I think you're underestimating the usefulness of scapegoats. If biased results exist, a lot of people look more effective. If such a bias is discovered, only a subset of those people will be blamed.