Abstract Sample nonresponse and selection biases that may occur in survey research such as contingent valuation applications are discussed and tested. Correction mechanisms for these types of biases are demonstrated. Results indicate the importance of testing and correcting for unit and item nonresponse bias in contingent valuation survey data. When sample nonresponse and selection bias go uncorrected, welfare measures may be overestimated or underestimated contributing to potential errors in resource policy and management decisions.