Rethinking Response Rates: New Evidence of Little Relationship Between Survey Response Rates and Nonresponse Bias

BACKGROUND: Federally funded evaluation research projects typically strive for an 80% survey response rate, but the increasing difficulty and expense in reaching survey respondents raises the question of whether such a threshold is necessary for reducing bias and increasing the accuracy of survey estimates. OBJECTIVES: This analysis focuses on a particular component of survey methodology: the survey response rate and its relationship to nonresponse bias. Following a review of the literature, new analysis of data from a large, multisite random assignment experiment explores the relationship between survey response rates and measured nonresponse bias. RESEARCH DESIGN: included 12,000 individuals who were fielded for 16 identical surveys as part of the Employment Retention and Advancement evaluation. RESULTS: The results suggest scant relationship between survey nonresponse bias and response rates. The results also indicate that the pursuit of high response rates lengthens the fielding period, which can create other measurement problems. CONCLUSIONS: The costly pursuit of a high response rate may offer little or no reduction of nonresponse bias. Achieving such a high rate of response requires considerable financial resources that might be better applied to methods and techniques shown to have a greater effect on the reduction of nonresponse bias.