Imputation based mean estimators in case of missing data utilizing robust regression and variance-covariance matrices

• Published in: Communications in statistics. Simulation and computation Vol. 51; no. 8; pp. 4276 - 4295
• Main Authors: Shahzad, Usman, Al-Noor, Nadia H., Hanif, Muhammad, Sajjad, Irsa, Muhammad Anas, Malik
• Format: Journal Article
• Language: English
• Published: Philadelphia Taylor & Francis 03.08.2022; Taylor & Francis Ltd
• Subjects: Covariance matrix; Estimators; Imputation methods; Missing data; Random sampling; Regression; relative mean square error; robust regression; robust variance-covariance matrices; Robustness (mathematics); simple random sampling; Statistical analysis; Statistical inference; Variance
• ISSN: 0361-0918; 1532-4141
• DOI: 10.1080/03610918.2020.1740266

Missing data is a common problem in sample surveys and statisticians have recognized that statistical inference can be spoiled in the presence of non-response. Kadilar and Cingi built up a class of estimators for assessing the population mean under simple random sampling scheme when there are missing observations in the data set. This article firstly, proposes a class of estimators in light of Zaman and Bulut work, and after that defines another class of regression type estimators utilizing robust regression tools, robust variance-covariance matrices and supplementary information. The use of robust techniques in Zaman and Bulut ratio type estimators enable us to estimate the population mean in several cases of missing observations. The hypothetical mean square error equations are also derived for adapted and proposed estimators. These hypothetical discoveries are assessed by the numerical illustration, in support of present work.