Sequential Logistic Regression: A Method to Reveal Subtlety in Self-Efficacy

This paper uses self-efficacy to predict the success of women in introductory physics. We show how sequential logistic regression demonstrates the predictive ability of self-efficacy, and reveals variations with type of physics course. Also discussed are the sources of self-efficacy that have the largest impact on predictive ability.

Comparison of Some Improved Estimators for Linear Regression Model under Different Conditions

Multiple linear regression model plays a key role in statistical inference and it has extensive applications in business, environmental, physical and social sciences. Multicollinearity has been a considerable problem in multiple regression analysis. When the regressor variables are multicollinear, it becomes difficult to make precise statistical inferences about the regression coefficients. There are some statistical methods that can be used, which are discussed in this thesis are ridge regression, Liu, two parameter biased and LASSO estimators. Firstly, an analytical comparison on the basis of risk was made among ridge, Liu and LASSO estimators under orthonormal regression model. I found that LASSO dominates least squares, ridge and Liu estimators over a significant portion of the parameter space for large dimension. Secondly, a simulation study was conducted to compare performance of ridge, Liu and two parameter biased estimator by their mean squared error criterion. I found that two parameter biased estimator performs better than its corresponding ridge regression estimator. Overall, Liu estimator performs better than both ridge and two parameter biased estimator.

Kerr-induced nonlinearity reduction in coherent optical OFDM by low complexity support vector machine regression-based equalization

We experimentally demonstrate 7-dB reduction of nonlinearity penalty in 40-Gb/s CO-OFDM at 2000-km using support vector machine regression-based equalization. Simulation in WDM-CO-OFDM shows up to 12-dB enhancement in Q-factor compared to linear equalization.

Outcomes after Unrelated Umbilical Cord Blood Transplantation for Children with Osteopetrosis.

Hematopoietic stem cell transplantation (HSCT) is the only curative treatment for most children with osteopetrosis (OP). Timing of HSCT is critical; therefore, umbilical cord blood transplantation (UCBT) is an attractive option. We analyzed outcomes after UCBT in 51 OP children. Median age at UCBT was 6 months. Seventy-seven percent of the cord blood grafts had 0 or 1 HLA disparity with the recipient. Conditioning regimen was myeloablative (mostly busulfan-based in 84% and treosulfan-based in 10%). Antithymocyte globulin was given to 90% of patients. Median number of total nucleated and CD34(+) cells infused was 14 × 10(7)/kg and 3.4 × 10(5)/kg, respectively. Median follow-up for survivors was 74 months. Cumulative incidence (CI) of neutrophil recovery was 67% with a median time to recovery of 23 days; 33% of patients had graft failure, 81% of engrafted patients had full donor engraftment, and 19% had mixed donor chimerism. Day 100 CI of acute graft-versus-host disease (grades II to IV) was 31% and 6-year CI of chronic graft-versus-host disease was 21%. Mechanical ventilation was required in 28%, and veno-occlusive disease was diagnosed in 16% of cases. Six-year overall survival rate was 46%. Comparative studies with other alternative donors should be performed to evaluate whether UCBT remains a valid alternative for children with OP without an HLA-matched donor.

Distributed Feature Selection in Large n and Large p Regression Problems

Fitting statistical models is computationally challenging when the sample size or the dimension of the dataset is huge. An attractive approach for down-scaling the problem size is to first partition the dataset into subsets and then fit using distributed algorithms. The dataset can be partitioned either horizontally (in the sample space) or vertically (in the feature space), and the challenge arise in defining an algorithm with low communication, theoretical guarantees and excellent practical performance in general settings. For sample space partitioning, I propose a MEdian Selection Subset AGgregation Estimator ({\em message}) algorithm for solving these issues. The algorithm applies feature selection in parallel for each subset using regularized regression or Bayesian variable selection method, calculates the `median' feature inclusion index, estimates coefficients for the selected features in parallel for each subset, and then averages these estimates. The algorithm is simple, involves very minimal communication, scales efficiently in sample size, and has theoretical guarantees. I provide extensive experiments to show excellent performance in feature selection, estimation, prediction, and computation time relative to usual competitors.

While sample space partitioning is useful in handling datasets with large sample size, feature space partitioning is more effective when the data dimension is high. Existing methods for partitioning features, however, are either vulnerable to high correlations or inefficient in reducing the model dimension. In the thesis, I propose a new embarrassingly parallel framework named {\em DECO} for distributed variable selection and parameter estimation. In {\em DECO}, variables are first partitioned and allocated to m distributed workers. The decorrelated subset data within each worker are then fitted via any algorithm designed for high-dimensional problems. We show that by incorporating the decorrelation step, DECO can achieve consistent variable selection and parameter estimation on each subset with (almost) no assumptions. In addition, the convergence rate is nearly minimax optimal for both sparse and weakly sparse models and does NOT depend on the partition number m. Extensive numerical experiments are provided to illustrate the performance of the new framework.

