On Improving Population Mean Estimation under Non-Response: A Simulation Study

Abstract

Estimating the finite population parameters is primary objective in many studies, especially when some units fail to respond. These units are known as missing values for specific variables. Some important reasons for missing data may be due to incomplete data entry by the surveyor, broken equipment, misplaced observations and numerous other causes. Keeping this fact in mind and to cope with the impact of missingness, this study aims to develop an improved hybrid imputation method for estimating mean of finite population. Further, we derive an enhanced resultant point estimator from this imputation technique and the resultant point estimator can be significantly used to enhance the precision of population mean estimates. The important properties of the proposed method of finite population mean are discussed and derived. The results show that the framed imputation method achieves the lowest mean squared error (MSE) and the highest relative efficiency as compared to the competing estimators. Hence, the proposed imputation method for handling missing values in a data set is more efficient and is recommended for practical applications.