Saudi Cultural Missions Theses & Dissertations
Permanent URI for this communityhttps://drepo.sdl.edu.sa/handle/20.500.14154/10
Browse
3 results
Search Results
Item Restricted Judging a book by its cover: Investigating the influence of test takers' physical attractiveness in synchronous oral communication assessment(Iowa State University, 2025) Aseeri, Fatimah Mohammed; Gary, OckeyThe assessment of Oral communication (OC) is an important measure of students' English language proficiency. In OC assessments, raters are commonly exposed to the visual appearance of the test takers, and depending on the test taker's physical attractiveness (PA), they may award students higher or lower scores than their actual ability. However, due to challenges associated with measuring social appearance values, research on this topic has only scratched the surface of this critical issue. Therefore, this study aims to investigate potential impacts of test takers’ physical attractiveness on their oral assessment scores. In phase I, a small exploratory study was conducted to define the PA construct and develop a PA scale. Phase II followed a concurrent embedded research design where quantitative and qualitative measures were applied. Quantitative measures including Many Facet Rasch Measurement and regression were used to investigate several aspects. First, the study investigated whether raters assign different scores to the test takers in two video conferencing modes: when exposed to test takers physical appearance (Face-to-Face) and when not exposed to it (Avatar-Mediated). Second, it explored whether raters show systematic bias in their evaluations based on the presence or absence of the test takers physical appearance. Then, it aimed at finding whether physical attractiveness is a potential explanatory variable that predicts test takers' OC scores in the F2FVC mode. Qualitative methods were conducted to investigate raters' perceptions about their rating process in the two modes (Face-to-Face vs. Avatar-Mediated), and whether they believe that physical attractiveness of the test takers may have affected their rating in the Face-to-Face video conferencing mode. The Many-facet Rasch Measurement analysis showed that 30 test takers received slightly higher scores in the Face-to-Face video conferencing mode compared to the Avatar-Mediated video conferencing mode. Yet, the difference was not significant (X2 = 3.6, d.f. =1, p = 0.06). Although not significant, four out of the 30 test takers failed the exam in Avatar-Mediated video conferencing mode, and would have passed the exam if the score difference between the two modes would have been awarded. Bias analysis found that out of the five expert raters, one showed significant bias towards the F2FVC mode (t (237) = −2.30, p < .05). Simple regression analysis of 41 test takers OC scores and PA scores revealed that test takers' PA significantly predicted their pronunciation scores within the OC construct, with female participants scores being more impacted than male participants. Findings from the qualitative analysis showed that the raters perceived the mode of delivery as having no impact on their rating decisions. They also believed that physical attractiveness of the test takers should not impact their rating, and if it did, it is unintentional. The findings of this research provide evidence of PA being a source of construct irrelevant variance in test takers' scores. Although this research does not end the controversy, it suggests that Patzer’s (1985) statement that “people do judge a book by its cover despite their claim to do the contrary" is true, at least for some people.22 0Item Restricted Fair and Accurate Machine Learning in Dynamic and Multi-domain Settings(Rutgers, The State University of New Jersey, 2024-05-01) Almuzaini, Abdulaziz; Singh, Vivek; Pennock, DavidA multitude of decision-making tasks, such as content moderation, medical diagnosis, misinformation detection, and recidivism prediction, are increasingly being automated due to recent developments in machine learning (ML). While ML models demonstrate superior capabilities in large-scale data processing and complex pattern recognition compared to humans, the decisions they make can profoundly impact individuals' opportunities and lives, necessitating the assurance of their accuracy and fairness. Besides developing ML models in controlled lab environments, automated machine learning tasks are often used in real-world settings where the concept of stationarity (i.e., the independent and identically distributed i.i.d. assumption) is frequently violated, leading to a notable decrease in the effectiveness of machine learning models. Specifically, real-world ML models can be trained on particular domains and deployed in dissimilar domains. These domains encompass diverse time points, heterogeneous population groups, or disparate tasks demanding careful, dynamic, ethical, and knowledge-transferring model development techniques. Due to the dynamic nature of many machine learning tasks and their continuous evolution, a previously trained model may become unfair or erroneous over time. Additionally, machine learning applications can be particularly challenging due to limited data or computational resources, which often require developers to leverage knowledge from other domains. In this dissertation, we investigate these issues and suggest ways to mitigate the challenges of maintaining the goals of fairness and accuracy in dynamic and multi-domain settings. Particularly, to mitigate the impact of the dynamic issue, we present a pair of anticipatory bias correction techniques that target fairness and accuracy simultaneously in temporally shifting and delayed labeling contexts, supporting the goals of timely and safe model adaptation. Furthermore, we leverage transfer learning methodology to study ML performance in developing a fair and accurate dermatological image processing task for skin cancer diagnosis using datasets gathered from various domains (i.e., locations) and models trained on different contexts (i.e., pre-trained image recognition model). Lastly, we explore the feasibility of combining diverse commercial pre-trained black-box models developed in various domains to jointly enhance fairness and accuracy for a sentiment analysis task. We present an overview of the observed results for each work, discuss the identified limitations, and propose future research directions. These results represent significant progress toward developing fair and accurate ML algorithms in dynamic and multi-domain settings.29 0Item Restricted The Responsibility of Non-Profit Organisations in Saudi Arabia for Terrorism Financing with Reference to Law and Practice in England and Wales.(University of Leeds, 2024-03-05) Alsalmi, Mohammad Abdulrahman; Walker, Clive; O'Reilly, ConorThe Kingdom of Saudi Arabia's (KSA) non-profit (NPO) sector has been implicated in allegations of providing funding to radical Islamic organizations. This association, voiced especially after 9/11, has raised critical questions about the KSA's understanding of, and response to, this risk. This thesis undertakes a comprehensive investigation into the policies, laws, and practices employed within the country to combat terrorism financing (CTF) through NPOs. Beyond assessing the effectiveness of these measures, this research delves into the fairness of the KSA's CTF approach towards NPOs. Addressing these core questions through documentary analysis and fieldwork, the thesis explores the deficiencies and barriers, not only legal but also political, cultural, and structural which hinder the KSA in effectively and fairly countering terrorism financing through NPOs. One of the significant identified factors pertains to the incorporation of Sharia within the KSA's legal system and the inherent challenges it poses for the implementation of CTF procedures. Sharia principles do not explicitly address CTF-related matters and also affect the impact of more recent legislation against CTF. Relying on a legal framework that prioritizes historical norms over contemporary legal requisites can impede efforts, whether in criminal law, civil law, or regulatory action in the treatment of the NPO sector. In addition to identifying the factors contributing to the shortcomings in the KSA's CTF approach involving NPOs, this thesis offers recommendations for reforming the KSA's NPO governance system having regard to notions of responsive regulation and an assemblage model. These recommendations draw insights from the governance experiences of England and Wales which form the basis for policy transfer.25 0