Introduction

As has been alluded to in various other chapters of this book, nonresponse is a prevalent problem in all human subject surveys. Nonresponse is generally the dominant cause of non-adherence to the sample design. Nonresponse is also a function of the survey method, as well as the quality of the survey instrument, the training of survey personnel, and the subject of the survey, among other things. Clearly, nonresponse is undesirable. Chapter 19 discussed the issue of weighting as a means to try to correct biases in the data. Here, again, the primary cause of bias is seen as nonresponse.

Nonresponse in human subject surveys may arise in two ways. First, there is survey nonresponse, also known as unit nonresponse. This arises when a sample unit fails to respond to the survey. For example, in a human population survey, unit or survey nonresponse occurs when a sampled person or household refuses to participate in the survey. In a telephone survey, this may arise when a qualified person answers the telephone but refuses to answer any questions and hangs up the telephone without providing any information, other than that there is a qualified individual or household at that telephone number. Similarly, in a face-to-face survey, this will occur when the survey respondent refuses to answer the interviewer’s questions and terminates any further discussion of the survey. In a postal survey, this occurs when the sampled person or household does not return the survey questionnaire, or returns a blank or spoiled survey. Similarly, in an internet survey, unit or survey nonresponse occurs when a person refuses to respond to any part of the internet survey, and perhaps even fails to visit the URL for the survey.