Sampling strategies for handling nonresponse and measurement errors in poststratified sampling scheme
| dc.contributor.author | Izunobi, Chinyeaka Hostensia | |
| dc.date.accessioned | 2026-10-19T12:41:57Z | |
| dc.date.available | 2026-10-19T12:41:57Z | |
| dc.date.issued | 2021-12-07 | |
| dc.description | This thesis is for the award of Doctor of Philosophy (PhD.) in Statistics | |
| dc.description.abstract | Nonresponse and measurement errors, often referred to as nonsampling errors, pose serious challenges in surveys. The focus of the present study is on the development of appropriate sampling strategies for handling of these problems under the poststratified sampling scheme. Three (3) broad cases considered in the study were the occurrence of nonresponse alone, the occurrence of measurement errors alone and the joint occurrence of nonresponse and measure ment errors. These three (3) broad cases were considered when each occurred on both the study and auxiliary variables, on only the study variable, and on only the auxiliary variable, giving rise to nine (9) distinct cases. Three (3) separate-type estimators of the population mean of the study variable were then proposed for each of the nine (9) distinct cases, giving a total of twenty-seven (27) sampling strategies. These three (3) types of estimators are the customary type sample mean estimators, the difference-type estimators, and a class of ratio/product-type estimators. Properties of the proposed estimators were obtained both for an achieved sample configuration (conditional argument) and for repeated samples of fixed size n (unconditional argument). The proposed customary-type and difference-type estimators were unbiased, while the proposed classes of ratio/product-type estimators were generally biased for the study vari able population mean. The results showed that there exists an optimum class of estimators among the proposed classes of estimators. The theoretical results were subjected to numeri cal validation and verification, using two datasets, namely, Dataset-1, for positively correlated study and auxiliary variables, and Dataset-2, for negatively correlated study and auxiliary variables. Both the theoretical and empirical results showed that the optimum estimators in the proposed classes of estimators performed better than the usual post stratified sample mean estimator (when there are no nonresponse and measurement errors). The optimum estimators also generally performed better than the proposed customary-type estimators and other proposed ratio/product-type estimators. The study recommends, among others, the use of auxiliary in formation when developing improved sampling strategies for the handling of the problem of nonresponse and measurement errors in poststratified sampling scheme. | |
| dc.identifier.citation | Izunobi, C. H. (2021). Sampling strategies for handling nonresponse and measurement errors in poststratified sampling scheme [Unpublished Doctoral Thesis]. Federal University of Technology, Owerri, Nigeria | |
| dc.identifier.uri | https://repository.futo.edu.ng/handle/20.500.14562/2856 | |
| dc.language.iso | en | |
| dc.publisher | Federal University of Technology, Owerri | |
| dc.rights | Attribution-NonCommercial-ShareAlike 4.0 International | en |
| dc.rights.uri | http://creativecommons.org/licenses/by-nc-sa/4.0/ | |
| dc.subject | Sampling strategies | |
| dc.subject | nonresponse | |
| dc.subject | measurement errors | |
| dc.subject | poststratified sam pling | |
| dc.subject | auxiliary information | |
| dc.subject | difference | |
| dc.subject | ratio | |
| dc.subject | product estimators | |
| dc.subject | Department of Statistics | |
| dc.title | Sampling strategies for handling nonresponse and measurement errors in poststratified sampling scheme | |
| dc.type | Doctoral Thesis |