Please use this identifier to cite or link to this item: http://hdl.handle.net/123456789/2549
Title: ESTIMATION MAXIMIZATION CONCEPT OF ASSESSING RESIDUALS: A STRUCTURAL EQUATION MODELLING APPROACH
Authors: Abdul-Aziz, Abdul-Rahaman
Issue Date: 2019
Abstract: The main objective of this study was to apply estimation maximization concept to assess residuals through structural equation modelling. This was achieved through simulation setup. Data was obtained from test scores of some selected course lecturers of Kumasi Technical University to validate the results from the simulation. The results showed that the estimated residuals of the measurement errors using all three estimators correlate negatively with the estimated residuals associated with measurement errors of items that load on the same factor. These correlations are strongest when using the Bartlett’s methodbased estimator and weakest when using the regression method-based estimator. Thus, the Bartlett’s method-based residual estimators are among the three estimators that achieved very close values. Also, it can be deduced from the results on the various simulation of quantile-quantile plots that all these methods demonstrate the ability to detect outliers and potential influential observation in a SEM framework. It is worth noting that the Anderson-Rubin method provided a quantile-quantile plot which was more efficient in terms of visual display for detecting outliers and potential influential observations as compared to the other class of residual estimators. Finally, it was therefore found from the comparative model fits information, by comparing among the three existing residual estimators, that the Bartlett’s based method gave better residual parameter estimates over the regression based method and the Anderson Rubin based method. However, the estimation maximization method gave better residual parameter estimates than the other three existing methods. It is therefore worth noting that this present study contribution to knowledge is demonstration of the fact that estimation maximization method could be a better residual estimator within the SEM framework compared to other existing methods.
Description: DOCTOR OF PHILOSOPHY IN APPLIED STATISTICS
URI: http://hdl.handle.net/123456789/2549
Appears in Collections:Faculty of Mathematical Sciences

Files in This Item:
File Description SizeFormat 
ABDUL-RAHAMAN ABDUL-AZIZ_2019.pdf5.73 MBAdobe PDFView/Open


Items in UDSspace are protected by copyright, with all rights reserved, unless otherwise indicated.