Abstract

Surveys are an important tool to assess quality of life, measure public opinion, gauge trust in institutions, and many other purposes. They are used in academia, health, education, statistics and marketing, and give citizens a voice. These same citizens, however, increasingly cannot or will not participate in surveys. Nonresponse is a serious threat to the reliability and validity of survey results.

Although it is clear that survey nonresponse is a problem, it is less clear how to calculate response rates, what are the causes of nonresponse, and when and why it is a problem. For instance, in a web survey on health, nonresponse can be caused by the web server temporarily being out of order or by the respondent not receiving the survey invitation because they are in hospital. In the first case, nonresponse is a nuisance; in the second case, nonresponse will result in an underestimate of sick people.

The entry provides guidance on how to enhance response rates and on how to analyze nonresponse and possibly to adjust for it. It also discusses the tension between efforts of survey agencies to get everyone to participate (and thus obtain good and useful data) and the right of potential respondents to be left alone.