Image: Public domain, via Wikimedia CommonsMilgram's obedience study, sixty years on
The advertisement in the New Haven newspaper in 1961 asked for men between 20 and 50 to take part in a study of memory and learning at Yale, for $4.50 and an hour of their time. Each man who answered was introduced to a second participant, a mild accountant in his late forties, and the two drew lots for the roles of teacher and learner. The draw was fixed. The accountant was an actor and always became the learner, strapped into a chair in the next room with an electrode on his wrist.
The teacher sat in front of a shock generator with 30 switches, labelled from 15 to 450 volts and, towards the end, "Danger: Severe Shock" and "XXX". He read word pairs, tested the learner, and for every mistake was to give a shock one step higher than the last. The learner made many mistakes. At 300 volts he pounded on the wall and stopped answering. If the teacher hesitated, the experimenter, a man in a grey technician's coat, used four prods in order: "Please continue", "The experiment requires that you continue", "It is absolutely essential that you continue", and "You have no other choice, you must go on". No shock was ever given.
The result
Stanley Milgram reported the first condition in 1963 in the Journal of Abnormal and Social Psychology, under the title "Behavioral Study of Obedience"; the paper is at the journal. Of 40 men, 26, which is 65%, went to the last switch. Every one of the 40 went to at least 300 volts. Before the study, Milgram had asked psychiatrists and students to predict the outcome, and the psychiatrists had estimated that about one in a thousand would go to the end.
The men who obeyed did not do so calmly. The paper describes sweating, trembling, stuttering, and nervous laughter, and one participant who had a seizure. Most protested, asked who would take responsibility, and were told by the experimenter that he would. Then they continued. That combination of distress and compliance is what made the study famous: it was a study of what ordinary people do when an authority asks them to do something they believe is wrong, and the people were visibly unwilling.
The variations
Milgram ran more than twenty variations over 1961 and 1962, with about 40 participants each, and reported them in his 1974 book Obedience to Authority. The rates tell more than the baseline does.
- When the learner was in the same room, 40% went to the end; when the teacher had to press the learner's hand onto the electrode himself, 30%.
- When the experimenter gave his orders by telephone from another room, 20% obeyed, and several participants gave lower shocks than ordered while telling him they were following instructions.
- When the study was moved from Yale to a run-down office in Bridgeport with no university connection, 47.5% obeyed, lower than at Yale but not significantly so.
- When two other teachers, both actors, refused to go on, 10% of participants went to the end. When the participant chose the shock level himself, one man in 40 went to the maximum.
- Women obeyed at the same rate as men, 65%.
The pattern is consistent. Obedience fell as the victim became closer and as the authority became more distant or less legitimate, and it collapsed when the participant saw others refuse. Those are the findings that the social psychology in Unit 7 of the masterclass is built on, together with Asch's conformity lines.
The ethics
In 1964 Diana Baumrind published a critique in the American Psychologist, free at the journal, arguing that the participants had been put through severe stress without consent, that the debriefing could not undo it, and that the deception damaged their trust in psychologists. Milgram replied that 84% of participants, surveyed afterwards, said they were glad to have taken part, and that a psychiatrist who interviewed the most distressed found no lasting harm.
The archives have complicated his reply. Gina Perry, who read the Yale records for her 2012 book Behind the Shock Machine, found that many participants were not told the shocks were fake until months later, when a report was posted to them, and that some left the laboratory believing they had hurt a man. Whatever the right verdict on the study, it could not be run today in its original form, and the rules of informed consent and debriefing that every ethics board now applies were written partly in response to it.
The 2009 replication
Jerry Burger found a way round the ethical problem. Milgram's data showed that almost everyone who went past 150 volts, the point at which the learner first demanded to be released, went on to the end. So Burger ran the procedure at Santa Clara University in 2006, with careful screening and a clinical psychologist present, and stopped every participant at 150 volts. Of 70 participants, 70% were prepared to continue past that point, against 82.5% at the same point in Milgram's comparable condition, a difference that was not statistically significant. The paper, "Replicating Milgram: Would People Still Obey Today?", appeared in the American Psychologist in 2009 and is at the journal. Forty-five years, a different coast, and a far more sceptical public had changed the rate by about 12 percentage points.
What the recordings say
Milgram explained his results by an agentic state: the participant hands responsibility to the authority and stops judging the act. The recordings tell a different story. Alexander Haslam, Stephen Reicher, and colleagues analysed what happened after each prod and found that the fourth, the only one that is a bare order, "You have no other choice, you must go on", was never obeyed: every participant who heard it stopped. The prods that worked were the ones that appealed to the experiment and to science. Participants who continued did not think of themselves as following orders; in their post-study comments, those who obeyed most fully were the ones who most identified with the scientific goal of the study.
Their reading, called engaged followership, is that Milgram's participants were not passive instruments of authority. They were doing what they believed a worthwhile enterprise required, persuaded by a man who seemed to know what he was doing. The argument is set out in a 2012 paper in PLOS Biology, "Contesting the Nature of Conformity", which is free to read, and it applies to the Stanford prison experiment as well. It makes the result more disturbing, because people who have accepted a cause are harder to stop than people who are only afraid.
What stands
The basic finding has held up across replications, countries, and decades: a majority of ordinary people, in that situation, will go further than anyone predicts. The explanation has moved from obedience to persuasion and identification, the variations matter more than the headline figure, and the ethics of the study are no longer defensible by the standards it helped to create. All three are part of the lesson, and a student who knows only the 65% knows the least interesting part.


