Student evaluations are heavily influenced by perception of the professor's personality. Look on Ninja Courses and you'll see what I'm talking about. Coward has a cult of influence that makes him seem like a close friend to his students, and this no doubt skews his ratings and potentially distorts students' perception of the professor's performance as a teacher (rather than his personality).
There are professors whose teaching quality is actually proficient and sometimes excellent, but whose student evaluations are marred by the fact that they can't speak English well, they have social deficiencies (e.g. Asperger's syndrome), etc. Student evaluations are flawed in the same way intelligence tests are flawed; student evaluations favor professors who speak and communicate most closely to the popular consciousness of students at the time in the same way intelligence tests are biased against non-native test-takers (some questions posed on intelligence tests may use culturally native objects).
There are professors whose teaching quality is actually proficient and sometimes excellent, but whose student evaluations are marred by the fact that they can't speak English well, they have social deficiencies (e.g. Asperger's syndrome), etc. Student evaluations are flawed in the same way intelligence tests are flawed; student evaluations favor professors who speak and communicate most closely to the popular consciousness of students at the time in the same way intelligence tests are biased against non-native test-takers (some questions posed on intelligence tests may use culturally native objects).