Optimal imputation of missing data for estimation of population mean

Abstract

In this article, we have addressed the issue of optimality under imputation by using Searls (1964) idea. We have proposed three new Searls-type difference (STD) methods for imputation of missing data. The resultant STD estimators are better than the estimators proposed by Kadilar and Cingi (2008) and Diana and Perri (2010). An analytical comparison shows that the proposed optimal methods are always better that the respective regression methods of imputation. Theoretical results are supported by an empirical study.

Keywords:

• Missing data
• Imputation
• Searls-type estimators