Made available in DSpace on 2016-04-29T13:18:02Z (GMT). No. of bitstreams: 1 Daniel Carvalho.pdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Previous issue date: 2007-04-04 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / The present research had two goals: (a) to test the efficacy of a delayed prompt procedure to develop errorless discrimination between letters and their rotations, and (b) to test the effects of the discrimination training when MTS trials with the trained letters, theirs rotations, and similar letters were presented. Participated in the study 9 children (aged 3 to 6) who did not have a perfect performance when discrimination between pairs of graphically similar letters and their rotations (b, d, n, u, p, q, and their 270o rotations, and the letter B, D, N, U, P, Q and their 180o rotations). A test with identity-matching-to-sample trials involving the S+ letters as sample stimuli and the same letters, their rotations and similar letters as comparison stimuli was first conducted. A discrimination training of each testes letter (S+) and its rotation (S_) was then conducted: During training, colors (previously established as S+ and S-) were superimposed on the new S+ (letter) and S- (letter-rotation) with longer delays after each successful trial, until the child systematically responded choosing S+ before the color was presented. After training the initial identity MTS test was replicated, followed by a test of arbitrary MTS: trials involving letters and colors were presented, to test the emergency of conditional discrimination. Results suggested that 4 children developed conditional discriminations indicating the emergency of arbitrary stimulus classes. Results also indicated that although the discrimination training was effective in producing simple discrimination, it was not sufficient to guarantee good performance when identity-matching-to-sample trials were presented / No presente trabalho um procedimento de discriminação simples simultânea sem erro - dica atrasada - entre pares de letras graficamente semelhantes e suas inversões. Testou-se também os possíveis efeitos do treino discriminativo em tentativas de emparelhamento de acordo com o modelo (matching to sample) de identidade e arbitrário em que as letras que participaram do treino discriminativo eram estímulos comparação. Nove crianças (de 3 a 6 anos) com dificuldades nas discriminações entre pares de letras graficamente semelhantes e suas inversões (b-d, n-u, p-q e inversões em 270o e B-D, N-U, P-Q e inversões em 180 o) participaram da pesquisa. Depois de um pré-teste de MTS de identidade em que eram estímulos modelo letras e estímulos comparação as mesmas letras, letras semelhantes e suas inversões, as crianças selecionadas passaram por treinos discriminativos simples simultâneos, envolvendo letras (S+) e inversões (S-). No treino, cores foram estabelecidas como estímulos discriminativos (S+) ou como S-. A partir daí tais estímulos foram sobrepostos a novos estímulos e tal sobreposição foi atrasada a cada tentativa bem sucedida até que o participante escolhesse o sistematicamente o novo S+ (letra) antes da sobreposição de cores entre letras e suas entre cores (previamente após o treino, foi feito um pós- teste semelhante ao pré teste (MTS) de identidade) e um teste de matching arbitrário envolvendo as letras treinadas, suas inversões e as cores como estímulos modelo e comparação, para verificar a possível emergência de discriminações condicionais entre letras e cores, o que seria indicativo da formação de classes de estímulos equivalentes. Todos os participantes tiveram desempenhos bem sucedidos no treino discriminativo - quase sem erros. Quatro participantes tiveram desempenhos nos testes de MTS arbitrário que atestariam a emergência de discriminações condicionais que indicariam a possibilidade de formação de classes de estímulos. Os resultados dos pós-testes de MTS de identidade indicaram que para vários participantes o treino de discriminação simples não foi suficiente para garantir um bom desempenho na situação de MTS. Finalmente, em geral o desempenho dos participantes foi melhor quando os estímulos foram as letras maiúsculas

