Essays on international lending and increasing returns to scale

Standard economic theory suggests that capital should flow from rich countries to poor countries. However, capital has predominantly flowed to rich countries. The three essays in this dissertation attempt to explain this phenomenon. The first two essays suggest theoretical explanations for why capital has not flowed to the poor countries. The third essay empirically tests the theoretical explanations.^ The first essay examines the effects of increasing returns to scale on international lending and borrowing with moral hazard. Introducing increasing returns in a two-country general equilibrium model yields possible multiple equilibria and helps explain the possibility of capital flows from a poor to a rich country. I find that a borrowing country may need to borrow sufficient amounts internationally to reach a minimum investment threshold in order to invest domestically.^ The second essay examines how a poor country may invest in sectors with low productivity because of sovereign risk, and how collateral differences across sectors may exacerbate the problem. I model sovereign borrowing with a two-sector economy: one sector with increasing returns to scale (IRS) and one sector with diminishing returns to scale (DRS). Countries with incomes below a threshold will only invest in the DRS sector, and countries with incomes above a threshold will invest mostly in the IRS sector. The results help explain the existence of a bimodal world income distribution.^ The third essay empirically tests the explanations for why capital has not flowed from the rich to the poor countries, with a focus on institutions and initial capital. I find that institutional variables are a very important factor, but in contrast to other studies, I show that institutions do not account for the Lucas Paradox. Evidence of increasing returns still exists, even when controlling for institutions and other variables. In addition, I find that the determinants of capital flows may depend on whether a country is rich or poor.^

Control of MNC environmental performance and the challenges of subsidiary network dimensions

The trend of green consumerism and increased standardization of environmental regulations has driven multinational corporations (MNCs) to seek standardization of environmental practices or at least seek to be associated with such behavior. In fact, many firms are seeking to free ride on this global green movement, without having the actual ecological footprint to substantiate their environmental claims. While scholars have articulated the benefits from such optimization of uniform global green operations, the challenges for MNCs to control and implement such operations are understudied. For firms to translate environmental commitment to actual performance, the obstacles are substantial, particularly for the MNC. This is attributed to headquarters' (HQ) control challenges (1) in managing core elements of the corporate environmental management (CEM) process and specifically matching verbal commitment and policy with ecological performance and by (2) the fact that the MNC operates in multiple markets and the HQ is required to implement policy across complex subsidiary networks consisting of diverse and distant units. Drawing from the literature on HQ challenges of MNC management and control, this study examines (1) how core components of the CEM process impact optimization of global environmental performance (GEP) and then uses network theory to examine how (2) a subsidiary network's dimensions can present challenges to the implementation of green management policies. It presents a framework for CEM which includes (1) MNCs' Verbal environmental commitment, (2) green policy Management which guides standards for operations, (3) actual environmental Performance reflected in a firm's ecological footprint and (4) corporate environmental Reputation (VMPR). Then it explains how an MNC's key subsidiary network dimensions (density, diversity, and dispersion) create challenges that hinder the relationship between green policy management and actual environmental performance. It combines content analysis, multiple regression, and post-hoc hierarchal cluster analysis to study US manufacturing MNCs. The findings support a positive significant effect of verbal environmental commitment and green policy management on actual global environmental performance and environmental reputation, as well as a direct impact of verbal environmental commitment on green policy management. Unexpectedly, network dimensions were not found to moderate the relationship between green management policy and GEP.

Treatment moderation and secondary outcomes: Results from a randomized clinical trial

The present study pursued two objectives in the context of a randomized clinical trial of cognitive-behavioral therapy with parent (CBT/P) and group (GCBT) involvement. The first objective was to examine the variability in treatment outcome. There were three specific aims within the first objective, to evaluate: (1) youth characteristics (age, depressive, and externalizing disorders) as moderators of treatment outcome; (2) the differential outcome of the treatment approaches as a function of youth characteristics; and (3) the relative efficacy of the treatment approaches at each level of the moderators. ^ The second objective was to evaluate the efficacy of anxiety treatments along secondary depressive symptoms and externalizing behaviors. There were five specific aims within the second objective, to evaluate: (1) whether anxiety treatment yields reductions in secondary problems, (2) the efficacy of anxiety treatments in reducing secondary problems as a function of approach and youth characteristics, (3) whether reductions in anxiety symptoms significantly mediate changes in secondary problems, (4) the directionality of change in the hypothesized mediated relations, and (5) whether the hypothesized mediated relations are moderated by treatment approach and youth characteristics. The specific aims were pursued using data collected from 183 youth and their mothers. Research questions were tested using multiple regressions and structural equation modeling. ^ Age, depressive, and externalizing disorders were significant moderators. CBT/P relative to GCBT lowered anxiety more for younger than older youth. GCBT relative to CBT/P lowered anxiety more for older than younger youth. GCBT relative to CBT/P lowered anxiety more for depressed youth than non-depressed youth. GCBT relative to CBT/P lowered anxiety less for externalizing youth than non-externalizing youth. Treatment reduced depressive symptoms and externalizing problem behaviors. Reductions in anxiety mediated changes in depressive symptoms and externalizing problem behaviors. Reversed directionality was found in the relation between social anxiety and depressive symptoms. In CBT/P the direction of change was from depressive to social anxiety. The opposite was true in GCBT. Reductions in social anxiety mediated posttreatment changes in depressive symptoms in GCBT but not CBT/P. The reverse was true at follow-up. Reductions in social anxiety mediated changes in depressive symptoms for girls but not boys.^

