Objetivos: Identificar a distribuição do índice de massa corporal (IMC), da circunferência da cintura (Ccintura), da razão cintura-quadril (RCQ) e do índice de conicidade (IC), segundo variáveis demográficas, socioeconômicas, relacionadas à saúde e comportamentais na população de 30 anos e mais, residente no município de Ribeirão Preto - SP, em 2006; avaliar o efeito modulador das regiões geográficas do município sobre os fatores associados aos indicadores antropométricos e estimar a contribuição do nível agregado para esses desfechos. Métodos: Estudo epidemiológico transversal, de base populacional, com amostragem em três estágios de sorteio. Pesos amostrais foram calculados para a recomposição do total de elegíveis e correção da taxa de não-resposta, em cada setor censitário, originando amostra ponderada de 2.197 participantes. Médias e intervalos de confiança (95%) dos indicadores antropométricos foram calculados, segundo sexo, nas categorias das variáveis independentes. Análise de variância, com um critério de classificação foi utilizada para a comparação das médias dos desfechos nas categorias das variáveis independentes. Testes de tendência linear foram aplicados para variáveis com mais de duas categorias. O nível de significância adotado foi ?=5%. Para a identificação de fatores associados aos indicadores antropométricos foram construídos modelos lineares multinível de efeitos fixos com dois níveis. Os 81 setores censitários sorteados foram agrupados em quatro regiões geográficas do município (nível agregado). A análise multinível seguiu modelo conceitual hierarquizado para avaliação do efeito direto de variáveis de nível individual e a contribuição do nível agregado sobre os desfechos (rho). Medidas de efeito (?) foram estimadas por pontos e por intervalos com 95% de confiança para as variáveis independentes. As medidas de associação foram calculadas, inicialmente, em modelos parciais que incluíram as variáveis de cada bloco hierárquico, ajustadas para as de blocos precedentes. Nos modelos finais, permaneceram as variáveis que mantiveram significância estatística (p < 0,05), após ajustamento simultâneo para os blocos precedentes. Para cada um dos desfechos considerados, calcularam-se médias ajustadas nas categorias das variáveis independentes, segundo regiões geográficas do município. Diferenças das médias ajustadas entre os extremos das categorias das variáveis independentes foram calculadas para avaliar o efeito modulador das regiões geográficas. Todas as estimativas calculadas levaram em consideração o efeito de desenho amostral. Resultados: Em relação ao IMC e em ambos os sexos, médias de maior magnitude foram detectadas entre os que referiram antecedentes familiares de excesso de peso, história pessoal de obesidade, utilizaram medicamentos nos últimos 15 dias e seguiram dieta alimentar para perda de peso. No sexo feminino, médias mais elevadas de IMC foram detectadas entre as mais velhas, com menor escolaridade, maior tempo de residência no município e que referiram antecedentes familiares de acidente vascular cerebral (AVC). O elenco de fatores associados ao IMC foi constituído por: sexo (?= -0,814; IC95% : -1,514 - -0,113), idade (?= 0,050; IC95% : 0,014 - 0,086), escolaridade (?= -0,101; IC95% : -0,206 - -0,005) , antecedentes familiares de excesso de peso (?= 1,214; IC95% : 0,468 - 1,961), história pessoal de obesidade (?= 6,422; IC95%: 5,724 - 7,119) e consumo de lipídios (?= 0,029; IC95% : 0,008 - 0,051), considerando-se a contribuição de 11% do nível agregado. As menores diferenças entre as médias ajustadas do IMC foram detectadas, principalmente, nos extremos das categorias das variáveis: sexo, faixas etárias, escolaridade e consumo de lipídios nas regiões Norte e Oeste do município. Médias de maior magnitude para os três indicadores de obesidade central (Ccintura, RCQ e IC), foram observadas entre os participantes mais velhos, com menor escolaridade, classificados nos estratos mais baixos de renda, com maior tempo de residência no município, com antecedentes familiares de excesso de peso, história pessoal de obesidade e entre aqueles classificados como \"pré-obesos\" e \"obesos\". O elenco de fatores associados à Ccintura foi constituído por: sexo (?= -8,686; IC95% : -9,439 - -7,932), idade (?= 0,169; IC95% : 0,122 - 0,216), antecedentes familiares de AVC (?= 0,689; IC95% : 0,051 - 1,327), história pessoal de obesidade (?= 2,363; IC95% : 0,320 - 4,410), IMC (?= 1,689; IC95% : 1,437 - 1,941), nº de medicamentos (?= 0,259; IC95% : 0,078 - 0,440), tempo de tabagismo (?= 0,035; IC95% : 0,008 - 0,061) e energia total da dieta (?= 0,084; IC95% : 0,012 - 0,157), considerando-se a contribuição de 12,4% do nível agregado. As menores diferenças entre as médias ajustadas da Ccintura foram detectadas nos extremos das categorias das variáveis: antecedentes familiares de AVC, tempo de tabagismo e energia total da dieta nas regiões Norte e Oeste. Em relação à RCQ, permaneceram associadas as variáveis: sexo (?= -0,099; IC95% : -0,107 - -0,090), idade (?= 0,002; IC95% : 0,001 - 0,003), IMC (?= 0,004; IC95% : 0,003 - 0,005), nº de medicamentos (?= 0,004; IC95% : 0,001 - 0,007), tempo de tabagismo (?= 0,0006; IC95% : 0,0004 - 0,0009), dependência de álcool (?= 0,013; IC95% : 0,003 - 0,022), dieta para perder peso (?= 0,017; IC95% : 0,004 - 0,030) e consumo de carboidratos (?= 0,0001; IC95% : 0,0001 - 0,0002). A contribuição do nível agregado foi equivalente a 12%. As menores diferenças entre as médias ajustadas da RCQ foram detectadas nos extremos das categorias das variáveis: nº de medicamentos e consumo de carboidratos na região Oeste. O elenco de fatores associados ao IC foi composto por: sexo (?