Orne and Holland (1966) argue that Milgram’s experiment lacks internal validity and participants believed that the experiment was set up. This could lead to demand characteristics and therefore we don’t get a true reflection of obedience from he results. This is supported by Perry (2013) when she listens to the recording ion the participants responses. She said many of them expressed their disbelief that the confederate was really being shocked. Milgram himself argues this and says he believes at least 70% of his participants believed they were administering real shocks.

Sheridan and King (1972) support Milgram with a replication of his study using puppies and real shocks. From their results they found that 54% of male participants reached the maximum of 450 V and 100% of female participants.

As Milgram’s experiment was a laboratory experiment we would tend to argue it could have good internal validity and can sow cause and effect due to its controls, standardisation and replicability. But some argue that the experimenter would often go off script.

The external validity of this experiment is also highly debated as it was a laboratory experiment and lacked ecological validity as this was not a real life scenario. However, others argue this point as the research focus was on the relationship of the participant and authority figure and that is what is seen in his experiment.

Hofling et al (1966) devised an experiment whereby nurses, on a night shift, were given instructions over the phone by an unknown doctor “Dr Smith” to give an unknown drug “Astroten” to a patient called Mr James. They were told to give 20mg to the patient when on the label of the drug gives maximum dosage at 10mg. Dr Smith also told the nurses that he was in a hurry and would sign the authorisation later when he came to see the patient. Nurses have strict guidelines and should never take orders over the phone and should be mindful of drugs they do not know and dosage limits. Even with this 21 out of 22 nurses went to give this fake drug to the patient.

The drug was a placebo and harmless but this shows the social pressure and power of those as authoritarian figures and how people may obey even with strict guidelines. The experiment was highly standardised and reliable with the orders over the phone followed by a script.  This experiment was compared to a control group of 33 nurses in a different hospital and a questionnaire on how they would respond to the request. 31 out of the 33 said they would not comply with the instructions. We also have to be mindful about social desirability bias when using questionnaires. 

Rank and Jacobson (1977) argue that Hofling’s experiment also lacks ecological validity as this is not a real life situation. They ran a similar study but this time the doctors gave orders in person and a drug they were used to was used, valium. So when doctors asked nurses to give 3 times the recommended dosage only 2 out of 18 nurses obeyed. Nurses were stopped before they could give the drugs to any patients.