Host and network optimizations for performance enhancement and energy efficiency in data center networks

Modern data centers host hundreds of thousands of servers to achieve economies of scale. Such a huge number of servers create challenges for the data center network (DCN) to provide proportionally large bandwidth. In addition, the deployment of virtual machines (VMs) in data centers raises the requirements for efficient resource allocation and find-grained resource sharing. Further, the large number of servers and switches in the data center consume significant amounts of energy. Even though servers become more energy efficient with various energy saving techniques, DCN still accounts for 20% to 50% of the energy consumed by the entire data center. The objective of this dissertation is to enhance DCN performance as well as its energy efficiency by conducting optimizations on both host and network sides. First, as the DCN demands huge bisection bandwidth to interconnect all the servers, we propose a parallel packet switch (PPS) architecture that directly processes variable length packets without segmentation-and-reassembly (SAR). The proposed PPS achieves large bandwidth by combining switching capacities of multiple fabrics, and it further improves the switch throughput by avoiding padding bits in SAR. Second, since certain resource demands of the VM are bursty and demonstrate stochastic nature, to satisfy both deterministic and stochastic demands in VM placement, we propose the Max-Min Multidimensional Stochastic Bin Packing (M3SBP) algorithm. M3SBP calculates an equivalent deterministic value for the stochastic demands, and maximizes the minimum resource utilization ratio of each server. Third, to provide necessary traffic isolation for VMs that share the same physical network adapter, we propose the Flow-level Bandwidth Provisioning (FBP) algorithm. By reducing the flow scheduling problem to multiple stages of packet queuing problems, FBP guarantees the provisioned bandwidth and delay performance for each flow. Finally, while DCNs are typically provisioned with full bisection bandwidth, DCN traffic demonstrates fluctuating patterns, we propose a joint host-network optimization scheme to enhance the energy efficiency of DCNs during off-peak traffic hours. The proposed scheme utilizes a unified representation method that converts the VM placement problem to a routing problem and employs depth-first and best-fit search to find efficient paths for flows.

Leadership Styles, Behaviors, and Outcomes of Athletic Training Education Program Directors

Context: With the increase in athletic training education programs, the demand for a highly qualified faculty member to serve as program director (PD) and fill the multiple roles and responsibilities of the position has increased. A successful PD must possess leadership skills crucial for the evolving academic environment. Research suggests that educational leaders must provide both transactional and transformational leadership if athletic trainers are to secure a legitimate place as healthcare providers. Objective: To describe the leadership styles and behaviors of athletic training education PDs and to describe the associations between leadership style, behavior, outcome, and experience. Design: We will utilize a survey design to identify the leadership styles, behaviors, outcomes, and experiences of athletic training education PDs. Setting: On-line questionnaire. Participants: The population of this study will be limited to the academic PDs of the 360 accredited entry-level athletic training education programs in the United States. Intervention: The investigation will utilize the Multifactor Leadership Questionnaire (MLQ). The MLQ is a validated tool composed of 45 items that identify and measure key leadership and effectiveness behaviors shown to be strongly linked with both individual and organizational success. In addition to the leader, it is recommended that all persons working above, below, and directly at the same organizational level as the leader rate the leader. Raters evaluate how frequently, or to what degree, they have observed the leader engage in 32 specific behaviors. Main Outcome Measure(s): Statistical analysis will be utilized to describe the associations between leadership styles, behaviors, outcomes, and experiences.

