arXiv subjects

Steffen Eikenberry

Publications and source records attributed to Steffen Eikenberry.

3 recordsLinked to original sources

arXiv2026

Structured Disagreement in Health-Literacy Annotation: Epistemic Stability, Conceptual Difficulty, and Agreement-Stratified Inference

Annotation pipelines in Natural Language Processing (NLP) commonly assume a single latent ground truth per instance and resolve disagreement through label aggregation. Perspectivist approaches challenge this view by treating disagreement as potentially informative rather than erroneous. We present a large-scale analysis of graded health-literacy annotations from 6,323 open-ended COVID-19 responses collected in Ecuador and Peru. Each response was independently labeled by multiple annotators using proportional correctness scores, reflecting the degree to which responses align with normative public-health guidelines, allowing us to analyze the full distribution of judgments rather than aggregated labels. Variance decomposition shows that question-level conceptual difficulty accounts for substantially more variance than annotator identity, indicating that disagreement is structured by the task itself rather than driven by individual raters. Agreement-stratified analyses further reveal that key social-scientific effects, including country, education, and urban-rural differences, vary in magnitude and in some cases reverse direction across levels of inter-annotator agreement. These findings suggest that graded health-literacy evaluation contains both epistemically stable and unstable components, and that aggregating across them can obscure important inferential differences. We therefore argue that strong perspectivist modeling is not only conceptually justified but statistically necessary for valid inference in graded interpretive tasks.

cs.CL↗

arXiv2023

Effect of Methane Mitigation on Global Temperature under a Permafrost Feedback

Earth systems may fall into an undesirable system state if 1.5 degrees celsius (C) of warming is exceeded. Carbon release from substantial permafrost stocks vulnerable to near-term warming represents a positive climate feedback that may increase the risk of 1.5 C warming or greater. Methane (CH4) is a short-lived but powerful greenhouse gas with a global warming potential 28.5 times that of carbon dioxide (CO2) over a 100 year time span. Because permafrost thaw in the coming centuries is partly determined by the warming in the 21st century, rapid reductions in methane emissions early in the 21st century could have far reaching effects. We use a reduced complexity carbon cycle model and a permafrost feedback module to explore the possibility that accelerating reductions in methane emissions could help avoid long-term warming by limiting permafrost melt. We simulate 3 extended Representative Concentration Pathway (RCP) emission scenarios (RCP 2.6, 4.5, and 6) through the year 2300 and impose methane mitigation strategies where we reduce CH4 emissions by 1%, 5% or 10% annually until the long-term scenario emission level is reached. We find that accelerated rates of methane mitigation do not sufficiently alter the global temperature anomaly to prevent or delay a permafrost feedback, nor do they result in meaningful long term reductions in temperatures. We find that the long-term magnitude of methane mitigation (i.e., long-term emission level) and not the rate of reduction, corresponds to long-term temperature change. Therefore, policy and mitigation efforts should emphasize durable decreases in methane emissions over rapidity of implementation.

physics.ao-ph↗

arXiv2020

Mathematical assessment of the impact of non-pharmaceutical interventions on curtailing the 2019 novel Coronavirus

A novel Coronavirus pandemic emerged in December of 2019, causing devastating public health impact across the world. In the absence of a safe and effective vaccine or antiviral, strategies for mitigating the burden of the pandemic are focused on non-pharmaceutical interventions, such as social-distancing, contact-tracing, quarantine, isolation and the use of face-masks in public. We develop a new mathematical model for assessing the population-level impact of these mitigation strategies. Simulations of the model, using data relevant to COVID-19 transmission in New York state and the entire US, show that the pandemic will peak in mid and late April, respectively. The worst-case scenario projections for cumulative mortality (based on the baseline levels of anti-COVID non-pharmaceutical interventions considered in the study) in New York State and the entire US decrease dramatically by 80% and 64%, respectively, if the strict social-distancing measures implemented are maintained until the end of May or June, 2020. This study shows that early termination of strict social-distancing could trigger a devastating second wave with burden similar to that projected before the onset of strict social-distance. The use of efficacious face-masks (efficacy greater than 70%) could lead to the elimination of the pandemic if at least 70% of the residents of New York state use such masks consistently (nationwide, a compliance of at least 80% will be required using such masks). The use of low efficacy masks, such as cloth masks (of efficacy less than 30%), could also lead to significant reduction of COVID-19 burden (albeit, they are not able to lead to elimination). Combining low efficacy masks with improved levels of other anti-COVID-19 intervention measures can lead to elimination of the pandemic. The mask coverage needed to eliminate COVID-19 decreases if mask-use is combined with strict social-distancing.

q-bio.PE↗