IMPUTATION USING REGRESSION ESTIMATORS FOR ESTIMATING POPULATION MEAN IN TWO-PHASE SAMPLING
Abstract
This paper presents the estimation of mean in presence of missing data under two-phase sampling design using regression estimators as a tool for imputation while the size of responding ( ) 1 R and non-responding ( ) 2 R group is considered as a random variable. The bias and mean squared error of suggested estimators are derived in the form of population parameters using the concept of large sample approximation. Numerical study is performed over two populations by using the expressions of bias and mean squared error and efficiency compared with existing estimators.