Examining the validity of the GRE General Test scores and undergraduate GPA for predicting success in graduate school at a large racially and ethnically diverse public university in Southeast Florida

The purpose of the study was to determine the degree of relationships among GRE scores, undergraduate GPA (UGPA), and success in graduate school, as measured by first year graduate GPA (FGPA), cumulative graduate GPA, and degree attainment status. A second aim of the study was to determine whether the relationships between the composite predictor (GRE scores and UGPA) and the three success measures differed by race/ethnicity and sex. A total of 7,367 graduate student records (masters, 5,990; doctoral: 1,377) from 2000 to 2010 were used to evaluate the relationships among GRE scores, UGPA and the three success measures. Pearson's correlation, multiple linear and logistic regression, and hierarchical multiple linear and logistic regression analyses were performed to answer the research questions. The results of the correlational analyses differed by degree level. For master's students, the ETS proposed prediction that GRE scores are valid predictors of first year graduate GPA was supported by the findings from the present study; however, for doctoral students, the proposed prediction was only partially supported. Regression and correlational analyses indicated that UGPA was the variable that consistently predicted all three success measures for both degree levels. The hierarchical multiple linear and logistic regression analyses indicated that at master's degree level, White students with higher GRE Quantitative Reasoning Test scores were more likely to attain a degree than Asian Americans, while International students with higher UGPA were more likely to attain a degree than White students. The relationships between the three predictors and the three success measures were not significantly different between men and women for either degree level. Findings have implications both for practice and research. They will provide graduate school administrators with institution-specific validity data for UGPA and the GRE scores, which can be referenced in making admission decisions, while they will provide empirical and professionally defensible evidence to support the current practice of using UGPA and GRE scores for admission considerations. In addition, new evidence relating to differential predictions will be useful as a resource reference for future GRE validation researchers.

Syndemic health disparities and resilience processes related to HIV transmission risk among African American/Black men in south Florida

Rates of HIV infection continue to climb among minority populations and men who have sex with men (MSM), with African American/Black MSM being especially impacted. Numerous studies have found HIV transmission risk to be associated with many health and social disparities resulting from larger environmental and structural forces. Using anthropological and social environment-based theories of resilience that focus on individual agency and larger social and environmental structures, this dissertation employed a mixed methods design to investigate resilience processes among African American/Black MSM.^ Quantitative analyses compared African American/Black (N=108) and Caucasian/White (N=250) MSM who participated in a previously conducted randomized controlled trial (RCT) of sexual and substance use risk reduction interventions. At RCT study entry, using past 90 day recall periods, there were no differences in unprotected sex frequency, however African American/Black MSM reported higher frequencies of days high (P<0.000), and drugs and sex used in combination (P<0.000), and substance dependence (P<0.000) and lower levels of social support (P<0.024) compared to Caucasian/White MSM. At 12- month follow-up, multi-level statistical models found that African American/Black MSM reduced their frequencies of days high and unprotected sex at greater rates than Caucasian/White MSM (P<0.001).^ Qualitative data collected among a sub-sample of African American/Black MSM from the RCT (N=21) described the men's experiences of living with multiple health and social disparities and the importance of RCT study assessments in facilitating reductions in risk behaviors. A cross-case analysis showed different resilience processes undertaken by men who experienced low socioeconomic status, little family support, and homophobia (N=16) compared to those who did not (N=5).^ The dissertation concludes that resilience processes to HIV transmission risk and related health and social disparities among African American/Black MSM varies and are dependent on specific social environmental factors, including social relationships, structural homophobia, and access to social, economic, and cultural capital. Men define for themselves what it means to be resilient within their social environment. These conclusions suggest that both individual and structural-level resilience-based HIV prevention interventions are needed.^

Executive Characteristics and Going Concern Opinions

Auditors have come under increased scrutiny over the past several years about the growing number of client failures without a warning in the form of a going-concern modified (GCM) audit opinion. Statement on Auditing Standards No. 59 requires auditors to evaluate whether substantial doubt exists on an audit client’s ability to continue as a going concern (AICPA 1988). My dissertation consists of three essays. For the three essays, I empirically investigate issues related to GCM audit opinions and executive characteristics. Specifically, I examine the impact of executive tenure and gender on the issuance of GCM audit opinions. In addition, my dissertation addresses two other unique issues. Given that the Sarbanes-Oxley Act represents an important watershed event in the history and regulation of the accounting profession, I provide evidence about auditors’ propensities to issue GCM audit opinions in the post-SOX period. Further, I also expand extant research in this area by using multiple outcomes and thus go beyond the traditional use of bankruptcy alone as a tool to evaluate auditors’ GCM opinion. The results indicate that, after controlling for other financial characteristics, GCM audit opinions are significantly more likely for firms that have CFOs with short tenure and/or for firms with a female CFO or CEO. However, when examining the association between executive characteristics and two types of reporting errors, the results vary with the type of reporting error. Overall, the results provide evidence that executive characteristics are associated with auditors' reporting decisions.

