15064 research outputs found
Sort by
Applying authoritative knowledge to better understand preparation for breastfeeding
Get PDFIntroduction: In this qualitative study, we employ the construct of authoritative knowledge to better understand how birthing people prepare for breastfeeding experiences postpartum. This construct has seldom been applied to the postpartum period, despite its application by reproductive anthropologists to pregnancy and childbirth experiences cross-culturally. Consistent with these applications, we define authoritative knowledge domains by the purveyors. We aimed to characterize the acquisition and valuation of information sources participants used to prepare for breastfeeding. Methods: Twenty-five participants were recruited from a hospital-based pregnancy study in Chicagoland, Illinois, USA to complete interviews between November 2020 and March 2021. Audio recorded interviews were coded using a priori themes and iterative code development. Codes were used to characterize information sources and the designation of three domains of authoritative knowledge: biomedical, social network, and lived experience. Results: All participants received information about breastfeeding from both biomedical and social network domains, with those with prior child rearing experiences also using the personal experience domain. Use of online resources like pregnancy tracking apps and social media platforms resulted in the domains of authoritative knowledge overlapping. Participants valued information from health care providers the most but found social network information was more accessible and fulfilled their desire for experiential information. Discussion: In this first application of authoritative knowledge within the context of infant feeding, participants consistently cited biomedical sources as the most accurate and important. However, they cited barriers to gaining this information such as the short duration of prenatal appointments and the challenge of completing prenatal education courses. Many participants sought evidence-based information about breastfeeding on apps, social media, and websites, however content and quality across platforms varies significantly. This may be an avenue to improve access to reliable and helpful breastfeeding information.</p
Infinite-Dimensional Inference and Learning via Optimal Transport
Get PDFThis thesis studies new frameworks and methods for infinite-dimensional statistical inference and learning problems through the lens of optimal transport. Optimal transport is a branch of mathematics concerning how to transport one probability distribution to another while minimizing certain costs, which, as this thesis shows, can provide principled guidance for sophisticated modern data science problems. There are two overarching themes in this thesis: (1) the dynamic viewpoint of optimal transport that concerns the optimal path of transporting one probability distribution to another, and (2) quadratic extensions of the classical linear optimal transport problem. This thesis effectively utilizes these themes to delve into central problems in modern data science, including hypothesis testing, online learning, simulation-based inference, and missing value imputation. The results of this thesis contribute to the growing body of literature on optimal transport and its applications in statistics and machine learning, providing new insights and tools for researchers and practitioners in these fields
Fueling Change: What Makes U.S. Universities Prime for Successful Student-Led Fossil Fuel Divestment?
Get PDFIn response to the ongoing climate crisis, a new generation of college students is joining the global environmental movement. By creating fossil fuel divestment campaigns (FFDCs) on university campuses, students pressure their institution to divest funds from the fossil fuel industry. Due to the novelty of FFDCs, there is limited research on institutional decision-making aspects of the movement. This thesis will reframe the FFD movement quantitatively by examining institutional characteristics that can shape whether these campaigns succeed. Drawing on the Global Fossil Fuel Divestment Commitments Database (GFFDCD) and the National Center for Education Statistics’ Integrated Postsecondary Education Data System (IPEDS), I examine a selection of variables to address the question: What characteristics of U.S. universities, if any, predict successful student-led fossil fuel divestment campaigns? Overall, the statistical analysis revealed that: (1) Public universities are more likely to divest than private ones; (2) Universities in the Northeast and West are more likely to divest than Midwest institutions, with institutions located in Southern or suburban areas being least likely to divest from fossil fuels; (3) Universities with a higher percentage of African American/Black students are less likely to divest from fossil fuels, whereas a higher percentage of Asian students or students with two or more races is associated with a greater likelihood of divestment; and (4) More selective schools (those with lower acceptance rates) are more likely to divest. Future researchers should examine FFDCs qualitatively to better understand the unique circumstances that each campaign faces in their pursuit of divestment
