The following study utilizes data from the High School and Beyond Study in order to predict mathematics achievement using both student characteristics and school level characteristics. Utilizing Hierarchical Linear Modeling, this study extends the body of literature by exploring how race, socio-economic status, and gender, as well as the percentage of minority students in a school, whether or not the school is Catholic, the proportion of students in the academic track, and the mean socioeconomic status of the school all affect mathematics achievement. Through this methodology, it was possible to see the direct effects of both student level and school level variables on achievement, as well as the cross-level interaction of all of these variables. Findings suggest that there are discrepancies in how different types of students achieve, as well as how those students achieve in varying contexts. Many of the variables were statistically significant in their effect on mathematics achievement. Implications for this research are discussed and considerations for future research are presented.

The following study utilizes data from the High School and Beyond Study in order to predict mathematics achievement using both student characteristics and school level characteristics. Utilizing Hierarchical Linear Modeling, this study extends the body of literature by exploring how race, socio-economic status, and gender, as well as the percentage of minority students in a school, whether or not the school is Catholic, the proportion of students in the academic track, and the mean socioeconomic status of the school all affect mathematics achievement. Through this methodology, it was possible to see the direct effects of both student level and school level variables on achievement, as well as the cross-level interaction of all of these variables. Findings suggest that there are discrepancies in how different types of students achieve, as well as how those students achieve in varying contexts. Many of the variables were statistically significant in their effect on mathematics achievement. Implications for this research are discussed and considerations for future research are presented.

The current study evaluated the performance of traditional versus modern MDTs in the estimation of fixed-effects and variance components for data missing at the second level of an hierarchical linear model (HLM) model across 24 different study conditions. Variables manipulated in the analysis included, (a) number of Level-2 variables with missing data, (b) percentage of missing data, and (c) Level-2 sample size. Listwise deletion outperformed all other methods across all study conditions in the estimation of both fixed-effects and variance components. The model-based procedures evaluated, EM and MI, outperformed the other traditional MDTs, mean and group mean substitution, in the estimation of the variance components, outperforming mean substitution in the estimation of the fixed-effects as well. Group mean substitution performed well in the estimation of the fixed-effects, but poorly in the estimation of the variance components. Data in the current study were modeled as missing completely at random (MCAR). Further research is suggested to compare the performance of model-based versus traditional MDTs, specifically listwise deletion, when data are missing at random (MAR), a condition that is more likely to occur in practical research settings.

No description available.

This dissertation has multiple aims in studying hierarchical linear models in biomedical data analysis. In Chapter 1, the novel idea of studying the durations of adolescent growth spurts as a predictor of adulthood obesity is defined, established, and illustrated. The concept of contextual effects modeling is introduced in this first section as we study secular trend of adulthood obesity and how this trend is mitigated by the durations of individual adolescent growth spurts and the secular average length of adolescent growth spurts. It is found that individuals with longer periods of fast height growth in adolescence are more prone to having favorable BMI profiles in adulthood. In Chapter 2 we study the estimation of contextual effects in a hierarchical generalized linear model (HGLM). We simulate data and study the effects using the higher level group sample mean as the estimate for the true mean versus using an Empirical Bayes (EB) approach (Shin and Raudenbush 2010). We study this comparison for logistic, probit, log-linear, ordinal and nominal regression models. We find that in general the EB estimate lends a parameter estimate much closer to the true value, except for cases with very small variability in the upper level, where it is a more complicated situation and there is likely no need for contextual effects analysis. In Chapter 3 the HGLM studies are made clearer with large-scale simulations. These large scale simulations are shown for logistic regression and probit regression models for binary outcome data. With repetition we are able to establish coverage percentages of the confidence intervals of the true contextual effect. Coverage percentages show the percentage of simulations that have confidence intervals containing the true parameter values. Results confirm observations from the preliminary simulations in the previous section of this paper, and an accompanying example of adulthood hypertension shows how these results can be used in an application.

