Asymmetric dominance effects in choice experiments and contingent valuation