A chain regression exponential type imputation method for mean estimation in the presence of missing data

Imputation methods deal with item nonresponse to solve the missing data problem. A new imputation method and corresponding point estimators for population mean have been proposed under two situations: using the response rate and the constant that gives the minimum mean square error for the estimator...

Full description

Saved in:
Bibliographic Details
Published in:Wārasān Songkhlā Nakharin 2022-08, Vol.44 (4), p.1109-1118
Main Authors: Kanisa Chodjuntug, Nuanpan Lawson
Format: Article
Language:English
Subjects:
bias
imputation
mean square error
missing data
regression estimator
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Imputation methods deal with item nonresponse to solve the missing data problem. A new imputation method and corresponding point estimators for population mean have been proposed under two situations: using the response rate and the constant that gives the minimum mean square error for the estimator. The biases and mean square errors of the proposed estimators are derived. The performance of this method is compared with some existing methods via simulations and an application to fine particulate matter data. The results show that the proposed estimator, which uses the optimum value of a constant, performs the best. It performs the second best when using the response rate in the estimator, which is free of known parameters. The estimated fine particulate matter in Kanchana Phisek Road in Bangkok using the best method is equivalent to 42.22 micrograms per cubic meter with a mean square error of 0.34 micrograms per cubic meter squared.
ISSN:0125-3395
DOI:10.14456/sjst-psu.2022.144