A Bayesian approach to estimating the regression coefficients of a multinomial logit model

A Bayesian approach to estimation of the regression coefficients of a multinominal logit model with ordinal scale response categories is presented. A Monte Carlo method is used to construct the posterior distribution of the link function. The link function is treated as an arbitrary scalar function. Then the Gauss-Markov theorem is used to determine a function of the link which produces a random vector of coefficients. The posterior distribution of the random vector of coefficients is used to estimate the regression coefficients. The method described is referred to as a Bayesian generalized least square (BGLS) analysis. Two cases involving multinominal logit models are described. Case I involves a cumulative logit model and Case II involves a proportional-odds model. All inferences about the coefficients for both cases are described in terms of the posterior distribution of the regression coefficients. The results from the BGLS method are compared to maximum likelihood estimates of the regression coefficients. The BGLS method avoids the nonlinear problems encountered when estimating the regression coefficients of a generalized linear model. The method is not complex or computationally intensive. The BGLS method offers several advantages over Bayesian approaches. ^

Using spatial linear models with SAR and CAR structure to examine Texas lung cancer incidence rates

Scholars have found that socioeconomic status was one of the key factors that influenced early-stage lung cancer incidence rates in a variety of regions. This thesis examined the association between median household income and lung cancer incidence rates in Texas counties. A total of 254 individual counties in Texas with corresponding lung cancer incidence rates from 2004 to 2008 and median household incomes in 2006 were collected from the National Cancer Institute Surveillance System. A simple linear model and spatial linear models with two structures, Simultaneous Autoregressive Structure (SAR) and Conditional Autoregressive Structure (CAR), were used to link median household income and lung cancer incidence rates in Texas. The residuals of the spatial linear models were analyzed with Moran's I and Geary's C statistics, and the statistical results were used to detect similar lung cancer incidence rate clusters and disease patterns in Texas.^

BAYESIAN STATISTICAL METHODS IN GENE-ENVIRONMENT AND GENE-GENE INTERACTION STUDIES

Complex diseases such as cancer result from multiple genetic changes and environmental exposures. Due to the rapid development of genotyping and sequencing technologies, we are now able to more accurately assess causal effects of many genetic and environmental factors. Genome-wide association studies have been able to localize many causal genetic variants predisposing to certain diseases. However, these studies only explain a small portion of variations in the heritability of diseases. More advanced statistical models are urgently needed to identify and characterize some additional genetic and environmental factors and their interactions, which will enable us to better understand the causes of complex diseases. In the past decade, thanks to the increasing computational capabilities and novel statistical developments, Bayesian methods have been widely applied in the genetics/genomics researches and demonstrating superiority over some regular approaches in certain research areas. Gene-environment and gene-gene interaction studies are among the areas where Bayesian methods may fully exert its functionalities and advantages. This dissertation focuses on developing new Bayesian statistical methods for data analysis with complex gene-environment and gene-gene interactions, as well as extending some existing methods for gene-environment interactions to other related areas. It includes three sections: (1) Deriving the Bayesian variable selection framework for the hierarchical gene-environment and gene-gene interactions; (2) Developing the Bayesian Natural and Orthogonal Interaction (NOIA) models for gene-environment interactions; and (3) extending the applications of two Bayesian statistical methods which were developed for gene-environment interaction studies, to other related types of studies such as adaptive borrowing historical data. We propose a Bayesian hierarchical mixture model framework that allows us to investigate the genetic and environmental effects, gene by gene interactions (epistasis) and gene by environment interactions in the same model. It is well known that, in many practical situations, there exists a natural hierarchical structure between the main effects and interactions in the linear model. Here we propose a model that incorporates this hierarchical structure into the Bayesian mixture model, such that the irrelevant interaction effects can be removed more efficiently, resulting in more robust, parsimonious and powerful models. We evaluate both of the 'strong hierarchical' and 'weak hierarchical' models, which specify that both or one of the main effects between interacting factors must be present for the interactions to be included in the model. The extensive simulation results show that the proposed strong and weak hierarchical mixture models control the proportion of false positive discoveries and yield a powerful approach to identify the predisposing main effects and interactions in the studies with complex gene-environment and gene-gene interactions. We also compare these two models with the 'independent' model that does not impose this hierarchical constraint and observe their superior performances in most of the considered situations. The proposed models are implemented in the real data analysis of gene and environment interactions in the cases of lung cancer and cutaneous melanoma case-control studies. The Bayesian statistical models enjoy the properties of being allowed to incorporate useful prior information in the modeling process. Moreover, the Bayesian mixture model outperforms the multivariate logistic model in terms of the performances on the parameter estimation and variable selection in most cases. Our proposed models hold the hierarchical constraints, that further improve the Bayesian mixture model by reducing the proportion of false positive findings among the identified interactions and successfully identifying the reported associations. This is practically appealing for the study of investigating the causal factors from a moderate number of candidate genetic and environmental factors along with a relatively large number of interactions. The natural and orthogonal interaction (NOIA) models of genetic effects have previously been developed to provide an analysis framework, by which the estimates of effects for a quantitative trait are statistically orthogonal regardless of the existence of Hardy-Weinberg Equilibrium (HWE) within loci. Ma et al. (2012) recently developed a NOIA model for the gene-environment interaction studies and have shown the advantages of using the model for detecting the true main effects and interactions, compared with the usual functional model. In this project, we propose a novel Bayesian statistical model that combines the Bayesian hierarchical mixture model with the NOIA statistical model and the usual functional model. The proposed Bayesian NOIA model demonstrates more power at detecting the non-null effects with higher marginal posterior probabilities. Also, we review two Bayesian statistical models (Bayesian empirical shrinkage-type estimator and Bayesian model averaging), which were developed for the gene-environment interaction studies. Inspired by these Bayesian models, we develop two novel statistical methods that are able to handle the related problems such as borrowing data from historical studies. The proposed methods are analogous to the methods for the gene-environment interactions on behalf of the success on balancing the statistical efficiency and bias in a unified model. By extensive simulation studies, we compare the operating characteristics of the proposed models with the existing models including the hierarchical meta-analysis model. The results show that the proposed approaches adaptively borrow the historical data in a data-driven way. These novel models may have a broad range of statistical applications in both of genetic/genomic and clinical studies.

