Your argument here largely boils down to, "you are biased because anecdotal evidence." If you knew nothing about heroin, and then watched a bunch of people get addicted, you wouldn't say, "Well were they rich or poor? Could have been a biased sample set."