= -0,068; IC95% : -0,076 - -0,056), idade (?= 0,003; IC95% : 0,002 - 0,004), antecedentes familiares de AVC (?= 0,008; IC95% : 0,001 - 0,016), IMC (?= 0,005; IC95% : 0,004 - 0,006), acesso a serviços de saúde (?= -0,011; IC95% : -0,020 - -0,002), nº de medicamentos (?= 0,004; IC95% : 0,002 - 0,007), tempo de tabagismo (?= 0,0004; IC95% : 0,0001 - 0,0007), dependência de álcool (?= 0,011; IC95% : 0,001 - 0,020), dieta para perder peso (?= 0,015; IC95% : 0,002 - 0,030) e consumo de ácidos graxos monoinsaturados (?= -0,0005; IC95% : -0,0010 - -0,0001), com contribuição do nível agregado equivalente a 14,1%. As menores diferenças entre as médias ajustadas do IC foram detectadas nos extremos das categorias das variáveis: acesso a serviços de saúde e dependência de álcool, na região Norte. Conclusões: No município de Ribeirão Preto, as regiões Norte e Oeste foram identificadas como regiões obesogênicas. Medidas de promoção de hábitos saudáveis, bem como de prevenção da obesidade, no município, não devem ser direcionadas apenas para mudanças de hábitos individuais, visto que o impacto de tais intervenções pode ser minimizado ou refreado pelo efeito de contexto destas regiões. / Objectives: To identify the distribution of the body mass index (BMI), waist circumference (waistC), waist-to-hip ratio (WHR) and conicity index (CI), according to demographic, socioeconomic, health-related and behavioral variables, in the population aged 30 years and older, living in Ribeirão Preto - SP, Brazil, in 2006; to assess shaping effect of geographical regions on associated factors related to anthropometric indexes as well as to estimate the contribution of the ecological level to the outcomes. Methods: Population-based cross-sectional epidemiological study with multistage sampling. Sampling weights were calculated to compose the total number of eligible participants and correct for the non-response rate, in each census tract, resulting in a weighted sample of 2,197 participants. Means and confidence intervals (95%) were calculated for the anthropometric indexes, according to gender, in the categories of the independent variables. To compare the mean outcomes in the categories of the independent variables, ANOVA (one criterion) was used. Linear trend tests were applied for variables with more than two categories. Significance level was 5%. To identify anthropometric indexes correlates, two-level fixed-effects multilevel linear models were applied. The 81 census tracts drawn were grouped in four geographical regions of the city (ecological level). Multilevel analysis followed a hierarchical conceptual model to assess the direct effect of individual variables and the contribution of the ecological level to the outcomes (rho). Point and confidence intervals (95%) for the effect measures (?) were calculated for independent variables. Firstly, these effect measures were calculated in partial models that included the variables of each hierarchical block, adjusted for the preceding blocks\' variables. In the final models, statistically significant variables (p< 0.05) were kept, after adjusting for the preceding blocks. For the outcomes, the adjusted means were calculated in the categories of the independent variables, according to geographical regions. Adjusted means differences between extreme categories of independent variables were calculated to assess the shaping effect of the geographical regions. All estimates took into account the design effect. Results: Higher mean of BMI were detected, in both gender, among those who reported familial antecedents of overweight, personal history of obesity, took medicines in the last 15 days, and adopted a weight loss diet. Among women, higher mean of BMI were found among the older, those with lower education level, who had lived in the city for longer time and reported familial antecedents of stroke (STR). The BMI correlates were: gender (?= -0.814; CI95% : -1.514 - -0.113), age (?= 0.050; CI95% : 0.014 - 0.086), education (?= -0.101; CI95% : -0.206 - -0.005), familial antecedents of overweight (?= 1.214; CI95% : 0.468 - 1.961), personal antecedents of overweight (?= 6.422; CI95% : 5.724 - 7.119) and total fat consumption (?= 0.029; CI95% : 0.008 - 0.051), considering the 11% contribution of the ecological level. The smallest adjusted means differences of BMI were mainly detected at the extreme categories of the following variables: gender, age strata, education and total fat consumption in the North and West regions. Higher means for the three central obesity indexes (waistC, WHR and CI) were observed among the older, with a lower education level, classified in the lowest income groups, who had lived in the city for longer time, with familial antecedents of overweight, personal antecedents of obesity, and among those classified as \"pre-obese\" and \"obese\". The waistC correlated included: gender (?= -8.686; CI95% : -9.439 - -7.932), age (?= 0.169; CI95% : 0.122 - 0.216), familial antecedents of STR (?= 0.689; CI95% : 0.051 - 1.327), personal antecedents of obesity (?= 2.363; CI95% : 0.320 - 4.410), BMI (?= 1.689; CI95% : 1.437 - 1.941), nº of medicines taken (?= 0.259; CI95% : 0.078 - 0.440), smoking time (?= 0.035; CI95% : 0.008 - 0.061), and total energy of diet (?= 0.084; CI95% : 0.012 - 0.157), considering the 12.4% contribution of the ecological level. The smallest adjusted means differences of waistC were detected at the extreme categories of the following variables: familial antecedents of STR, smoking time, and total energy of diet in the North and West. The correlates of WHR were: gender (?= -.099; CI95% : -0.107 - -0.090), age (?