For datasets with both large sample sizes and high dimensionality, I propose a new "divided-and-conquer" framework {\em DEME} (DECO-message) by leveraging both the {\em DECO} and the {\em message} algorithm. The new framework first partitions the dataset in the sample space into row cubes using {\em message} and then partition the feature space of the cubes using {\em DECO}. This procedure is equivalent to partitioning the original data matrix into multiple small blocks, each with a feasible size that can be stored and fitted in a computer in parallel. The results are then synthezied via the {\em DECO} and {\em message} algorithm in a reverse order to produce the final output. The whole framework is extremely scalable.

Estimation of Sample Size and Power For Quantile Regression

Quantile regression (QR) was first introduced by Roger Koenker and Gilbert Bassett in 1978. It is robust to outliers which affect least squares estimator on a large scale in linear regression. Instead of modeling mean of the response, QR provides an alternative way to model the relationship between quantiles of the response and covariates. Therefore, QR can be widely used to solve problems in econometrics, environmental sciences and health sciences. Sample size is an important factor in the planning stage of experimental design and observational studies. In ordinary linear regression, sample size may be determined based on either precision analysis or power analysis with closed form formulas. There are also methods that calculate sample size based on precision analysis for QR like C.Jennen-Steinmetz and S.Wellek (2005). A method to estimate sample size for QR based on power analysis was proposed by Shao and Wang (2009). In this paper, a new method is proposed to calculate sample size based on power analysis under hypothesis test of covariate effects. Even though error distribution assumption is not necessary for QR analysis itself, researchers have to make assumptions of error distribution and covariate structure in the planning stage of a study to obtain a reasonable estimate of sample size. In this project, both parametric and nonparametric methods are provided to estimate error distribution. Since the method proposed can be implemented in R, user is able to choose either parametric distribution or nonparametric kernel density estimation for error distribution. User also needs to specify the covariate structure and effect size to carry out sample size and power calculation. The performance of the method proposed is further evaluated using numerical simulation. The results suggest that the sample sizes obtained from our method provide empirical powers that are closed to the nominal power level, for example, 80%.

Characterising granuloma regression and liver recovery in a murine model of schistosomiasis japonica

For hepatic schistosomiasis the egg-induced granulomatous response and the development of extensive fibrosis are the main pathologies. We used a Schistosoma japonicum-infected mouse model to characterise the multi-cellular pathways associated with the recovery from hepatic fibrosis following clearance of the infection with the anti-schistosomal drug, praziquantel. In the recovering liver splenomegaly, granuloma density and liver fibrosis were all reduced. Inflammatory cell infiltration into the liver was evident, and the numbers of neutrophils, eosinophils and macrophages were significantly decreased. Transcriptomic analysis revealed the up-regulation of fatty acid metabolism genes and the identification of Peroxisome proliferator activated receptor alpha as the upstream regulator of liver recovery. The aryl hydrocarbon receptor signalling pathway which regulates xenobiotic metabolism was also differentially up-regulated. These findings provide a better understanding of the mechanisms associated with the regression of hepatic schistosomiasis.

On the Use of Regression and Verbal Protocol Analysis in Modeling Analysts’ Behavior in an Unstructured Task Environment: A Methodological Note

This paper discusses areas for future research opportunities by addressing accounting issues faced by management accountants practicing in hospitality organizations. Specifically, the article focuses on the use of the uniform system of accounts by operating properties, the usefulness of allocating support costs to operated departments, extending our understanding of operating costs and performance measurement systems and the certification of practicing accountants.

