DIGITAL ACCESS TO

This dissertation uses large-scale administrative data and quasi-experimental methods to examine three interrelated dimensions of public policy and K–12 education finance: the fiscal effects of policy changes, families’ and students’ behavioral responses, and the academic and labor market consequences of these policies.

Chapter 1 examines how increases in school spending influence families’ decisions between public and private education. Leveraging court-mandated school finance reforms (SFRs) as exogenous shocks to district funding, I estimate how changes in funding and spending affect public school enrollment and racial segregation. Consistent with prior research, I find that reforms substantially increased spending in low-income school districts and that these gains persisted over time. Overall, the additional resources did not lead to broad shifts toward public schooling, suggesting that higher spending alone is insufficient to draw families into the public sector. Only in low-income suburban districts do I find evidence that increased spending modestly raised the share of students attending traditional public schools. However, low-income districts experienced increases in the school-age population in the medium and long run following reform implementation, even as the share of students attending traditional public schools remained unchanged.

Chapter 2 addresses a long-standing question in social science research and education policy: “Does school spending matter?” While recent causal studies document positive effects of increased school spending on student outcomes, context-specific evidence remains important to policymakers. This study exploits discontinuities in the state aid formula introduced by the 1993 Massachusetts Education Reform Act to estimate the effects of increased school funding on academic and labor market outcomes. I replicate prior work and show that state aid increases fourth- and eighth-grade achievement on state standardized tests. I then extend this analysis to examine long-run academic and labor market outcomes, including high school graduation, college enrollment, and early-career earnings, using cohort-level data from the Massachusetts Longitudinal Data System. In contrast to the short-run achievement results, the estimated impacts on long-run outcomes are small and statistically insignificant.

Chapter 3 examines the impact of public subsidies for private development projects on K–12 school finance. A large empirical literature finds little evidence that publicly financed professional sports facilities generate broad-based economic growth. Yet far less is known about the opportunity cost of these subsidies for public services. I focus on publicly financed stadiums and arenas for professional sports franchises in the NFL, NBA, MLB, and NHL and link existing data on facility financing to district-level school finance records to construct a novel panel dataset. Using quasi-experimental variation in the timing of subsidy approval, I show these subsidies generate a two-stage reduction resources for K-12 education. First, causing an immediate decline in state funding followed by reductions in revenues at the local level.

Metastatic dissemination is responsible for a vast majority of cancer-related deaths, yet the biological mechanisms that enable tumor cells to survive during early metastatic spread remain incompletely understood. Lymph nodes represent a common early site of metastasis in breast cancer, among other cancer types (e.g., melanoma, pancreatic cancers) and provide a distinct metabolic microenvironment that may influence tumor cell survival and metastatic fitness. Emerging evidence suggests that ferroptosis, a regulated form of cell death driven by iron-dependent lipid peroxidation, acts as a critical barrier to metastatic progression in the bloodstream, but not lymphatic metastasis. However, the extent to which the metabolic microenvironment of the lymph node microenvironment influences tumor cell survival remains poorly defined.

This dissertation investigates how the lymph node metabolic environment influences breast cancer survival. First, using murine and human breast cancer models, including spontaneous metastasis models and lymph node-derived tumor cell isolations, this work characterizes the metabolic features of the lymph node microenvironment that promotes breast cancer cell survival. To experimentally model lymph node conditions ex vivo, we designed LyMe (lymph-like medium), a physiologically informed culture system that recapitulates the key metabolic features of the lymph node microenvironment. LyMe enabled the interrogation of metabolic dependencies that influence cancer cell survival in in vitro cell culture contexts, thus facilitating an opportunity to more closely model the metabolic features of the lymph node microenvironment in standard cell culture.

To assess the extent to which metabolic features observed in our in vitro and in vivo experimental models are reflected in human disease, a clinical study of human breast cancer fine-needle aspiration biopsy samples was initiated and completed as part of this dissertation work. Metabolomic and lipidomic profiling of lymph node-positive and lymph node-negative patient plasma and lymph node aspirates revealed distinct metabolic signatures associated with lymph node metastasis. These analyses identified enrichment of pathways related to lipid metabolism, purine metabolism, and immune-associated scavenger receptor signaling, including involvement of scavenger class F receptors. Together, these findings support the translational relevance of the experimental systems and suggest that lymph node metastases undergo coordinated metabolic adaptations that may promote tumor cell survival during metastatic progression.