Physical properties and sedimentology of 4 profiles fom the South Atlantic

Ultrasonic P wavc transmission seismograms recorded on sediment cores have been analyzed to study the acoustic and estimate the clastic properties of marine sediments from different provinces dominated by terrigenous, calcareous, amI diatomaceous sedimentation. Instantaneous frequencies computed from the transmission seismograms are displayed as gray-shaded images to give an acoustic overview of the lithology of each core. Ccntirneter-scale variations in the ultrasonic waveforms associated with lithological changes are illustrated by wiggle traces in detail. Cross-correlation, multiple-filter, and spectral ratio techniques are applied to derive P wave velocities and attenuation coefficients. S wave velocities and attenuation coefficients, elastic moduli, and permeabilities are calculated by an inversion scheme based on the Biot-Stoll viscoelastic model. Together wilh porosity measurements, P and S wave scatter diagrams are constructed to characterize different sediment types by their velocity- and attenuation-porosity relationships. They demonstrate that terrigenous, calcareous, and diatomaceous sediments cover different velocity- and attenuation-porosity ranges. In terrigcnous sediments, P wave vclocities and attenuation coefficients decrease rapidly with increasing porosity, whereas S wave velocities and shear moduli are very low. Calcareous sediments behave similarly at relatively higher porosities. Foraminifera skeletons in compositions of terrigenous mud and calcareous ooze cause a stiffening of the frame accompanied by higher shear moduli, P wave velocities, and attenuation coefficients. In diatomaceous ooze the contribution of the shear modulus becomes increasingly important and is controlled by the opal content, whereas attenuation is very low. This leads to the opportunity to predict the opal content from nondestructive P wave velocity measurements at centimeter-scale resolution.

Influence of the sputtering flow regime on the structural properties and magnetic behavior of Fe-Ga thin films (Ga ∼ 30 at.%)

In this paper we analyze the structure of Fe-Ga layers with a Ga content of ∼30 at.% deposited by the sputtering technique under two different regimes. We also studied the correlation between the structure and magnetic behavior of the samples. Keeping the Ar pressure fixed, we modified the flow regime from ballistic to diffusive by increasing the distance between the target and the substrate. X-ray diffraction measurements have shown a lower structural quality when growing in the diffusive flow. We investigated the impact of the growth regime by means of x-ray absorption fine structure (XAFS) measurements and obtained signs of its influence on the local atomic order. Full multiple scattering and finite difference calculations based on XAFS measurements point to a more relevant presence of a disordered A2 phase and of orthorhombic Ga clusters on the Fe-Ga alloy deposited under a diffusive regime; however, in the ballistic sample, a higher presence of D0_3/B2 phases is evidenced. Structural characteristics, from local to long range, seem to determine the magnetic behavior of the layers. Whereas a clear in-plane magnetic anisotropy is observed in the film deposited under ballistic flow, the diffusive sample is magnetically isotropic. Therefore, our experimental results provide evidence of a correlation between flow regime and structural properties and its impact on the magnetic behavior of a rather unexplored compositional region of Fe-Ga compounds.

Redeeming Faustus: Tracing the pacts of Mariken and Faust from the 1500s to the present

This dissertation uncovers and analyzes the complicated history of the devil’s pact in literature from approximately 1330 to 2015, focusing primarily on texts written in German and Dutch. That the tale of the pact with the devil (the so-called Faustian bargain) is one of the most durable and pliable literary themes is undeniable. Yet for too long, the success of Johann Wolfgang von Goethe’s Faust I (1808) decisively shaped scholarship on early devil’s pact tales, leading to a misreading of the texts with Goethe’s concerns being projected onto the earliest manifestations. But Goethe’s Faust really only borrows from the original Faust his name; the two characters could not be more different. Furthermore, Faustus was not the only early pact-maker character and his tale was neither limited to the German language nor to the Protestant faith. Among others, tales written in Dutch about a female, Catholic, latemedieval pact-maker, Mariken van Nieumeghen (1515), illustrate this. This dissertation seeks to redeem the early modern Faustus texts from its misreading and to broaden the scholarship on the literature of the devil’s pact by considering the Mariken and Faust traditions together.

