AI has quickly grown from being a vast concept to an emerging technology that many companies are looking to integrate into their businesses, generally considered an ongoing “revolution” transforming science and society altogether. Researchers and organizations agree that AI and the recent rapid developments in machine learning carry huge potential benefits. At the same time, there is an increasing worry that ethical challenges are not being addressed in the design and implementation of AI systems. As a result, AI has sparked a debate about what principles and values should guide its development and use. However, there is a lack of consensus about what values and principles should guide the development, as well as what practical tools should be used to translate such principles into practice. Although researchers, organizations and authorities have proposed tools and strategies for working with ethical AI within organizations, there is a lack of a holistic perspective, tying together the tools and strategies proposed in ethical, technical and organizational discourses. The thesis aims to contribute with knowledge to bridge this gap by addressing the following purpose: to explore and present the different tools and methods companies and organizations should have in order to build machine learning applications in a fair and transparent manner. The study is of qualitative nature and data collection was conducted through a literature review and interviews with subject matter experts. In our findings, we present a number of tools and methods to increase fairness and transparency. Our findings also show that companies should work with a combination of tools and methods, both outside and inside the development process, as well as in different stages of the machine learning development process. Tools used outside the development process, such as ethical guidelines, appointed roles, workshops and trainings, have positive effects on alignment, engagement and knowledge while providing valuable opportunities for improvement. Furthermore, the findings suggest that it is crucial to translate high-level values into low-level requirements that are measurable and can be evaluated against. We propose a number of pre-model, in-model and post-model techniques that companies can and should implement in each other to increase fairness and transparency in their machine learning systems. / AI har snabbt vuxit från att vara ett vagt koncept till en ny teknik som många företag vill eller är i färd med att implementera. Forskare och organisationer är överens om att AI och utvecklingen inom maskininlärning har enorma potentiella fördelar. Samtidigt finns det en ökande oro för att utformningen och implementeringen av AI-system inte tar de etiska riskerna i beaktning. Detta har triggat en debatt kring vilka principer och värderingar som bör vägleda AI i dess utveckling och användning. Det saknas enighet kring vilka värderingar och principer som bör vägleda AI-utvecklingen, men också kring vilka praktiska verktyg som skall användas för att implementera dessa principer i praktiken. Trots att forskare, organisationer och myndigheter har föreslagit verktyg och strategier för att arbeta med etiskt AI inom organisationer, saknas ett helhetsperspektiv som binder samman de verktyg och strategier som föreslås i etiska, tekniska och organisatoriska diskurser. Rapporten syftar till överbrygga detta gap med följande syfte: att utforska och presentera olika verktyg och metoder som företag och organisationer bör ha för att bygga maskininlärningsapplikationer på ett rättvist och transparent sätt. Studien är av kvalitativ karaktär och datainsamlingen genomfördes genom en litteraturstudie och intervjuer med ämnesexperter från forskning och näringsliv. I våra resultat presenteras ett antal verktyg och metoder för att öka rättvisa och transparens i maskininlärningssystem. Våra resultat visar också att företag bör arbeta med en kombination av verktyg och metoder, både utanför och inuti utvecklingsprocessen men också i olika stadier i utvecklingsprocessen. Verktyg utanför utvecklingsprocessen så som etiska riktlinjer, utsedda roller, workshops och utbildningar har positiva effekter på engagemang och kunskap samtidigt som de ger värdefulla möjligheter till förbättringar. Dessutom indikerar resultaten att det är kritiskt att principer på hög nivå översätts till mätbara kravspecifikationer. Vi föreslår ett antal verktyg i pre-model, in-model och post-model som företag och organisationer kan implementera för att öka rättvisa och transparens i sina maskininlärningssystem.

The thesis focuses on the way the secondary students of an international British curriculum school in Spain portray gender equality. It analyses the themes emerging from a focus group discussion with a close consideration of the concepts of gender, gender equality, and gender mainstreaming. A crucial element underpinning the research is the concept of children’s voices, as the study aims at creating space for those voices to be expressed and heard. This paper addresses various notions constituting inextricable parts of the identified themes: the notions of fairness, responsibility, control, and heteronormativity. I argue that this kind of research is a necessity due to insufficient consideration of children’s voices in policy-making processes in education. I denounce the paucity of adults’ knowledge of school children’s needs and opinions. This paper aims at understanding how students define gender equality, how they associate it with the notions of responsibility, control and fairness, and how my assumptions, as an adult and as a teacher, differ from the perception of secondary students in question.

