6 November، 2022
Discussion of a master’s thesis in the College of Computer Science and Mathematics – Department of Statistics and Informatics entitled: Proposed Robust Methods for Structural Equations Modeling – A Comparative Study
Discussion of a master’s thesis in the College of Computer Science and Mathematics – Department of Statistics and Informatics entitled: Proposed Robust Methods for Structural Equations Modeling – A Comparative StudyIt was discussed in the discussion room at the Faculty of Computer Science and Mathematics at the University of Mosul on Sunday 6/11/2022Master’s thesis by student Saif Ramzi Ahmed Shayal Al-Alam , under the supervision of Asst. prof.Dr. Bashar Abd Al-Aziz Al-TALIBThis study took idea of the study is to robust the path coefficients in the structural equation model using the robust correlation coefficients. Three methods of the robust correlation as well as the Pearson correlation were compared. A comparison has been made between the proposed method and the traditional method with a causal model proposed by the researcher, and many simulation scenarios have been applied on the experimental side. To test the efficiency of the generated models, samples of different sizes were used, and the explanatory variables in the causal model were polluted with different percentages of contamination. Path parameters were estimated using the correlation matrix for four methods and the results were compared with some statistical criteria. In addition to applying this to real data collected by the researcher through a questionnaire distributed to a segment of students applying for postgraduate studies on knowledge of the factors affecting the success of electronic tests at the University of Mosul. The number of data reached 205 observations and includes 6 latent variables that represent the factors affecting these tests. The thesis concluded that the path coefficients method using the robust correlation gave the best results compared to the traditional methods.The study touched outliers and values that move away from the permissible limit to be anomalous values greatly affect the estimation process for the coefficients of statistical models because of data entry or that there is a small or large percentage of the data that is far from its large or small value in relation to the rest of the data (anomalous values). Since its presence greatly affects the estimation process, the researcher must process the data before starting the estimation process. Therefore, in this thesis, the data were treated and purified from the outliers using robust methods in the structural equation modeling process.This study aims to treat data from the presence of anomalies and use robust methods for modeling structural equations through some methods that relied on employing the use of some strong correlation coefficients in immunizing the path coefficients in the structural equations model. And study the efficiency of the proposed methods using a set of immune coefficients, as well as Pearson’s correlation coefficient. And the comparison between the proposed methods and the traditional methods with a causal model that was formulated in the style of modeling with structural equations, and applying this to experimental data of different sizes and applied data and comparing the results obtained in the two cases.The discussion committee consists of:
| Prof. Dr. Taha Hussein Ali – Salahaldeen University- Chairman |
| Assis. Prof. Dr. Heyam Abd Al-Majeed Hayawi – member |
| Lecturer Dr. Omar Salim Ibraheem – member |
| Assis. Prof. Dr. Bashar Abd Al-Aziz Al-TALIB – member and supervisor
Contact the university 
|