Comparing linear and non linear wind flow models

Assessing wind conditions on complex terrain has become a hard task as terrain complexity increases. That is why there is a need to extrapolate in a reliable manner some wind parameters that determine wind farms viability such as annual average wind speed at all hub heights as well as turbulence intensities. The development of these tasks began in the early 90´s with the widely used linear model WAsP and WAsP Engineering especially designed for simple terrain with remarkable results on them but not so good on complex orographies. Simultaneously non-linearized Navier Stokes solvers have been rapidly developed in the last decade through CFD (Computational Fluid Dynamics) codes allowing simulating atmospheric boundary layer flows over steep complex terrain more accurately reducing uncertainties. This paper describes the features of these models by validating them through meteorological masts installed in a highly complex terrain. The study compares the results of the mentioned models in terms of wind speed and turbulence intensity.

Fate and Transport of Naproxen in a Sandy Aquifer Material: Saturated Column Studies and Model Evaluation

Naproxen-C14H14O3 is a nonsteroidal anti-inflammatory drug which has been found at detectable concentrations in wastewater, surface water, and groundwater. Naproxen is relatively hydrophilic and is in anionic form at pH between 6 and 8. In this study, column experiments were performed using an unconsolidated aquifer material from an area near Barcelona (Spain) to assess transport and reaction mechanisms of Naproxen in the aquifer matrix under different pore water fluxes. Results were evaluated using HYDRUS-1D, which was used to estimate transport parameters. Batch sorption isotherms for Naproxen conformed with the linear model with a sorption coefficient of 0.42 (cm3 g−1), suggesting a low sorption affinity. Naproxen breakthrough curves (BTCs) measured in soil columns under steady-state, saturated water flow conditions displayed similar behavior, with no apparent hysteresis in sorption or dependence of retardation (R, 3.85-4.24) on pore water velocities. Soil sorption did not show any significant decrease for increasing flow rates, as observed from Naproxen recovery in the effluent. Sorption parameters estimated by the model suggest that Naproxen has a low sorption affinity to aquifer matrix. Most sorption of Naproxen occurred on the instantaneous sorption sites, with the kinetic sorption sites representing only about 10 to 40% of total sorption.