No / People, systems, and things in the cities generate large amount of data which is considered to be the most scalable asset of any smart city. Linux users are rapidly increased in last few years, and many large multinational organizations are deploying long distance high bandwidth (LDHB) cloud networks for centralizing the data from various smart cities on a central location. TCP is responsible for reliable communication of data in these cloud networks. For reliability communication among various smart cities, a number of TCP congestion control mechanisms have been developed in the past. TCP Compound, TCP Fusion, and TCP CUBIC are the default TCP congestion control mechanisms for Microsoft Windows, Sun Solaris, and Linux operating systems respectively. The response function of TCP CUBIC is higher than the response function of Standard TCP, which is a trademark congestion control mechanism. As a result, TCP CUBIC does not behave friendly with Standard TCP in LDHB cloud networks. The Congestion Window (cwnd) reduction and growth of TCP CUBIC is very aggressive, which causes high packet loss rate and unfair share of available link bandwidth among competing flows from various smart cities. The aim of this research is to design a new TCP congestion control mechanism for Linux operating system to achieve maximum performance in LDHB cloud networks being used by smart cities. In this paper, congestion control module for slow start (CCM-SS) is designed by increasing the lower boundary limit of cwnd size in slow start phase of communication. Congestion control module for loss event (CCM-LE) is designed by increasing the cwnd reduction rate at each packet loss event and finally Advance Response Function for TCP CUBIC (ARFC) is proposed to design a new congestion control mechanism for Linux operating system. NS-2 is used to compare the performance of TCP CUBIC* with TCP CUBIC in short distance high bandwidth (SDHB) and long distance high bandwidth (LDHB) cloud networks. Results show that TCP CUBIC* has outperformed in LDHB networks, at least by a factor of 18% as compared to TCP CUBIC.

People are increasingly turning to online communities for entertainment, information, and social support, among other uses and gratifications. Online communities include traditional online social networks (OSNs) such as Facebook but also specialized online health communities (OHCs) where people go specifically to seek social support for various health conditions. OHCs have obvious health ramifications but the use of OSNs can also influence people's mental health and health behaviors. The use of online communities has been widely studied but in the health context their exploration has been more limited. Not only are online communities being extensively used for health purposes, but there is also increasing concern that the use of online communities can itself affect health. Therefore, there is a need to better understand how such technologies influence people's health and health behaviors. The research in this dissertation centers on examining how online community use influences health and health behaviors. There are three studies in this dissertation. The first study develops a conceptual model to explain the process whereby the characteristics of a request from an OHC user for social support is answered by a wounded healer, who is a person leveraging their own experiences with health challenges to help others. The second study investigates how algorithmic fairness, accountability, and transparency of an OSN newsfeed algorithm influence the users' attitudes and beliefs about childhood vaccines and ultimately their vaccine hesitancy. The third study examines how OSN social overload, through OSN use, can lead to psychological distress and received social support. The research contributes theoretical and practical insights to the literature on the use of online communities in the health context. / Doctor of Philosophy / People use online communities to socialize and to seek out information and help. Online social networks (OSNs) such as Facebook are large communities on which people segregate into smaller groups to discuss joint interests. Some online communities cater to specific needs, such as online health communities (OHCs), which provide platforms for people to talk about the health challenges they or their loved ones are facing. Online communities do not intentionally seek controversy, but because they welcome all perspectives, they have contributed to phenomena such as vaccine hesitancy. Moreover, social overload from the use of OSNs can have both positive and negative psychological effects on users. This dissertation examines the intersection of online communities and health. The first study explains how the interaction of the characteristics of a request for social support made by an OHC user and the characteristics of the wounded healer drive the provision of social support. The model that is developed shows the paths through which the empathy of the wounded healer and the characteristics of the request lead to motivation to provide help to those in need on an OHC. In the second study, the role of characteristics of a newsfeed algorithm, specifically fairness, accountability, and transparency (FAT), in the development of childhood vaccine hesitancy is examined. The findings show that people's perceptions of the newsfeed algorithm's FAT increase their negative attitudes toward vaccination and their perceived behavioral control over vaccination. The third study examines how different uses of OSNs can influence the relationships between social overload and psychological distress and received social support. The findings show how OSN use can be tailored to decrease negative and increase positive psychological consequences without discontinuing use.

