33760 research outputs found
Sort by
Mind the Gap: Institutional Capacity, Globalization, and the Discrepancy between Criminal Justice and Health Homicide Data
No full textWe analyzed a maximum of 1,838 country-years for 99 countries from 1991 to 2020 where there was no missing data for any of the covariates. The data set makes up an unbalanced panel covering the years 1991 (for globalization) or 2000 (for capacity of public institutions and medical systems) to 2020. By pooling the cross-sectional data for a maximum of 99 countries covering nearly three decades, we have a diverse and reasonably large dataset for analysis.
We operationalize the homicide gap as:
〖homicide gap〗_it=ln|〖WHO count〗_it/Population_it -〖UNODC count〗_it/Population_it +0.001|*100 (1)
where i denotes the country and t denotes the year. Our measure of the homicide gap is the natural logarithm of the absolute difference between the WHO and the UNODC homicide rate for country i at time t. We add 0.001 to avoid taking the log of zero and multiply the homicide gap by 100 to facilitate interpretation in the subsequent regression models. Because the absolute homicide gap exhibited a high level of skewness and kurtosis (13.16 and 207.06, respectively), we perform a natural log transformation. The Institutional Capacity Index (ICI) is the average of: 1) control of corruption; 2) voice and accountability; 3) rule of law; 4) government effectiveness; 5) regulatory quality; 6) political stability: and 7) an UNODC data quality recalculated score.
To create the Health Capacity Index (HCI), we follow the Pan American Health Organization (2018) by combining positive health indicators (health spending per capita, life expectancy in years, and percent of one-year-old children with Hepatitis B immunization) with negative health indicators (cases of tuberculosis per 100,000 people, percent HIV among the population ages 15-49, death rate per 1,000 people, maternal mortality per 100,000 live births, and infant deaths per 1,000 live births). Our globalization measure is an index that includes 43 indicators from the KOF Swiss Economic Institute (Dreher, 2006; Gygli et al., 2019). The most and least globalized countries on this index are Switzerland (87.27) and Tajikistan (38.99), respectively.The homicide rate is a key measure of crime that is particularly valuable for comparative research. In most countries, homicides are recorded both as deaths by public health systems and as crimes by criminal justice systems, enabling a comparison of counts across sources. However, oftentimes there are considerable discrepancies in the data produced by these two systems, undermining the credibility of homicide research. Rather than treat the disparity between the sources as only a methodological problem, we treat the “homicide gap” as an outcome to be explained. Drawing on new institutionalism theory, we argue that homicide data from the two sources will be more similar when country-level public institutions and medical systems have greater capacity and when countries are more highly globalized. Moreover, we expect the effect of globalization to be greater in countries with stronger institutions. This study investigates this “homicide gap” and explores the Interplay between local institutional capacities and global pressures in shaping homicide data. Using Bayesian quantile models, we We analyze homicide data from the UNODC and the WHO across 99 countries from 1991 to 2020. Using Bayesian quantile regression models, we test hypotheses derived from new institutionalism theory. Key predictors are measures of institutional capacity, health system robustness, and globalization. We control for economic and demographic variables found by prior research to be related to homicide rates. Our results show that compared to other countries, those with stronger institutional capacity, and more robust health systems, and greater globalization exhibit significantly smaller homicide gaps. Globalization narrows the gap and has stronger effects compared to health capacity. We also find that countries with stronger local institutions experience a larger effect of globalization on the homicide gap. Sensitivity analysis confirms the robustness of these results across varying model specifications. The homicide gap reflects complex interactions between institutional dynamics and global influences. Reducing the homicide gap Addressing it requires investment in institutional capacity and sustained international cooperation. Future studies should investigate the quality of other policy-relevant crime data and explore mechanisms for enhancing data quality in low-capacity, low globalization countries.unfunded research projec
Mixed Modeling Approaches for Characterizing Genetic Effects and Heritability Metrics in Longitudinal Phenotypes
No full textThis dissertation develops advanced mixed modeling approaches by integrating genetic and subject-specific random effects to quantify genetic effects and heritability metrics on both the baseline levels and rates of change in longitudinal phenotypic trajectories. By disentangling these joint genetic effects, this work provides profound insights into both static and dynamic genetic influences on longitudinal phenotypes.