For Love and Money: Relational Work, Precarity, and Show Business
No full textThis study investigates how career hair and makeup artists (HMUs) in the elite film, television, theater, and opera gig economy sustain middle-class livelihoods amid a precarious labor market. Existing scholarship on creative labor often frames workers as either entrepreneurial members of the “creative class” or bohemian artists complicit in their own exploitation, a dichotomy that mirrors debate over whether nonstandard employment offers entrepreneurial opportunity or precarity. Drawing on Viviana Zelizer’s relational work framework, I analyze how HMUs leverage social and institutional relationships to nurture creative collaboration, stabilize wages, and secure employment benefits uncommon in gig work, expecting intrinsic artistic rewards and economic returns for their labor exchange. Data come from twenty interviews with veteran unionized HMUs, autoethnography, and archival analysis, comparing routine industry jobless cycles and prolonged unemployment during the pandemic and the 2023 Hollywood strikes. Findings reveal that peer friendships are a key site of labor market exchange, where gig information and respect are currency, making relational and emotional labor vital to securing work. Success depends on both technical skill and the ability to curate a diverse portfolio of “scripted” (e.g., union contracts or long-term work friendships) and “unscripted” gigs (e.g., nonunion gigs or new work relationships), with the latter involving more negotiation and greater financial and emotional uncertainty. HMUs may end work relationships that fail to balance respect, trust, and artistic and economic reward, challenging portrayals of arts workers as passive in the face of exploitation. HMUs often strategically outsource relational work with producers to unions and labor law, brokering standard employment relationships, wages, and employer-earmarked benefits, mitigating precarity. However, recent boom-and-bust cycles have battered HMUs’ financial security and mental health, with no lateral career exit available, producing a form of “golden handcuffs” in formalized precarious work wherein the increasing lack of steady, reliable work offsets creative and economic rewards. This study adds to the literature on creative, precarious, and relational labor by describing middle-class precarity in show business and exploring the role of unions and legal brokerage of employment relationships in gig work
Old Bones in New Databases: Historical Insights Into Race, Statistics, and Ancestry Estimation in Anthropology
Get PDFThis article explores the persistence of race in biological anthropology, particularly in the context of ancestry estimation using the Fordisc software. Despite efforts to move away from race-based typologies since the mid-20th century, historical notions of race continue to shape scientific methods and technologies in anthropology. By tracing the “data journey” of a skeletal collection within Fordisc's database, we reveal how early 20th-century race science shaped statistical methods used in contemporary anthropology and how typological notions of race persist today. Our interdisciplinary approach, combining history of science and science and technology studies, highlights the need to historicize and critically examine the methods and technologies that underpin anthropological practices. This analysis demonstrates that issues of race in science are deeply rooted in the material practices of data collection, analysis, and statistical methods. Recognizing and dismantling these legacies is central to creating more ethical scientific practices. We argue that addressing the trouble with race in anthropology requires a comprehensive reevaluation of scientific practices, its methods and technologies, and would benefit from interdisciplinary collaboration within anthropology and beyond. Este artículo explora la persistencia de raza en la antropología biológica, particularmente en el contexto de estimación de linaje usando el software Fordisc. A pesar de los esfuerzos para distanciarse de las tipologías basadas en raza desde mediados del siglo XX, las nociones históricas de raza continúan estructurando los métodos científicos y las tecnologías de la antropología. Mediante el rastreo del “viaje de datos” de una colección esqueletal dentro de la base de datos Fordisc, revelamos cómo la ciencia de la raza a principios del siglo XX estructuró los métodos estadísticos en la antropología contemporánea y cómo las nociones tipológicas de raza persisten hoy en día. Nuestra aproximación interdisciplinaria, combinando la historia de la ciencia y los estudios de la ciencia y la tecnología, enfatiza la necesidad de historizar y examinar críticamente los métodos y las tecnologías que apoyan las prácticas antropológicas. Este análisis demuestra que cuestiones de raza en la ciencia están enraizadas profundamente en las prácticas materiales de la recolección de datos, el análisis y los métodos estadísticos. Reconocer y desmantelar estos legados es central para crear prácticas científicas más éticas. Argumentamos que abordar el problema con raza en la antropología requiere una reevaluación exhaustiva de las prácticas científicas, sus métodos y tecnologías, y se beneficiaría de la colaboración interdisciplinaria dentro de la antropología y más allá.</p
