查看文章

In this paper we information on contingent valuation mail survey respondents and non-respondents and find non-response bias with respect to several determinants of willingness to pay. We use data from a combination phone/mail sampling and survey approach and find significant differences between early and late respondents, late and non-respondents to the mail survey and all respondents and non-respondents to the mail survey. We fail to find support for the contention that if early and late respondents are similar, respondents and non-respondents are also similar. The range of aggregate benefit estimates narrows as response rates increase and non-response bias decreases. A weighting approach is used to correct for some of the non-response bias in aggregate benefits.