= 0.002; CI95%: 0.001 - 0.003), BMI (?= 0.004; CI95% : 0.003 - 0.005), nº of medicines taken (?= 0.004; CI95% : 0.001 - 0.007), smoking time (?= 0.0006; CI95% : 0.0004 - 0.0009), alcohol addiction (?= 0.013; CI95% : 0.003 - 0.022), weight loss diet (?= 0.017; CI95% : 0.004 - 0.030) and carbohydrate consumption (?= 0.0001; CI95% : 0.0001 - 0.0002). The ecological level contributed with 12%. The smallest adjusted means differences of WHR were detected at the extreme categories of the variables: nº of medicines taken and carbohydrate consumption in the West of the city. The CI correlates included: gender (?= -0.068; CI95% : -0.076 - -0.056), age (?= 0.003; CI95% : 0.002 - 0.004), familial antecedents of STR (?= 0.008; CI95% : 0.001 - 0.016), BMI (?= 0.005; CI95% : 0.004 - 0.006), access to health services (?= -0.011; CI95% : -0.020 - -0.002), nº of medicines taken (?= 0.004; CI95% : 0.002 - 0.007), smoking time (?= 0.0004; CI95% : 0.0001 - 0.0007), alcohol addiction (?= 0.011; CI95% : 0.001 - 0.020), weight loss diet (?= 0.015; CI95% : 0.002 - 0.030) and MUFA´s consumption (?= -0.0005; CI95% : -0.0010 - -0.0001), with 14.1% of contribution from the ecological level. The smallest adjusted means differences of CI were detected at the extreme categories of the variables: access to health services and alcohol addiction, in the North region. Conclusions: In the Ribeirão Preto city, the North and West regions were identified as obesogenic areas. Thus to promoting health behaviors and preventing the obesity in the city it should be taken into account that interventions might be applied not only for changing individual behaviors, since the impact of those interventions may be minimized or constrained by the contextual effect imposed by those regions.

Submitted by Alexandre Pignanelli (alexandre.pignanelli@fgv.br) on 2012-09-14T04:50:22Z No. of bitstreams: 1 Tese Alexandre Pignanelli.pdf: 10725491 bytes, checksum: 294362f3f6f0edd963b319bb62eb292c (MD5) / Approved for entry into archive by Suzinei Teles Garcia Garcia (suzinei.garcia@fgv.br) on 2012-09-14T11:31:58Z (GMT) No. of bitstreams: 1 Tese Alexandre Pignanelli.pdf: 10725491 bytes, checksum: 294362f3f6f0edd963b319bb62eb292c (MD5) / Made available in DSpace on 2012-09-14T11:56:44Z (GMT). No. of bitstreams: 1 Tese Alexandre Pignanelli.pdf: 10725491 bytes, checksum: 294362f3f6f0edd963b319bb62eb292c (MD5) Previous issue date: 2011-12-21 / The mechanisms leading to superior performance and the factors that explain them are subjects of interest to many areas in business research. The variance decomposition of performance has been used for such purposes, initially trying to understand the contribution of idiosyncratic organizational factors (firm effect) and factors related to the economic sector (industry effect); more recently other factors have gained interest, such as the contributions of corporation and country to firm performance. This doctoral thesis has introduced a new factor in studies of variance decomposition of performance: the supply chain effect, quantifying the influence of affiliation to a given supply chain to firm performance. Another contribution of this thesis is a comprehensive panorama of the structure of performance variance in Brazil, expanding previous research in terms of sample size, a more suitable method of analysis (multilevel or hierarchical linear models) and performance dimensions used. The empirical analysis considered indicators of profit and growth, with the broader sample containing 592,905 observations from 77,468 Brazilian firms and 485 industries, in a period of 10 years. Data were obtained from annual structural economic surveys conducted by Brazilian Institute of Geography and Statistics (IBGE). Three-levels hierarchical linear models, with observations of performance, firms and industries, showed that the firm effect accounts for the greatest portion of explained variance; these models also pointed to peculiar characteristics of the Brazilian context, such as the differences in the structure of variance when the different sectors are analyzed separately, mainly in terms of intensity of the industry effect – more relevant, for example, for service firms than for firms from other industries – and the differences in the structure of variance for the various regions of Brazil, being firms from the North, Northeast and Midwest regions more dependent on industry’s contribution to their performance than firms from the South and Southeast. With the introduction of a fourth level – the supply chain – to the model, it was found that the magnitude of the supply chain effect reached 15% to 25% of the explained variability, measured by the square root of the variance components, representing about 50% to 90% of the industry effect magnitude. Beyond highlighting the importance of supply chain management, the findings suggest a new understanding of the industry effect, as they indicate that the benefits traditionally attributed to the industry are partly resulting from the firm affiliation to a supply chain, and not only from the similarity of activities it shares with other firms in the same industry. / Os mecanismos que levam ao desempenho superior e, por consequência, os fatores que os explicam, são focos de várias áreas de pesquisa em Administração de Empresas. A decomposição da variabilidade do desempenho vem sendo utilizada para esse fim, inicialmente procurando entender a contribuição de fatores intrínsecos às organizações (efeito empresa) e de fatores relacionados ao setor econômico (efeito setor); mais recentemente outros fatores ganharam interesse, como é o caso das contribuições da corporação e do país para o desempenho. Esta