News to Numbers: A Comparative Analysis of Traditional and API-driven LLMs in Stock Return Prediction
No full textThis thesis investigates the comparative effectiveness of two large language model (LLM) deployment paradigms—zero-shot inference via commercial APIs and supervised fine-tuning of transformer-based models—for stock return prediction using financial news. Leveraging a novel dataset of over 30,000 Dow Jones Newswire (DJN) articles aligned with firm-level returns from 1989 to 2020, the study systematically evaluates OpenAI’s GPT-4 in a zero-shot setting against fine-tuned BERT-based models trained on identical text–return pairs. The analysis employs a rolling-window cross-validation framework to ensure robustness across market regimes and emphasizes not only predictive accuracy but also operational feasibility, cost efficiency, and scalability. Results demonstrate that both modeling approaches produce statistically significant predictive power above random baselines, yet with distinct trade-offs. GPT-4 achieves 53.38% directional accuracy and a Sharpe ratio of 1.076, highlighting its ability to extract meaningful signals without domain-specific tuning. Its operational simplicity, low setup cost, and per-call flexibility make it particularly attractive for rapid deployment and lightweight financial applications. By contrast, the supervised BERT model delivers markedly stronger financial performance, attaining an annualized Sharpe ratio of 4.08 and mean daily returns of 12.81 basis points. These superior outcomes underscore the value of domain-specific adaptation, but they come at the cost of substantial GPU resources, specialized expertise, and ongoing maintenance requirements. Temporal analysis further reveals that both approaches maintain predictive robustness across diverse economic conditions—including the dot-com era, the 2008 financial crisis, and post-crisis market environments—though performance metrics vary with shifts in efficiency and volatility. Importantly, GPT-4 provides consistent positive alphas despite its modest accuracy, while fine-tuned BERT models extract higher-magnitude signals during volatile periods, reinforcing the economic relevance of tailored architectures. The findings contribute to the growing literature at the intersection of financial economics and natural language processing by offering the first unified empirical framework to compare API-driven and locally fine-tuned LLMs for financial prediction. They highlight a fundamental trade-off in institutional strategy: API-based models democratize access to advanced language capabilities with minimal infrastructure, while fine-tuned models yield superior risk-adjusted returns for institutions capable of sustaining the computational investment. This research advances both theory and practice by demonstrating that textual data continues to contain exploitable information for asset pricing and that the operational context—budget, infrastructure, and regulatory considerations—should guide the choice of modeling paradigm. Ultimately, the results suggest that financial institutions must navigate between accessibility and performance, as the integration of LLMs into quantitative trading evolves from experimental adoption toward strategic deployment
Development of an alcohol biosensor non-wear algorithm: Laboratory-based machine learning and field-based deployment
Get PDFWrist-worn alcohol biosensors can continuously track alcohol consumption, but their measurements are disrupted when the device is removed. Left unaddressed, non-wear data compromises observations of alcohol use and subsequent predictions of intoxication. To advance beyond commonly used temperature cutoffs and enable more precise detection of non-wear, we trained a random forest algorithm using laboratory ground truth data. Participants in Study One (N = 36) wore a wrist-worn alcohol biosensor (BACtrack Skyn) across 61 five-hour laboratory sessions, generating ground truth non-wear by removing and re-applying the device at specified times. Algorithm features included temperature, motion, and their time-series quadratic coefficients. According to device-based cross-validation, the algorithm performed with excellent sensitivity to detect non-wear (0.96) and specificity to confirm wear (0.99), out-performing all univariable temperature cutoffs from 25 to 30 °C. The algorithm was then used to evaluate biosensor adherence in Study Two, a four-week field study where participants (N = 114) wore the Skyn and self-reported non-wear intervals each day. The algorithm detected 1.6 h of daily non-wear per participant and had more agreement with self-report compared with the temperature cutoff method. This non-wear algorithm can assess biosensor adherence in field studies and may also facilitate precise data imputation, resulting in more objective models of alcohol-related outcomes
Regulatory Uncertainty Pricing in Digital Economy
