Estimation of variance from missing data

摘要

Any large scale survey may be prone to nonresponse problems. No exact formulation of the nature of nonresponse in surveys is available. So, several methods of handling nonresponse problems are proposed by survey statisticians. In this paper, the problems of estimation of the population total and its variance have been studied in the presence of nonresponse. The proposed methodology is based on the assumption that the set of respondents (comprises response sample) is a Poisson sub-sample from the initial sample. The proposed method is more general than the method presented by Sarndal (1992) since one can use auxiliary information in estimating the unknown response probabilities. The relative efficiencies of the proposed estimator are compared by simulation studies.