Cette thèse étudie l'affectation centralisée des enseignants aux écoles et un nouveau modèle d'appariement inspirée par cette dernière.Dans le premier chapitre, nous développons un modèle théorique de réaffectation afin d'étudier le problème de réaffecter des enseignants titulaires enseignant au sein d'un établissement et demandant une mutation. Le problème est similaire à celui d'affecter des élèves dans des écoles. Dans ce cas, l'algorithme à Acceptation Différée a été identifié comme étant le seul algorithme qui: i) est stable, ii) efficace et qui iii) incite les élèves à soumettre sincèrement leurs préférences. La différence principale avec le problème d'affecter des élèves aux écoles est que les enseignants ont déjà une position initiale au sein d'un établissement. On doit donc prendre en compte une contrainte additionnelle, la Rationalité Individuelle (RI): un enseignant doit être affecté dans un établissement qu'il préfère faiblement à son établissement d'origine. Pour prendre en compte cette contrainte, une modification de l'algorithme à Acceptation Différée a été identifiée dans la littérature académique et utilisé en pratique pour affecter les enseignants aux écoles en France. Nous montrons que cet algorithme modifié souffre d'un important défaut: il n'est pas efficace au sens fort. Il est en effet possible de réaffecter les enseignants aux écoles de telle sorte que: i) les enseignants obtiennent une école qu'ils préfèrent et ii) les écoles obtiennent des enseignants mieux classés. Partant de ce constat, nous identifions la classe de tous les algorithmes, les algorithmes Block-Exchange (BE), qui ne souffrent pas de ce défaut. Parmi eux, nous montrons qu'il en existe un unique qui incite les enseignants à soumettre leurs préférences sincèrement: le Teacher Optimal Block-Exchange algorithm (TO-BE). En utilisant un modèle de marché large, nous montrons théoriquement que ces algorithmes ont de meilleures performances en termes de mouvement et de bien-être des enseignants que l'algorithme actuel. Nous utilisons ensuite une base de données sur l'affectation des enseignants aux écoles du secondaire en France en 2013 pour quantifier les gains possibles que nos algorithmes peuvent apporter. Dans un cadre de réaffectation pur sans enseignant néotitulaire et places vacantes, nous montrons qu'il est possible de plus que double le nombre d'enseignants obtenant une nouvelle affectation.Dans le second chapitre, nous concevons un algorithme pratique, inspiré de nos résultats du chapitre précédent, pour la procédure française d'affectation des enseignants du secondaire. Plus globalement, cette conception a également pour but de fournir un outil face à deux problèmes importants communs aux pays de l'OCDE: i) le manque d'attractivité de la profession enseignants et ii) les importantes inégalités de réussites des élèves issus de milieux sociaux différents. Nous considérons l'ensemble du marché français composé des enseignants titulaires demandant une réaffectation, les enseignants sans affectation initiale et des places vacantes. Améliorer la mobilité des enseignants permet de leur donner de meilleures perspectives de carrière ce qui peut potentiellement attirer plus d'entrants dans la profession. Mais cette mobilité accrue peut entrainer l'affectation de plus d'enseignants peu expérimentés au sein d'académies déjà très défavorisées, affectant in fine la réussite des élèves au sein de celles-ci. Nous proposons un algorithme flexible qui permet de mieux contrôler le mouvement et la distribution des enseignants au sein de régions, notamment celles très désavantagées. En utilisant les données françaises d'affectation de 2013, nous simulons plusieurs scénarios contre factuels et montrons que notre algorithme peut prendre en compte plusieurs objectifs de politique publique. / This thesis studies the centralized assignment of teachers to schools and a new matching framework inspired by it. In the first chapter, we develop a theoretical model of reassignment to study the problem of reassigning tenured teachers who already have a position and are willing to move to another school. The problem is similar to the one of assigning students to schools. In this case, the well known Deferred Acceptance algorithm has been identified as the only algorithm that: i) is stable ii) efficient and iii) gives incentives to students to report their true preferences. The main difference with the problem of assigning students to schools is that teachers have an initial assignment. One has to consider an additional constraint, Individual Rationality (IR): a teacher must receive a school that he weakly prefers to his initial one. To incorporate this constraint, a modification of the Deferred Acceptance algorithm has been identified in the academic literature and used in practice to assign teachers to schools in France. We show that this modified algorithm has a serious drawback: it is not efficient in a strong sense. Indeed, it is possible to reassign teachers to schools such that both: i) teachers obtain a school that they prefer and ii) schools are assigned teachers that they rank higher. Thus, we identify the class of all algorithms, the Block-Exchange (BE) algorithms, that do