A few years ago I watched a short documentary on the Stanford Prison experiment. It's purpose was to simulate the psychological affects that occur to prisoners. The head researcher interviewed dozens of young males so that he could end up with a group of 24, perfectly sane, middle class men.
He divided them randomly into prisoners and guards, and it was done without bias. The, "prisoners" were arrested by the actual police department in Palo Alto, and then they were brought to the basement of the psychology building, where the experiment took place. Immediately the prisoners were put into their characters. At first they didn't take it too seriously, and neither did the guards. However, after the first day both groups were already believing their roles. Prisoners started to stress out, and the guards became more abusive.
Over a period of six days(which was the premature ending), The prisoners had lost all sense of identity, became submissive, and forgot they were in an experiment. The guards had become abusive, and committed enforcement against the prisoners in the middle of the night when the researchers weren't around. I remember in the documentary, that one of the prisoners had to threaten to sue in order to get out. This does not surprise me, seeing as the head researcher had begun emerging into his own role.
My question is, how valid or ethical can some research be if the situation is simulated and all of the participants and researchers are absorbed in their role. The experiment didn't have a party that simply observed, seeing as the creator of the experiment was the superintendent of the prison. The data was interesting, and paralleled a lot of prisons, but can the results really have the same contexts and reasoning behind them? How safe can the participants be if the researchers can't maintain control of the situation and differentiate reality from acting. It's just something to think about for a lot of experimentation