QSpace: Queen's Scholarship & Digital Collections
QSpace is an open access repository for scholarship and research produced at Queen's University. QSpace offers faculty, students, staff, and researchers a free and secure home to preserve and present their scholarship.
ItemThree Essays on Sustainable Finance: Canadian Physical Costs of Climate Change, Global Carbon Prices and the Costs of Climate Change, and, Environmental and Disclosure Performance: A Study of CA 100+ CompaniesSustainable finance has been increasingly gaining prominence in mainstream financial studies. The consideration of sustainability concerns within the sphere of finance has significant economic implications for society, investors, and beyond. In the past decade, the severity of climate change and its associated costs has spurred conversations about how to safeguard our prosperity in the face of this growing concern. These questions form the fundamental motivation for the study of how global climate change influences economic outcomes and firm performance and cost of equity. Specifically, this thesis is motivated by areas within sustainable finance that demand further research. The results from the first essay provide insight into the economic costs to Canada across several temperature outcomes leading into 2100. Using the Dynamic Integrated Climate and Economic (DICE) model, we project that climate change mitigation efforts that lead to a 2oC outcome more than pay for themselves in avoided climate costs from physical damages. The second essay investigates the climate change outcomes under varying global average carbon prices. We find that a global carbon price, while playing a critical role, will not be sufficient to meet our Paris Climate Agreement goals of limiting our global temperature increase to 1.5-2oC above pre-industrial levels by 2100. We also project significant differences in global physical costs, highlighting the urgency of taking action to mitigate global warming. The third essay investigates how environmental and disclosure performance influences firm’s return performance and cost of equity. We examine CA 100+ and find that they outperform other world indices in returns at a lower standard deviation, indicating stochastic dominance performance. We further find that the 2022 cost of equity (CoE) estimates for CA 100+ firms are in line with expectations after adjusting for country and industry impacts. We also find that, among CA 100+ firms, those that performed the worst environmentally and regarding disclosures had a lower CoE, contrary to expectations. ItemConsequence, Impact, and Washback: Examining Test Preparation as a Unique Dimension of WashbackTesting consequence, impact, and washback are crucial ongoing evidence that testing professionals collect to support the validity of the interpretation and use of test scores. Test preparation is frequently discussed within the context of washback. Guided by Green’s (2007a) predictive model of washback, this study adopts a two-phase consecutive mixed-method design to investigate the washback effects of TOEFL iBT test preparation on integrated English skills for academic purposes among Chinese test takers. A total of 1758 survey responses underwent analysis using descriptive statistics, t-tests, CFA, and SEM to evaluate the statistical relationship between perception of TOEFL iBT test design, test value, and their test preparation practices. Ten semi-structured interviews were analyzed through thematic analysis to explore the impact of TOEFL iBT test preparation on cognitive, metacognitive, and affective learning process and outcomes. The results were combined to examine washback direction, variability, and intensity on learning integrated English skills. The results contribute positive validity evidence to the TOEFL iBT, demonstrating that test preparation for TOEFL iBT has positive washback on what and how Chinese test-takers learn integrated English skills. This positive influence extends to test-takers’ knowledge and skills, academic English proficiency, language learning strategies, and overall test performance. TOEFL iBT test preparation represents a unique form of self-regulated learning, enabling Chinese test-takers to autonomously plan, engage, and monitor their learning. Washback variations were identified in test-takers’ perceptions of test value and design, test preparation choices, and associated characteristics, which have been theoretically and statistically confirmed to be linked to diverse test preparation behaviors, ultimately resulting in different washback outcomes. Washback intensity varies among test-takers and is linked to their perceptions of test importance, usefulness, and difficulty. Chinese test-takers who view the test as more important and useful, with manageable difficulty, are more likely to experience positive impacts. The study provides an in-depth understanding of integrated English skills and offers theoretical, methodological, and empirical implications for future research. The intricate dynamics of learner washback revealed in this study highlight the pivotal role of test-takers and associated factors in shaping washback effects and their connection to test’s validity. ItemOn the Use of Robotic Technology to Quantify Sensorimotor Impairments After StrokeIt is well recognized that sensory feedback plays an important role in allowing us to move and interact in our complex world. However, sensory feedback can be impaired following stroke potentially reducing an individual’s independence and increasing their fall risk. Commonly used clinical assessments do not have the ability to accurately quantify these impairments. The objective of this dissertation is to use