The first chapter outlines the beginnings of pact literature as a Catholic phenomenon, considering the tales of Theophilus and Pope Joan alongside Mariken of Nijmegen. The second chapter turns to the original Faust tale, the Historia von D. Johann Fausten (1587), best read as a Lutheran response to the Catholic pact literature in the wake of the Reformation. In the third chapter, this dissertation offers a new, united reading of the early modern Faust tradition. The fourth and fifth chapters trace the literary preoccupation with the pacts of both Mariken and Faustus from the late early modern to the present.

The dissertation traces the evolution of these two bodies of literature and provides an in-depth analysis and comparison of the two that has not been done before. It argues for a more global literary scholarship that considers texts across multiple languages and one that takes into consideration the rich body of material of the pact tradition.

Data to Decision in a Dynamic Ocean: Robust Species Distribution Models and Spatial Decision Frameworks

Human use of the oceans is increasingly in conflict with conservation of endangered species. Methods for managing the spatial and temporal placement of industries such as military, fishing, transportation and offshore energy, have historically been post hoc; i.e. the time and place of human activity is often already determined before assessment of environmental impacts. In this dissertation, I build robust species distribution models in two case study areas, US Atlantic (Best et al. 2012) and British Columbia (Best et al. 2015), predicting presence and abundance respectively, from scientific surveys. These models are then applied to novel decision frameworks for preemptively suggesting optimal placement of human activities in space and time to minimize ecological impacts: siting for offshore wind energy development, and routing ships to minimize risk of striking whales. Both decision frameworks relate the tradeoff between conservation risk and industry profit with synchronized variable and map views as online spatial decision support systems.

For siting offshore wind energy development (OWED) in the U.S. Atlantic (chapter 4), bird density maps are combined across species with weights of OWED sensitivity to collision and displacement and 10 km2 sites are compared against OWED profitability based on average annual wind speed at 90m hub heights and distance to transmission grid. A spatial decision support system enables toggling between the map and tradeoff plot views by site. A selected site can be inspected for sensitivity to a cetaceans throughout the year, so as to capture months of the year which minimize episodic impacts of pre-operational activities such as seismic airgun surveying and pile driving.

Routing ships to avoid whale strikes (chapter 5) can be similarly viewed as a tradeoff, but is a different problem spatially. A cumulative cost surface is generated from density surface maps and conservation status of cetaceans, before applying as a resistance surface to calculate least-cost routes between start and end locations, i.e. ports and entrance locations to study areas. Varying a multiplier to the cost surface enables calculation of multiple routes with different costs to conservation of cetaceans versus cost to transportation industry, measured as distance. Similar to the siting chapter, a spatial decisions support system enables toggling between the map and tradeoff plot view of proposed routes. The user can also input arbitrary start and end locations to calculate the tradeoff on the fly.

Essential to the input of these decision frameworks are distributions of the species. The two preceding chapters comprise species distribution models from two case study areas, U.S. Atlantic (chapter 2) and British Columbia (chapter 3), predicting presence and density, respectively. Although density is preferred to estimate potential biological removal, per Marine Mammal Protection Act requirements in the U.S., all the necessary parameters, especially distance and angle of observation, are less readily available across publicly mined datasets.

In the case of predicting cetacean presence in the U.S. Atlantic (chapter 2), I extracted datasets from the online OBIS-SEAMAP geo-database, and integrated scientific surveys conducted by ship (n=36) and aircraft (n=16), weighting a Generalized Additive Model by minutes surveyed within space-time grid cells to harmonize effort between the two survey platforms. For each of 16 cetacean species guilds, I predicted the probability of occurrence from static environmental variables (water depth, distance to shore, distance to continental shelf break) and time-varying conditions (monthly sea-surface temperature). To generate maps of presence vs. absence, Receiver Operator Characteristic (ROC) curves were used to define the optimal threshold that minimizes false positive and false negative error rates. I integrated model outputs, including tables (species in guilds, input surveys) and plots (fit of environmental variables, ROC curve), into an online spatial decision support system, allowing for easy navigation of models by taxon, region, season, and data provider.

