I'm sorry, this experiment has been reproduced numerous times and the initial study that produced this
WAS rigorously tested
AND peer reviewed. Milgram performed 19 variations of his experiment before his peer reviewed journal article was ever published, it studied the distance from authority, the distance from the subject and the authority level to measure what percentage of people performed inhumane acts.
The variables ranged from the authority being virtually over the subjects shoulder to being on a telephone. The proximity to the victim ranged from a one-way mirror to only being able to hear the person over a speaker. The level of authority went from a white coat to blue collar.
The subjects were initially polled to see how many believed they would administer the lethal voltage (1.2%), and also polled his colleagues (who came up with a similar percentage on how many people they expected would go all the way).
A later meta-analysis performed on multiple repeats of the study found the percentages to remain steady at 61-66%.
Zimbardo (who performed the Stanford Prison Experiment) made an excellent point: none of the participants who refused to administer the final shocks insisted that the experiment itself be terminated, nor left the room to check the health of the victim without requesting permission to leave.
Despite disobeying authority not to hurt someone, they still required permission of the authority to help someone, which is quite possibly the more worrying aspect that is never highlighted.