Nonresponse Rates

Responses

Cases that are treated as responses can also be divided into two groups: complete responses and partial responses. Standards that are widely used to define partial responses versus complete responses include the following: (a) the proportion of all applicable questions completed; (b) the proportion of essential questions completed; and (c) the proportion of all questions administered. Essential questions may vary, depending upon the purposes of a survey. In the case of the previously mentioned survey about public approval of the new war plan, the crucial variables may include attitudes toward the plan, party identification, and so on. Survey researchers should define what constitutes a complete versus a partial response based on one of the standards or a combination of the standards prior to data collection. Complete responses, for example, could be defined as cases that have answers to 95% or more of essential questions, whereas partial responses might be defined as cases with answers to 50–94% of such questions. Cases with fewer than 50% of the essential questions answered could be treated as breakoffs, which is a form of nonresponse. By this definition, only cases with data for at least 50% of crucial questions would be deemed as responses or partial responses.

Non Responses

Nonresponses can result from noncooperation or refusals, noncontacts, and from other factors. The situations in which instances of nonresponse occur vary across surveys and across different sampling or data collection modes. Refusals happen in random-digit dialing (RDD) telephone or in-person surveys of households when a household has been contacted and either a responsible household member or the designated respondent has refused to participate in the study. Breakoffs refer to a premature termination from an interview. In mail or online surveys, refusals occur when contact has been made with a specifically named respondent or with a household or organization where the respondent works or lives, and the respondent or a responsible member of the household or organization has refused to participate. However, researchers often are not certain when this has happened as they typically receive no direct evidence to this effect. A breakoff occurs when partially completed questionnaires are returned with some notification suggesting that the respondent refused to complete the rest of the questionnaire. The distinction between breakoffs and partial responses should be pre-determined by researchers using the criteria defining a complete response versus a partial response, as previously mentioned.

...