not suffer from this drawback. Among them, we show that there is a unique one that gives good incentives to teachers to report their true preferences, the Teacher Optimal Block-Exchange algorithm (TO-BE). In using a large market setting, we theoretically show that these algorithms perform better in terms of movement and welfare for teachers than the currently used one. We then use a dataset on the assignment of teachers to schools in France in 2013 to quantify the possible gains that can bring our algorithms. In a reassignment setting with no newly tenured teachers or empty seats, we show that we can more than double the number of teachers obtaining a new assignment. In the second chapter, we aim to design a practical algorithm, inspired by our findings in the previous chapter, for the French assignment system of teachers to schools. More generally, this design also aims to provide a tool about two important issues common to OECD countries: i) the lack of attractiveness of the teaching profession and ii) the high achievement inequality between students from different social backgrounds. We consider the complete French market composed of tenured teachers looking for a reassignment, newly tenured teachers with no initial assignment and empty positions. In improving the mobility of teachers, one can give them better career perspectives and so potentially attract more teachers into the profession. But in doing so, it can also hurt deprived regions in assigning more teachers with low experience to them and ultimately the students from these regions. We propose a flexible algorithm that allows to better control the movement and distribution of teachers across regions, especially deprived ones. Using the data of the French assignment of teachers in 2013, we simulate several counter factuals and show that our algorithm can accommodate a wide range of policy objectives.

This dissertation examines how matching -- in marriage markets and labor markets -- can change under certain market circumstances and under different information provisions. The first two chapters analyze marriage market, with a particular focus on the impacts of cross-border marriage in marriage markets. Given the severely male-biased sex ratios in many Asian countries including China and India, demands for foreign brides are expected to grow in the near future. In the first chapter, I theoretically investigate the impacts of cross-border marriage on marital patterns and surplus division of couples. I use a frictionless transferable utility matching framework to analyze how cross-border marriage affects matching patterns and marital shares for couples. In the second chapter, I test the model's predictions, focusing on Taiwan (a wealthier side with male biased sex ratios) and Vietnam (a poorer side with balanced sex ratios in the marriage market). I find that cross-border marriages are predominantly made up of Taiwanese men and Vietnamese women; Taiwanese men are selected from the middle level of the socioeconomic status distribution, and Vietnamese women are positively selected. Moreover, cross-border marriage significantly affects men and women who stay in their own countries without engaging in cross-border marriage, by altering marriage rate, matching partners, and intra-household allocations within the households. My results suggest that changes in trade and immigration policies can have far-reaching implications on marital outcomes and women's bargaining power. The third chapter investigates job and jobseeker matching in labor market. Specifically, it explores whether inaccurate expectations of job seekers about their competitiveness contribute to poor job matching in developing countries. We utilize the largest online job portal in the Middle East and North Africa region to evaluate the effect of an intervention providing information about own competitiveness to job applicants. Providing information about the relative fit of an applicant's background for a particular job causes job seekers to apply for jobs that are better matches given their background. The effects of information are the largest among entry-level workers with higher levels of education, who generally face the highest unemployment rates in the region. The findings are consistent with the hypothesis that changes over time in demand for skills in the job market may lead to inaccurate expectations that hinder labor market matching. Improving the efficiency of online job search may be particularly welfare-enhancing in the Middle East and North Africa region given that the young, highly-educated subpopulation that faces the greatest labor market hurdles also has the highest level of internet connectedness.