tese introduziu um novo fator aos estudos de decomposição da variabilidade do desempenho: o efeito cadeia de suprimentos, quantificando a influência da afiliação a uma determinada cadeia para o desempenho da empresa. Outra contribuição desta tese é o amplo mapeamento da estrutura de variância de desempenho das empresas brasileiras, expandindo pesquisas anteriores em termos de tamanho de amostra, método de análise mais adequado (modelagem multinível) e dimensões de desempenho utilizadas. As análises empíricas consideraram indicadores de lucro e de crescimento, com a amostra mais ampla contendo 592.905 observações de 77.468 empresas brasileiras e 485 setores de negócios, em um período de 10 anos. Os dados foram obtidos das bases de dados das pesquisas econômicas estruturais do Instituto Brasileiro de Geografia e Estatística (IBGE). Modelos de 3 níveis – observações de desempenho, empresas e setores – mostraram que o efeito empresa individual responde pela maior parcela da variância do desempenho; esses modelos também apontaram para características peculiares da realidade brasileira, como as diferenças das estruturas de variância quando os diversos setores são analisados separadamente, principalmente em termos da intensidade do efeito setor – mais relevante, por exemplo, para as empresas de serviços do que para as empresas dos demais setores – e como as diferenças nas estruturas de variância para as diversas regiões do Brasil, sendo as empresas das regiões Norte, Nordeste e Centro-Oeste mais dependentes da contribuição do setor para seu desempenho do que as empresas da região Sul e Sudeste. Ao se introduzir um quarto nível – a cadeia de suprimentos – ao modelo, foi possível identificar que a magnitude do efeito cadeia alcança entre 15% a 25% da variabilidade explicada, medida pela raiz quadrada dos componentes de variância, representando cerca de 50% a 90% da magnitude do efeito setor. Além de evidenciar a importância da gestão das cadeias de suprimentos, os achados apontam para uma nova compreensão do efeito setor, já que indicam que os benefícios tradicionalmente atribuídos ao setor econômico são em parte decorrentes da afiliação da empresa a uma determinada cadeia, e não à similaridade das atividades que ela compartilha com outras empresas do mesmo setor.

[pt] Neste trabalho apresentamos algumas contrubuições ao estudo dos modelos de avaliação estatística usados nas ciências sociais. As contribuições originais são: i ) uma descrição unificada sobre como a teoria da medição evoluiu nas diversas disciplinas científicas; ii ) uma resenha abrangente sobre os métodos de estimação por máxima verossimilhança empregados na medição estatística; iii ) uma formulação geral do métodos da máxima verossimilhan ça tendo em vista a aplicação em modelos não-lineares; e principalmente, iv ) a apresentação do método da aproximação estocástica na estimação dos modelos estatísticos de avaliação e medição. Os modelos não-lineares ocorrem freqüentemente nas ciências sociais onde é importante a modelagem de variáveis de resposta dicotômicas ou ordinais. Em particular, este trabalho trata dos modelos da teoria da resposta ao item, dos modelos de regressão logística e dos modelos de componentes aleatórias em geral. A estimação destes modelos ainda é objeto de intensa pesquisa. Não se pode afirmar que exista um método de estimação inteiramente confiável. Os métodos aproximados produzem estimativas com viés acentuado nas componentes de variância, enquanto os métodos de integração numérica e os métodos bayesianos podem apresentar problemas de convergência em muitos casos. O método da aproximação estocástica se baseia na maximização da verossimilhança e emprega o algoritmo de Robbins- Monro para resolver a equação do escore. Como um método estocástico ele gera um processo de Markov que se aproxima das estimativas desejadas e portanto pode ser considerado um MCMC (Monte Carlo Markov chain) freqüentista. Nas simulações realizadas o método apresentou um bom desempenho, produzindo estimativas com viés pequeno, precisão razoável e raros problemas de convergência. / [en] This work presents a study of statistical models used for assessment and measurement in the social sciences. The main contributions are: i ) a unified description of how evaluation, assessment, and the theory of measurement evolved within several branches of science; ii ) a review of estimation methods currently employed in nonlinear models; iii ) a general formulation of the maximum likelihood estimation method; and particularly, iv the presentation of the stochastic approximation method for estimation of non linear statistical models in measurement and assessment. Non linear models occurs frequently in the social sciences where it is important to model binary or ordinal response variables. This work deals with item response theory models, logistic regression models and general models with random components. The estimation of these models has been the subject of several recent simulation studies. One cannot say there is a best estimation method. The approximate methods are known to produce biased estimates, numerical integration methods and bayesian methods can present convergence problems in many cases. Stochastic approximation method is a maximum likelihood method that uses the Robbins-Monro algorithm to solve the score equation. As a stochastic approximation method it generates a Markov chain that converges to the desired estimates and can be considered a frequentist MCMC. A simulation study and a comparative estimation study show a good performance, the method producing small bias for the estimates, good precision, and very rare convergence problems.

