Difference-in-Differences on a Censored Rating Scale Can Manufacture an Effect: Evidence from a Pre-Registered LLM-Judge Audit
cs.CL, cs.AI, cs.CY
Submitted: 2026-08-27
Updated: 2026-10-04
Terminology
Sources
- Rethinking LLM-Judged Helpfulness as a Pedagogy Signal: A Pre-Registered Audit Across Tutor Models
- Pre-registration for Predictive Modeling
- Prestige over merit: An adapted audit of LLM bias in peer review
- LearnLM: Improving Gemini for Learning
- Grading Scale Impact on LLM-as-a-Judge: Human-LLM Alignment Is Highest on 0-5 Grading Scale
- Intersectional Sycophancy: How Perceived User Demographics Shape False Validation in Large Language Models
- The AI Teacher Test: Measuring the Pedagogical Ability of Blender and GPT-3 in Educational Dialogues
- Difference-in-Differences for Ordinal Outcomes: Application to the Effect of Mass Shootings on Attitudes toward Gun Control
Related papers
- Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
- Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
- Subliminal Steering: Stronger Encoding of Hidden Signals
- MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
- The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
- Untangling the Mechanisms of Misleading Context in Medical Question Answering