A sound sampling plan for a survey includes extensive effort to obtain useable data for each unit selected into the sample. Resources are allocated to develop a good sampling frame, design a good questionnaire, good interviewer's training and other data collection procedures such as how to gain cooperation of respondents. However, in spite of such efforts, all surveys encounter missing data which could occur either due to noncoverage or nonresponse. In this paper, we will discuss missing data due to nonresponse and methods to adjust for it. It occurs when some or all responses to the questions on a questionnaire are not obtained. This may be due to the respondents inability or unwillingness to answer.

Researchers have been striving to reduce nonresponse. For example, they have done this by better designing and testing questionnaires thoroughly for complete and accurate answers before fielding the survey, providing respondents aids to keep better records, giving respondents gifts (cash or kind) to gain their cooperation and finding ways to improve training given to the data collection staff. Researchers are also heavily involved in improving the methods to account for missing data. Two approaches commonly used are imputation and weighting adjustment.