Le continuel développement des outils statistiques permet aujourd’hui la modélisation de nombreux phénomènes, toujours plus complexes. En combinant l’approche offerte par des modèles statistiques spécifiques, dit "multiniveaux", et leurs applications à différentes problématiques médicales, cette thèse s’inscrit à la croisée de divers domaines : celui des statistiques, de par la méthodologie sur laquelle reposent les résultats ; mais aussi, celui de l’économie de la santé et de la santé publique en général, au travers des applications présentées.La première partie de cette thèse s’intéresse aux aspects théoriques, et plus particulièrement à l’évolution des méthodologies, du modèle de régression linéaire simple aux modèles multiniveaux pour des liens non nécessairement linéaires. Le déroulé historique de la modélisation mais également les hypothèses, le principe, la stratégie d'analyse et enfin les limites y sont abordés.La seconde partie s’articule autour de deux applications multiniveaux distinctes. La première concerne les déterminants de l'observance et des interruptions de traitement, chez les personnes infectées par le virus du VIH/Sida suivant un traitement par antirétroviraux, dans le contexte du Cameroun. La seconde, quant à elle, s'intéresse au recours aux soins dentaires chez les personnes âgées de 60 ans et plus et vivant en domicile ordinaire. Ces deux applications sont comparables dans leurs méthodologies puisqu’il s’agit de déterminer les comportements étudiés par des variables individuelles habituellement retenues, mais aussi des variables de "contexte" (caractéristiques de l’offre de soins).La troisième partie est consacrée aux applications sur le panel de médecins généralistes et traite deux études autonomes. La première expose les réticences des médecins, et le rôle de leurs aprioris, face aux Recommandations de Bonnes Pratiques (RBP). Elle nous permet de déboucher sur quelques pistes pour améliorer l’usage des RBP en médecine de ville. La seconde analyse la similarité – ou parfois la dissimilarité – entre le cycle d’activité des médecins généralistes et la saisonnalité des épidémies, dans le but d’identifier les facteurs favorisant l’ajustement des médecins généralistes aux variations saisonnières des besoins des patients. Cette étude pourrait permettre, notamment, d’anticiper et de mieux gérer des situations de crise sanitaire, avec l’appui effectif de la médecine de ville. / The continual development of statistical tools allows the modelling of numerous phenomena, including the complex ones. Using a set of statistical techniques and applications, based on the so-called “multilevel” modelling, this thesis deals with different aspects related to the statistical methodology and applications as per health economics and public health.The first part reconsiders the evolution of methodology, starting from the simple linear regression techniques to the more complex multilevel modelling as applied to both the linear and non-linear relations. It addresses issues related to the historical development, the hypotheses, the strategy of analysis, and the scope of applications. The second part presents two distinct multilevel studies. The first concerns the determinants of observance and interruptions of treatment for persons infected by the HIV/AIDS and treated with antiretroviral in Cameroon. The second one focuses on the use of dental services for the elderly. The two studies are methodologically comparable in that, besides integrating the usually retained individual variables, the analyses examine health seeking behaviours, particularly, the utilisation of health care services, while accounting for contextual determinants such as the characteristics of health supply (clinic, department or region).The third part is dedicated to the applications on GPs' panel and contains two different studies. The first one explains the GPs’ behaviours and the role of their aprioris vis-à-vis Clinical Practice Guidelines (CPG), and suggests solutions to overcome their negative attitudes. The second study analyses the similarity – sometimes the differences – between the cycle of GPs’ activity and the seasonality of epidemics with the aim of better understanding determinants favouring the adjustment of the GP in the seasonal variations of the patients needs. This study helps anticipate and manage situations of sanitary crisis, with the effective support of general practitioners.

