a time. The famous 65% turned out to be one point in a wide range:
- Distance from the learner. With the learner audibly protesting from the next room, 62.5% went to the end. With him sitting in the same room, 40%. When the teacher had to press the learner’s hand onto a shock plate, 30%.
- Distance from authority. When the experimenter left and was reachable only by phone, 9 of 40 (22.5%) went to the end.
- Prestige. In a run-down office in Bridgeport, away from the Yale campus, 47.5% went to the end.
- Peers. When two other “teachers” (both actors) refused to go on, only 4 of 40 (10%) continued. When two others kept going, 72.5% did.
- Who pulls the lever. When the participant only read the words and someone else gave the shocks, 92.5% stayed to the end.
- Women. Forty women in the baseline setup obeyed at the same rate as men, 65%.
A 2014 analysis of 21 of these conditions found an average obedience rate of 43.6%. Factors on the experimenter’s side and on the learner’s side mattered about equally.
What really happened (replications and critiques)
The archives tell a messier story. Psychologist Gina Perry worked through Milgram’s papers and tapes at Yale and interviewed former participants for her book Behind the Shock Machine. Her findings, and those of other researchers working in the archive, include:
- The script was not always followed. Stephen Gibson (2013), studying the session recordings, found that in three sessions of one condition the experimenter went beyond the four prods, for example leaving the room to “check on” the learner, a tactic that Milgram and the experimenter negotiated as they went along. Gibson argues that this worked as a ruse rather than an order, so it is hard to read as obedience to authority.
- Not everyone believed it. Milgram wrote in 1963 that participants were convinced the situation was real. In 2019 Perry and colleagues reanalysed an unpublished 1962 report from Milgram’s own team. In 18 of 23 conditions, people who fully believed the shocks were real gave lower shocks than people who doubted. Participants who believed most strongly were about two and a half times more likely to defy the experimenter than the rest. The effect is modest, and the authors note that belief is hard to measure.
- Debriefing came late. According to the same paper, Milgram did not formally debrief participants until about three-quarters of them had already taken part, and only about a quarter received the full spoken debriefing. Many learned the truth from a written report months later.
Why did people obey? Milgram’s own answer, set out in his later writing, was the “agentic state”: people hand responsibility to the authority and treat themselves as its instrument. Psychologists Alex Haslam and Stephen Reicher argue nearly the opposite. In their reading, participants went on when they identified with the experimenter and believed they were helping science, not because they had stopped thinking. They point out that the most direct order, the “no other choice” prod, usually led people to refuse. They call this “engaged followership”. Perry and colleagues accept that it explains some behaviour but argue that participants had many different motives. This remains an open debate.
Modern partial replications. A full replication would not be approved today, so researchers have built safer versions:
- United States, 2009. Jerry Burger at Santa Clara University tested 70 adults and stopped at 150 volts, the point where the learner first demands to be released. He chose that point because, in Milgram’s data, 79% of those who passed it went all the way. He screened participants and told them repeatedly they could leave and keep the $50. In the condition comparable to Milgram’s, 70% were ready to continue past 150 volts, against 82.5% in Milgram’s study, a difference that was not statistically significant. Seeing another person refuse did not reduce obedience much.
- Poland, 2017. Dariusz Doliński and colleagues at SWPS University tested 80 people aged 18 to 69 with ten buttons at lower “shock” levels. 90% pressed the highest button. Refusals were about three times as common when the learner was a woman, though the authors caution that a sample this size cannot settle the question.
Both studies stop far short of Milgram’s endpoint, so they show that many people will start down the path, not how far they would go.
Ethics
The study relied on deception at every step: the cover story, the rigged draw, the actor and the fake generator. Its distress was visible and was described in the paper itself.
In 1964 developmental psychologist Diana Baumrind published a sharp critique. She argued that the distress should have led Milgram to stop, that a researcher’s responsibility does not end with consent, that the link to Nazi camps was unconvincing, and that such studies damage public trust in psychology.
Milgram replied the same year. He argued that every participant was free to refuse, and he cited a follow-up questionnaire in which, as later textbooks report it, 84% said they were glad to have taken part and 1.3% were very sorry. A psychiatrist who later interviewed a group of participants reportedly found no signs of lasting harm. Perry’s archive work complicates that reassuring picture: some participants were still troubled decades later, and many had left the lab without being told the truth.
What it means today
The studies do not show that people are machines that follow orders, and they do not show that most people would commit atrocities. They do show something sobering and specific. In a setting that feels legitimate, step by step, with someone else claiming responsibility, many people do things they would have predicted they never would.