Multi-Agent Planning (MAP) is a topic of growing interest that deals with the problem of automated planning in domains where multiple agents plan and act together in a shared environment. In most cases, agents in MAP are cooperative (altruistic) and work together towards a collaborative solution. However, when rational self-interested agents are involved in a MAP task, the ultimate objective is to find a joint plan that accomplishes the agents' local tasks while satisfying their private interests. Among the MAP scenarios that involve self-interested agents, non-cooperative MAP refers to problems where non-strictly competitive agents feature common and conflicting interests. In this setting, conflicts arise when self-interested agents put their plans together and the resulting combination renders some of the plans non-executable, which implies a utility loss for the affected agents. Each participant wishes to execute its plan as it was conceived, but congestion issues and conflicts among the actions of the different plans compel agents to find a coordinated stable solution. Non-cooperative MAP tasks are tackled through non-cooperative games, which aim at finding a stable (equilibrium) joint plan that ensures the agents' plans are executable (by addressing planning conflicts) while accounting for their private interests as much as possible. Although this paradigm reflects many real-life problems, there is a lack of computational approaches to non-cooperative MAP in the literature. This PhD thesis pursues the application of non-cooperative games to solve non-cooperative MAP tasks that feature rational self-interested agents. Each agent calculates a plan that attains its individual planning task, and subsequently, the participants try to execute their plans in a shared environment. We tackle non-cooperative MAP from a twofold perspective. On the one hand, we focus on agents' satisfaction by studying desirable properties of stable solutions, such as optimality and fairness. On the other hand, we look for a combination of MAP and game-theoretic techniques capable of efficiently computing stable joint plans while minimizing the computational complexity of this combined task. Additionally, we consider planning conflicts and congestion issues in the agents' utility functions, which results in a more realistic approach. To the best of our knowledge, this PhD thesis opens up a new research line in non-cooperative MAP and establishes the basic principles to attain the problem of synthesizing stable joint plans for self-interested planning agents through the combination of game theory and automated planning. / La Planificación Multi-Agente (PMA) es un tema de creciente interés que trata el problema de la planificación automática en dominios donde múltiples agentes planifican y actúan en un entorno compartido. En la mayoría de casos, los agentes en PMA son cooperativos (altruistas) y trabajan juntos para obtener una solución colaborativa. Sin embargo, cuando los agentes involucrados en una tarea de PMA son racionales y auto-interesados, el objetivo último es obtener un plan conjunto que resuelva las tareas locales de los agentes y satisfaga sus intereses privados. De entre los distintos escenarios de PMA que involucran agentes auto-interesados, la PMA no cooperativa se centra en problemas que presentan un conjunto de agentes no estrictamente competitivos con intereses comunes y conflictivos. En este contexto, pueden surgir conflictos cuando los agentes ponen en común sus planes y la combinación resultante provoca que algunos de estos planes no sean ejecutables, lo que implica una pérdida de utilidad para los agentes afectados. Cada participante desea ejecutar su plan tal como fue concebido, pero las congestiones y conflictos que pueden surgir entre las acciones de los diferentes planes fuerzan a los agentes a obtener una solución estable y coordinada. Las tareas de PMA no cooperativa se abordan a través de juegos no cooperativos, cuyo objetivo es hallar un plan conjunto estable (equilibrio) que asegure que los planes de los agentes sean ejecutables (resolviendo los conflictos de planificación) al tiempo que los agentes satisfacen sus intereses privados en la medida de lo posible. Aunque este paradigma refleja muchos problemas de la vida real, existen pocos enfoques computacionales para PMA no cooperativa en la literatura. Esta tesis doctoral estudia el uso de juegos no cooperativos para resolver tareas de PMA no cooperativa con agentes racionales auto-interesados. Cada agente calcula un plan para su tarea de planificación y posteriormente, los participantes intentan ejecutar sus planes en un entorno compartido. Abordamos la PMA no cooperativa desde una doble perspectiva. Por una parte, nos centramos en la satisfacción de los agentes estudiando las propiedades deseables de soluciones estables, tales como la optimalidad y la justicia. Por otra parte, buscamos una combinación de PMA y técnicas de teoría de juegos capaz de calcular planes conjuntos estables de forma eficiente al tiempo que se minimiza la complejidad computacional de esta tarea combinada. Además, consideramos los conflictos de planificación y congestiones en las funciones de utilidad de los agentes, lo que resulta en un