The Threat of Nonresponse Bias to Survey Research

Abstract

Mail and telephone surveys have become the major research tools of social scientists, yet low response rates and the resulting nonresponse bias continue to be major detractors and concerns. This study quantifies the impact of nonresponse bias associated with a community survey. More specifically, an intensive restudy of nonrespondents to an earlier contingent valuation method survey was undertaken that increased the response rate from 59 to 87 percent. Traditionally used methodologies for assessing nonresponse bias are presented, followed by some newly proposed and more comprehensive procedures for determining differences among respondents and nonrespondents. On the basis of these new procedures, “willingness-to-pay” values were recalculated and then compared to the original estimates. The results indicate that nonresponse bias can be quantified and that, contrary to much of the present literature, nonrespondents can have valuations for a particular resource that are higher than those of respondents. Possible implications of these findings are advanced.