Milgram’s Obedience Experiment – Strengths and Limitations

Last Updated on October 9, 2026 by Karl Thompson

Milgram’s obedience experiment found that two-thirds of ordinary men would give what they believed was a potentially lethal electric shock to a stranger, simply because a man in a lab coat told them to. This post covers:

1. The aim, sample and procedure of the original 1963 experiment
2. The findings, and why they shocked everyone, Milgram included
3. Later repeats: Burger (2009) and two TV versions
4. The strengths of the study
5. Its limitations, including evidence from Milgram’s own archive that changes how it should be read
6. The ethical problems
7. How to use Milgram in A-level sociology

Milgram’s experiment is one of the two classic examples of a laboratory experiment used in the research methods part of A-level sociology. It is also useful for crime and deviance, particularly state crime.

Aim

Stanley Milgram, a psychologist at Yale University, wanted to find out how far ordinary people would go in obeying an instruction from an authority figure if it meant harming another person. He was interested in how ordinary people could be drawn into committing atrocities, with the Holocaust very much in mind.

Sample

Milgram recruited 40 men aged between 20 and 50 from the New Haven area of Connecticut, through newspaper adverts and direct mail. Their jobs ranged from unskilled to professional. They were paid $4.50 for turning up, and told they could keep the money whatever happened.

They believed they were taking part in a study of memory and learning. This was the first deception.

Procedure

Each participant met a second “volunteer”, who was in fact working for Milgram. The two drew lots to decide who would be the “teacher” and who the “learner”. The draw was rigged so the real participant was always the teacher.

The learner was strapped into a chair in a separate room with electrodes attached to his wrist. The teacher sat in front of a shock generator with 30 switches, running from 15 volts (labelled “slight shock”) to 450 volts (labelled “danger – severe shock” and, beyond that, simply “XXX”). The experimenter, a man in a grey lab coat, stayed in the room with the teacher.

The teacher read out word pairs, then tested the learner on them. Every time the learner got one wrong, the teacher had to give him a shock, going one switch higher each time.

The learner gave mostly wrong answers on purpose. No shocks were real, but the teacher did not know that. At 300 volts the learner banged on the wall; after 315 volts he went silent and stopped answering.

If the teacher hesitated, the experimenter used a set series of four “prods”:

1. “Please continue.”
2. “The experiment requires that you continue.”
3. “It is absolutely essential that you continue.”
4. “You have no other choice, you must go on.”

If the teacher still refused after the fourth prod, the experiment stopped. Otherwise it ended when the teacher had given the 450-volt shock three times.

The video below is quite long, but you can selectively watch it to get an idea of the procedure (which is outlined below)

Findings

26 of the 40 men (65%) went all the way to 450 volts. Every one of them went to at least 300 volts.

Nobody expected this. Before running the study, Milgram asked students and colleagues to predict what would happen, and the general view was that only a tiny minority — around 1 to 3% — would go to the end.

The participants did not obey calmly. Milgram described them sweating, trembling, stuttering, biting their lips and digging their nails into their flesh. Many had fits of nervous laughter, and three had full seizures. Many protested and said they wanted to stop — and then carried on when told to.

Milgram went on to run a series of variations, changing one part of the situation at a time to see how it affected obedience. Obedience fell when the learner was in the same room, when the teacher had to physically hold the learner’s hand on the shock plate, and when the experimenter gave his orders by telephone rather than in person. In other words, the closer people were to the victim and the further from the authority figure, the less they obeyed.

What Milgram concluded

Milgram concluded that ordinary people will go a very long way in obeying an authority they see as legitimate, even when it means seriously harming someone. He later argued that people enter what he called an agentic state: they stop seeing themselves as responsible for their actions and see themselves instead as carrying out someone else’s wishes.

From a sociological point of view, the important thing is that obedience is learned. We are socialised into obeying legitimate authority in the family, at school and at work, and the participants were responding to a social role — the scientist at a respected university — rather than to anything about their own personalities.

Later repeats of the Milgram experiment

Burger (2009)

Jerry Burger ran the most important modern version, with 70 adults (29 men and 41 women). Ethical rules no longer allowed the full procedure, so Burger stopped every session at 150 volts — the point where, in Milgram’s study, the learner first cries out in protest and most participants hesitate. 70% of Burger’s participants were willing to continue past that point, compared with 82.5% in the equivalent Milgram condition, a difference Burger found was not statistically significant. He also found no difference between men and women.

BBC Horizon (2009)

The BBC’s Horizon repeated the experiment in an episode called “How Violent Are You?”. Of 12 participants, 9 reportedly went all the way to 450 volts.

Derren Brown, The Heist (2006)

Derren Brown ran a version as part of The Heist, first shown on Channel 4 in January 2006, to test how his participants responded to authority before deciding who to take forward to the “robbery”. Interestingly, one participant says they can’t carry on because they’ve heard of the experiment (even though they seem to have started it), while another who hadn’t heard of it comments that they needed more buttons on the shock machine. Brown also decided not to take one participant further because she took too long to recover.

Strengths of Milgram’s experiment

Control and cause and effect. The procedure was the same for every participant: the same script, the same prods, the same responses from the learner. That allowed Milgram to vary one thing at a time across his variations and show what caused obedience to rise or fall.

High reliability. Because the procedure was standardised, it can be repeated, and it has been. Burger’s 2009 results were very close to Milgram’s, almost fifty years later.