[ES] En las últimas décadas, la intensificación del uso de las tecnologías de la información y la comunicación (TIC) ha supuesto grandes cambios en nuestra forma de vida. En este contexto de intensa y creciente digitalización, esta tesis doctoral estudia el papel que juegan las TIC como un factor determinante del rendimiento académico de los estudiantes de educación secundaria, así como los factores que propician el uso de las TIC en el aula por parte de los docentes. La tesis se compone de tres capítulos: (1) en el primero de ellos, se analiza la relación entre distintos tipos de uso de las TIC en el contexto social y educativo y el rendimiento académico; (2) en el capítulo dos, se centra la atención en el impacto que tiene sobre el rendimiento académico el uso de las TIC en el aula para realizar tareas y ejercicios; (3) y en el capítulo tres se analizan los factores que determinan la frecuencia de uso de las TIC en el aula por parte de los docentes. Para realizar estos análisis, se estudian datos procedentes de evaluaciones educativas internacionales y nacionales mediante la aplicación de distintos métodos estadísticos: modelos multinivel, método de variables instrumentales, método de emparejamiento por puntaje de propensión, regresiones cuantílicas y técnica de imputación multivariante por ecuaciones encadenadas. Los resultados alcanzados en las distintas investigaciones proporcionan evidencia empírica novedosa que permite elaborar recomendaciones en materia de política educativa, así como abrir futuras líneas de investigación que permitirán complementar los resultados de esta tesis doctoral. / [CA] En les últimes dècades, la intensificació de l'ús de les tecnologies de la informació i la comunicació (TIC) ha suposat grans canvis en la nostra forma de vida. En aquest context d'intensa i creixent digitalització, aquesta tesi doctoral estudia el paper que juguen les TIC com un factor determinant del rendiment acadèmic dels estudiants d'educació secundària, així com els factors que propicien l'ús de les TIC a l'aula per part dels docents. La tesi es compon de tres capítols: (1) en el primer d'ells, s'analitza la relació entre diferents tipus d'ús de les TIC en el context social i educatiu i el rendiment acadèmic; (2) en el capítol dos, se centra l'atenció en l'impacte que té sobre el rendiment acadèmic l'ús de les TIC a l'aula per a fer tasques i exercicis; (3) i en el capítol tres s'analitzen els factors que determinen la freqüència d'ús de les TIC a l'aula per part dels docents. Per a realitzar aquestes anàlisis, s'estudien dades procedents d'avaluacions educatives internacionals i nacionals mitjançant l'aplicació de diferents mètodes estadístics: models multinivell, mètode de variables instrumentals, mètode d'aparellament per puntuació de propensió, regressió quantílica i tècnica d'imputació multivariant per equacions encadenades. Els resultats aconseguits en les diferents investigacions proporcionen evidència empírica nova que permet elaborar recomanacions en matèria de política educativa, així com obrir futures línies d'investigació que permetran complementar els resultats d'aquesta tesi doctoral. / [EN] In recent decades, the intensification of the use of information and communication technologies (ICT) has brought about major changes in our way of life. In this context of intense and increasing digitalization, this doctoral thesis studies the role of ICT as a determinant of the academic performance of secondary school students, as well as the factors that favour the use of ICT in the classroom by teachers. The thesis consists of three chapters: (1) in the first one, the relationship between different types of ICT use in the social and educational context and academic performance is analysed; (2) in chapter two, attention is focused on the impact on academic performance of the use of ICT in the classroom to carry out tasks and exercises; (3) and in chapter three, the factors that determine the frequency of ICT use in the classroom by teachers are analysed. In order to carry out these analyses, data from international and national educational assessments are studied by applying different statistical methods: multilevel models, instrumental variables method, propensity score matching method, quantile regressions and multivariate imputation technique by chained equations. The results achieved in the different investigations provide novel empirical evidence that allows us to elaborate recommendations for educational policy, as well as to open future lines of research that will allow us to complement the results of this doctoral thesis. / Mi agradecimiento al Ministerio de Universidades por su apuesta en la financiación de mi proyecto de investigación mediante el contrato FPU16/04571 y por permitirme dedicarme durante estos cuatro años exclusivamente a la investigación y a la docencia universitaria. Agradezco también al Ministerio de Economía y Competitividad y al doctor Jorge Calero, investigador principal del proyecto “Evaluación de intervenciones educativas para la mejora de la calidad educativa”, por permitirme participar como miembro del equipo de trabajo en el proyecto EDU2016-76414-R y financiar la presentación de mis investigaciones en congresos nacionales e internacionales. Igualmente, agradezco a la Fundación Sabadell por otorgarme una ayuda a la investigación científica en la convocatoria 2020-2021. / Gómez Fernández, NM. (2022). Use of Statistical Methods for the Analysis of Educational Data: the Role of ICTs in the Educational Context [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/181000 / TESIS

Ce mémoire a pour objectif d’étudier la relation entre la culture organisationnelle et la violence au travail. Plus spécifiquement, les résultats permettent de déterminer comment la perception des cultures organisationnelles de type groupal, développemental, hiérarchique et rationnel de l’approche des valeurs concurrentes de Cameron et Quinn (1999) s’associent aux conflits interpersonnels et au harcèlement physique et sexuel dans les organisations. Les données proviennent de l’Étude SALVEO, menée par l’Équipe de recherche sur le travail et la santé mentale de l’Université de Montréal. À notre connaissance, aucune étude n’a étudié les perceptions des cultures organisationnelles globales des entreprises. Les études antérieures se sont intéressées à certains traits spécifiques des cultures, telles que l’acceptation, la tolérance et la banalisation de la violence, sans considérer intégralement la culture organisationnelle. Il est possible d’utiliser le modèle de Cameron et Quinn (1999) avec l’échelle de Marchand, Haines et Dextras-Gauthier (2013) pour mesurer la perception que les travailleurs se font de leur culture organisationnelle pour pouvoir les associer avec les niveaux de conflits interpersonnels et de harcèlement physique et sexuel par la suite. Les analyses multiniveaux de cette recherche ont révélé que la culture groupale s’associe à des niveaux plus bas de conflits interpersonnels et la culture développementale à des niveaux plus élevés. Bien que les résultats ne soient pas significatifs pour tous les types de culture organisationnelle, les entreprises qui adoptent des caractéristiques de la culture groupale, telles que le soutien social, la participation des travailleurs et la justice organisationnelle, semblent mieux prévenir le phénomène de la violence au travail. D’autre part, l’intégration d’un grand nombre de variables contrôles a permis de déterminer que les facteurs individuels et organisationnels les plus associés à la violence sont : le fait d’être une femme, d’être jeune, d’être syndiqué, l’effort au travail et l’injustice organisationnelle. / This master’s thesis aims to study the relation between the organizational culture and workplace violence. More specifically, the results allow us to determine how the organizational culture of groupal, developmental, hierarchical and rational type of the competing values framework of Cameron and Quinn (1999) joins the interpersonal conflicts and physical or sexual harassment in organizations. The data results come from the SALVEO surveys led by the research team on work and mental health of University of Montreal (ERTSM). Most studies were interested in some specific features of culture, such as the acceptance and tolerance of violence without considering the organizational culture. It is possible to use the model of Cameron and Quinn (1999) with the Marchand, Haines and Dextras-Gauthier scale (2013) to calculate the perception that the workers have of their organizational culture to be able to measure them with the levels of interpersonal conflicts and physical or sexual harassment afterward. Multilevel analyses revealed that the perception of the groupal culture shows lower levels of interpersonal conflicts and the developmental culture, higher levels. Although the results are not significant for all types of organizational culture, companies which adopt the characteristics of the groupal culture, such as social support, workers participation and organizational justice, seem to better prevent the phenomenon of workplace violence. Moreover, the integration of a large number of control variables allows us to determine that the individual and organizational factors that are most associated with violence are : being a woman, being young, being a member of a union, having a work overload and the presence of organizational injustice.

