Nonresponse (spelled as a single word) is the term used in survey literature to describe missing data problems; encompasses the issues of unit nonresponse, item nonresponse, and nonresponse biases
Nonresponse (spelled as a single word) is the term used in survey literature to describe missing data problems. Three issues are associated with the generic term of nonresponse:
- Unit nonresponse is a situation when a sampled unit does not provide any survey data on any of the survey variables at all. (This looks like a blank line in the rectangular data set.)
- Item nonresponse is a situation when a sampled unit provides responses on some, but not all of the survey variables. (This looks like Swiss cheese holes in the rectangular data set.)
- Nonresponse bias arises when the standard analyses leads to the estimates that are biased for their (finite) population targets. In terms of the standard Rubin's missing data classification, this indicates that the data are not missing at random (NMAR), although the analogies are not exact as the NMAR concept is formulated for the data likelihood which is difficult to define in informative ways for sampling designs.
The issue of nonresponse and the biases that it may cause has been the central concern for survey statisticians in the past 20 or so years. Groves (2006) is the most cited paper in Public Opinion Quarterly, one of about five journals on survey methodology. Well-designed and well-executed surveys maintain relatively little bias with response rates as low as 10% (Pew 2012).
Unit nonresponse is usually corrected by survey weights. Valliant, Dever and Kreuter (2013) provide some guidance as well as an R package to implement some of the corrections.
Item nonresponse can usually be corrected, at least partially, by imputation of the missing data.
A measure of the relative importance of nonresponse is response rate, which is the ratio of the number of completed interviews over the number of eligible reporting units in the sample. The definition is fairly technical, and involves the multiple categories of the final survey dispositions. The main reference is Standard Definitions: Final Dispositions of Case Codes and Outcome Rates for Surveys by the American Association for Public Opinion Research (http://www.aapor.org/Publications-Media/AAPOR-Journals/Standard-Definitions.aspx). The relevant edition, as of mid 2016, is the 9th edition.