enfoque más realista. Bajo nuestro punto de vista, esta tesis doctoral abre una nueva línea de investigación en PMA no cooperativa y establece los principios básicos para resolver el problema de la generación de planes conjuntos estables para agentes de planificación auto-interesados mediante la combinación de teoría de juegos y planificación automática. / La Planificació Multi-Agent (PMA) és un tema de creixent interès que tracta el problema de la planificació automàtica en dominis on múltiples agents planifiquen i actuen en un entorn compartit. En la majoria de casos, els agents en PMA són cooperatius (altruistes) i treballen junts per obtenir una solució col·laborativa. No obstant això, quan els agents involucrats en una tasca de PMA són racionals i auto-interessats, l'objectiu últim és obtenir un pla conjunt que resolgui les tasques locals dels agents i satisfaci els seus interessos privats. D'entre els diferents escenaris de PMA que involucren agents auto-interessats, la PMA no cooperativa se centra en problemes que presenten un conjunt d'agents no estrictament competitius amb interessos comuns i conflictius. En aquest context, poden sorgir conflictes quan els agents posen en comú els seus plans i la combinació resultant provoca que alguns d'aquests plans no siguin executables, el que implica una pèrdua d'utilitat per als agents afectats. Cada participant vol executar el seu pla tal com va ser concebut, però les congestions i conflictes que poden sorgir entre les accions dels diferents plans forcen els agents a obtenir una solució estable i coordinada. Les tasques de PMA no cooperativa s'aborden a través de jocs no cooperatius, en els quals l'objectiu és trobar un pla conjunt estable (equilibri) que asseguri que els plans dels agents siguin executables (resolent els conflictes de planificació) alhora que els agents satisfan els seus interessos privats en la mesura del possible. Encara que aquest paradigma reflecteix molts problemes de la vida real, hi ha pocs enfocaments computacionals per PMA no cooperativa en la literatura. Aquesta tesi doctoral estudia l'ús de jocs no cooperatius per resoldre tasques de PMA no cooperativa amb agents racionals auto-interessats. Cada agent calcula un pla per a la seva tasca de planificació i posteriorment, els participants intenten executar els seus plans en un entorn compartit. Abordem la PMA no cooperativa des d'una doble perspectiva. D'una banda, ens centrem en la satisfacció dels agents estudiant les propietats desitjables de solucions estables, com ara la optimalitat i la justícia. D'altra banda, busquem una combinació de PMA i tècniques de teoria de jocs capaç de calcular plans conjunts estables de forma eficient alhora que es minimitza la complexitat computacional d'aquesta tasca combinada. A més, considerem els conflictes de planificació i congestions en les funcions d'utilitat dels agents, el que resulta en un enfocament més realista. Des del nostre punt de vista, aquesta tesi doctoral obre una nova línia d'investigació en PMA no cooperativa i estableix els principis bàsics per resoldre el problema de la generació de plans conjunts estables per a agents de planificació auto-interessats mitjançant la combinació de teoria de jocs i planificació automàtica. / Jordán Prunera, JM. (2017). Non-Cooperative Games for Self-Interested Planning Agents [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/90417

<p dir="ltr">Amidst the rising popularity of digital marketplaces, addressing issues such as non-<br>payment/non-delivery crimes, centralization risks, hacking threats, and the complexity of<br>ownership transfers has become imperative. Many existing studies exploring blockchain<br>technology in digital marketplaces and asset management merely touch upon various application scenarios without establishing a unified platform that ensures trustworthiness and<br>efficiency across the product life cycle. In this thesis, we focus on designing a reliable and efficient e-commerce model to trade various assets. To enhance customer engagement through<br>consensus, we utilize the XGBoost algorithm to identify loyal nodes from the platform entities pool. Alongside appointed nodes, these loyal nodes actively participate in the consensus<br>process. The consensus algorithm guarantees that all involved nodes reach an agreement on<br>the blockchain’s current state. We introduce a novel consensus mechanism named Modified-<br>Practical Byzantine Fault Tolerance (M-PBFT), derived from the Practical Byzantine Fault<br>Tolerance (PBFT) protocol to minimize communication overhead and improve overall efficiency. The modifications primarily target the leader election process and the communication<br>protocols between leader and follower nodes within the PBFT consensus framework.</p><p dir="ltr"><br>In the domain of tangible assets, our primary objective is to elevate trust among various<br>stakeholders and bolster the reputation of sellers. As a result, we aim to validate secondhand<br>products and their descriptions provided by the sellers before the secondhand products are<br>exchanged. This validation process also holds various entities accountable for their actions.