That is why the writer specified "on average", which clearly remains true, at least in the case that the decisionmaker is part of the affected group. The optimistic part is in assuming that latter.