Comparative study of different imputation methods

Abstract

An improved class of imputation procedures are proposed for estimating the finite population mean by using the known population mean and higher order moments of the auxiliary variable. Mathematical expressions for bias and mean square error of proposed and existing imputation procedures are derived up to first order approximation. The theoretical results are supported by numerical study of five empirical populations and three hypothetical populations by taking different values of ρ_yx at varying response.

Keywords:

• Higher order moments
• Imputation
• Mean squared error
• Non-response

Mathematical classification:

• 62D05