A common problem for analysts of survey data. There are two types: complete non-response in which the individual selected for interview is absent, dead, or uncooperative; and item non-response where the interviewee fails to answer some (but not all) questions. Provided the lack of response is not a consequence of the questions being asked, neither type causes bias in the subsequent analysis (this is called ignorable non-response). However, if the reason for the non-response is the question being asked (for example, the very rich may not wish to disclose exactly how rich they are), then this is described as non-ignorable non-response. Imputation provides a partial solution to item non-response.
Subjects: Probability and Statistics — Sociology.