Matching market design studies the fundamental problem of how to allocate scarce resources to individuals with varied needs. In recent years, the theoretical study of matching markets such as medical residency, public housing and school choice has greatly informed and improved the design of such markets in practice. Impactful work in matching market design frequently makes use of techniques from computer science, economics and operations research to provide end–to-end solutions that address design questions holistically. In this dissertation, I develop tools for optimization in market design by studying matching mechanisms for school choice, an important societal problem that exemplifies many of the challenges in effective marketplace design. In the first part of this work I develop frameworks for optimization in school choice that allow us to address operational problems in the assignment process. In the school choice market, where scarce public school seats are assigned to students, a key operational issue is how to reassign seats that are vacated after an initial round of centralized assignment. We propose a class of reassignment mechanisms, the Permuted Lottery Deferred Acceptance (PLDA) mechanisms, which generalize the commonly used Deferred Acceptance school choice mechanism and retain its desirable incentive and efficiency properties. We find that under natural conditions on demand all PLDA mechanisms achieve equivalent allocative welfare, and the PLDA based on reversing the tie-breaking lottery during the reassignment round minimizes reassignment. Empirical investigations on data from NYC high school admissions support our theoretical findings. In this part, we also provide a framework for optimization when using the prominent Top Trading Cycles (TTC) mechanism. We show that the TTC assignment can be described by admission cutoffs, which explain the role of priorities in determining the TTC assignment and can be used to tractably analyze TTC. In a large-scale continuum model we show how to compute these cutoffs directly from the distribution of preferences and priorities, providing a framework for evaluating policy choices. As an application of the model we solve for optimal investment in school quality under choice and find that an egalitarian distribution can be more efficient as it allows students to choose schools based on idiosyncracies in their preferences. In the second part of this work, I consider the role of a marketplace as an information provider and explore how mechanisms affect information acquisition by agents in matching markets. I provide a tractable “Pandora's box” model where students hold a prior over their value for each school and can pay an inspection cost to learn their realized value. The model captures how students’ decisions to acquire information depend on priors and market information, and can rationalize a student’s choice to remain partially uninformed. In such a model students need market information in order to optimally acquire their personal preferences, and students benefit from waiting for the market to resolve before acquiring information. We extend the definition of stability to this partial information setting and define regret-free stable outcomes, where the matching is stable and each student has acquired the same information as if they had waited for the market to resolve. We show that regret-free stable outcomes have a cutoff characterization, and the set of regret-free stable outcomes is a non-empty lattice. However, there is no mechanism that always produces a regret-free stable matching, as there can be information deadlocks where every student finds it suboptimal to be the first to acquire information. In settings with sufficient information about the distribution of preferences, we provide mechanisms that exploit the cutoff structure to break the deadlock and approximately implement a regret-free stable matching.

O reconhecimento aproximado de cadeias de texto é um problema recorrente em diversas aplicações onde o computador é utilizado como meio de processamento de uma massa de dados sujeita a imprecisões, erros e distorções. Existem inúmeras metodologias, técnicas e métricas criadas e empregadas na resolução deste tipo de problema, mas a maioria delas é inflexível em pelo menos um dos seguintes pontos: arquitetura, métrica utilizada para aferir o erro encontrado ou especificidade na aplicação. Esse trabalho propõe e analisa a utilização dos Autômatos Finitos Nebulosos para a resolução desse tipo de problema. A teoria nebulosa oferece uma base teórica sólida para o tratamento de informações inexatas ou sujeita a erros, enquanto o modelo matemático dos autômatos finitos é uma ferramenta consolidada para o problema de reconhecimento de cadeias de texto. Um modelo híbrido não só oferece uma solução flexível para a resolução do problema proposto, como serve de base para a resolução de inúmeros outros problemas que dependem do tratamento de informações imprecisas. / The approximate string matching problem is recurring in many applications where computer is used to process imprecise, fuzzy or spurious data. An uncountable number of methods, techniques and metrics to solve this class of problem are available, but many of them are inflexible at least in one of following: architecture, metric or application specifics. This work proposes and analyzes the use of Fuzzy Finite State Automata to solve this class of problems. The fuzzy theory grants a solid base to handle imprecise or fuzzy information; the finite state automata is a classic tool in string matching problems. A hybrid model offers a flexible solution for this class of problem and can be a base for other problems related with imprecise data processing.