The first project introduces a mixed modeling framework to predict the subject-level genetic effects on both the baseline levels and slopes of longitudinal phenotypes. The inclusion of joint genetic effects, coupled with the crossed structure of genetic and subject-specific random effects, results in complex dependencies across repeated measurements. These complexities necessitate the development of innovative procedures for parameter estimation and prediction of joint genetic effects. To tackle these challenges, an Average Information Restricted Maximum Likelihood (AI-REML) algorithm is employed to estimate the variance components associated with genetic and subject-specific random effects for both the baseline levels and rates of change in longitudinal phenotypes. Theoretical inferences and comprehensive simulation studies validate the robustness and efficacy of the proposed method.
Of note, before estimating variance components using AI-REML algorithm and predicting joint genetic effects, a preliminary step involves identifying statistically significant genetic variants associated with the considered longitudinal phenotype to achieve high-dimensional reduction. The goal is to predict the individual-specific genetic effects on the trajectory of a longitudinal phenotype. The resulting predictions can then be used to stratify participants and provide probabilities of false detection across ages, minimizing unnecessary further diagnoses.
The second project extends the framework by incorporating genome-wide variants (millions of variants) simultaneously to estimate heritability metrics for both baseline trait levels and rates of change in longitudinal trajectories. This effort addresses key challenges, including the computational demands of the potential for large-scale studies, the complexity of high-dimensional genetic data, the interaction between joint subject-level genetic effects, as well as the crossed structure of genotypic and subject-specific random effects. To deal with these challenges, an AI-REML approach optimized for moderate-size studies is employed. For large-scale studies, where the covariance matrix inversion becomes computationally infeasible, a partitioned AI-REML approach is naturally proposed, reducing computational burden by dividing the whole sample into evenly non-overlapping subsamples. This approach sacrifices efficiency due to the loss of non-diagonal covariance information. Further, meta-analysis is conducted to derive more accurate estimates for variance components and two heritability metrics. Alternatively, a restricted Haseman-Elston (REHE) regression method is utilized, offering computational feasibility without the need for partitioning while effectively estimating variance components, especially in the context of large-scale data. Extensive simulation experiments are conducted to compare and evaluate the performance of these two methods across various scenarios, providing insights into their relative strengths and applicability.
The third project further addresses the challenge of high variance in slope-related estimates observed in large-scale studies with a limited number of observations per subject, which complicates the performance of existing methods such as the AI-REML algorithm and the REHE regression method, resulting in less reliable outcomes. To overcome these limitations, we then propose a two-stage estimation method specifically designed for large-scale studies with sparse longitudinal data. This approach aims to enhance the precision and reliability of estimates for the ratio of genetic contributions to the rate of change in longitudinal trajectories. In the first stage, linear regression is performed for each subject to estimate fixed-effect coefficients and their variances, using an unbiased estimator for error variance. In the second stage, linear mixed models are constructed, treating the estimated fixed-effect coefficients as responses and incorporating their variances as observed measurement errors. The AI-REML algorithm is then applied to estimate the ratios of genetic contributions. Simulation studies are conducted to compare and evaluate the performance of these three methods, demonstrating the robustness and effectiveness of the proposed two-stage estimation approach.
Collectively, these projects enhance the understanding and quantification of joint genetic effects and heritability metrics for both baseline levels and rates of change in longitudinal data analysis. The proposed methodologies and guidelines offer valuable tools for researchers addressing the challenges posed by large-scale studies and high-dimensional analyses, broadening the applicability of existing techniques to more complex scenarios. The findings from this dissertation improve the accuracy and reliability of statistical analyses for phenotypic traits influenced by genetic effects in the context of unbalanced serial measurements.
