Improved imputation methods in survey sampling

A well-known problem in the field of survey sampling is the problem of missing data due to a number of kinds of non-responses. In this thesis, I reviewed existing imputation methods and proposed new improved imputation methods to when the data is missing completely at random responses (MCAR). In general, MCAR happens when an interviewer fails to meet an interviewee on a random attempt either through personal visit at home, telephone or a mail survey. In this thesis, I compared the proposed imputation methods with its competitors analytically as well as through simulation studies