For predicting cetacean density within the inner waters of British Columbia (chapter 3), I calculated density from systematic, line-transect marine mammal surveys over multiple years and seasons (summer 2004, 2005, 2008, and spring/autumn 2007) conducted by Raincoast Conservation Foundation. Abundance estimates were calculated using two different methods: Conventional Distance Sampling (CDS) and Density Surface Modelling (DSM). CDS generates a single density estimate for each stratum, whereas DSM explicitly models spatial variation and offers potential for greater precision by incorporating environmental predictors. Although DSM yields a more relevant product for the purposes of marine spatial planning, CDS has proven to be useful in cases where there are fewer observations available for seasonal and inter-annual comparison, particularly for the scarcely observed elephant seal. Abundance estimates are provided on a stratum-specific basis. Steller sea lions and harbour seals are further differentiated by ‘hauled out’ and ‘in water’. This analysis updates previous estimates (Williams & Thomas 2007) by including additional years of effort, providing greater spatial precision with the DSM method over CDS, novel reporting for spring and autumn seasons (rather than summer alone), and providing new abundance estimates for Steller sea lion and northern elephant seal. In addition to providing a baseline of marine mammal abundance and distribution, against which future changes can be compared, this information offers the opportunity to assess the risks posed to marine mammals by existing and emerging threats, such as fisheries bycatch, ship strikes, and increased oil spill and ocean noise issues associated with increases of container ship and oil tanker traffic in British Columbia’s continental shelf waters.

Starting with marine animal observations at specific coordinates and times, I combine these data with environmental data, often satellite derived, to produce seascape predictions generalizable in space and time. These habitat-based models enable prediction of encounter rates and, in the case of density surface models, abundance that can then be applied to management scenarios. Specific human activities, OWED and shipping, are then compared within a tradeoff decision support framework, enabling interchangeable map and tradeoff plot views. These products make complex processes transparent for gaming conservation, industry and stakeholders towards optimal marine spatial management, fundamental to the tenets of marine spatial planning, ecosystem-based management and dynamic ocean management.

Bayesian Inference in Large-scale Problems

Many modern applications fall into the category of "large-scale" statistical problems, in which both the number of observations n and the number of features or parameters p may be large. Many existing methods focus on point estimation, despite the continued relevance of uncertainty quantification in the sciences, where the number of parameters to estimate often exceeds the sample size, despite huge increases in the value of n typically seen in many fields. Thus, the tendency in some areas of industry to dispense with traditional statistical analysis on the basis that "n=all" is of little relevance outside of certain narrow applications. The main result of the Big Data revolution in most fields has instead been to make computation much harder without reducing the importance of uncertainty quantification. Bayesian methods excel at uncertainty quantification, but often scale poorly relative to alternatives. This conflict between the statistical advantages of Bayesian procedures and their substantial computational disadvantages is perhaps the greatest challenge facing modern Bayesian statistics, and is the primary motivation for the work presented here.

Two general strategies for scaling Bayesian inference are considered. The first is the development of methods that lend themselves to faster computation, and the second is design and characterization of computational algorithms that scale better in n or p. In the first instance, the focus is on joint inference outside of the standard problem of multivariate continuous data that has been a major focus of previous theoretical work in this area. In the second area, we pursue strategies for improving the speed of Markov chain Monte Carlo algorithms, and characterizing their performance in large-scale settings. Throughout, the focus is on rigorous theoretical evaluation combined with empirical demonstrations of performance and concordance with the theory.

One topic we consider is modeling the joint distribution of multivariate categorical data, often summarized in a contingency table. Contingency table analysis routinely relies on log-linear models, with latent structure analysis providing a common alternative. Latent structure models lead to a reduced rank tensor factorization of the probability mass function for multivariate categorical data, while log-linear models achieve dimensionality reduction through sparsity. Little is known about the relationship between these notions of dimensionality reduction in the two paradigms. In Chapter 2, we derive several results relating the support of a log-linear model to nonnegative ranks of the associated probability tensor. Motivated by these findings, we propose a new collapsed Tucker class of tensor decompositions, which bridge existing PARAFAC and Tucker decompositions, providing a more flexible framework for parsimoniously characterizing multivariate categorical data. Taking a Bayesian approach to inference, we illustrate empirical advantages of the new decompositions.

Latent class models for the joint distribution of multivariate categorical, such as the PARAFAC decomposition, data play an important role in the analysis of population structure. In this context, the number of latent classes is interpreted as the number of genetically distinct subpopulations of an organism, an important factor in the analysis of evolutionary processes and conservation status. Existing methods focus on point estimates of the number of subpopulations, and lack robust uncertainty quantification. Moreover, whether the number of latent classes in these models is even an identified parameter is an open question. In Chapter 3, we show that when the model is properly specified, the correct number of subpopulations can be recovered almost surely. We then propose an alternative method for estimating the number of latent subpopulations that provides good quantification of uncertainty, and provide a simple procedure for verifying that the proposed method is consistent for the number of subpopulations. The performance of the model in estimating the number of subpopulations and other common population structure inference problems is assessed in simulations and a real data application.