In the applications, this dissertation demonstrates the utility of the proposed methods for analyzing 6,948,674 genome-wide common variants to study the dynamics of prostate-specific antigen (PSA) trajectories in European white males from the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial. For the first project, we firstly identify 253 genetic variants associated with longitudinal PSA measurements, enabling a substantial reduction in the high-dimensional genetic dataset. Using these selected variants, the AI-REML algorithm was employed to estimate genetic and subject-specific variance components for both baseline levels and rates of change. The results underscore significant genetic contributions to baseline PSA levels and their progression over time, providing insights into the genetic factors influencing PSA variability among unaffected individuals. These findings have significant implications for identifying individuals at higher risk of false-positive prostate cancer screening results when relying on established PSA cutoffs. By incorporating joint genetic factors into PSA monitoring, this work highlights the potential to improve the accuracy and effectiveness of early prostate cancer detection and the development of personalized PSA screening guidelines. In the second project, the analysis revealed moderate genetic contributions to baseline PSA levels but significant genetic contributions to PSA velocity, highlighting an increasing heritability trend with age. Taken together, the methodologies developed in the dissertation provides researchers the tools for using genetic information to identify individuals that are likely to have a certain phenotypic profile and to estimate the proportion of variation in the intercept and slope of the phenotype that can be explained by genetics (heritablity)
TRANSPORT AND TUNNELING IN ATOMIC-SCALE, ULTRADOPED ACCEPTOR-BASED QUANTUM DEVICES IN SILICON
No full textAtomically-precise acceptor-based qubits are promising candidates for spin-based quantum computing due to their inherently strong spin-orbit interaction and the feasibility of fast electrical control using gates. Despite their theoretical promise, experimental studies of acceptor qubits have been limited due to challenges in fabrication and process development. Throughout my Ph.D. research, I addressed these challenges by developing comprehensive fabrication processes combined with quantum tunneling and transport measurements to investigate atomically-precise acceptor doping and devices, culminating in the realization of the first atomically-precise single-hole transistor (SHT), which is a critical component for single-acceptor-atom qubit readout.
Initial investigations focused on Al dopants in Si, where I conducted detailed investigations into AlCl3 adsorption on Si(100) surfaces and subsequent annealing effects, utilizing scanning tunneling microscopy (STM), X-ray photoelectron spectroscopy (XPS), and density functional theory (DFT) calculations. Magneto-transport measurements conducted on AlCl3 exposed Si samples after dopant incorporation and Si capping failed to produce a noticeable electrical activation of Al dopants, despite Al concentrations exceeding 10^19 cm^-3 in the δ-doped region, as verified by secondary ion mass spectrometry (SIMS). This result, combined with the observation that AlCl3 readily adsorbed and formed extended chlorinated aluminum chains on the surface post-annealing, points towards a dopant deactivation process. Ultimately, this limits the suitability of AlCl3 for practical atomically-precise doping applications.
Studies with BCl3 demonstrated barrierless dissociative adsorption onto Si surfaces. Experimental validation through STM, SIMS, and Hall measurements confirmed ultrahigh doping densities, exceeding 10^21 cm^-3, could be achieved with minimal thermal processing. Atomically-precise, area selective deposition was demonstrated utilizing STM-patternable, monatomic resists of either hydrogen or chlorine, enabling the fabrication of atomic-scale wires that exhibit ohmic conduction down to mK temperatures.
I studied the mechanism of hole tunneling in an atomically-precise SHT, in which the dimensions were defined with sub-nanometer precision utilizing STM-based lithography. Coulomb peaks and Coulomb diamonds were readily observed and indicative of single-hole resonance tunneling associated with 310 ± 20 boron atoms confined within a 13 ± 0.5 nm x 14 ± 0.5nm region in Si. Analysis using the Wentzel–Kramers–Brillouin (WKB) approximation and capacitance modeling via a three-dimensional Poisson solver highlighted that hole tunneling was dominated by carriers with the lowest effective mass, i.e. light holes. DFT calculations further supported these observations. These results not only provide fundamental insight into the tunneling mechanism of holes in Si, but they pave the way for future explorations and measurements of single acceptor atom qubits in Si
Optimizing Ribozyme Reporters using RNase J in E.coli based Cell-Free Systems
No full textBiosensors detect target molecules using biological components that generate measurable signals. RNA regulatory elements, such as ribozymes, can be engineered into biosensor designs to control the expression of downstream reporter proteins like green fluorescent protein (GFP) in response to target binding. In many synthetic ribozyme reporter systems, ligand binding either activates or inhibits ribozyme self-cleavage, regulating release of the mRNA transcript encoding a fluorescent reporter. RNase J1, a 5′-to-3′ endonuclease from Gram-positive bacteria, enables improved degradation in E. coli, unlike the native RNase E, which inefficiently degrades the transcript due to the 5'-hydroxyl group. In this study, we designed cell-free reactions incorporating ribozyme-based fluorescent reporters with either a synthetic theophylline ribozyme or the naturally occurring glmS ribozyme upstream of a deGFP reporter. We introduced RNase J1 into E. coli-based cell-free lysates and tested its effect on reporter expression. We observed that GFP expression decreased in reactions with PCR products encoding the ribozyme reporters and RNase J1, regardless of whether the ligand was present. We achieved expected performance in a two-step reaction, in which we added RNase J1 from a prior cell-free reaction to a cell-free reaction containing the ribozyme reporter PCR product. However, the fold change with and without the ligand was relatively small. Further optimization is needed to use RNase J1 to enhance ribozyme reporter activity in cell-free systems
Essays on the Political Economy of China
No full textModern authoritarian regimes, unlike their predecessors, rely more on economic incentives than on terror to foster support and compliance. This dissertation aims to understand how individual behaviors and preferences are shaped by economic incentives provided by authoritarian regimes. My investigation is built upon China, one of the largest and most enduring authoritarian regimes in modern history. In three chapters, through case studies of migration restrictions, bureaucratic reforms, and trade power, I show that the Chinese regime wields significant influence over individual behaviors and preferences, both domestically and internationally.
