I think the biggest unaccounted-for factor is subtypes. In my opinion, that's why individual people tend to have their own idiosyncratic common mistypings like, say, Kindred instead of Benefit, or Supervisor instead of Supervisee, as in your example. A C-LII might be more commonly mistyped as ILE than LSI, whereas an N-LII might be more commonly mistyped as LSI. Maybe someday I'll try to untangle that mess and write a more complicated version of this.

Overall, though, I think my list there is pretty accurate when looking at the big picture of common mistypings. Individuals may differ, but there are certain overall trends that tend to remain pretty constant, at least that I've observed. So I wouldn't say it's an oversimplification so much as a useful generalization.