How robust are predictions of galaxy clustering?