The problem I have with something like a trimmed mean is that the data is already very limited. Taking 2 games out of an 8 game sample seems counterproductive in terms of analyzing how a player is playing.
You mentioned not having Kupp as a potential reason there was a drop off in Goff's production. I don't agree with that because he had his best game as a pro against KC when Kupp wasn't in the lineup. No Gurley, Anderson filled in nicely. O-line not playing as well? Based on what? The numbers don't support a significant drop off.
I certainly appreciate considering these other things but each of them don't really hold water IMO. So, I'm left with my theory. Much, if not most of Goff's success since McVay arrived was due to the pre snap communication tactic. Which there's nothing wrong with using. Many teams also use it but it seems that Goff still has much to learn when not having that resource available.
When teams took that away and the Rams weren't able to run plays up tempo, Goff struggled. Much like when he was a rookie. It seems like a plausible lynch pin to me. Take away that pre snap communication and Goff's reads, audibles and line calls/protection calls are going to suffer. So, is it more reasonable to think that there is a list of 6 or 7 different things failing throughout the coaches and players or that this one thing happened which had rippling effects?