Sample Composition Bias and Response Bias in a Mail Survey: A Comparison of Inducement Methods

Do techniques that improve response rates in mail surveys improve the accuracy of survey results? Are the increases in sample size (and corresponding reductions in sampling error) obtained at the cost of introducing one or more systematic biases? These questions were addressed through an experimental investigation of response rate, sample composition bias, and response bias effects related to manipulations of survey sponsorship, cover letter message, notification method, and questionnaire format. Recent studies have shown that a number of inducement methods increase response rates in mail surveys (Heberlein and Baumgartner 1978; Kanuk and Berenson 1975; Linsky 1975). However, several researchers (Houston and Ford 1976; Jones and Linda 1978; Peterson 1975) have pointed out that response rate alone is an inadequate criterion with which to judge the effectiveness of inducement techniques. Methods that increase response rates may do more harm than good if they foster response bias (Houston and Ford 1976) or sample composition bias. Response bias distorts survey results by decreasing the validity of individual subjects' responses to questionnaire items. Sample composition bias also is a threat to survey