Raros autores estudam as características das empresas e dos seus setores de atuação na explicação dos retornos das ações em períodos exclusivamente de crise. A escassez de trabalhos em períodos de crise pode ser considerada uma importante lacuna na literatura acadêmica, tendo em vista que as perdas são substanciais nestes períodos. O objetivo do presente trabalho é identificar características das empresas e dos seus setores de atuação que expliquem a queda dos preços das ações das empresas listadas na Bovespa durante a crise financeira de 2008. O período de crise escolhido começa em 20 de maio de 2008 (pico do Ibovespa) e termina em 27 de outubro de 2008 (vale do Ibovespa), com queda de 60%. São estudadas 135 empresas não financeiras, com informações disponíveis e eliminados outliers. Utilizou-se neste trabalho uma técnica multinível, Modelos Hierárquicos Lineares, para endereçar claramente a interação entre os dois níveis envolvidos na análise: empresas (1º nível: objeto) e setores (2º nível: contexto). Dada a pouca utilização desta técnica em estudos em administração, sua aplicação também é um diferencial do trabalho. Os resultados indicam a pertinência da escolha por esta técnica, pois se identificou que a variabilidade total dos retornos tem origem (i) em características das empresas (1º nível), correspondendo a 76,9% da variabilidade total e (ii) em características dos setores (2º nível), correspondendo a 23,1% da variabilidade total. O modelo final explica 39,9% da variabilidade total. As características das empresas que têm influência significativa no retorno das ações são: livro / mercado (valor contábil do patrimônio líquido / valor de mercado do patrimônio líquido), tamanho e iliquidez. As características dos setores que têm influência significativa no retorno das ações das empresas são: beta desalavancado, crescimento histórico da receita e ter ou não a tarifa regulada. Por fim, identificou-se que a característica setorial beta desalavancado modera a influência da característica da empresa livro / mercado no retorno das ações das empresas. Em outras palavras, o coeficiente angular da variável livro / mercado é diferente para os diversos setores, sendo que o impacto da variável livro / mercado no retorno é menos acentuado para empresas de setores com alto beta desalavancado. / Few authors study the role of firms and industries\' characteristics in explaining stock\'s returns exclusively in periods of crisis. The scarcity of such studies can be considered an important gap in the academic literature, given the substantial losses that one can experience during such periods. The objective of this study is to identify firms and industries\' characteristics that explain the decline in prices of stocks of companies listed in Bovespa during the 2008 financial crisis. The crisis period chosen begins on May 20, 2008 (Ibovespa\'s peak) and ends on October 27, 2008 (Ibovespa\'s valley), representing a decline of 60%. 135 non-financial companies, with information available and after the exclusion of outliers were studied. A multilevel technique was adopted: Hierarchical Linear Models, to clearly address the interaction between the two levels involved in the analysis: firms (1st level: object) and industries (2nd level: context). Given the low utilization of this technique in studies in business administration, its adoption is also a differential of this study. The results indicate the relevance of the technique\'s choice. It was identified that (i) 76.9% of the total variability is due to firms\' characteristics and (ii) 23.1% of the total variability is due to industries\' characteristics. The final model explains 39.9% of the total variability. Firms\' characteristics that have significant influence on stock returns are: book / market (book value of equity / market value of equity), size and illiquidity. Industries\' characteristics that have significant influence on stock returns are: unlevered beta, historical sales growth and whether or not the industry has a regulated tariff. Finally, it was found that industries\' characteristic unlevered beta moderates the influence of the firms\' characteristic book / market in stock returns. In other words, slope coefficient for the firms\' characteristic book / market is different between industries, with the impact of the variable book / market on stock return being less pronounced for companies in sectors with high unlevered beta.

