Yes, you´re mistaken. What you need is a sample that is representative of the whole gamer population. A minority is indeed enough to generalize from when that minority is comparable to the majority in diversity and composition.
So, if you poll ENworld, you can generalize stuff about hard-core D&D DMs and players that post on the internet. Not the D&D gamer population as a whole.
My group play 3e and 4e, with a little M&M2e on the down-low.
WoD actually has a much larger reader-base than player-base. This is subjective, of course, but from personal experience I see WoD books fly off the shelves and almost never being played.
WoD actually has a much larger reader-base than player-base.