<br>We employ validators based on their location and qualifications to validate the products’<br>descriptions and generate validation certificates for the products, which are then securely<br>recorded on the blockchain. To incentivize the participation of validator nodes and up-<br>hold honest validation of product quality, we introduce an incentive mechanism leveraging<br>Stackelberg game theory.</p><p dir="ltr"><br>On the other hand, for optimizing intangible assets management, we employ Non-Fungible<br>Tokens (NFT) technology to tokenize these assets. This approach enhances traceability of<br>ownership, transactions, and historical data, while also automating processes like dividend<br>distributions, royalty payments, and ownership transfers through smart contracts. Initially,<br>sellers mint NFTs and utilize the InterPlanetary File System (IPFS) to store the files related<br>to NFTs, NFT metadata, or both since IPFS provides resilience and decentralized storage solutions to our network. The data stored in IPFS is encrypted for security purposes.<br>Further, to aid sellers in pricing their NFTs efficiently, we employ the Stackelberg mechanism. Furthermore, to achieve finer access control in NFTs containing sensitive data and<br>increase sellers’ profits, we propose a Popularity-based Adaptive NFT Management Scheme<br>(PANMS) utilizing Reinforcement Learning (RL). To facilitate prompt and effective asset<br>sales, we design a smart contract-powered auction mechanism.</p><p dir="ltr"><br>Also, to enhance data recording and event response efficiency, we introduce a weighted<br>L-H index algorithm and transaction prioritization features in the network. The weighted<br>L-H index algorithm determines efficient nodes to broadcast transactions. Transaction prior-<br>itization prioritizes certain transactions such as payments, verdicts during conflicts between<br>sellers and validators, and validation reports to improve the efficiency of the platform. Simulation experiments are conducted to demonstrate the accuracy and efficiency of our proposed<br>schemes.<br></p>

<p dir="ltr">The pursuit of fairness in machine learning (ML) systems is a critical challenge in today’s world that relies heavily on AI systems. However, computing and mitigating the bias necessitates substantial computational resources and time when evaluating across entire datasets. This research introduces an innovative approach to estimate fairness in ML systems by leveraging data characteristics and constructing a metafeatures dataframe. Using our methodology enables the prediction of fairness with significantly reduced computational cost and expedited analysis times. Furthermore, our approach is scalable to different distributions and requires minimal training to deal with out of sample data. This approach not only enhances the efficiency of fairness assessments in ML systems but also provides a scalable framework for future fairness evaluation methodologies. Our findings suggest that using data characteristics to estimate fairness is not only feasible but also effective, offering a promising avenue for developing more equitable ML systems with reduced resource consumption.</p>

Dans cette thèse, les travaux constitutifs demandent "Qu'est-ce qui est gagné ou perdu lorsque nous faisons varier le nombre de paramètres ?". Cette question est de plus en plus pertinente à l'ère de la recherche scientifique où la mise à l'échelle des réseaux est une recette largement utilisée pour garantir des gains de performance, mais où l'on ne comprend pas bien comment les changements de capacité modifient les propriétés de généralisation. Cette thèse mesure la généralisation selon plusieurs dimensions différentes, englobées par des questions telles que \textit{``Certains types d'exemples ou de classes sont-ils affectés de manière disproportionnée par la compression ?''} et \textit{``La variation du nombre de poids amplifie-t-elle la sensibilité aux erreurs corrompues ? contributions?''}. Pour explorer l'impact de la variation du nombre de poids, nous exploitons l'élagage - une classe de techniques de compression largement utilisée qui introduit un niveau de parcimonie des poids. Cela nous permet de faire varier précisément le nombre de poids apprenables dans les réseaux que nous comparons. Nous constatons à la fois dans la computer vision et dans NLP, à travers différents ensembles de données et tâches, que la parcimonie amplifie l'impact disparate sur les performances du modèle entre les sous-groupes de données minoritaires et majoritaires, de sorte que les \textit{les riches deviennent plus riches et les pauvres s'appauvrissent}. Même si l’erreur moyenne globale reste largement inchangée lorsqu’un modèle est compressé, les attributs sous-représentés encourent une part disproportionnée de l’erreur. Les modèles clairsemés \emph{cannibalisent} les performances sur les attributs protégés sous-représentés pour préserver les performances sur les attributs majoritaires. La compression amplifie également la sensibilité à certains types de perturbations. Nous trouvons également quelques mises en garde importantes : dans les contextes à faibles ressources où les données sont très spécialisées et distinctes des tâches en aval, la rareté aide en freinant la mémorisation et en induisant l'apprentissage d'une