One of my best friend's and fellow Ribeiro black belt Jason Clarke got into a discussion about the effectiveness of a Brazilian Jiu-Jitsu armbar in a self defense situation. This all started off with a video of a parking lot fight where one person attempted an armbar on another person and the one person escaped because they didn't know a counter when someone stood out of the armbar.
Jason's take on the situation was, like him, pretty straight forward. He said "Armbar's don't end fights, just matches". I thought that it was a pretty strong take on the move, but his point was that the video we were watching wasn't a fight, it was a "monkey dance". The people who were fighting weren't really trying to kill each other, they were drunk and peacocking or in a primitive sense showing off for their mates. His point was that this wasn't a violent criminal assault where one or more people were trying to kill or steal from someone else. If this were a life or death situation, then the armbar wouldn't be effective because a criminal could fight through the pain and still continue to get weapons.
I always value Jason's opinion highly. He has been training BJJ since 1994 and was a blue belt when I started training at the University of Michigan club. He was also an army ranger and has been in life and death situations. Now I thought that while an armbar isn't the perfect move for every situation, it is has been shown to be highly effective and to discount it just because it won't kill or make your opponent unconscious was a bit too much. So I pointed out the military study where 20% of infantryman ended up in a close quarters situation, 75% of them used grappling in that situation and close to 20% of those people used an armbar, which was more than than used chokes and way more than all striking techniques combined.
Jason didn't think the study was perfect because army situations usually used overwhelming force so hopefully there were more US soldiers in a situation than enemies. The environment a soldier will use hands on techniques is usually quite different than what a civilian who is legally defending themselves will be in. A soldier will most likely be operating in teams of two or more (Fire Team = 3 - 5 people, Squad = ~12 people). Very rarely are they operating alone or far from support of more teammates. A civilian on the other hand will most likely be alone and the odds of them being outnumbered in the direct engagement are greater based on what is known of the criminal assault paradigm. So, typically, the Soldier usually has superiority in numbers where the civilian will either be equal matched or at a numerical disadvantage. But I don't think you are ever going to get a study where you have a civilians vs criminals and the civilians are trained enough to show the value of said training. He also pointed out again that techniques are limited by the possibility of multiple attackers, weapons, pain intolerance, or that a certain percentage of the population are not affected.
My final point was that there is a ton of evidence showing that this move works on resisting opponents, even when there are the group dynamics and weapons of armed combat. Not every self defence situation is going to require a lethal response and you are going to have to make use of force decisions as a civilian. We can get into the hypothetical of every criminal being an armed pcp junkie with friends behind every corner, but then the only response is going to be a possibly lethal one. I think you have to train moves that have shown to be effective, that you can practice repeatedly against fully resisting opponents. Worrying about every hypothetical can lead you down a pretty crazy rabbit hole of self defense that I don't think is great use of time.