Unit Nonresponse

Unit nonresponse in a survey occurs when an eligible sample member fails to respond at all or does not provide enough information for the response to be deemed usable (not even as a “partial completion”). Unit nonresponse can be contrasted to item nonresponse (missing data) wherein the sample member responds but does not provide a usable response to a particular item or items. Unit nonresponse can be a source of bias in survey estimates, and reducing unit nonresponse is an important objective of good survey practice.

Reasons for unit Nonresponse

Despite the best efforts of the survey practitioner, there are reasons why unit nonresponse will still occur.

• Refusal: The sample member may refuse to participate in the survey. Often refusals are divided into hard refusals and soft refusals, depending on the intensity with which the sample member refuses to participate.
• Sample member cannot be found: The sample member (whether a person, a household, or an establishment) may have moved or otherwise cannot be located. In the case of a telephone survey, the telephone may ring repeatedly with no answer or be picked up by an answering machine with an uninformative message.
• Sample member may be temporarily away: The sample member may be known to be travelling and unavailable during the data collection period.
• Communication difficulty: The sample member may speak a language that none of the interviewers speaks or into which the survey instrument has not been translated. In other instances, the sample member may have a physical or mental disability that interferes with communication.
• Sample member provides inadequate data: The sample member provides some data, but after editing, the data are deemed inadequate to constitute a valid response.
• Other: There are other reasons for unit nonresponse, such as quarantines, which although unlikely, do occur. (Detailed discussion of these other instances is available at http://www.aapor.org.)

It is important to distinguish sample members that are unit nonrespondents from those that are ineligible (out of scope). In the case of sample members that cannot be contacted, it may not be possible to know for certain if a particular sample member is a nonre-spondent or ineligible, but the proportion of nonrespondents among the “noncontacts” can be estimated. An example is a random-digit dial telephone survey of households in which a selected telephone number is called repeatedly with no answer. The telephone number may belong to a household or an establishment, or it may not be an assigned number. Only in the first case is the sample member a unit nonrespondent.