Click Here For Latest Update Please

Behavioural Study of Obedience

Stanley Milgram


Many atrocities had been committed in the Second World War. Many culprits were put on trial for their lives in Nuremberg, at the end of the war. The chief defence was that they were only following orders from somebody above. This argument, followed to it's extreme would absolve every person involved in an atrocity, leaving only Hitler as the only person who could be found guilty (as he took orders from no one)! Naturally, with feelings running high at the end of the war, this defence was rejected outright, and many defendants were found guilty and hanged. Popular opinion for some time after the end of the Second World War was that there was something in the German character that made them particularly cruel. This is a dispositional view. People are genetically determined to act in a certain fashion, or are brought up to act in a certain fashion.

Milgram questioned this dispositional view of the German character. He felt that the situation that many people had found themselves in had led to their cruel behaviour. It would follow from this, that most people, regardless of nationality, would perform cruel acts upon another, under certain situations. The defence for many of the war criminals had been they had been following orders. Milgram, reasoned that people could commit atrocities when they are given orders by somebody in authority.

Details of Experiment

Milgram tested his hypothesis by using a laboratory experiment. The subjects were chosen from volunteers who had responded to a newspaper article.

This means the sample was self-selecting. We must question whether or not Milgram had a representative sample, by using this study. Milgram chose to study only men, but from a variety of backgrounds and different ages. You might say that by using men this produced a sample that was biased, or did not reflect the general population. Men are thought to be more aggressive than women, so it would make sense to begin a series of experiments with them. Many of the war criminals had been men, so Milgram's sample was representative of the target population. Milgram did in a later experiment use only women, and achieved similar results to those produced by the men.

The men were paid to participate in the experiment. Again you would be misguided to say that the subjects were only obeying in order to receive the payment. Milgram's experimenter (Milgram had no direct contact with the subjects) made it quite clear that they were paid the money just for appearing at the laboratory, and they were free to leave at anytime without forfeiting the money. Having said this though, the experimenter did not remind the subjects of their right to leave subsequent to this initial assurance. Later in the experiment the subjects were to be told by the experimenter the exact opposite "You have no choice, you must go on". This conflicts with ethical guidelines; subjects have the right to withdraw from any experiment without censure at anytime.

The experiment was conducted at Yale University. Again we should question this as a prestigious setting could add to the authority of the experimenter, and the willingness for the subject to obey. Milgram was aware of this possibility, and in a subsequent variation of his experiment, located in a shabby downtown office, found that the level of obedience was slightly less than the level reported in this original experiment.

A stern looking male experimenter was used (a secondary school teacher). A mild mannered man played the part of the victim. Milgram chose a likeable inoffensive character to play the part of the victim, probably so there could be no apparent motive to harm him. We should question this, and ask whether appearing mild-mannered might encourage bullying behaviour. Perhaps Milgram should have chosen a victim who appeared to be able to take care of himself, yet appear friendly at the same time (a likeable hulk!).