Raros autores estudam as características das empresas e dos seus setores de atuação na explicação dos retornos das ações em períodos exclusivamente de crise. A escassez de trabalhos em períodos de crise pode ser considerada uma importante lacuna na literatura acadêmica, tendo em vista que as perdas são substanciais nestes períodos. O objetivo do presente trabalho é identificar características das empresas e dos seus setores de atuação que expliquem a queda dos preços das ações das empresas listadas na Bovespa durante a crise financeira de 2008. O período de crise escolhido começa em 20 de maio de 2008 (pico do Ibovespa) e termina em 27 de outubro de 2008 (vale do Ibovespa), com queda de 60%. São estudadas 135 empresas não financeiras, com informações disponíveis e eliminados outliers. Utilizou-se neste trabalho uma técnica multinível, Modelos Hierárquicos Lineares, para endereçar claramente a interação entre os dois níveis envolvidos na análise: empresas (1º nível: objeto) e setores (2º nível: contexto). Dada a pouca utilização desta técnica em estudos em administração, sua aplicação também é um diferencial do trabalho. Os resultados indicam a pertinência da escolha por esta técnica, pois se identificou que a variabilidade total dos retornos tem origem (i) em características das empresas (1º nível), correspondendo a 76,9% da variabilidade total e (ii) em características dos setores (2º nível), correspondendo a 23,1% da variabilidade total. O modelo final explica 39,9% da variabilidade total. As características das empresas que têm influência significativa no retorno das ações são: livro / mercado (valor contábil do patrimônio líquido / valor de mercado do patrimônio líquido), tamanho e iliquidez. As características dos setores que têm influência significativa no retorno das ações das empresas são: beta desalavancado, crescimento histórico da receita e ter ou não a tarifa regulada. Por fim, identificou-se que a característica setorial beta desalavancado modera a influência da característica da empresa livro / mercado no retorno das ações das empresas. Em outras palavras, o coeficiente angular da variável livro / mercado é diferente para os diversos setores, sendo que o impacto da variável livro / mercado no retorno é menos acentuado para empresas de setores com alto beta desalavancado. / Few authors study the role of firms and industries\' characteristics in explaining stock\'s returns exclusively in periods of crisis. The scarcity of such studies can be considered an important gap in the academic literature, given the substantial losses that one can experience during such periods. The objective of this study is to identify firms and industries\' characteristics that explain the decline in prices of stocks of companies listed in Bovespa during the 2008 financial crisis. The crisis period chosen begins on May 20, 2008 (Ibovespa\'s peak) and ends on October 27, 2008 (Ibovespa\'s valley), representing a decline of 60%. 135 non-financial companies, with information available and after the exclusion of outliers were studied. A multilevel technique was adopted: Hierarchical Linear Models, to clearly address the interaction between the two levels involved in the analysis: firms (1st level: object) and industries (2nd level: context). Given the low utilization of this technique in studies in business administration, its adoption is also a differential of this study. The results indicate the relevance of the technique\'s choice. It was identified that (i) 76.9% of the total variability is due to firms\' characteristics and (ii) 23.1% of the total variability is due to industries\' characteristics. The final model explains 39.9% of the total variability. Firms\' characteristics that have significant influence on stock returns are: book / market (book value of equity / market value of equity), size and illiquidity. Industries\' characteristics that have significant influence on stock returns are: unlevered beta, historical sales growth and whether or not the industry has a regulated tariff. Finally, it was found that industries\' characteristic unlevered beta moderates the influence of the firms\' characteristic book / market in stock returns. In other words, slope coefficient for the firms\' characteristic book / market is different between industries, with the impact of the variable book / market on stock return being less pronounced for companies in sectors with high unlevered beta.

The Supreme Court of the United States has recently determined that assigning students to schools and classrooms based on racial identity is unconstitutional. However, it also left the door open for further and different rulings. If researchers are able to show that lack of consideration of race has deleterious effects on federally mandated programs and initiatives, the ruling may be modified or opened up to specific circumstances. Among its many consequences, this ruling brings a focus onto the question of student-teacher matching in classrooms. Over the years, there has been a great deal of discussion in the literature about matching teacher and student by race, ethnicity, gender, and language. Some people claim that matching is crucial for student success while others dispute this claim. The current study examines the question of racial and ethnic matching empirically in the context of a large-scale randomized controlled study of an innovation for middle school mathematics learners. It extends the literature by (1) focusing on the relationship between student-teacher match and a specific, heavily documented situation with targeted learning goals, (2) adding information about Hispanic students to the discussion, and (3) helping evaluate factors that may be important in determining the validity of large-scale experiments. Alone and in conjunction with other similar empirical evidence, it will also have a significant effect on federal and state educational policy. The sample consists of the 92 teachers and 1576 7th grade students on 76 school campuses throughout 8 Texas regions who participated in the Scaling-Up SimCalc project. Teachers and students either used SimCalc Mathworlds™ curriculum and technology or a control for a two-week replacement unit. The crux of the current analysis was a match between aggregated and individual teacher and student characteristics and an inquiry into how these matches influence student math performance in the classroom within and between our experimental and control group. Hierarchical Linear Modeling (HLM) analysis was used to investigate the differences in student mathematics performance, modeled as students nested in classrooms nested in schools. / Ph. D.

