Imputation Methods for Missing Values in Estimation of Population Mean under Diagonal Systematic Sampling Scheme

Abstract

In survey sampling, the use of auxiliary information to enhance estimators of population parameters under simple random sampling stratified random sampling and systematic sampling has been widely discussed. Similarly, some existing estimators were modified using regression imputation approach to obtain two imputation schemes and estimators that impute the responses non-respondents thereby eliminating difficulties in data presentation, compilation. The theoretical properties (estimators, biases and mean squared errors) of the proposed imputation scheme were derived so as to assess their robustness and efficiency. The theoretical findings were supported by simulation studies on population generated using four distributions namely; Beta, Gamma, Exponential and Uniform distributions. The averages of biases, MSEs and PREs of the estimators in comparison to the existing estimators were computed from the simulated data and the results showed that on average, the estimators of the proposed imputation scheme have minimum biases, minimum MSEs and higher PREs compared to the traditional unbiased estimators. These results imply that the estimators of the proposed schemes are more efficient and robust than the conventional unbiased estimators.