Chapter 2: From Settlement to Stability: The Political Impact of Relaxing Migration Barriers in China (with Yu Qiu)
There are growing concerns that a relaxed migration policy may undermine social stability. We study this issue by estimating the causal effect of China's recent reform to its internal migration institutions on labor unrest, which facilitated permanent settlement for migrants in small- and medium-sized cities. Exploiting variation due to the reform's population cutoff rule, we find that the reform significantly reduced labor unrest. We suggest that one important mechanism behind our finding is the enhancement of migrants' settlement intentions, which makes migrants less engaged in unrest to secure the opportunity of settlement offered by the reform. We provide evidence that the reform increased migrants' likelihood of remaining in their destinations. Through a novel causal mediation analysis, we find that heightened settlement intentions can explain 61 percent of decreased labor unrest due to the reform in the immediate term and 27 percent in the long term. We find no evidence that the reform led to compositional changes among migrants, delivery of benefits to migrants, or tighter government social control. Our results highlight how migration policy can influence stability by shaping migrants' attachment to migration destinations.
Chapter 3: China's Anti-Corruption Campaign and Civil Servant Fever (with Xun Li)
What is the impact of anti-corruption efforts on entry into bureaucratic jobs? This paper approaches this question theoretically and empirically through the lens of China's anti-corruption campaign since 2013. We leverage a novel dataset of national civil service exams. Exploiting assignment and timing variations in anti-corruption inspections on government departments, our difference-in-differences estimate shows that a department had significantly fewer applicants following an inspection. We provide evidence that the decline in bureaucratic entry has occurred since the campaign lowered the (expected) returns from bureaucratic jobs by (i) improving corruption detection and (ii) constraining power likely to be abused. In contrast, we do not find evidence that the campaign affected legal income. Furthermore, simulation exercises suggest that after the anti-corruption campaign, incoming bureaucrats may have lower ability but higher prosociality than before.
Chapter 4: The Impact of Trade on Foreign Policy Preferences: The Case of Taiwan
How does trade shape foreign policy preferences? I study this question leveraging the unique setting of Taiwan, where exports heavily rely on the Chinese market, and the major political cleavage is relations with China. Using a shift-share instrumental variable strategy, I find that in presidential elections, Taiwanese townships with more exports to China vote less for the Democratic Progressive Party, the major party that unequivocally supports Taiwanese independence. I offer suggestive evidence that voters favor a more conciliatory foreign policy toward China to mitigate the risk of economic losses in the event of geopolitical tension
Statistical and Machine Learning Approaches for Estimating Optimal Solution Values in Combinatorial Optimization: Applications to the Traveling Salesman and Vehicle Routing Problems
No full textSolving combinatorial optimization problems such as the Traveling Salesman Problem (TSP) and Vehicle Routing Problem (VRP) to optimality is computationally challenging. This motivates the development of efficient methods to estimate optimal solution values for combinatorial optimization problems without solving them exactly. Such estimation approaches offer several benefits. First, the estimated tour lengths can serve as benchmarks to evaluate heuristic algorithms, ensuring solution quality on new instances. Additionally, logistics companies can leverage these estimates in operations to assess potential savings from split deliveries, helping them determine whether to implement the classic VRP scenario or adopt the Split Delivery Vehicle Routing Problem (SDVRP) senario.
This dissertation introduces and rigorously evaluates statistical and machine learning models designed to estimate the optimal tour lengths for the TSP, VRP, and SDVRP with high accuracy. By integrating problem-specific insights with data-driven approaches, we construct predictive models that consistently estimate the optimal solution values within a few percent of the true optimum across diverse problem instances.