Transferability of Regional Permafrost Disturbance Susceptibility Modelling Using Generalized Linear and Generalized Additive Models

To effectively assess and mitigate risk of permafrost disturbance, disturbance-p rone areas can be predicted through the application of susceptibility models. In this study we developed regional susceptibility models for permafrost disturbances using a field disturbance inventory to test the transferability of the model to a broader region in the Canadian High Arctic. Resulting maps of susceptibility were then used to explore the effect of terrain variables on the occurrence of disturbances within this region. To account for a large range of landscape charac- teristics, the model was calibrated using two locations: Sabine Peninsula, Melville Island, NU, and Fosheim Pen- insula, Ellesmere Island, NU. Spatial patterns of disturbance were predicted with a generalized linear model (GLM) and generalized additive model (GAM), each calibrated using disturbed and randomized undisturbed lo- cations from both locations and GIS-derived terrain predictor variables including slope, potential incoming solar radiation, wetness index, topographic position index, elevation, and distance to water. Each model was validated for the Sabine and Fosheim Peninsulas using independent data sets while the transferability of the model to an independent site was assessed at Cape Bounty, Melville Island, NU. The regional GLM and GAM validated well for both calibration sites (Sabine and Fosheim) with the area under the receiver operating curves (AUROC) N 0.79. Both models were applied directly to Cape Bounty without calibration and validated equally with AUROC's of 0.76; however, each model predicted disturbed and undisturbed samples differently. Addition- ally, the sensitivity of the transferred model was assessed using data sets with different sample sizes. Results in- dicated that models based on larger sample sizes transferred more consistently and captured the variability within the terrain attributes in the respective study areas. Terrain attributes associated with the initiation of dis- turbances were similar regardless of the location. Disturbances commonly occurred on slopes between 4 and 15°, below Holocene marine limit, and in areas with low potential incoming solar radiation

Comparison of virulence gene profiles of Escherichia coli strains isolated from healthy and diarrheic swine

A combination of uni- and multiplex PCR assays targeting 58 virulence genes (VGs) associated with Escherichia coli strains causing intestinal and extraintestinal disease in humans and other mammals was used to analyze the VG repertoire of 23 commensal E. coli isolates from healthy pigs and 52 clinical isolates associated with porcine neonatal diarrhea (ND) and postweaning diarrhea (PWD). The relationship between the presence and absence of VGs was interrogated using three statistical methods. According to the generalized linear model, 17 of 58 VGs were found to be significant (P < 0.05) in distinguishing between commensal and clinical isolates. Nine of the 17 genes represented by iha, hlyA, aidA, east1, aah, fimH, iroN(E).(coli), traT, and saa have not been previously identified as important VGs in clinical porcine isolates in Australia. The remaining eight VGs code for fimbriae (F4, F5, F18, and F41) and toxins (STa, STh, LT, and Stx2), normally associated with porcine enterotoxigenic E. coli. Agglomerative hierarchical algorithm analysis grouped E. coli strains into subclusters based primarily on their serogroup. Multivariate analyses of clonal relationships based on the 17 VGs were collapsed into two-dimensional space by principal coordinate analysis. PWD clones were distributed in two quadrants, separated from ND and commensal clones, which tended to cluster within one quadrant. Clonal subclusters within quadrants were highly correlated with serogroups. These methods of analysis provide different perspectives in our attempts to understand how commensal and clinical porcine enterotoxigenic E. coli strains have evolved and are engaged in the dynamic process of losing or acquiring VGs within the pig population.

Finite deformation model of simple shear of fault with microrotations: apparent strain localisation and en-echelon fracture pattern

Strain localisation is a widespread phenomenon often observed in shear and compressive loading of geomaterials, for example, the fault gouge. It is believed that the main mechanisms of strain localisation are strain softening and mismatch between dilatancy and pressure sensitivity. Observations show that gouge deformation is accompanied by considerable rotations of grains. In our previous work as a model for gouge material, we proposed a continuum description for an assembly of particles of equal radius in which the particle rotation is treated as an independent degree of freedom. We showed that there exist critical values of the model parameters for which the displacement gradient exhibits a pronounced localisation at the mid-surface layers of the fault, even in the absence of inelasticity. Here, we generalise the model to the case of finite deformations characteristic for the gouge deformation. We derive objective constitutive relationships relating the Jaumann rates of stress and moment stress to the relative strain and curvature rates, respectively. The model suggests that the pattern of localisation remains the same as in the linear case. However, the presence of the Jaumann terms leads to the emergence of non-zero normal stresses acting along and perpendicular to the shear layer (with zero hydrostatic pressure), and localised along the mid-line of the gouge; these stress components are absent in the linear model of simple shear. These additional normal stresses, albeit small, cause a change in the direction in which the maximal normal stresses act and in which en-echelon fracturing is formed.

