UnitNonresponse

Unit nonresponse is a form of missing data in survey research in which a sampled unit does not participate at all, so no data are collected for that unit. It is distinct from item nonresponse, where a respondent provides answers to some questions but not others. Unit nonresponse can result from noncontact (the interviewer cannot reach the unit) or refusal (the unit declines to participate), and it may also involve misclassification of eligibility during screening.

Measuring unit nonresponse typically involves the response rate, defined as the number of completed interviews divided by the number of eligible sampled units. The unit nonresponse rate is the complement of the response rate, though accurate calculations require correct determination of eligibility. Nonresponse bias depends on whether the probability of participation is related to the survey outcomes of interest.

The impact of unit nonresponse depends on the relationship between response propensity and key variables. If respondents resemble nonrespondents on important characteristics, bias is limited; if they differ systematically, estimates can be biased, particularly for subgroups or highly variable measures. The risk is also influenced by the overall response rate and the quality of the sampling frame.

Mitigation strategies include design and fieldwork practices such as multiple contact attempts, mixed modes of data collection, incentives, and efforts to update contact information. In analysis, weighting adjustments—such as response propensity weighting, calibration, and post-stratification—are commonly used to correct for differential nonresponse. Paradata on contact attempts and cooperation can inform these adjustments. Imputation is generally applied to item nonresponse rather than unit nonresponse.