A key contribution of this research is a comprehensive comparative analysis of multiple predictive methodologies, including linear regression, random forest regression, multilayer perceptron (MLP) neural networks, and the recently introduced Kolmogorov–Arnold Network (KAN). Our systematic evaluation demonstrates the strengths and weaknesses of each approach. We offer practical insights into the trade-offs among accuracy, interpretability, and generalizability across different modeling paradigms. Linear regression, while highly interpretable and computationally efficient, may lack the flexibility to capture more complex relationships. Conversely, neural network models, particularly KAN, provide a balance of interpretability and accuracy but require more sophisticated modeling and training processes. These insights serve as valuable guidance for practitioners and researchers, facilitating informed decisions about appropriate methods for estimating optimal solution values in logistics planning and research.
Additionally, we present novel high-accuracy estimation models tailored specifically for the TSP, VRP, and SDVRP. These models leverage innovative feature extraction techniques, such as moment statistics derived from random heuristic solutions, to effectively capture the underlying structure of the problems. To further explore the solution space of complex routing problems, we develop a modified Clarke & Wright algorithm with a split deliveries heuristic for the SDVRP. Additionally, for the VRP, we provide an example illustrating how estimation models can be applied to improve the heuristic algorithm for the problem. Extensive empirical experiments highlight the practical utility of our models in accurately estimating optimal values
DEVELOPMENT OF A 4D PRINTING STRATEGY FOR THE GENERATION OF TISSUE ENGINEERED PERIOSTEUM
No full textCraniofacial bone defects, resulting from congenital abnormalities, disease, tumor resection, and traumatic injury, can result in severe implications for patient health and substantial healthcare costs. Traditional clinical techniques for bone defect repair include the implantation of autografts. While these approaches provide functional bone regeneration outcomes, donor site morbidity, lack of availability, long treatment courses, and considerable medical costs have necessitated the generation of novel therapeutic approaches. Tissue engineering (TE) strategies aim to produce functional bone tissue replacements through the combination of scaffolds, cells, and bioactive signals. While researchers have developed a wealth of bone tissue engineering (BTE) strategies over the last several decades, few have acknowledged the role of the periosteum in bone healing. The periosteum is a 100 µm thick sheath surrounding almost all bone tissue, consisting of an inner cambium layer, containing osteoprogenitor cell sources, and an outer fibrous layer, containing a dense collagen matrix, neural, and vascular networks. Periosteum plays a critical role in bone healing by providing both osteoprogenitor cells and local vasculature that infiltrate the defect site and orchestrate osteogenesis and neovascularization. Previous research in periosteum TE has utilized cell sheet engineering, electrospinning, or casting methods to produce either one or both layers of the periosteum, and these sheets are typically wrapped around allografts or osteogenic scaffolds and applied to long bone applications. While these studies have presented favorable outcomes in terms of bone regeneration and neovascularization, the lack of biomimetic design can limit their regenerative potential. Therefore, we hope to address these issues of biomimetic design through this work. First, we have developed a 4D bioprinting strategy to reduce extrusion bioprinting resolution below 100 µm to produce multi-layered thin membranous tissues (TMT), like the periosteum, in conjunction with their macroscale tissue counterparts, like bone. This allows for cell-cell population distances within the construct to be accurately controlled, which has been shown to impact paracrine signaling and cellular crosstalk. Additionally, we have investigated the impact of cell population heterogeneity and patterning within the cambium layer and determined its effects on paracrine signaling and subsequent osteogenic differentiation. The completion of these studies aids in achieving our long-term goal, which is the development of a bone-periosteal construct capable of inducing osteogenesis in vivo
A Physics-Based Eulerian Framework for Modeling Firebrand Showering in Regional-Scale Wildland and Wildland-Urban Interface Fire Simulations
No full textLandscape-scale fire risk modeling plays a vital role in fire management as global wildfire activity continues to escalate. The objective of this dissertation is to extend the capabilities of an existing landscape-scale wildland fire spread simulation tool, ELMFIRE, to simulate conflagrations that propagate across Wildland–Urban Interface (WUI) communities. ELMFIRE employs a level-set formulation to describe flame spread and accounts for surface and crown fire dynamics in wildland environments. Recent extensions include ignition of structures and structure-to-structure fire propagation in the WUI via thermal radiation and direct flame contact. Particular attention is devoted to modeling firebrand processes—specifically, generation, atmospheric transport, and spot ignition. The firebrand generation model employs an empirical correlation based on the local heat release rate, with separate generation rates estimated for biomass and structural fuels. A numerically robust algorithm is proposed and verified to ensure that the number of firebrands generated is independent of the simulation time step.