Ce mémoire a pour objectif d’étudier la relation entre la culture organisationnelle et la violence au travail. Plus spécifiquement, les résultats permettent de déterminer comment la perception des cultures organisationnelles de type groupal, développemental, hiérarchique et rationnel de l’approche des valeurs concurrentes de Cameron et Quinn (1999) s’associent aux conflits interpersonnels et au harcèlement physique et sexuel dans les organisations. Les données proviennent de l’Étude SALVEO, menée par l’Équipe de recherche sur le travail et la santé mentale de l’Université de Montréal. À notre connaissance, aucune étude n’a étudié les perceptions des cultures organisationnelles globales des entreprises. Les études antérieures se sont intéressées à certains traits spécifiques des cultures, telles que l’acceptation, la tolérance et la banalisation de la violence, sans considérer intégralement la culture organisationnelle. Il est possible d’utiliser le modèle de Cameron et Quinn (1999) avec l’échelle de Marchand, Haines et Dextras-Gauthier (2013) pour mesurer la perception que les travailleurs se font de leur culture organisationnelle pour pouvoir les associer avec les niveaux de conflits interpersonnels et de harcèlement physique et sexuel par la suite. Les analyses multiniveaux de cette recherche ont révélé que la culture groupale s’associe à des niveaux plus bas de conflits interpersonnels et la culture développementale à des niveaux plus élevés. Bien que les résultats ne soient pas significatifs pour tous les types de culture organisationnelle, les entreprises qui adoptent des caractéristiques de la culture groupale, telles que le soutien social, la participation des travailleurs et la justice organisationnelle, semblent mieux prévenir le phénomène de la violence au travail. D’autre part, l’intégration d’un grand nombre de variables contrôles a permis de déterminer que les facteurs individuels et organisationnels les plus associés à la violence sont : le fait d’être une femme, d’être jeune, d’être syndiqué, l’effort au travail et l’injustice organisationnelle. / This master’s thesis aims to study the relation between the organizational culture and workplace violence. More specifically, the results allow us to determine how the organizational culture of groupal, developmental, hierarchical and rational type of the competing values framework of Cameron and Quinn (1999) joins the interpersonal conflicts and physical or sexual harassment in organizations. The data results come from the SALVEO surveys led by the research team on work and mental health of University of Montreal (ERTSM). Most studies were interested in some specific features of culture, such as the acceptance and tolerance of violence without considering the organizational culture. It is possible to use the model of Cameron and Quinn (1999) with the Marchand, Haines and Dextras-Gauthier scale (2013) to calculate the perception that the workers have of their organizational culture to be able to measure them with the levels of interpersonal conflicts and physical or sexual harassment afterward. Multilevel analyses revealed that the perception of the groupal culture shows lower levels of interpersonal conflicts and the developmental culture, higher levels. Although the results are not significant for all types of organizational culture, companies which adopt the characteristics of the groupal culture, such as social support, workers participation and organizational justice, seem to better prevent the phenomenon of workplace violence. Moreover, the integration of a large number of control variables allows us to determine that the individual and organizational factors that are most associated with violence are : being a woman, being young, being a member of a union, having a work overload and the presence of organizational injustice.

No description available.

