The biggest flaw in that logic is that, to test people's supposed general reactions, playing music really loudly and messing with lighting is... basically just being an asshole to a bunch of people trying to get dinner. So instead of getting an idea how people react in bad situations, you get how they deal with co-workers nobody likes. 

And the most ineffective part is doing this somewhere you'll NOT get a serious response. It's not like people are going to react to loud music in a cafeteria as a serious threat.