Made available in DSpace on 2016-04-25T18:40:16Z (GMT). No. of bitstreams: 1 Claudio Junior Bernardo.pdf: 362107 bytes, checksum: 15a103dae1eca502a54a5453cd2d708c (MD5) Previous issue date: 2016-01-27 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / This study aimed to examine the influence of institutional factors in determining the capital structure of Latin American firms and to analyze if the significance of institutional factors to explain the firms' capital structure is changed considering the short and long term debt. The sample was composed by public companies from six Latin American countries: Argentina, Brazil, Chile, Colombia, Mexico and Peru, analyzed during 2009-2014. Linear hierarchical models (or multilevel regression) for processing the data were used. Six leverage indicators were used as dependent variables. As explanatory variables, firm variables (business characteristics) and country variables (macroeconomic and institutional factors) were considered. These variables have been identified in the literature as important determinants of firms' capital structure. The main results show that both variables (firm and country characteristics) are important determinants of firms' capital structure. However, firm variables explain a much larger percentage of variance. These results can be derived from the similarity of economic contexts of the six countries analysed. Probably, in future work, if countries with very different macroeconomic and institutional features are inserted in the analysis, the results can change significantly. Thus, it is emphasized that much remains to be done to analyze the effects of institutional factors on the capital structure of companies. It is expected that this study has generated new contributions to national literature on capital structure, by using a theoretical approach, as well as econometric, still little explored in the literature and offer suggestions for future work on the subject, thereby contributing to the academy. It is also expected that the research will contribute to the agents of the capital market by analyze the determinants of the capital structure considering the institutional aspects as well as the relevance of these variables in the financing decision process / Esta pesquisa teve por objetivo examinar a influência de fatores institucionais na determinação da estrutura de capital de empresas latino-americanas, bem como analisar se a significância de fatores institucionais para explicar a estrutura de capital das empresas é alterada considerando a decomposição do financiamento em curto e longo prazos. A amostra investigada foi composta por companhias abertas pertencentes a seis países latino americanos: Argentina, Brasil, Chile, Colômbia, México e Peru, analisadas durante o período 2009-2014. Foram utilizados modelos hierárquicos lineares (ou regressão multinível) para tratamento dos dados. Como variáveis dependentes, foram considerados seis indicadores de alavancagem e como variáveis explicativas, foram consideradas variáveis de firma (características das empresas) e país (fatores macroeconômicos e institucionais) identificadas na literatura como importantes determinantes da estrutura de capital. Os principais resultados evidenciam que, tanto as variáveis representativas de características de firma, quanto as variáveis representativas de países, são importantes determinantes da estrutura de capital das empresas. No entanto, as variáveis de firma explicam um percentual de variância muito maior. Estes resultados podem ser derivados da similaridade dos contextos econômicos dos seis países analisados. Provavelmente, em trabalhos futuros, caso sejam inseridos na análise países com características macroeconômicas e institucionais muito distintas, o resultado possa se alterar significativamente. Assim, ressalta-se que ainda há muito a ser feito para análise dos efeitos de fatores institucionais sobre a estrutura de capital das empresas. Espera-se que este estudo tenha gerado novas contribuições para a literatura nacional sobre estrutura de capital, por utilizar uma abordagem teórica, e também econométrica, ainda pouco exploradas na literatura da área, fornecendo subsídios para futuros trabalhos sobre o tema, contribuindo, dessa forma, para a academia. Também se espera que a pesquisa contribua para os agentes do mercado de capitais ao analisar os determinantes da estrutura de capital considerando os aspectos institucionais, bem como a relevância dessas variáveis quando da decisão de financiamento por parte das empresas

The central research question is the extent to which gifted programming effects student academic outcomes of gifted as compared to not-gifted students and how this differs by race/ethnicity and/or poverty status. Since the identification of elementary school students as gifted is not random, propensity score matching is used to remove this bias in the estimates of the effects. A matched sample of North Carolina middle school students based on individual level data of both gifted and not-gifted students of varied racial/ethnic groups and income levels is used for this analysis. This enables a comparison of sixth, seventh, and eighth grade student achievement to determine the extent to which participating in gifted programming differentiates effects by race/ethnicity and poverty status. I show the additional test score gain, if any, from being in gifted programming compared to students not participating in gifted programs. Variations in gifted program effects across race/ethnicity and income are assessed. This research adds empirical evidence to the more qualitatively focused gifted debate by analyzing differences in student outcomes between gifted and not-gifted students in North Carolina. Since black and lower income students are less likely to participate in gifted programs, they disproportionately encounter less experienced teachers, lower expectations, and fewer resources. The extent to which these additional learning supports translate to differences in student outcomes are analyzed.