On the Holocaust link, Haslam and Reicher point to historians who describe Nazi perpetrators such as Adolf Eichmann as committed and inventive, not merely obedient, and Perry and colleagues argue that the Holocaust framing narrowed what researchers looked for. Both views take the experiments seriously while disagreeing about what they explain.
The variations also contain the hopeful part. Obedience fell when the victim was close, when the authority was distant or less legitimate, and above all when others refused. Some cautious takeaways, beyond what Milgram himself tested:
- Small steps matter. Noticing where you would draw the line before you start makes it easier to hold it.
- “It’s not my responsibility” is a warning sign, not a defence.
- One or two people saying no can change what everyone else does.
Would you have passed?
Coming soon. We will not simulate shocks: a screen cannot recreate the pressure of a real room, and pretending otherwise would be misleading. Instead, a short reflection quiz asks where you think you would stop, then shows you what Milgram’s participants actually did in each variation, next to the 1.2% that Yale students predicted. Most people, like those students, expect to stop much earlier than the people in the lab did.
Questions people ask
What percentage of people obeyed in the Milgram experiment?
In the first published study, 26 of 40 men (65%) went to the final 450-volt switch. Across all of Milgram's variations the rate ranged from almost no one to nearly everyone; one analysis of 21 conditions found an average of 43.6%.
Has the Milgram experiment been replicated?
Only partially, because the full procedure would not be approved today. A 2009 US study stopped at 150 volts and found obedience only slightly below Milgram's. A 2017 study in Poland, using lower 'shock' levels, found 90% of participants pressed the highest button.
What made people less likely to obey?
Being closer to the victim, having the experimenter leave the room, an ordinary location instead of Yale, and above all seeing two other people refuse. Direct commands without a reason also tended to backfire.
Was the Milgram experiment ethical?
It is one of the most debated studies in psychology. Diana Baumrind argued in 1964 that the distress was not justified. Milgram replied that participants were free to stop and that most later said they were glad to have taken part. Archive research has since shown that many were not told the truth for months.
Sources
- Milgram, S. (1963). Behavioral study of obedience. Journal of Abnormal and Social Psychology, 67(4), 371–378. doi:10.1037/h0040525
- Milgram, S. (1964). Issues in the study of obedience: A reply to Baumrind. American Psychologist, 19(11), 848–852. doi:10.1037/h0044954
- Milgram, S. (1965). Liberating effects of group pressure. Journal of Personality and Social Psychology, 1(2), 127–134.
- Milgram, S. (1974). Obedience to authority: An experimental view. Harper & Row.
- Baumrind, D. (1964). Some thoughts on ethics of research: After reading Milgram's 'Behavioral study of obedience.' American Psychologist, 19(6), 421–423.
- Burger, J. M. (2009). Replicating Milgram: Would people still obey today? American Psychologist, 64(1), 1–11. doi:10.1037/a0010932
- Doliński, D., Grzyb, T., Folwarczny, M., Grzybała, P., Krzyszycha, K., Martynowska, K., & Trojanowski, J. (2017). Would you deliver an electric shock in 2015? Obedience in the experimental paradigm developed by Stanley Milgram in the 50 years following the original studies. Social Psychological and Personality Science, 8(8), 927–933. doi:10.1177/1948550617693060
- Haslam, N., Loughnan, S., & Perry, G. (2014). Meta-Milgram: An empirical synthesis of the obedience experiments. PLoS ONE, 9(4), e93927. doi:10.1371/journal.pone.0093927
- Haslam, S. A., & Reicher, S. D. (2012). Contesting the 'nature' of conformity: What Milgram and Zimbardo's studies really show. PLoS Biology, 10(11), e1001426. doi:10.1371/journal.pbio.1001426
- Perry, G. (2013). Behind the shock machine: The untold story of the notorious Milgram psychology experiments. The New Press.
- Perry, G., Brannigan, A., Wanner, R. A., & Stam, H. (2020). Credibility and incredulity in Milgram's obedience experiments: A reanalysis of an unpublished test. Social Psychology Quarterly, 83(1), 88–106. doi:10.1177/0190272519861952
- Gibson, S. (2013). 'The last possible resort': A forgotten prod and the in situ standardization of Stanley Milgram's voice-feedback condition. History of Psychology, 16(3), 177–194.
- Uzelac, E., & Yang, Y.-F. Obedience to authority. In R.-M. Rahal (Ed.), Open Social Psychology (FORRT), chapter 8. forrt.org/open-social-psychology/chapter4.html
- The Open University. Psychological research, obedience and ethics (OpenLearn course), section 2.3. www.open.edu/openlearn/society-politics-law/sociology/psychological-research-obedience-and-ethics/content-section-2.3