You can not just say to a subject that you are going to test their readiness to obey an authority figure. A typical subject might well deliberately put up a resistance to any pressure put upon him to obey. It was therefore necessary to deceive the subject, by telling the subject that this was an experiment into the effects of punishment upon learning. Again this breaks the ethical guidelines, but we should not be too hard on Milgram unless you can think of a feasible alternative way of overcoming this problem of 'Demand Characteristics'. So at the beginning of the experiment the real subject is introduced to another 'subject' (really the victim who is working for Milgram). It is explained that one needs to be the teacher and the other the learner (or victim). A draw is fixed so that the real subject plays the part of the teacher. The victim is strapped into a chair. His wrists are covered with an electrode paste, and electrodes are placed upon the paste. The subject (teacher) is left in no doubt that the learner (victim) can not escape receiving electric shocks. The subject is taken into an adjoining room and shown the 'shock generator'. The subject is led to believe that this machine can deliver shocks from 15 volts through to 450 volts to the victim, when really it produces nothing except an impressive electrical noise and a flashing blue light. The 'shock generator' has a switch for every voltage between 15 and 450 volts increasing in steps of 15 volts. Each switch is labelled, so as to give an impression of how severe the shock is (for example, 'danger: severe shock'). The subject is offered a real sample shock of 45 volts (delivered from a battery connected to the appropriate switch on the generator). All of the foregoing is designed to convince the teacher that he is really giving shocks to the subject. The instructions are simple. The teacher reads a list of word pairs. The learner has to learn these (known as 'paired-associate learning'). The teacher then tests the learner by giving him one of the words in a pair. The learner has to select the given word's paired word from four alternatives given by the teacher. If the learner performs at chance level, then he should get one in every four answers correct. If the learner gets it right then the next question is given without any punishment. If the learner gets the answer wrong, then he is administered a shock that is 15 volts higher than the voltage of the last shock delivered. Note the voltages are not decreased after a series of correct answers are given. You can work out that under these instructions it would be only a matter of time before any learner, regardless of ability, would receive the maximum shock of 450 volts! The victim has been asked by Milgram prior to the experiment to perform at chance level. For those interested in the technicalities of this experiment (and that should be all of you Oxford and Cambridge students), the strapped down victim can indicate his answer by pressing one of four conveniently placed levers. Each lever is connected to one of four indicator lights situated in the teacher's room. In this original experiment the victim is unable to talk to the teacher because of the intervening wall. The teacher communicates with the learner with a one way intercom. A preliminary run is given. The learner gets only three correct out of ten and thus receives seven shocks. The maximum being 7 times 15 volts, which is 105 volts. In any experiment it is usually a good idea to allow your subject to have a dummy run. This allows the subject to get used to following the instructions, and gives them a chance to warm up and produce a reasonable performance. The experimenter can also make sure that the subject is following his instructions carefully, and the experimenter is given a chance to correct any problems. The victim gives responses after each question, until the 300 volt shock. At this point pounding on the wall is heard and no response to the question is received from the victim. As you might have expected subjects turned to the experimenter for guidance at this point. The experimenter instructs the teacher to treat the absence of a response as an incorrect answer. The victim pounds on the wall (you will gather by now his legs are not bound!) at 315 volts, and continues not to answer the questions. At higher voltages the victim gives no response whatsoever, giving the impression that he is at the very least unconscious, and at the worse dead! Naturally, subjects will turn to the experimenter for guidance, before administering shocks of greater than 300 volts. The experimenter gives one of four scripted 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.

The experimenter would initially use prod one, but if the subject refused to continue, the experimenter would try prod 2, and then prod 3 and finally prod 4, breaking off as soon as the subject continued. (How does this accord with the ethical point of allowing the subject to withdraw?) In the event of the subject asking about any danger to the victim's health, the experimenter would give the following scripted prod: "Although the shocks may be painful, there is no permanent tissue damage, so please go on". If the teacher pointed out that it would seem the learner wanted to withdraw from the experiment, the experimenter would say "Whether the learner likes it or not, you must go on until he has learned all the word pairs correctly. So please go on". The experiment would end either when the 450-volt shock had been administered, or when the subject walked out.

Ask yourself the following:

The dependent measure in this experiment is the maximum voltage the subject will go to before walking out, with a maximum voltage of 450 volts recorded for obedient subjects.


Psychology undergraduates were asked what percentage of subjects would continue to give shocks up to the maximum of 450 volts. The mean percentage given was 1.2%. In fact, 26 out of the 40 subjects continued to 450 volts. Only 5 dropped out at 300 volts when the pounding on the wall was heard. A further four dropped out at 315 volts.

On a 14 point scale, subjects indicated that the intensity of shock was 13.42 (mean), which was labelled 'extremely painful'.

Many subjects became extremely nervous. Evidence for this was sweating, trembling, stuttering, biting lips, groans, digging fingernails into their flesh. Fourteen subjects demonstrated nervous laughter. Three subjects had seizures. Further qualitative data reported by Milgram were the comments made by the subjects. In short, although many subjects administered shocks up to 450 volts, they experienced acute stress. It is interesting to draw a parallel with the Nazi execution squads, who were given extra rations of alcohol, presumably to counter the stress that their terrible acts produced within themselves.

Discussion points

It would seem that it is the situation that has produced these results and not the disposition of the subjects. The Yale undergraduates who predicted that only very few (perhaps psychopathic) individuals would administer shocks of 450 volts, were guided by their understanding of the morality that guides human behaviour. They were not in the situation. Observers looking at the experiment in progress could not believe what they were seeing; again, they were not in the situation. We judge people outside of the situation surrounding their action(s).

Other key studies that support these results are Zimbardo and Nisbett. Zimbardo also provides evidence to suggest it is the situation that people find themselves in rather than their disposition that best explains their actions. Nisbett provides evidence to suggest we view the actions of others as though they are guided by their disposition; yet we view our own actions as being the result of the situation we were in. The main conclusion, however, simply is that people tend to obey others in authority. We should not be unduly concerned nor surprised about this fact, because if few people obeyed orders, very little would get done, and a relatively complex society could not exist.