In contingency table analysis, sparse data is frequently encountered for even modest numbers of variables, resulting in non-existence of maximum likelihood estimates. A common solution is to obtain regularized estimates of the parameters of a log-linear model. Bayesian methods provide a coherent approach to regularization, but are often computationally intensive. Conjugate priors ease computational demands, but the conjugate Diaconis--Ylvisaker priors for the parameters of log-linear models do not give rise to closed form credible regions, complicating posterior inference. In Chapter 4 we derive the optimal Gaussian approximation to the posterior for log-linear models with Diaconis--Ylvisaker priors, and provide convergence rate and finite-sample bounds for the Kullback-Leibler divergence between the exact posterior and the optimal Gaussian approximation. We demonstrate empirically in simulations and a real data application that the approximation is highly accurate, even in relatively small samples. The proposed approximation provides a computationally scalable and principled approach to regularized estimation and approximate Bayesian inference for log-linear models.

Another challenging and somewhat non-standard joint modeling problem is inference on tail dependence in stochastic processes. In applications where extreme dependence is of interest, data are almost always time-indexed. Existing methods for inference and modeling in this setting often cluster extreme events or choose window sizes with the goal of preserving temporal information. In Chapter 5, we propose an alternative paradigm for inference on tail dependence in stochastic processes with arbitrary temporal dependence structure in the extremes, based on the idea that the information on strength of tail dependence and the temporal structure in this dependence are both encoded in waiting times between exceedances of high thresholds. We construct a class of time-indexed stochastic processes with tail dependence obtained by endowing the support points in de Haan's spectral representation of max-stable processes with velocities and lifetimes. We extend Smith's model to these max-stable velocity processes and obtain the distribution of waiting times between extreme events at multiple locations. Motivated by this result, a new definition of tail dependence is proposed that is a function of the distribution of waiting times between threshold exceedances, and an inferential framework is constructed for estimating the strength of extremal dependence and quantifying uncertainty in this paradigm. The method is applied to climatological, financial, and electrophysiology data.

The remainder of this thesis focuses on posterior computation by Markov chain Monte Carlo. The Markov Chain Monte Carlo method is the dominant paradigm for posterior computation in Bayesian analysis. It has long been common to control computation time by making approximations to the Markov transition kernel. Comparatively little attention has been paid to convergence and estimation error in these approximating Markov Chains. In Chapter 6, we propose a framework for assessing when to use approximations in MCMC algorithms, and how much error in the transition kernel should be tolerated to obtain optimal estimation performance with respect to a specified loss function and computational budget. The results require only ergodicity of the exact kernel and control of the kernel approximation accuracy. The theoretical framework is applied to approximations based on random subsets of data, low-rank approximations of Gaussian processes, and a novel approximating Markov chain for discrete mixture models.

Data augmentation Gibbs samplers are arguably the most popular class of algorithm for approximately sampling from the posterior distribution for the parameters of generalized linear models. The truncated Normal and Polya-Gamma data augmentation samplers are standard examples for probit and logit links, respectively. Motivated by an important problem in quantitative advertising, in Chapter 7 we consider the application of these algorithms to modeling rare events. We show that when the sample size is large but the observed number of successes is small, these data augmentation samplers mix very slowly, with a spectral gap that converges to zero at a rate at least proportional to the reciprocal of the square root of the sample size up to a log factor. In simulation studies, moderate sample sizes result in high autocorrelations and small effective sample sizes. Similar empirical results are observed for related data augmentation samplers for multinomial logit and probit models. When applied to a real quantitative advertising dataset, the data augmentation samplers mix very poorly. Conversely, Hamiltonian Monte Carlo and a type of independence chain Metropolis algorithm show good mixing on the same dataset.

Inter-site and inter-scanner diffusion MRI data harmonization.