Cette thèse doctorale poursuit l’objectif de mieux comprendre le rôle joué par la profession réglementée en tant que déterminant de la détresse psychologique de la population en emploi au Québec et au Canada. Ceci, dans un contexte où plusieurs ordres professionnels représentant des professions réglementées, s’inquiètent de la santé mentale de leurs membres et de la pression considérable exercée sur eux dans une économie caractérisée par des pénuries de main-d’oeuvre importantes. Cette thèse fut également inspirée par les nombreuses limites constatées à la suite d’une revue de la littérature sur la santé mentale au travail, alors que les risques différenciés auxquels seraient soumis ces professionnels, comparativement à l’ensemble de la population en emploi, demeurent largement à documenter. La profession réglementée s’associe-t-elle directement à l’expérience de détresse psychologique? Quelles sont les conditions de travail susceptibles de conduire au développement ou à l’aggravation de la détresse psychologique pour ces professions? Dans le but de mieux comprendre le rôle joué par la profession réglementée en matière de détresse psychologique, nous avons eu recours à un modèle théorique multidimensionnel qui postule que les contraintes et les ressources découlent d’un ensemble de structures sociales incluant la profession, le travail, la famille, le réseau social hors-travail et les caractéristiques personnelles. Ce modèle découle des théories micro et macro en sociologie (Alexander et al., 1987; Ritzer, 1996), de l’approche agent-structure(Archer, 1995; Giddens, 1987) ainsi que de la théorie du stress social (Pearlin,1999). Trois hypothèses sont soumises à l’étude à travers ce modèle. La première hypothèse, est à l’effet que la profession réglementée, les conditions de travail, la famille ainsi que le réseau social hors-travail et les caractéristiques individuelles, contribuent directement et conjointement à l’explication du niveau de détresse psychologique. La seconde hypothèse induite par le modèle proposé, pose que le milieu de travail médiatise la relation entre la profession réglementée et le niveau de détresse psychologique. La troisième et dernière hypothèse de recherche, postule enfin que la relation entre le milieu de travail et le niveau de détresse psychologique est modérée par les caractéristiques individuelles ainsi que par la famille et le réseau social hors-travail. Ces hypothèses de recherche furent testées à partir des données longitudinales de l’Enquête nationale sur la santé de la population (ENSP) (cycles 1 à 7). Les résultats obtenus sont présentés sous forme de 3 articles, soumis pour publication, lesquels constituent les chapitres 5 à 7 de cette thèse. Dans l’ensemble, le modèle théorique proposé obtient un soutien empirique important et tend à démontrer que la profession réglementée influence directement les chances de vivre de la détresse psychologique au fil du temps, ainsi que le niveau de détresse psychologique lui-même. Les résultats indiquent que les professions réglementées sont soumises à des risques différenciés en termes de conditions de travail susceptibles de susciter de la détresse psychologique. Notons également que la contribution du milieu de travail et de la profession réglementée s’exerce indépendamment des autres dimensions du modèle (famille, réseau social hors-travail, caractéristiques personnelles). Les résultats corroborent l’importance de considérer plusieurs dimensions de la vie d’un individu dans l’étude de la détresse psychologique et mettent à l’ordre du jour l’importance de développer de nouveaux modèles théoriques, mieux adaptés aux contextes de travail au sein desquels oeuvrent les travailleurs du savoir. Cette thèse conclue sur les implications de ces résultats pour la recherche, et sur les retombées qui en découlent pour le marché du travail ainsi que pour le développement futur du système professionnel québécois et canadien. / This doctoral thesis aims to understand the role played by the regulated occupations as a determinant of psychological distress of the working population in Quebec and Canada. This, in a context where several professional organizations, representing regulated occupations, are concerned about the mental health of their members and the pressure exerted on them in an economy characterized by important shortages of labor. This thesis was also inspired by the many limitations observed after a literature review on work and mental health, whereas the differentiated risks which would be subjected to these professionals, compared to the total working population, remains largely undocumented. Is there a direct link between the regulated occupation and the experience of psychological distress? What working conditions contribute to the development or accentuate psychological distress for these regulated occupations? In order to better understand the role played by the regulated occupations in psychological distress, this thesis proposes a multidimensional theoretical model which postulates that the constraints and resources are generated by a set of social structures including the regulated occupation, the working conditions, family, social network outside of work and personal characteristics. This model stems from the micro and macro theories in sociology (Alexander et al., 1987; Ritzer, 1996), the agent-structure approach (Archer, 1995; Giddens, 1987) as well as the social stress theory (Pearlin, 1999). Three hypotheses are subject to analysis through the model. The first hypothesis assumes that regulated occupations, work conditions, family, social network outside the workplace, and individual characteristics contribute directly and jointly to explaining the level of psychological distress. The second hypothesis induced by the proposed model postulates that the workplace mediates the relationship between regulated occupations and psychological distress levels. The third and final research hypothesis postulates that the relationship between the workplace and psychological distress levels is moderated by individual characteristics, as well as by family and the social network outside the workplace.These hypotheses have been validated using longitudinal data from the National population health survey (NPHS) (cycles 1 to 7). The results of these analyses are presented in three articles submitted for publication, which are the chapters 5-7 of this thesis. Overall, the theoretical model gets an important empirical support and suggests that the regulated occupations directly influence the chances of living psychological distress over time as well as the level of psychological distress itself. The results also suggest that the regulated occupations are exposed to differentiated risks in terms of working conditions likely to generate psychological distress. The contribution of the workplace and regulated occupations is exercised independently of other dimensions of the model (family, social network outside of work, personal characteristics). The results also corroborate the importance in considering many dimensions of the life of an individual in the psychological distress and to put on the agenda the importance of developing new theoretical models, better suited to the realities characterizing today’s working environments in which knowledge workers work. This thesis concludes on the implications of these findings for research, and the benefits it brings to the labor market and for the future development of the professional system in Quebec and Canada.

No description available.