Quantitative, precise results. The voltage reached gives a clear, comparable measure of obedience.

It is still a useful wake-up call. The findings challenge the comforting idea that only unusual or evil people do terrible things on orders, and they sit awkwardly with theories such as the flight from deference — the claim that people in modern societies no longer simply defer to authority.

Limitations of Milgram’s experiment

Low external validity. The setting was completely artificial. When were you last asked by a scientist in a lab coat to give electric shocks to a stranger? Authority in real life comes from teachers, police officers, managers and officials, in settings people understand, and people may respond quite differently. A Yale laboratory also carried a lot of prestige.

Did they believe it? Milgram reported that about three-quarters of participants believed the shocks were real — but that figure includes the 24% who said they had “some doubts”. If a large share of participants suspected a set-up, the results say less about obedience than they appear to.

An all-male, single-town sample. All 40 participants in the original study were men from one American town in the early 1960s. Milgram did later test women, and Burger found no gender difference, but generalising from the original sample to “people” is still a leap.

Awareness of the study. The Hawthorne effect — people changing their behaviour because they know they are being studied — is nicely illustrated in the Derren Brown clip above. (Seriously, has not EVERYONE heard of the Milgram experiment by now? Or is that just my middle-class cultural bias?) Any modern repeat runs into the fact that the study is famous.

It may not be about obedience at all. Haslam, Reicher, Millard and McDonald (2015) went through participants’ replies to Milgram’s follow-up questionnaire in the Yale archive. They found that most participants were glad to have taken part, because they had been persuaded they were contributing to important science. On this reading, people went on not because they blindly obeyed, but because they identified with the scientist’s goals — what the researchers call engaged followership. People do harmful things not because they don’t know they are doing wrong, but because they have been convinced it is right. That is an even more sociological conclusion than Milgram’s.

Ethical problems with Milgram’s experiment

Deception and no informed consent. Participants were told it was a study of learning, the draw was rigged, and the shocks were fake. They could not consent to something they had been misled about.

Harm. The study caused extreme stress, as Milgram’s own account of sweating, trembling and seizures shows. Diana Baumrind criticised it on exactly these grounds in American Psychologist in 1964.

The right to withdraw was undermined. The prods were designed to stop people leaving. “You have no other choice, you must go on” is the opposite of telling someone they can withdraw at any time.

HOWEVER, there is a defence. All participants were told afterwards that the learner had not been hurt. Milgram’s follow-up found most said they were glad to have taken part, which Haslam and colleagues’ archive work confirms. One participant later wrote to Milgram that the experience had made him realise how easily he submitted to authority, and that it led him to apply for conscientious objector status rather than be drafted. Milgram argued that what the study revealed justified the distress it caused. Whether that argument works is itself a good evaluation point.

[IMAGE: retain existing — milgram-experiment-ethics.jpg · alt text: “A participant in Milgram’s obedience experiment showing signs of distress”] One of the original participants showing signs of distress

Using Milgram in A-level sociology

For research methods, Milgram is the standard example of a laboratory experiment. It shows the strengths (control, reliability, quantitative data) and every major limitation (external validity, demand characteristics, sampling, ethics) in one study. Pair it with the Bobo doll experiment and you have the two classic lab examples. For the method in general, see Laboratory Experiments in Sociology.

For crime and deviance, it helps explain state crime and crimes carried out within organisations: ordinary people can be drawn into serious harm when instructions come from a legitimate authority, which is why the “only following orders” defence matters so much in discussions of genocide and torture.

For theory, it supports the idea that behaviour is shaped by social roles and socialisation rather than individual personality. The engaged followership reading pushes this further: people need to believe in what they are doing, which is a point about meaning that Interpretivists would recognise.

Theory and Methods A Level Sociology Revision Bundle

Theory and Methods A-level sociology revision bundle cover

If you like this sort of thing, then you might like my Theory and Methods Revision Bundle – specifically designed to get students through the theory and methods sections of A-level sociology papers 1 and 3.

Contents include:

1. 74 pages of revision notes
2. 15 mind maps on various topics within theory and methods
3. Five theory and methods essays
4. ‘How to write methods in context essays’

Signposting and Related Posts

This post has been written primarily for students studying research methods as part of the AQA’s A-level sociology course.

Milgram’s experiment is an example of a laboratory experiment, so the place to start is Laboratory Experiments in Sociology, which covers the strengths and limitations of the method in full.

The Bobo doll experiment is the other classic laboratory experiment used in sociology.

The Stanford Prison Experiment is a third, and a good example of how a laboratory study can go wrong.

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.

Sources: Baumrind, D. (1964) American Psychologist (critique of Milgram’s ethics). Burger, J. M. (2009) ‘Replicating Milgram: would people still obey today?’, American Psychologist. Haslam, S. A., Reicher, S. D., Millard, K. and McDonald, R. (2015) ‘”Happy to have been of service”: the Yale archive as a window into the engaged followership of participants in Milgram’s “obedience” experiments’, British Journal of Social Psychology, 54, pp. 55–83. Milgram, S. (1963) ‘Behavioral study of obedience’, Journal of Abnormal and Social Psychology, 67, pp. 371–378. Milgram, S. (1974) Obedience to Authority: An Experimental View, Harper and Row.

Please click here to return to the homepage – ReviseSociology.com