Review
Jul 2026
Auditing Question-Order Effects in Large Language Models with the QQ Equality: Mechanism Characterization and a Saturation Caveat
It is argued that saturation screening and label counterbalancing should precede structural interpretation in distribution-level audits of LLM judgments because next-token probabilities should not be interpreted as survey-response distributions without first establishing adequate dispersion.
Pilsung Kang
· arXiv.org · 1 citation