2026-10-28 · ArXiv
arXiv:2610.37070v1 Announce Type: new
Abstract: Restricting chain-of-thought (CoT) faithfulness evaluation to correctly answered instances changes the population being measured. We study this selection step in a seven-variable structural causal model and derive the exact observed gap $\Delta = \operatorname{Cov}(A, C)/P(C=1)$. The model distinguishes difficulty confounding from shortcut-induced collider bias and states the conditional-independence assumption needed to express the gap through latent response propensities. A protocol review of 18 papers distinguishes answer-correctness restrictions from agreement-based selection, benchmark construction, and training-data filtering. We evaluate four sub-ceiling deployments on arithmetic tasks and observe positive selection gaps in all four. The primary GLM-5.3 run yields a +7.4 percentage-point gap ($n = 1000$, $p
原文链接：https://arxiv.org/abs/2610.37070
查看更多资讯请见 AI 行业资讯。