As you might recall, the Stanford Prison Experiment pitted several student against each other as randomly chosen prison guards or prisoners. The end result was quite gruesome. Guards enforced authoritarian measures and even inflicted psychological torture on the prisoners. The experiment is a classic note in psychology texts and has often been used to explain the Nazis, Abu Ghraib, and other forms of inhumane practices.
Apparently, there are serious flaws with the study. This post at the British Psychological Society details some of the issues:
The SPE was criticised back in the 70s, but that criticism has noticeably escalated and widened in recent years. New details to emerge show that Zimbardo played a key role in encouraging his “guards” to behave in tyrannical fashion. Critics have pointed out that only one third of guards behaved sadistically (this argues against the overwhelming power of the situation). Question marks have also been raised about the self-selection of particular personality types into the study. Moreover, in 2002, the social psychologists Steve Reicher and Alex Haslam conducted the BBC Prison Study to test the conventional interpretation of the SPE. The researchers deliberately avoided directing their participants as Zimbardo had his, and this time it was the prisoners who initially formed a strong group identity and overthrew the guards.
Given that the SPE has been used to explain modern-day atrocities, such as at Abu Ghraib, and given that nearly two million students are enrolled in introductory psychology courses in the US, Richard Griggs, professor emeritus at the University of Florida, says “it is especially important that coverage of it in our texts be accurate.”