This thesis consists of three theoretical chapters, all related to the response of unemployment to shocks and the role of active and passive labour market policies. Throughout the thesis, unemployment is assumed to evolve as a result of the uncoordinated nature of the labour market along the lines outlined in the Diamond-Mortensen-Pissarides equilibrium search and matching model. Chapter 2 examines the effects of employment policies on vacancy creation and allocation decisions of firms and unemployment across workers with different skills. We develop a partial equilibrium model with heterogeneous high- and low-tech jobs and with skilled and unskilled workers, which we motivate by the stark evidence on the incidence of cross-skill employment (which crowds out unskilled workers, e.g. evidence for the US, the UK and the EU put these at 58%, 32%, and 35%, respectively). We show that certain employment protection policies could, in fact, lead to a reduction in job creation and might alter the allocation of vacancies across low- and high-tech job type. We find that: (i) skilled workers benefit while unskilled workers experience high jobless rate; (ii) policy effects differ when they are skill-specific; (ii) stricter policies can have more severe consequences; and (iv) vacancy creation subsidy can play a key role in reducing unemployment across worker type as well as alleviating the cross-skill crowding out of jobs. Against conventional wisdom, we demonstrate that severance compensation can have a ‘real’ effect on job creation decision, provided there is some degree of strictness in its enforcement. Motivated by the extensive use of fiscal stimulus policies and labour market reforms during the last economic crisis, in Chapter 3 we study the implications of labour market regulations in driving the sensitivity of an economy to fiscal spending shocks, in a Dynamic Stochastic General Equilibrium (DSGE) model with job search frictions. We demonstrate that less rigidity in the labour market reduces the impact of fiscal demand shock on job creation and employment, both at extensive and intensive margins, whereas higher rigidity amplifies it. We also establish that the extent to which government spending promotes economic activity, job creation and employment depends on the degree of substitutability between private and public consumption. Higher substitutability dampens economic activity and reduces the sizes of output and employment multipliers. Labour market-oriented fiscal spending is found to be the most potent policy instruments for promoting employment – especially in the presence of high labour market rigidities. Finally, in Chapter 4, we study how openness to international trade and capital mobility and their interactions with labour market policies affect the behaviour of an economy, in particular with respect to its unemployment level. We show that the degree of openness to international capital flow is crucial for understanding the response of unemployment to different shocks. In isolation, by raising the incentive to invest, a reduction in capital mobility barriers leads to lower unemployment, both in the long-run and the dynamic short-run. With limited restrictions to capital movement, unemployment responds faster and with greater magnitude to a domestic productivity shock, and this is further enhanced the more the economy is open to international trade. A striking finding of this study is that while a higher degree of capital mobility enhances the adjustment of unemployment in response to a domestic productivity shock, it dampens its adjustment to a foreign demand shock. By contrast, higher openness to international trade enhances the adjustment effects of both shocks on unemployment. Finally, we find that heterogeneity in the welfare state systems in the EU can generate substantial differentials in the adjustment of unemployment to various shocks.

