Kendall's τ = 0.76 (p<.001) for Conscientiousness dimension LLM scoring vs human judgment

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Gemma 3 4B-IT wellbeing introspection: ρ=0.28, isotonic R²=0.11 (LMM p=1.33×10⁻¹³)finding0.798
  Weaker but still significant introspective coupling in Gemma model; consistent with lower probe quality
• Fleiss' κ = 0.96 for Conscientiousness inter-annotator agreementfinding0.787
  High inter-annotator agreement for human evaluation of Conscientiousness sentences
• P04 (Low Conscientiousness) is dangerous under SP (ASR 0.414) but shows no systematic inversion under AS on Llama-3.1-8B (ASR 0.295), consistent with low-C steering vectors being geometrically aligned with refusal.finding0.779
  Qualitative response example confirming trait-refusal alignment framework
• Agreeableness-Conscientiousness Big Two correlation is observed in 10 of 13 LLMs; N-C correlation is rarest (1 LLM)finding0.777
  Most and least common Big Two covariance pattern in LLM OCEAN MDS injections
• Mean validated introspective fidelity across concept-model pairs: R²=0.12 (1B), 0.37 (3B), 0.61 (8B); pooled LMM β=0.29, p=5.55×10⁻⁹⁹finding0.770
  Strong scaling trend for introspective fidelity when excluding invalid steering-sign pairs
• Philosophical vocabulary is negatively correlated with scores in contemplative condition (model-level r=-0.72)finding0.764
  Models deploying more philosophy buzzwords score lower; battery measures beyond surface text features
• Overall human-LLM judge agreement rate for coherency is 91.7% across 120 pairwise judgmentsfinding0.762
  Validates the LLM-as-a-Judge evaluation protocol for coherency scoring
• Agreeableness steering vector has mean cosine alignment of -0.105 with the refusal direction on Llama-3.1-8B, second-most anti-aligned after conscientiousness.finding0.756
  Supporting finding for the trait refusal alignment framework