Furthermore, through targeted metabolic perturbation experiments, we demonstrate that specific nutrient pools contribute differentially to tumor cell survival in lymph-like conditions. In particular, cystine availability emerged as a critical determinant of tumor cell survivability, consistent with its central role in regulating ferroptosis through glutathione metabolism and redox homeostasis. These findings expose the metabolic features of lymph node microenvironments that influence breast cancer survival.

Finally, complementary collaborative work examined metabolic regulation of ferroptosis in metastatic cancer, including the role of lysosomal iron handling and dietary lipid metabolism in shaping ferroptotic susceptibility. Collectively, these studies highlight ferroptosis and lipid peroxidation as metabolically regulated vulnerabilities during metastatic progression. Overall, this dissertation demonstrates that microenvironmental metabolic conditions within lymph nodes can influence tumor cell survival and ferroptosis susceptibility and introduces a metabolic media matching lymph conditions as a platform to study cancer biology of cancer cells residing in lymph nodes under physiologically relevant conditions in vitro. Together, these findings provide a foundation for understanding how tissue-specific metabolic environments support metastatic adaptation and reveal metabolic vulnerabilities that may be therapeutically leveraged to limit cancer cell survival within lymph nodes.

Empirical analyses of insurance, care utilization, and health outcomes are often constrained by fragmented health administrative data and limited follow-up in national surveys, yielding only a partial view of how these outcomes may change over longer time horizons. In this dissertation, I develop microsimulation modeling approaches to estimate the cumulative consequences of the fragmented US health system across the full course of childhood. Chapter 1 examines health insurance coverage; the remaining chapters expand this focus to routine care utilization (Chapter 2) and dental caries outcomes (Chapter 3), highlighting the implications of insurance fragmentation and the medical-dental divide. Chapter 1 estimates the cumulative health insurance experiences over childhood under post-Affordable Care Act (ACA), pre-pandemic policy conditions. The insurance landscape in the United States is fragmented across public and private options with substantial state variation. Using natality records and longitudinal national surveys including the Survey of Income and Program Participation (SIPP) and the Medical Expenditure Panel Survey (MEPS), I create a nationally representative synthetic cohort and develop nonparametric matching algorithms to simulate individual-level trajectories from birth until 18th birthday across five health insurance types: 1) Medicaid or Children’s Health Insurance Program (CHIP); 2) uninsured; 3) employment-based private; 4) Marketplace; and 5) other. Under the assumed policy conditions, I estimate that 61% of US children are ever enrolled in Medicaid or CHIP and 42% are ever uninsured by age 18. Among children born in Medicaid or CHIP, 36% experience any insurance loss in Medicaid adult expansion states compared with 59% in non-expansion states. These findings demonstrate the broad reach of Medicaid and CHIP, the common experience of coverage gaps, and substantial state-level variation by Medicaid policy restrictiveness over childhood. Chapter 2 investigates joint patterns of pediatric routine care utilization across primary care and dental settings and quantifies the impact on care continuity in two medical-dental integration scenarios under perfect implementation: 1) delivering dental services in primary care; 2) delivering primary care services in dental care. Using longitudinal MEPS data, I develop nonparametric matching algorithms to simulate individual-level trajectories of dental insurance coverage, primary care well-child visits, and routine dental visits for the synthetic cohort described in Chapter 1. Relative to recommended routine care schedules in each setting, I estimate that US children accumulate a mean of 11 years of unmet dental needs and 6 years of unmet primary care needs over childhood. Under perfect implementation, medical-dental integration could improve continuity of services in all insurance groups but does not fully eliminate unmet needs. Furthermore, the greatest gains accrue to insurance groups with the lowest unmet needs. Chapter 3 estimates cumulative dental caries outcomes and examines disparities by childhood insurance experience. As children’s dentition evolves through primary tooth eruption, shedding, and permanent tooth emergence, tooth surfaces become susceptible to dental caries, the most common chronic disease of childhood in the United States. Leveraging longitudinal electronic dental records and cross-sectional National Health and Nutrition Examination Survey (NHANES) data, I develop, calibrate, and validate a dental caries microsimulation model that incorporates dentition changes, insurance dynamics, and dental utilization (routine and symptom-driven) for the synthetic cohort described in Chapters 1 and 2. I estimate that 72% of US children develop any caries (primary or permanent) by their 18th birthday, with an average of 5.5 affected tooth surfaces. Children with high health insurance gaps (>10% of childhood) and those insured with >75% Medicaid/CHIP experience the highest cumulative burden of caries, whereas those insured with >75% employment-based health & dental coverage have the lowest. Although primary caries is more common, disparities by childhood insurance experience are more pronounced for permanent caries.

