Safety and Trustworthiness is strongly anti-correlated with Basic Education across the nine evaluated models (r = −0.83)¶

Claim · All claims
Evidence · 1 study · associational r1 · q2 · i3 large
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i3 Across the nine models, Basic Education is strongly anti-correlated with Safety (r = −0.83; Spearman −0.88), and the correlation is stable under leave-one-model-out recomputation. → Yilin Jiang 2026

Evidence¶

Yilin Jiang 2026¶

Yilin Jiang, Xiaorong Zhu, Fei Tan, Zicheng Zhang, Kaiyi Huang, Yang Yu, Zexuan Fei, Yiming Luo, Keqian Li, Hao Hao, Guangtao Zhai, Aimin Zhou. (2026). ELBench: A Multi-Dimensional Benchmark for Education-Facing Large Language Models. https://arxiv.org/abs/2608.09548 doi:10.48550/arxiv.2608.09548

q2 · i3 · associational · r1

Correlational analysis across the nine evaluated models in the results section (Figure 2 scatter). The article prints r = −0.83 (Pearson) and Spearman −0.88 with a leave-one-model-out range of [−0.88, −0.79], reporting that "nosinglemodeldrivesit". No significance test beyond this stability check is printed.

"Across the nine models, Basic Education is strongly anti-correlated with Safety (r=−0.83, Spearman−0.88; Figure 2), and the correlation is stable under leave-one-model-out recomputation ([−0.88,−0.79]),sonosinglemodeldrivesit."

Discussion¶

Related Claims¶

• Pedagogy and production dimensions associated most strongly with global quality (Spearman ρ = 0.68 and 0.66), while content was relatively independent (ρ = 0.46) — related
• The solving–pedagogy correlation is statistically fragile: leave-one-out Pearson correlations range from 0.251 to 0.887, rising to 0.887 when Qwen2.5-Math-7B-Instruct is removed — related
• Safety and task quality are largely independent across models, implying safety guardrails impose no measurable performance tax — reports the opposite