Human resource management and organizational performance:evidence from the retail banking sector

Underpinned by the resource-based view (RBV), social exchange theory (SET), and a theory of intrinsic motivation (empowerment), I proposed and tested a multi-level model that simultaneously examines the intermediate linkages or mechanisms through which HPWS impact individual and organizational performance. First and underpinned by RBV, I examined at the unit level, collective human capital and competitive advantage as path-ways through which the use of HPWS influences – branch market performance. Second and-, underpinned by social exchange (perceived organizational support) and intrinsic motivation (psychological empowerment) theories, I examined cross and individual level mechanisms through which experienced HPWS may influence employee performance. I tested the propositions of this study with multisource data obtained from junior and senior customer contact employees, and managers of 37 branches of two banks in Ghana. Results of the Structural Equation Modeling (SEM) analysis revealed that (i) collective human capital partially mediated the relationship between management-rated HPWS and competitive advantage, while competitive advantage completely mediated the influence of human capital on branch market performance. Consequently, management-rated HPWS influenced branch market performance indirectly through collective human capital and competitive advantage. Additionally, results of hierarchical linear modeling (HLM) tests of the cross-level influences on the motivational implications of HPWS revealed that (i) management-rated HPWS influenced experienced HPWS; (ii) perceived organizational support (POS) and psychological empowerment fully mediated the influence of experienced HPWS on service-oriented organizational citizenship behaviour (OCB), and; (iii) service-oriented OCB mediated the influence of psychological empowerment and POS on service quality and task performance. I discuss the theoretical and practical implications of these findings.

Gaussian regression and optimal finite dimensional linear models

The problem of regression under Gaussian assumptions is treated generally. The relationship between Bayesian prediction, regularization and smoothing is elucidated. The ideal regression is the posterior mean and its computation scales as O(n³), where n is the sample size. We show that the optimal m-dimensional linear model under a given prior is spanned by the first m eigenfunctions of a covariance operator, which is a trace-class operator. This is an infinite dimensional analogue of principal component analysis. The importance of Hilbert space methods to practical statistics is also discussed.

Clustering web documents using hierarchical representation with multi-granularity

Web document cluster analysis plays an important role in information retrieval by organizing large amounts of documents into a small number of meaningful clusters. Traditional web document clustering is based on the Vector Space Model (VSM), which takes into account only two-level (document and term) knowledge granularity but ignores the bridging paragraph granularity. However, this two-level granularity may lead to unsatisfactory clustering results with “false correlation”. In order to deal with the problem, a Hierarchical Representation Model with Multi-granularity (HRMM), which consists of five-layer representation of data and a twophase clustering process is proposed based on granular computing and article structure theory. To deal with the zero-valued similarity problemresulted from the sparse term-paragraphmatrix, an ontology based strategy and a tolerance-rough-set based strategy are introduced into HRMM. By using granular computing, structural knowledge hidden in documents can be more efficiently and effectively captured in HRMM and thus web document clusters with higher quality can be generated. Extensive experiments show that HRMM, HRMM with tolerancerough-set strategy, and HRMM with ontology all outperform VSM and a representative non VSM-based algorithm, WFP, significantly in terms of the F-Score.

Leveraging employee creativity through high performance work systems:a multilevel perspective

Research has looked at single rather than a configuration of human resource management (HRM) practices to influence creativity so it is not yet clear how these practices synergistically facilitate creativity and organisational performance. I address this significant but unanswered question in a three-part study. In Study 1, I develop a high performance work system (HPWS) for creativity scale. I use Study 2 sample to test the validity of the new scale. In Study 3, I test a multilevel model of the intervening processes through which branch HPWS for creativity influences creativity and branch performance. Specifically, at the branch level, I draw on social context theory and hypothesise that branch HPWS for creativity relates to climate for creativity which, in turn, leads to creativity, and ultimately, to profit. Furthermore, I hypothesise environmental dynamism as a boundary condition of the creativity-profit relationship. At the individual level, I hypothesise a cross-level effect of branch HPWS for creativity on employee-perceived HPWS. I draw on self-determination theory and argue that perceived HPWS for creativity relate to need satisfaction and the psychological pathways of intrinsic motivation and creative process engagement to predict creativity. I also hypothesise climate for creativity as a cross-level moderator of the intrinsic motivation-creativity and creative process engagement-creativity relationships. Results of hierarchical linear modeling (HLM) indicate that ten out of the fifteen hypotheses were supported. The findings of this study respond to calls for HPWS to be designed around a strategic focus by developing and providing initial validity evidence of an HPWS for creativity scale. The results reveal the underlying mechanisms through which HPWS for creativity simultaneously influences individual and branch creativity leading to profit. Lastly, results indicate environmental dynamism to be an important boundary condition of the creativity-profit relationship and climate for creativity as a cross-level moderator of the creative process engagement-creativity.