UnBecoming: The Poetics of Rupture in Visions of Black Girlhood examines twentieth and twenty first century American art and visual culture—specifically the work of artists Faith Ringgold, Deborah Roberts, and Clarissa Sligh—to reveal how discourses of race, gender, and vision shape perceptions of Black girls and their civic sovereignty. This dissertation proposes a study of rupture as a method of constructing meaning in art history to center the critical and elusive space of raced childhood. I contend that as these artists foregrounded race and childhood, they evince Black girlhood as a subject that is central to questions of nationhood, citizenship, and ultimately what it means to be human.

For this project, I developed the term “poetics of rupture” to name and interrogate the artists’ use of collage, constructed photography, and quilting to disrupt and effectively reorder the visual narratives surrounding childhood. Here poetics, or poesis, describes the process by which something is conjured or brought into existence. My consideration of rupture builds upon Fred Moten’s theorization of the break as a site of “radical breakdown” that generates new possibilities. Together, the poetics of rupture captures the productive practice of unmaking, the breaking open that allows for the emergence of something new.

In the introduction, I chart the emergence of Black girls onto the public stage through the visual economy of the civil rights. The body of the dissertation moves across three artist-specific case studies. Chapter One, “Sutured Visions: The Aesthetics of Redress in the Work of Faith Ringgold,” examines Faith Ringgold’s paintings and story quilts as a Black feminist postmodern practice of aesthetic redress that centers the Black girl child. Chapter Two, “Remnants and Returns: Intimate Fragments of Girlhood, turns to Clarissa Sligh, arguing that her constructed photographs and artist’s books mobilize fragmentation, annotation, and archival return to challenge the narratives that secure childhood innocence and family coherence. Chapter Three, “Refiguring Citizenship: Deborah Roberts’s Countervisual Archive of Black Girlhood,” argues that Deborah Roberts’s collaged figurations, text based works, and installations construct a countervisual archive that fractures dominant regimes of beauty, vision, and citizenship, repositioning Black girls at the center of American civic imagination. I conclude with a consideration Freedom Square: The Black Girlhood Altar (2023-2024).

UnBecoming bridges art history, Black studies, and feminist studies to further scholarship on vision as a racialized regime of domination that Black women and girls subvert through art making. The subjects of this project put forth ways of seeing and being that contest the limits of citizenship and national belonging—a discursive move necessary for the advancement of art history. UnBecoming aims to show how art history and visual culture can better register acts of repossession—often of one’s image or narrative—to expand discourse around the visual rhetoric of selfhood and belonging in the United States.

Many scientific problems require inference at a finer scale than the scale at which data are observed. In public health, ecology, survey statistics, policy analysis, and graph-based risk modeling, responses are often available only as aggregate counts, rates, margins, or group-level labels, while the scientific target is a latent fine-scale field. This dissertation studies such problems as aggregate-supervised inverse inference: inference about fine-scale structure from observations that have passed through a known aggregation mechanism.

The central difficulty is that aggregation is many-to-one. A model may fit all observed aggregates while assigning risk, mass, labels, or intensity to the wrong fine-scale units. Fine-scale inference therefore requires more than aggregate prediction. It requires an explicit forward operator, scientifically meaningful structure, uncertainty representations that acknowledge non-identifiability, and evaluation at the scale of the latent target.

The dissertation develops this perspective through mathematical foundations, a synthesis of ecological inference, small-area estimation, weak supervision, Bayesian inverse problems, graph learning, and generative modeling, and two original methodological developments. The first, DisDiffEM, couples aggregate observations with graph-conditioned diffusion-style denoising in an approximate EM framework. The second, DiffRes, introduces a residual latent-variable formulation for graph-indexed aggregate count data. DiffRes separates predictable graph- and covariate-driven structure from lower-dimensional residual ambiguity, concentrating probabilistic and generative modeling on the part of the latent signal not already explained by structured prediction.

Empirical benchmarks distinguish aggregate agreement, fine-scale recovery, residual recovery, and uncertainty behavior. The results show that aggregate fit alone can be misleading, that the usefulness of expressive priors depends on residual dimension and sample size, and that inference-time refinement helps only when the prior and aggregate likelihood are sufficiently aligned. Overall, the dissertation argues that aggregates can support fine-scale scientific inference, but only when the model is explicit about what the data observe, what the assumptions supply, and what uncertainty remains unresolved.