Milgram's experiment breaks several of the ethical guidelines. Milgram deceives his subjects, by misinforming them about the true purpose of the experiment and by making them believe they are administering real electric shocks to a real subject. We would have to balance any criticisms with a consideration as to the necessity of deception. The main problem is one of demand characteristics, whereby if a subject knows the true purpose of an experiment, he might behave differently. It was good that Milgram stated at the start that the money paid to the subjects was theirs regardless of whether they continued with the experiment; However, during the experiment the prods used suggested that withdrawal was not possible for the subject. This is ethically incorrect. Even so, we should consider whether the experiment would have been valid if the experimenter kept reminding the subject about his right to withdraw. Many of the subjects were suffering from an enormous amount of stress, and this conflicts with the ethical principle of protecting the participant. Perhaps Milgram should have stopped the experiment as soon as the subject appeared to be suffering; But what sort of results could be recorded? Remember the perception of suffering is to some degree subjective, and therefore would not be a reliable indicator, if the voltage reached at abandonment of the experiment was to be used as the dependent variable. Some will argue that Milgram was not expecting the results that he did get, since many of his colleagues and students doubted whether many subjects would continue through to 450 volts. This is a weak argument, as this may explain why one or two subjects were allowed to suffer stress, but this does not explain why all forty subjects were allowed to suffer, or why Milgram repeated his experiment many times. In one of Milgram's subsequent experiments the subject was asked to force an arm of the struggling victim onto the electrodes!

We should consider whether the experiment was ecologically valid. The subjects may well have obeyed the experimenter because they accepted that he knew best. Remember the subjects were reassured that the shocks were not harmful. A more ecologically valid experiment by Hofling et al, 1966, suggested that Milgram's results were valid. In Hofling's experiment, nurses in a hospital were asked over the phone by a bogus doctor to administer an overdose of a drug without obtaining authorisation. Twenty-one out of twenty-two nurses attempted to administer the drug (which, unknown to the nurses, was really glucose).

Milgram's experiment was deliberately ethnocentric, as he was interested to see whether Americans would be as obedient as German subjects. The results obtained from the American sample meant it wasn't necessary to test the German sample, as the Americans were expected to refuse the orders of the experimenter. Other psychologists have tested subjects from many different cultures and found that generally people regardless of culture tend to obey people in authority.

Ethical considerations in Milgram are:

Consent - not informed.  Participants volunteered for a learning experiment not one on obedience.

Deception - As to nature of experiment.  Shocks not real. Victim not really a participant.

Asking colleagues - At first colleagues were asked and they said that the experiment would not cause any psychological problems.  However, Milgram
could have stopped at this point, but he went on to conduct numerous other trials.

Protection of participants - the participants displayed symptoms similar to that of a nervous breakdown.

Withdrawal from the investigation - physically yes the participants could, but with strong social pressure on them this was unlikely.  The prods used,
such as 'you have no alternative, you must go on' achieved this.

Observation - participants were filmed without their consent.

However, Milgram did debrief the participants.  I trust Milgram was confidential and the participant's names that he used were made up.

Your questions

1)      Evaluate the problems of generalising in psychology, using Milgrams`
> study of obedience as an example.
> Just think of the artificiality of Milgram's study and you begin to see
the problem of generalising.  Additionally think of the self-selecting
sample as one that is more likely to do as they are told.

2)      After a subject in his obedience experiment repeatedly protested and questioned the experimenter, Milgram commented that, ''he thinks he is killing someone, yet uses the language of the tea table''. What four factors help explain why an individual, such as this man, could become morally disengaged from the consequences of his actions

1 .†† The subject is placed in a position where he must respond to the competing demands of two people: the experimenter and the victim; satisfaction of their demands is mutually exclusive.Also, the resolution must take the form of a highly visible action, either continuing to shock or breaking off the experiment.The subject is forced, therefore, into a public conflict for which there is no completely satisfactory solution.

2.††† While the demands of the experimenter carry the weight of scientific authority, the demands of the victim spring from his Personal experience of pain and suffering.The two claims need not be seen as equally pressing and legitimate, since the experimenter seeks an abstract scientific datum, while the victim cries out for relief from physical suffering caused by the subjectís actions.


3.††† The experimenter gives the subject little time for reflection: it is only minutes after the experiment begins that the victim begins his protests.Moreover, the first Protests are heard after the subject has only gone through two thirds of the shock levels, implying that the conflict will be persistent and may well become more intense as increasingly powerful shocks are given.This may well be a source of tension to the subject.


4.††† More generally, the conflict stems from the opposition of two deeply ingrained behaviour dispositions: first, the tendency not to harm other people, and second, the tendency to obey those whom we perceive to be legitimate authorities.

Return to Gary Sturt's Homepage _