I think there are several flaws with the poll, but I've been working 12+ hour days so I may not be reading / thinking right.
First : Asking whether units were underpowered / useful / overpowered, without mentioning whether people should be compared to normal, recruitable troops, or considering summoned creatures and whatnot. ( On the one paw, a person might think it was normal recruitable troops, but on the other paw, banefire archers were included in the poll. )
Second : All the questions about "what damage should X be :" were .... Well, lets just say that I didn't have time or energy to look up what their damages
currently were, so simply picked choice #1 for each and every one of those. From a methodology POV, too many questions is a
bad thing, especially when all questions need to be answered to see the results. Especially when giving any old answer for them gives the instant gratification of seeing the results. Bad wolf, no cherry. Bad Cherry, no wolf.