Sample bias in contingent valuation: A comparison of the correction methods