O reconhecimento aproximado de cadeias de texto é um problema recorrente em diversas aplicações onde o computador é utilizado como meio de processamento de uma massa de dados sujeita a imprecisões, erros e distorções. Existem inúmeras metodologias, técnicas e métricas criadas e empregadas na resolução deste tipo de problema, mas a maioria delas é inflexível em pelo menos um dos seguintes pontos: arquitetura, métrica utilizada para aferir o erro encontrado ou especificidade na aplicação. Esse trabalho propõe e analisa a utilização dos Autômatos Finitos Nebulosos para a resolução desse tipo de problema. A teoria nebulosa oferece uma base teórica sólida para o tratamento de informações inexatas ou sujeita a erros, enquanto o modelo matemático dos autômatos finitos é uma ferramenta consolidada para o problema de reconhecimento de cadeias de texto. Um modelo híbrido não só oferece uma solução flexível para a resolução do problema proposto, como serve de base para a resolução de inúmeros outros problemas que dependem do tratamento de informações imprecisas. / The approximate string matching problem is recurring in many applications where computer is used to process imprecise, fuzzy or spurious data. An uncountable number of methods, techniques and metrics to solve this class of problem are available, but many of them are inflexible at least in one of following: architecture, metric or application specifics. This work proposes and analyzes the use of Fuzzy Finite State Automata to solve this class of problems. The fuzzy theory grants a solid base to handle imprecise or fuzzy information; the finite state automata is a classic tool in string matching problems. A hybrid model offers a flexible solution for this class of problem and can be a base for other problems related with imprecise data processing.

The purpose of this study was to explore the relationship of dual enrollment course participation by comparing first-time, full-time traditional community college students who participated in dual enrollment (N=246) to peers (N=986) that did not participate. Dual enrollment participation was defined as taking one or more dual enrollment courses. The population for this study (N=1,232) included first-time, full-time students who graduated from public high schools in the service area of Northeast State Community College over a five year span from 2008 through 2012. Propensity score matching eliminated self-selection bias by controlling for confounding covariates such as parental education, high school GPA, and ACT scores. The major findings of the study included the following: dual enrollment participants (a) were nearly four times less likely to take remediation than non-participants, (b) earned approximately 1 extra credit hour in the first semesters of college, (c) earned higher first semester GPAs, (d) were 2.5 times more likely to graduate in 2 years (100% of degree time) and, (e) were 1.68 times more likely to graduate in 3 years (150% of degree time). The study concluded that dual enrollment benefits community college students in Tennessee, both at the beginning and completion of college. This is a significant justification for the current investment in dual enrollment by the State of Tennessee and for further increasing access to dual enrollment for all students, especially for students that live in rural areas, experience poverty, or are underrepresented in higher education.

The PSUBOT (pronounced pea-es-you-bought) is an autonomous wheelchair robot for persons with certain disabilities. Its use of voice recognition and autonomous navigation enable it to carry out high level commands with little or no user assistance. We first describe the goals, constraints, and capabilities of the overall system including path planning and obstacle avoidance. We then focus on localization-the ability of the robot to locate itself in space. Odometry, a compass, and an algorithm which matches points to lines are each employed to accomplish this task. The matching algorithm (which matches "points" to "lines") is the main contribution to this work. The .. points" are acquired from a rotating sonar device, and the "lines" are extracted from a user-entered line-segment model of the building. The algorithm assumes that only small corrections are necessary to correct for odometry errors which inherently accumulate, and makes a correction by shifting and rotating the sonar image so that the data points are as close as possible to the lines. A modification of the basic algorithm to accommodate parallel lines was developed as well as an improvement to the basic noise removal algorithm. We found that the matching algorithm was able to determine the location of the robot to within one foot even when required to correct for as many as five feet of simulated odometry error. Finally, the algorithm's complexity was found to be well within the processing power of currently available hardware.

Objective: To assess the ability of propensity score methods to maintain covariate balance and minimize bias in the estimation of treatment effect in a time-to-event setting. Data Sources: Generated simulation model Study Design: Simulation study Data Collection: 6 scenarios with varying covariate relationships to treatment and outcome with 2 different censoring prevalences Principal Findings: As time lapses, balance achieved at baseline through propensity score methods between treated and untreated groups trends toward imbalance, particularly in settings with high rates of censoring. Furthermore, there is a high degree of variability in the performance of different propensity score models with respect to effect estimation. Conclusions: Caution should be used when incorporating propensity score analysis methods in survival analyses. In these settings, if model over-parameterization is a concern, Cox regression stratified on propensity score matched pairs often provides more accurate conditional treatment effect estimates than those of unstratified matched or IPT weighted Cox regression models.