Get PDFIn Chapter 1, I quantitatively assess the economic implications of two regulatory paradigms in the digital economy: data privacy laws and command-and-control regulations. To this end, I develop a production-based equilibrium model that (i) microfounds firms' technology adoption decisions, (ii) incorporates ``data emissions'' as negative externalities of excessive data collection and data sharing arising from the non-rival nature of digital capital, and (iii) accounts for potential model misspecifications introduced by regulatory changes. The model implies a decomposition of the risk price associated with increasing market concentration, driven by digital capital accumulation, into two components: short-term firm-level productivity gains from adopting data-driven technologies and long-term social costs stemming from data emissions. This theoretical implication aligns with empirical evidence showing that the corresponding equity risk premia in the US have turned negative since the early 2000s, coinciding with the rapid growth of the data-trading market and data-driven technologies over the past 20 years. The model further predicts that firms adopting data-driven technologies exhibit stock returns that co-move more with market concentration growth, resembling the return profiles of growth firms. Using a calibrated model informed by financial market data, I demonstrate that the marginal social cost of data emissions decreases as technology adoption scales and can be further mitigated by increases in total factor productivity or intensity of innovation. Finally, counterfactual analysis suggests that the most effective regulatory paradigm combines data privacy laws with command-and-control regulations. This hybrid paradigm, when enforced through protocols that reduce uncertainty in data emissions while embracing uncertainty in innovation dynamics, can enhance social welfare. Chapter 2 presents a novel Bayesian approach that incorporates financial frictions into a panel structural break model, utilizing economically informed priors from intermediary asset pricing theories. The data-driven prior selection method, adept at handling unbalanced panels, enhances the identification of regime shifts and the selection of return predictors, thus improving equity return forecasts. Validated through simulations and empirical analysis, this approach boosts out-of-sample cumulative returns and Sharpe ratios. Leveraging asset holdings data and intermediary-induced priors, the approach facilitates real-time regime change detection and provides Bayesian insights into the inconsistencies of risk prices associated with intermediary risks
Geometry of Expansion: Scrolls, Folds, and Concentric Shapes in Contemporary American and Chinese Long Poems
Get PDFGeometry of Expansion: Scrolls, Folds, and Concentric Shapes in Contemporary American and Chinese Long Poems treats recursive geometry as an analytic framework to study material, thematic, rhythmic, and epistemological innovations in recent long poems. Each chapter compares a book-length American poem with a Chinese counterpart, with the poems selected for their shared formal designs––the unwinding scroll, the self-embedding and self-duplicating folds, and the concentric squares and circles. The interest in recursive geometry, emerging in parallel in these two disparate literary traditions, situates the long poem within the tension between epic and lyric poetry, reconfigures many classic poetic concepts, and transforms the genre of the long poem into a tentative, piecemeal, yet committed engagement with contemporary reality. I further propose that the mathematical interpretation of these recursive shapes lends us critical vocabularies and insights, with which we can describe the poiesis of recent experimental works and conceive a hermeneutics not only for these specific poems but also for contemporary experience in general. The introduction traces the evolution of the modern long poem into a hybrid, porous, and attractive genre epitomizing the interplay between epic and lyric, formal constraints and artistic ingenuity, as well as nationalism and globalization. Drawing from the interdisciplinary applications of recursivity, I discuss why scrolls, folds, and concentric shapes appeal to many poets as the structure to uncover and overcome materialism, commercialism, alienation, and political repression. The first chapter examines two poems written on scrolls—Tape for the Turn of the Year (1965) by A. R. Ammons and Following Huang Gongwang in Touring the Fuchun Mountains (2015) by Zhai Yongming. I discuss how the scroll’s kinematic features enable the poets to transform an attention economy characterized by stimulations and incessant distractions into a sustainable commitment. The second chapter focuses on two elegies––The Midnight (2003) by Susan Howe and An Ideal Couple in the Water-Paint Garden: Mao Pijiang and Dong Xiaowan 1642––1651 (2019) by Bai Hua. Their rapid shifts in forms and themes reveal bifurcating, layered, and fractal patterns, which diffuse lyric sources and redemptions in everyday life. The last chapter concerns two family romances written in concentric shapes––Lyn Hejinian’s My Life (1980) and Yang Lian’s Narrative Poems (2011). I argue that these poems organize familial and cultural memories centripetally and centrifugally to envision how a decentered poetic subjectivity is still capable of social effects
Controlling Administrative Power Under Systems of Separation of Powers
Get PDFDeveloping a theory of the expansion of executive control over the administrative state under systems of separation of powers