robot-based behavioural tasks to accurately quantify impairments in the use of sensory information for guiding motor actions. The first study explored the use of a novel interception task to quantify impairments in the use of proprioceptive and visual feedback to generate rapid motor responses. The Fast Feedback Interception Task (FFIT) was developed to examine the ability to intercept a moving ball with a cursor. Sensory feedback was assessed by either mechanically perturbing the participant’s arm or shifting the visual location of the ball or cursor. Compared to healthy controls, 85% of individuals with stroke displayed impairments using their affected arm. Of note, 75% of individuals with stroke were impaired using their unaffected arm. The second study examined whether individuals with stroke could have distinct impairments in sensory feedback for action and perception. We compared performance on FFIT with a reaching task and an upper limb proprioceptive task. Most individuals with stroke were impaired in FFIT and reaching using their affected arm. Further, most FFIT and reaching task parameters were significantly correlated. We found only a few significant correlations between FFIT and proprioceptive task parameters. The third study examined spatial impairments for individuals with stroke using an 8-target reaching task. Of 265 individuals with stroke, 64% demonstrated impaired reaction time. 35% were impaired in initiating a movement towards a specific region of the workspace and 44% were impaired in all directions of reach. Further, lesions to specific brain regions or white matter tracts were associated with reaction time impairment. Our results highlight the high proportion of individuals with stroke impaired in the use of sensory feedback. Future research is required to understand how these findings can be implemented into clinical practice for guiding diagnosis and tracking stroke recovery. ItemA Novel Active Islanding Detection Method Based on Real-Time Wavelet Analysis in DC MicrogridsThis thesis presents a novel real-time active scheme for fast and reliable identifcation of islanding events in DC microgrids (MGs). The scheme is applicable in scenarios with or without power exchange between the main DC grid and MGs. In this approach, the bidirectional Dual Active Bridge (DAB) DC-DC converter’s controller injects a distinct single-tone periodic perturbation signal at predetermined intervals. The discrete nature of the perturbation signal considerably reduces its impact on power quality. A real-time wavelet analyzer determines whether to maintain or disconnect connections based on critical parameters it receives from decentralized detectors. The proposed method disconnects the main grid and common DC bus through converters in case of islanding events eliminating the need for complex DC circuit breakers (CBs). Detailed mathematical stability analysis proves and demonstrates that the proposed method, unlike some other active methods, guarantees the system’s stability. This analysis aligns seamlessly with the IEEE 1547 Standard and is validated by extensive simulation results. ItemNatural Language Processing for Justifiable Legal Practitioner AssistanceThe significant advancement of artificial intelligence (AI) has had a profound impact on many areas of applications. With Natural Language Processing (NLP) being a key player in this technological progress, exploring NLP within vital areas represents a pathway for immediate practical benefits while also challenging the existing AI solutions and helping to develop better models. This dissertation is centered around a comprehensive examination of NLP in the legal domain, a realm where the interpretation of language is of critical importance. In this dissertation, we investigate three main aspects of challenges: (i) The interpretability of NLP models, which is essential for legal practitioners to derive meaningful insights. To solve this challenge, we utilize post-hoc and ante-hoc methods of interpretability and show that the latter can be utilized to develop versatile interpretable solutions for several tasks in the legal domain. Specifically, we introduce a system that contains model-agnostic interpretable intermediate layers, a technique that proves to be effective for legal documents. (ii) The evaluation of existing datasets while identifying the disparity between the controlled experimental settings in academic research and the real-world legal problems. For this challenge, we propose a systematic annotation protocol that involves the expertise of legal professionals in dataset curation. We utilize weakly supervised learning by means of a curriculum learning strategy, effectively demonstrating the improved performance of a deep learning model. Then, we design a task with parallels to the thought-process of lawyers, i.e., with reference to both precedents and legislative provisions. Additionally, we introduce a prototype architecture to add interpretability, achieving strong performance while adhering to decision parameters used by lawyers. (iii) Lastly, the task of adapting general-purpose NLP models for specific legal tasks needs to be examined. We refactor conventional tasks on hierarchical datasets into a more indicative long-tail prediction task. We observe LLMs are prone to failure in these cases. To address these limitations, we propose the use of entailment-contradiction prediction in conjunction with LLMs, which allows for strong performance in a strict zero-shot setting. Our method does not require any parameter updates, a resource-intensive process and achieves strong performance across multiple datasets.
Communities in QSpace
Select a community to browse its collections.