Nonresponse in sample surveys: the problem and some solutions

Nonresponse is a common problem in sample surveys in the field of physical therapy. Several recently published articles in Physical Therapy illustrate how such nonresponse can yield information with a large potential bias. Assuming that a more personal contact with the subject is likely to increase the subject's willingness to participate in a survey, several techniques to increase the response rate are reviewed. Another method of dealing with nonresponse is to select a subsample of the nonrespondents for intensive follow-up. This technique is less well known than the personal contact techniques and is described in detail. Formulas are given for estimating a population mean by combining information from the initial respondents and from the subsample of nonrespondents. An example of this method is presented. Subsampling the nonrespondents is recommended for use in sample surveys in which a large nonresponse rate may be expected.