représentation plus générale. Les travaux inclus dans cette thèse suggèrent qu'il existe des rendements clairement décroissants pour une formule simple de paramètres d'échelle. Nos résultats ont de puissantes implications : la plupart des paramètres apprenables sont utilisés pour apprendre un ensemble de points de données qui portent une part disproportionnée de l'erreur. Nous appelons ces points de données Pruning Identified Exemplars (\texttt{PIEs}). Nous constatons que la majorité des poids sont nécessaires pour améliorer les performances sur ce petit sous-ensemble de la distribution de l’entraînement. Cela suggère qu'une petite fraction de la distribution de formation a un \textit{coût par capacité de données} beaucoup plus élevé. Cohérents dans tous les chapitres de cette thèse, nos résultats soutiennent la recommandation selon laquelle les modèles clairsemés doivent être soigneusement audités avant d'être déployés à grande échelle. L’un des principaux points à retenir de notre travail est que nos algorithmes ne sont pas impartiaux et que certains choix de conception peuvent amplifier les dommages. Il est essentiel de comprendre cet impact disparate compte tenu du déploiement généralisé de modèles compressés dans la nature. Nos résultats soutiennent la recommandation selon laquelle les modèles compressés font l'objet d'un audit supplémentaire avant leur déploiement dans la nature. / In this thesis, the constituent works ask “What is gained or lost as we vary the number of parameters?”. This question is increasingly relevant in an era of scientific inquiry where scaling networks is a widely used recipe to secure gains in performance but where there is not a good understanding of how changes in capacity alter generalization properties. This thesis measures generalization along several different dimensions, encompassed by questions such as “Are certain types of examples or classes disproportionately impacted by compression?” and “Does varying the number of weights amplify sensitivity to corrupted inputs?”. To explore the impact of varying the number of weights, we leverage pruning – a widely used class of compression techniques that introduce weight-level sparsity. We find in both computer vision and language settings, across different datasets and tasks that sparsity amplifies the disparate impact in model performance between minority and majority data subgroups such that the rich get richer and the poor get poorer. While the overall average error is largely unchanged when a model is compressed, underrepresented attributes incur a disproportionately high portion of the error. The work included in this thesis suggests that there are clearly decreasing returns to a simple formula of scaling parameters. Most learnable parameters are used to learn a set of data points which bears a disproportionately high portion of the error. We term these data points Pruning Identified Exemplars (PIEs). This suggests that a small fraction of the training distribution has a far higher per-data capacity cost. Consistent across all chapters of this thesis, our results support the finding that popular compression techniques are not impartial and can amplify harm. Our results support the recommendation that compressed models receive additional auditing scrutiny prior to deployment in the wild.

This dissertation examines correlations between religiousness and spirituality, to moral orientations using moral foundations theory as a framework. Using the 2012 Measuring Morality dataset, which provides a representative sample of the population of the United States, I create linear regressions which test associations between religiousness, spirituality, and each of the five moral foundations ((harm/care, fairness, in-group loyalty, respect for authority, and purity). I find that religiousness is negatively associated with concern for harm, and positively associated with respect for authority, a finding which implies that the moral behavior of religious people is rooted in respect for authority more than in any other moral concern. Spirituality is positively associated with concern for fairness. The implications of all findings are discussed, as well as limitations and recommendations for future research.

The dissertation examines the process of retrenchment in a public institution with reference to the execution of such a process by the Independent Electoral Commission (IEC). The aim is to establish whether or not the retrenchment was substantively and procedurally fair as required by legislation. Notwithstanding the fact that the staff may have been disadvantaged by the short retrenchment notice (the staff did not have representation prior to the announcement, and the swiftness of the process did not, under the circumstances, provide the staff with enough time to comprehensively apply their mind to the underlying issues), the dissertation finds that the retrenchments had been substantively fair given the fact that the IEC could not retain all staff because of budgetary constraints. The dissertation also finds that the process had been procedurally fair in accordance with section 189 of the Labour Relations Act, 66 of 1995. / Public Adminstration & Development Studies / M.A. (Public Administration)