Data-Adaptive Modeling using Convex Regression

Thesis (Ph.D.)--University of Washington, 2016-08

Adolescent boys’ health : managing emotions, masculinities and subjective social status

The health of adolescent boys is complex and surprisingly little is known about how adolescent boys perceive, conceptualise and experience their health. Thus, the overall aim of this thesis was to explore adolescent boys’ perceptions and experiences of health, emotions, masculinity and subjective social status (SSS). This thesis consists of a qualitative, a quantitative and a mixed methods study. The qualitative study aimed to explore how adolescent boys understand the concept of health and what they find important for its achievement. Furthermore, the adolescent boys’ views of masculinity, emotion management and their potential effects on wellbeing were explored. For this purpose, individual interviews were conducted with 33 adolescent boys aged 16-17 years. The quantitative study aimed to investigate the associations between pride, shame and health in adolescence. Data were collected through a cross-sectional postal survey with 705 adolescents. The purpose of the mixed methods study was to investigate associations between SSS in school, socioeconomic status (SES) and self-rated health (SRH), and to explore the concept of SSS in school. Cross-sectional data were combined with interview data in which the meaning of SSS was further explored. Individual interviews with 35 adolescents aged 17-18 years were conducted. In the qualitative study, data were analysed using Grounded Theory. In the quantitative study, statistical analyses (e.g., chi-square test and uni- and multivariable logistic regression analyses) were performed. In the mixed method study, a combination of statistical analyses and thematic network analysis was applied. The results showed that there was a complexity in how the adolescent boys viewed, experienced, dealt with and valued health. On a conceptual level, they perceived health as holistic but when dealing with difficult emotions, they were prone to separate the body from the mind. Thus, the adolescent boys experienced a difference between health as a concept and health as an experience (paper I). Concerning emotional orientation in masculinity, two main categories of masculine conceptions were identified: a gender-normative masculinity and a non-gender-normative masculinity (paper II). Gender-normative masculinity comprised two seemingly opposite emotional masculinity orientations, one towards toughness and the other towards sensitivity, both of which were highly influenced by contextual and situational group norms and demands, despite that their expressions are in contrast to each other. Non-gender-normative masculinity included an orientation towards sincerity, emphasising the personal values of the boys. Emotions were expressed more independently of peer group norms. The findings suggest that different masculinities and the expression of emotions are intricately intertwined and that managing emotions is vital for wellbeing. The present findings also showed that both shame and pride were significantly associated with SRH, and furthermore, that there seems to be a protective effect of experiencing pride for health (paper III). The results also demonstrated that SSS is strongly related to SRH, and high SRH is related to high SSS, and further that the positioning was done in a gendered space (paper IV). Results from all studies suggest that the emotional and relational aspects, as well as perceived SSS, were strongly related to SRH. Positive emotions, trustful relationships and having a sense of belonging were important factors for health and pride was an important emotion protecting health. Physical health, on the other hand, had a more subordinated value, but the body was experienced as an important tool to achieve health. Even though health was mainly perceived in a holistic manner by the boys, there were boys who were prone to dichotomise the health experience into a mind-body dualism when having to deal with difficult emotions. In conclusion, this thesis demonstrates that young, masculine health is largely experienced through emotions and relationships between individuals and their contexts affected by gendered practices. Health is to feel and function well in mind and body and to have trusting relationships. The results support theories on health as a social construction of interconnected processes. Having confidence in self-esteem, access to trustful relationships and the courage to resist traditional masculine norms while still reinforcing and maintaining social status are all conducive to good health. Researchers as well as professionals need to consider the complexity of adolescent boys’ health in which norms, values, relationships and gender form its social determinants. Those working with young boys should encourage them to integrate physical, social and emotional aspects of health into an interconnected and holistic experience.

Análise experimental do efeito da diversificação em carteiras de ações: a relação entre risco e quantidade de ativos sob a ótica da estratégia ingênua

The objective is to analyze the relationship between risk and number of stocks of a portfolio for an individual investor when stocks are chosen by "naive strategy". For this, we carried out an experiment in which individuals select actions to reproduce this relationship. 126 participants were informed that the risk of first choice would be an asset average of all standard deviations of the portfolios consist of a single asset, and the same procedure should be used for portfolios composed of two, three and so on, up to 30 actions . They selected the assets they want in their portfolios without the support of a financial analysis. For comparison we also tested a hypothetical simulation of 126 investors who selected shares the same universe, through a random number generator. Thus, each real participant is compensated for random hypothetical investor facing the same opportunity. Patterns were observed in the portfolios of individual participants, characterizing the curves for the components of the samples. Because these groupings are somewhat arbitrary, it was used a more objective measure of behavior: a simple linear regression for each participant, in order to predict the variance of the portfolio depending on the number of assets. In addition, we conducted a pooled regression on all observations by analyzing cross-section. The result of pattern occurs on average but not for most individuals, many of which effectively "de-diversify" when adding seemingly random bonds. Furthermore, the results are slightly worse using a random number generator. This finding challenges the belief that only a small number of titles is necessary for diversification and shows that there is only applicable to a large sample. The implications are important since many individual investors holding few stocks in their portfolios

Loan quality, ownership and efficiency of Indian banks : A bootstrap truncated regression approach