The Milgram Experiment Wasn't One Study. It Was Two Dozen.

Stanley Milgram's obedience study has been retold so often it has hardened into a single statistic: 65 percent of ordinary people, told to administer dangerous electric shocks to a stranger by a man in a lab coat, did so. The figure entered textbooks within a year of the original 1963 paper in the Journal of Abnormal and Social Psychology and became one of the most-cited results in social psychology.

The figure is real. It is also one of two dozen. Milgram ran 24 separate variations of the protocol at Yale between 1961 and 1962, with around 780 subjects in total. The 65 percent number comes from a single condition: 26 of 40 male volunteers fully obeyed when the experimenter sat in the room and the "learner" was in another room banging on the wall. Move the experimenter to a phone call and full obedience drops to about 21 percent. Run two confederate teachers who refuse, and it drops to ten. Pair the subject with a friend instead of a stranger as the victim — and obedience falls to roughly 15 percent. In one variation no one shocked to the maximum at all.

In 2012 the Australian psychologist Gina Perry published Behind the Shock Machine, based on Milgram's papers at Yale and on interviews with surviving participants. She found that Milgram had reported only a handful of his variations in detail and selected the most striking conditions for popular write-up. Tape recordings showed many subjects suspected the shocks were fake and pushed the lever anyway, complicating the clean "banality of evil" reading. She also documented that many participants left New Haven without proper debriefing, and some carried the belief that they had truly hurt someone for years.

The textbook version of Milgram is not wrong — it is one frame from a much longer film, presented as the whole picture.