HI!MVC: Hierarchical MVC Framework for WEB Applications

This paper presents Hi!MVC, a framework for developing high interactive web applications with a MVC Architecture. Nowadays, to manage, extend and correct web applications can be difficult due to the navigational paradigm they are based on. Hi!MVC framework helps to make these tasks easier. This framework allows building a web based interface, generating each page from the objects that represent its state. Every class to be showed in the interface is associated with two entities: its html representation (view) and its interactions in the view manager (controller). The whole html page is generated by composition of views according to the composition relationship of objects. Interactions between user and application are managed by the controller associated to the view which shows interaction elements (links or forms). Hi!MVC allows building web interface in a hierarchical and distributed way. There are other frameworks and APIs offering MVC architectures to web applications, but we think that they are not applying exactly the same concepts. While they keep on basing their architectures on the navigational paradigm we are offering a new point of view based on an innovator hierarchical model. First, we present the main ideas of our proposal. Next, we expose how to implement it using different Java technologies. Finally, we make a first approach to our hierarchical MVC model. We also compare shortly our proposal with the previously cited technologies.

Antecedents and consequences of percieved effective managerial coaching:an individual, team-level and cross-level examination

The major contribution of the study is the identification of a positive link between perceived effective managerial coaching (PEMC) and team task performance and also, the examination of PEMC adopting a multilevel research design and incorporating dual-source data. Specifically, drawing on social psychology, the thesis aims at developing and testing a comprehensive conceptual framework of the antecedents and consequences of PEMC for knowledge workers. The model takes into consideration intrapersonal, interpersonal and team-level characteristics, which relate to PEMC and, subsequently associate with important work outcomes. In this regard, the thesis identifies PEMC as a practice of dual nature in that it may be experienced not only as a one-on-one workplace developmental interaction, but also as a managerial practice that is experienced by each member of a team for co-ordination purposes. Adopting a cross-sectional survey research design, the hypotheses are tested in three organisations in Greece and the UK. In particular, hierarchical linear modelling of 191 employees nested in 60 teams yields that employees’ learning goal orientation (LGO) and high-quality exchanges between an employee and a manager (LMX) are positively related to effective MC, while a manager’s LGO moderates the relationship between employees’ LGO and PEMC. In turn, PEMC, as a one-on-one practice, is related to cognitive outcomes, such as information sharing, while as a shared team practice is related also to behavioural outcomes, including individual and team performance. Overall, the study contributes to a growing body of coaching and management literature that acknowledges PEMC as a core managerial practice.

The effects of servant leadership on follower performance and well-being:underlying mechanisms, boundary conditions, and the role of training

Based on a review of the servant leadership, well-being, and performance literatures, the first study develops a research model that examines how and under which conditions servant leadership is related to follower performance and well-being alike. Data was collected from 33 leaders and 86 of their followers working in six organizations. Multilevel moderated mediation analyses revealed that servant leadership was indeed related to eudaimonic well-being and lead-er-rated performance via followers’ positive psychological capital, but that the strength and di-rection of the examined relationships depended on organizational policies and practices promot-ing employee health, and in the case of follower performance on a developmental team climate, shedding light on the importance of the context in which servant leadership takes place. In addi-tion, two more research questions resulted from a review of the training literature, namely how and under which conditions servant leadership can be trained, and whether follower performance and well-being follow from servant leadership enhanced by training. We subsequently designed a servant leadership training and conducted a longitudinal field experiment to examine our sec-ond research question. Analyses were based on data from 38 leaders randomly assigned to a training or control condition, and 91 of their followers in 36 teams. Hierarchical linear modeling results showed that the training, which addressed the knowledge of, attitudes towards, and ability to apply servant leadership, positively affected leader and follower perceptions of servant leader-ship, but in the latter case only when leaders strongly identified with their team. These findings provide causal evidence as to how and when servant leadership can be effectively developed. Fi-nally, the research model of Study 1 was replicated in a third study based on 58 followers in 32 teams drawn from the same population used for Study 2, confirming that follower eudaimonic well-being and leader-rated performance follow from developing servant leadership via increases in psychological capital, and thus establishing the directionality of the examined relationships.