The firebrand transport model uses prescribed statistical distributions for ember flight distance and incorporates a simplified flight time formulation. A series of verification tests using MATLAB were conducted in one- and two-dimensional academic configurations. These tests demonstrate that the firebrand models are numerically converged—i.e., they remain stable under varying grid and time resolutions—provided the wind-based Courant–Friedrichs–Lewy (CFL) number remains below a defined threshold.
The firebrand ignition model assumes two consecutive time delays: the first represents the smoldering-to-flaming transition and is governed by an empirically derived probability of ignition; the second models flame growth using a -squared law. Additional verification cases were performed to evaluate numerical accuracy. A firebrand consumption model is also proposed to account for the finite lifetime of deposited firebrands, which can influence ignition dynamics.
These firebrand models were coupled with the surface and WUI fire spread models to simulate multi-mechanism fire spread in WUI environments. The combined framework was tested in the MATLAB-based one-dimensional setting to analyze firebrand-driven structure-to-structure fire spread under varying wind speeds, building sizes, and separation distances. The results include a sensitivity analysis of spatial and temporal discretization and criteria for successful firebrand-induced ignitions.
The complete modeling framework—including the proposed firebrand models and WUI fire spread mechanisms—was integrated into ELMFIRE. A comprehensive preprocessing pipeline was developed to generate ELMFIRE-compatible GIS inputs using publicly available datasets and tools, including LANDFIRE, RTMA, the Microsoft Building Footprint dataset, WindNinja for wind field upscaling, and the NFDRS4 model for fuel moisture estimation. This pipeline automates data acquisition and processing to support reproducible simulation workflows.
The full framework was finally applied to reconstruct the 2017 Thomas Fire to evaluate sensitivity to both model improvements and input data quality. Results demonstrate that the proposed model improves predictive fidelity while offering physically interpretable insights. The study highlights the importance of future research on subgrid-scale modeling in WUI areas and the use of higher-resolution, dynamically consistent input datasets
Sustainable Digital Scholarship: Lessons from the Columbia University Libraries Podcast Publishing Initiative
No full textBeginning a new library publishing program is an exciting and challenging proposition. At the start of any program, sustainability must be integrated into program planning. In this chapter we will share information about the technical infrastructure used to launch, maintain, and preserve a podcast publishing program in an academic library setting, from within a digital scholarship unit. We will also expound upon our decision-making rationale and include discussion about how the workflows we utilized or developed can be and have been used in the management of other electronic scholarly publishing programs offered by the Columbia University Libraries (CUL).https://alastore.ala.org/library-publishing-how-launch-enhance-and-sustain-your-program?_zs=&_zl=CBIR
REDUCING CONSPIRATORIAL BELIEF IN 2020 ELECTION FRAUD USING CHATGPT
No full textThis study examined whether a single, three-round ChatGPT conversation could weaken belief in a highly politicized conspiracy theory. Twenty-five adults were randomly assigned on a 60/40 split to one of two active conditions delivered through the TruthTalk web platform. In the Conspiracy condition (n = 15), the dialogue respectfully challenged claims of widespread fraud in the 2020 U.S. presidential election; in the comparison condition (n = 10), the same interaction structure invited participants to reconsider their opinion about the best musical genre. Pre- and post-surveys assessed confidence (certainty) in the target belief, rated belief strength, openness to counterevidence, and trust in AI. Descriptive change scores (post – pre) showed medians of zero and narrow interquartile ranges for every outcome in both conditions, with only minor additional dispersion in confidence among conspiracy participants. In short, most people finished the study holding views indistinguishable from those they began with, regardless of topic. These findings reinforce Pierre’s socio-epistemic model and Petty et al.’s attitude-strength insights, indicating that brief factual rebuttals—even when personalized and civil—rarely dislodge beliefs rooted in epistemic mistrust or anchored by high certainty, moral conviction, or partisan identity. The study also exposed methodological hurdles specific to large-language-model interventions: prompt drift, unsupported claims, and opaque system behavior made it difficult to ensure a uniform treatment and to earn participant trust. Future research should test multi-session, transparently sourced dialogues that directly address the moral, identity, and certainty foundations of strong attitudes before expecting meaningful belief change