We propose a novel method to harmonize diffusion MRI data acquired from multiple sites and scanners, which is imperative for joint analysis of the data to significantly increase sample size and statistical power of neuroimaging studies. Our method incorporates the following main novelties: i) we take into account the scanner-dependent spatial variability of the diffusion signal in different parts of the brain; ii) our method is independent of compartmental modeling of diffusion (e.g., tensor, and intra/extra cellular compartments) and the acquired signal itself is corrected for scanner related differences; and iii) inter-subject variability as measured by the coefficient of variation is maintained at each site. We represent the signal in a basis of spherical harmonics and compute several rotation invariant spherical harmonic features to estimate a region and tissue specific linear mapping between the signal from different sites (and scanners). We validate our method on diffusion data acquired from seven different sites (including two GE, three Philips, and two Siemens scanners) on a group of age-matched healthy subjects. Since the extracted rotation invariant spherical harmonic features depend on the accuracy of the brain parcellation provided by Freesurfer, we propose a feature based refinement of the original parcellation such that it better characterizes the anatomy and provides robust linear mappings to harmonize the dMRI data. We demonstrate the efficacy of our method by statistically comparing diffusion measures such as fractional anisotropy, mean diffusivity and generalized fractional anisotropy across multiple sites before and after data harmonization. We also show results using tract-based spatial statistics before and after harmonization for independent validation of the proposed methodology. Our experimental results demonstrate that, for nearly identical acquisition protocol across sites, scanner-specific differences can be accurately removed using the proposed method.

Advanced Aerogel Composites for Oil Remediation and Recovery

Oil spills in marine environments often damage marine and coastal life if not remediated rapidly and efficiently. In spite of the strict enforcement of environmental legislations (i.e., Oil Pollution Act 1990) following the Exxon Valdez oil spill (June 1989; the second biggest oil spill in U.S. history), the Macondo well blowout disaster (April 2010) released 18 times more oil. Strikingly, the response methods used to contain and capture spilled oil after both accidents were nearly identical, note that more than two decades separate Exxon Valdez (1989) and Macondo well (2010) accidents.

The goal of this dissertation was to investigate new advanced materials (mechanically strong aerogel composite blankets-Cabot® Thermal Wrap™ (TW) and Aspen Aerogels® Spaceloft® (SL)), and their applications for oil capture and recovery to overcome the current material limitations in oil spill response methods. First, uptake of different solvents and oils were studied to answer the following question: do these blanket aerogel composites have competitive oil uptake compared to state-of-the-art oil sorbents (i.e., polyurethane foam-PUF)? In addition to their competitive mechanical strength (766, 380, 92 kPa for Spaceloft, Thermal Wrap, and PUF, respectively), our results showed that aerogel composites have three critical advantages over PUF: rapid (3-5 min.) and high (more than two times of PUF’s uptake) oil uptake, reusability (over 10 cycles), and oil recoverability (up to 60%) via mechanical extraction. Chemical-specific sorption experiments showed that the dominant uptake mechanism of aerogels is adsorption to the internal surface, with some contribution of absorption into the pore space.

Second, we investigated the potential environmental impacts (energy and chemical burdens) associated with manufacturing, use, and disposal of SL aerogel and PUF to remove the oil (i.e., 1 m3 oil) from a location (i.e., Macondo well). Different use (single and multiple use) and end of life (landfill, incinerator, and waste-to-energy) scenarios were assessed, and our results demonstrated that multiple use, and waste-to-energy choices minimize the energy and material use of SL aerogel. Nevertheless, using SL once and disposing via landfill still offers environmental and cost savings benefits relative to PUF, and so these benefits are preserved irrespective of the oil-spill-response operator choices.

To inform future aerogel manufacture, we investigated the different laboratory-scale aerogel fabrication technologies (rapid supercritical extraction (RSCE), CO2 supercritical extraction (CSCE), alcohol supercritical extraction (ASCE)). Our results from anticipatory LCA for laboratory-scaled aerogel fabrication demonstrated that RSCE method offers lower cumulative energy and ecotoxicity impacts compared to conventional aerogel fabrication methods (CSCE and ASCE).

The final objective of this study was to investigate different surface coating techniques to enhance oil recovery by modifying the existing aerogel surface chemistries to develop chemically responsive materials (switchable hydrophobicity in response to a CO2 stimulus). Our results showed that studied surface coating methods (drop casting, dip coating, and physical vapor deposition) were partially successful to modify surface with CO2 switchable chemical (tributylpentanamidine), likely because of the heterogeneous fiber structure of the aerogel blankets. A possible solution to these non-uniform coatings would be to include switchable chemical as a precursor during the gel preparation to chemically attach the switchable chemical to the pores of the aerogel.

Taken as a whole, the implications of this work are that mechanical deployment and recovery of aerogel composite blankets is a viable oil spill response strategy that can be deployed today. This will ultimately enable better oil uptake without the uptake of water, potential reuse of the collected oil, reduced material and energy burdens compared to competitive sorbents (e.g., PUF), and reduced occupational exposure to oiled sorbents. In addition, sorbent blankets and booms could be deployed in coastal and open-ocean settings, respectively, which was previously impossible.